Characterization of the Denitrification-Associated Phosphorus Uptake Properties of “Candidatus Accumulibacter phosphatis” Clades in Sludge Subjected to Enhanced Biological Phosphorus Removal

Abstract

To characterize the denitrifying phosphorus (P) uptake properties of “Candidatus Accumulibacter phosphatis,” a sequencing batch reactor (SBR) was operated with acetate. The SBR operation was gradually acclimated from anaerobic-oxic (AO) to anaerobic-anoxic-oxic (A2O) conditions by stepwise increases of nitrate concentration and the anoxic time. The communities of “Ca. Accumulibacter” and associated bacteria at the initial (AO) and final (A2O) stages were compared using 16S rRNA and polyphosphate kinase genes and using fluorescence in situ hybridization (FISH). The acclimation process led to a clear shift in the relative abundances of recognized “Ca. Accumulibacter” subpopulations from clades IIA > IA > IIF to clades IIC > IA > IIF, as well as to increases in the abundance of other associated bacteria (Dechloromonas [from 1.2% to 19.2%] and “Candidatus Competibacter phosphatis” [from 16.4% to 20.0%]), while the overall “Ca. Accumulibacter” abundance decreased (from 55.1% to 29.2%). A series of batch experiments combined with FISH/microautoradiography (MAR) analyses was performed to characterize the denitrifying P uptake properties of the “Ca. Accumulibacter” clades. In FISH/MAR experiments using slightly diluted sludge (∼0.5 g/liter), all “Ca. Accumulibacter” clades successfully took up phosphorus in the presence of nitrate. However, the “Ca. Accumulibacter” clades showed no P uptake in the presence of nitrate when the sludge was highly diluted (∼0.005 g/liter); under these conditions, reduction of nitrate to nitrite did not occur, whereas P uptake by “Ca. Accumulibacter” clades occurred when nitrite was added. These results suggest that the “Ca. Accumulibacter” cells lack nitrate reduction capabilities and that P uptake by “Ca. Accumulibacter” is dependent upon nitrite generated by associated nitrate-reducing bacteria such as Dechloromonas and “Ca. Competibacter.”

INTRODUCTION

Enhanced biological phosphorus removal (EBPR) has been applied in many wastewater treatment plants to economically remove phosphorus, which causes eutrophication in surface waters. EBPR employs polyphosphate-accumulating organisms (PAOs), which are enriched through alternating anaerobic-oxic (AO) cycles (1). Culture-independent techniques such as PCR-clone libraries, fluorescence in situ hybridization (FISH), and microautoradiography (MAR) have highlighted yet uncultured “Candidatus Accumulibacter phosphatis” (henceforth referred to as “Ca. Accumulibacter”) in lab-scale EBPR reactors as well as full-scale EBPR plants to be the most frequently identified PAOs (2–11). Therefore, many previous studies have sought to investigate “Ca. Accumulibacter” from various EBPR systems and have shown the presence of morphologically and genetically diverse “Ca. Accumulibacter” members (5, 8, 12–17).

The polyphosphate kinase (ppk1) gene, which is involved in the production of polyphosphate, was suggested to be a phylogenetic marker for finer classification of the diverse “Ca. Accumulibacter” members (18). Diverse “Ca. Accumulibacter” populations were characterized into two main types (types I and II) using the ppk1 gene, and each type was classified into several subgroups (19, 20). Kim et al. (15) also differentiated the “Ca. Accumulibacter” populations into four “Ca. Accumulibacter” clades (Acc-SG1, Acc-SG2, Acc-SG3, and Acc-SG4) on the basis of the 16S rRNA gene sequences and linked the “Ca. Accumulibacter” clades with their morphological characteristics and ppk1 gene sequences after FISH–fluorescence-activated cell sorting.

The denitrifying ability of PAOs is a key factor in EBPR process designs for simultaneous denitrification and P removal that can lead to savings in plant operational costs. Many previous investigations have demonstrated that successful EBPR has been achieved using nitrate as a final electron acceptor (21–25), which suggests that “Ca. Accumulibacter” clades may have the ability to take up P using nitrate (8). Additionally, some research groups suggested the presence of two morphologically different “Ca. Accumulibacter” populations: denitrifying PAOs with a rod-type morphology and nondenitrifying PAOs with a coccus-type morphology (12, 26). Recently, Flowers et al. (27) and Oehmen et al. (28) demonstrated that among the “Ca. Accumulibacter” populations, clade IA with rod morphotypes might have P uptake capabilities when using nitrate as a final electron acceptor but clade IIA with coccus morphotypes might not.

A metagenomic analysis of EBPR sludge did not identify the nar gene encoding respiratory nitrate reductase in the “Ca. Accumulibacter” genome, even though the genes responsible for nitrite reduction to nitrogen were detected; however, the metagenomic analysis was primarily limited to the clade IIA-enriched metagenome (13). Moreover, the “Ca. Accumulibacter” genome of clade IA had no clearly identifiable respiratory nitrate reductase homologue (K. D. McMahon, personal communication). Therefore, in this study, we hypothesized that other unknown “Ca. Accumulibacter” clades with nitrate reduction capabilities might be present in denitrifying EBPR sludge; the alternative hypothesis was that “Ca. Accumulibacter” populations might be able to take up phosphorus when utilizing nitrite produced from nitrate by other associated species with nitrate reduction capabilities.

MATERIALS AND METHODS

Operation of an SBR.

A cylindrical vessel with a 4-liter working volume was used for operation as a sequencing batch reactor (SBR) initially inoculated with activated sludge from a wastewater treatment plant at the campus of POSTECH (Republic of Korea). Synthetic wastewater containing acetate as the sole carbon source was used for the feed. The feed contained 770 mg/liter sodium acetate, 40 mg/liter NH₄⁺ N, 12 mg/liter PO₄³⁻ P, and trace element mixtures (29). The SBR was initially operated for more than 100 days with 15 min of filling, 1 h 15 min of anaerobic reaction, 3 h 30 min of aerobic reaction, 45 min of settling, and 15 min of decanting at 20°C. The SBR operation was then gradually changed from AO to anaerobic-anoxic-oxic (A2O) conditions by stepwise increases in the nitrate concentration (NO₃⁻ N, 0 to 50 mg/liter) and anoxic time (0 to 2 h), as shown in Table 1. During the anaerobic and anoxic periods, the reactor was continuously stirred but kept under the anaerobic condition by gassing with nitrogen. Two liters of clarified supernatant was withdrawn at the end of the settling phase. The sludge retention time was maintained at approximately 10 days by withdrawing a small portion of sludge from the reactor at the end of the aerobic phase. Concentrations of soluble orthophosphate (ortho-P), nitrate, and nitrite were analyzed using an ICS-1000 ion chromatograph (Dionex).

Table 1.

Procedures used for acclimation to denitrifying P uptake conditions in the acetate-fed SBR

Acclimation stage	Anoxic and aerobic phasesa	Nitrate concn (mg NO3− N/liter)b	Operation time (days)
1	0 min anoxic, 210 min aerobic		∼100
2	60 min anoxic, 150 min aerobic	10	38
3	120 min anoxic, 90 min aerobic	20	53
4	120 min anoxic, 90 min aerobic	30	45
5	120 min anoxic, 90 min aerobic	50	50

^aThe anoxic and/or oxic phase was preceded by a 75-min anaerobic phase in all stages.

^bSodium nitrate solution was pumped into the SBR at the beginning of the anoxic phase.

DNA extraction and phylogenetic analysis.

To compare the communities of “Ca. Accumulibacter” under the AO (stage 1) and final A2O (stage 5) conditions using 16S rRNA and ppk1 gene sequences, two 10-ml sludge samples were collected from the aerobic phases of stages 1 and 5 by centrifugation (10,000 × g, 5 min). The precipitated sludge was resuspended in 0.75 ml of 50 mM Tris (pH 8.0) and 50 mM EDTA and then frozen at −80°C and thawed at 90°C three times. After that, total genomic DNA was extracted by using a FastDNA spin kit (MPbio) according to the manufacturer's instructions, and bacterial 16S rRNA gene clone libraries of “Ca. Accumulibacter” populations were constructed as described previously (15). Briefly, modified two-step PCR amplification was used to reduce chimeric products using primers 27f and 1492r (30, 31). PCR amplicons were ligated into the pCR2.1 vector using a TOPO cloning kit (Invitrogen) according to the manufacturer's instructions. After blue-white screening of the colonies, inserted 16S rRNA genes were evaluated using restriction fragment length polymorphism (RFLP) analysis after PCR amplification and HaeIII and HhaI double digestion, as described previously (32). All representative clones with unique RFLP patterns were partially sequenced using the M13 reverse primer, and the resulting sequences were submitted to GenBank for BLASTN searches. The “Ca. Accumulibacter”-related clones were more completely sequenced, and their chimeric properties were checked using the Bellerophon program (33). The resulting sequences were compared with available “Ca. Accumulibacter” 16S rRNA gene sequences from the GenBank database and were aligned using the greengenes NAST server (http://greengenes.lbl.gov/cgi-bin/nph-NAST_align.cgi) (34). A phylogenetic tree was constructed using the neighbor-joining (NJ) algorithm available in PHYLIP software, version 3.68 (35), and the resulting tree topology was evaluated using bootstrap analysis based on 1,000 resamplings.

“Ca. Accumulibacter” ppk1 gene homologs were amplified and analyzed using a previously described method (15). Briefly, “Ca. Accumulibacter” ppk1 gene homologs from the total genomic DNA were amplified using the “Ca. Accumulibacter”-specific primers ACCppk1-254F (TCACCACCGACGGCAAGAC) and ACCppk1-1376R (ACGATCATCAGCATCTTGGC), as described previously (18). The resulting PCR amplicons were inserted into the pCR2.1 vector. After blue-white screening of the colonies, inserted ppk1 genes were amplified with ACCppk1-254F and ACCppk1-1376R, and their PCR amplicons were analyzed using the RFLP approach described above. Clones were grouped according to their RFLP patterns, and representative clones with unique RFLP patterns were sequenced. Retrieved ppk1 gene homologs were compared to available sequences from the GenBank database. Sequences belonging to “Ca. Accumulibacter” ppk1 gene homologs were aligned, and a phylogenetic tree was constructed using the PROTDIST and Neighbor modules available in PHYLIP software, version 3.68 (35).

FISH probe design and FISH analyses.

Specific oligonucleotide FISH probes were designed to target a “Ca. Accumulibacter” clade (Acc-SG3) and members of Dechloromonas, which were identified to be dominant populations of the A2O sludge from the 16S rRNA gene sequencing analysis described above, using the probe design tool in the ARB software package (36). On the basis of comparative analysis of reliable sequences with >1,200 bp in the ARB software and our in-house clone sequences, specific regions were selected and their specificities were subsequently confirmed using the Probe Match tool on the website of the Ribosome Database Project II (RDP; release 10) (37).

The designed probes were synthesized and labeled at the 5′ end with fluorescein isothiocyanate (FITC) or 3-iodocyanine dye (Cy3) by Thermo (Ulm, Germany). These labeled probes were evaluated with paraformaldehyde (PFA)-fixed sludge. The formamide (FA) concentration for optimum probe stringency was determined empirically by performing a series of FISH experiments at 5% FA increments from 20% to 60% formamide at set hybridization and wash temperatures (38). Subsequently, these new probes were used simultaneously with other FISH probes, EUBmix and PAOmix, with different fluorescent dyes. The FISH probes, helpers, and competitors used in this study are listed in Table 2.

Table 2.

FISH oligonucleotides used in this study

Probea	Sequence (5′–3′)	Target organism	Helper(s) and/or competitor	Reference or source
EUB338I	GCTGCCTCCCGTAGGAGT	Eubacteria		39
EUB338II	GCAGCCACCCGTAGGTGT	Eubacteria		40
EUB338III	GCTGCCACCCGTAGGTGT	Eubacteria		40
NonEUB338	CGACGGAGGGCATCCTCA	Nilb		39
PAO651	CCCTCTGCCAAACTCCAG	“Ca. Accumulibacter”		4
PAO462	CCGTCATCTACWCAGGGTATTAAC	“Ca. Accumulibacter”		4
PAO846	GTTAGCTACGGCACTAAAAGG	“Ca. Accumulibacter”		4
Acc444	CCCAAGCAATTTCTTCCC	Acc-SG1	HAcc462 and HAcc426	15
HAcc466	CATCTACTCAGGGTATTAA	Acc-SG1		15
HAcc426	CGCCGAAAGAGCTTTACA	Acc-SG1		15
Acc184	GCTCCCAGAACGCAAGGT	Acc-SG2	CAcc184	15
CAcc184	GCTCCCAGAGCGCAAGGT	Acc-SG2		15
Acc623	CCAGCTGGACAGTCTCAA	Acc-SG3		This study
Acc119	GGATACGTTCCGATGCTT	Acc-SG4	HAcc99, HAcc139, and CAcc119	15
HAcc99	CTCACCCGTCCGCCACTC	Acc-SG4		15
HAcc139	GCTACGTTATCCCCCACTC	Acc-SG4		15
CAcc119	GGGCACGTTCCGATGCAT	Acc-SG4		15
GB	CGATCCTCTAGCCCACT	“Ca. Competibacter” GB1-7		16
GB429	CCCCACCTAAAGGGCTTT	“Ca. Competibacter” GB8		41
Dech453	GGGTATTCACCCATGCGA	Dechloromonas		This study

^aEUB338I, EUB338II, and EUB338III were used as Eubmix; PAO651, PAO462, and PAO846 were used as PAOmix; and GB and GB429 were used as GBmix. H and C, helper and competitor probes, respectively.

^bProbe NonEUB33, complementary to EUB338, was used to exclude nonspecific binding.

Two sludge samples collected from the aerobic phase of stages 1 and 5 were washed with phosphate-buffered saline (PBS) buffer (137 mM NaCl, 8.1 mM Na₂HPO₄, 2.68 mM KCl, 1.47 mM KH₂PO₄, pH 7.2) and fixed with 4% paraformaldehyde for 4 h at 4°C. Fixed sludge samples were washed twice in PBS, resuspended in PBS-ethanol solution (1:1, vol/vol), and then stored at −20°C until further FISH analysis. FISH analysis was carried out using methods described previously (38, 39, 41). Nonprobe controls were always prepared to examine autofluorescent bacteria, and the NonEUB338 probe was used for nonspecific oligonucleotide binding (39). FISH images were captured under ×1,000 magnification using a Zeiss Axiophot epifluorescence microscope equipped with an HBO100 mercury vapor short-arc lamp and band path filter sets no. 10 and no. 20 for fluorescein and Cy3, respectively, and an AxioCam MRm digital camera (Carl Zeiss, Germany). Enhanced FISH images were obtained using AxioVs40 software, version 4.7.1.0 (Zeiss), by combining two captured images. Nomarski photographs were also collected simultaneously with the FISH images. The relative abundance of each probe-targeted cell was quantified on the basis of all EUBmix-binding cells, using more than 20 FISH images per sample, directly by naked eye because sludge cells were highly heterogeneous in size.

FISH/MAR analyses.

All FISH and MAR analyses were performed in serum bottles to characterize the denitrifying P uptake properties of “Ca. Accumulibacter” clades according to methods described previously (8, 15, 42) with some modifications. To prepare the sludge dilution solution without nitrate and nitrite, bulk solution from the end of the anoxic phase was collected and mixed with a small amount of sodium acetate (2 mg/liter) to completely remove nitrate and nitrite. The mixture was incubated anaerobically for 1 h, and sludge was eliminated by filtration using a 0.2-μm-pore-size membrane (Whatman) under the anaerobic condition. Sludge samples were collected from the end of the anaerobic phase of stage 5. The sludge samples were homogenized and diluted to approximately 0.5 and 0.005 g (dry weight) sludge per liter in the previously prepared dilution solution. A small amount of Na₂SO₃ · 7H₂O (0.5 g/liter) was added to the diluted sludge solutions to completely remove trace oxygen, followed by 30 min of preincubation. Finally, the absence of oxygen and nitrate or nitrite in the sludge solutions was confirmed by measurements taken using Fibox 3 LCD trace (PreSens GmbH, Rogensburg, Germany) and FIAstar 5000 (FOSS, Sweden) analyzers, respectively. The sludge solutions were incubated with 25 μCi of ³³P_i (PerkinElmer) for 4 h under denitrifying conditions with nitrate (20 mg/liter NO₃⁻ N) or nitrite (20 mg/liter NO₂⁻ N). After cells were harvested by centrifugation, FISH was performed using the same FISH analysis method described above, except for the use of coverslips instead of poly-l-lysine-coated slides. Subsequent steps of the MAR protocol were carried out as detailed by Lee et al. (42). Kodak type NTB was used as an autoradiographic liquid emulsion according to the manufacturer's instructions. Views of probe-targeted cells were identified first under fluorescence excitation, and then bright-view photographs of the same field for the MAR signals of black silver particles were collected using the phase-contrast mode. Pasteurized sludge (80°C, 10 min) was used as a negative control and was included in all experiments, and no MAR-positive signals were found in these control samples.

Nucleotide sequence accession numbers.

The GenBank accession numbers for the 16S rRNA and ppk1 gene sequences determined in this study are JQ726360 to JQ726380 and JQ726381 to JQ726393, respectively.

RESULTS

EBPR performance of the SBR.

The microbial populations and denitrifying P uptake properties of “Ca. Accumulibacter” clades were characterized throughout the 5 stages of acclimation from anaerobic-oxic (AO) to anaerobic-anoxic-oxic (A2O) conditions in an SBR fed with sodium acetate as the sole carbon source. First, the SBR was operated for more than 100 days under the AO condition (stage 1) before introduction of the anoxic phase. The SBR exhibited good EBPR performance with rapid acetate consumption and ortho-P release under the anaerobic conditions and ortho-P uptake under the subsequent aerobic conditions (Fig. 1A). These profiles were similar to those from the previous SBR operation (43). The anoxic phase was introduced by pumping a sodium nitrate solution at the end of the anaerobic phase, and the SBR operation was gradually acclimated to denitrifying P uptake conditions by stepwise increases in the nitrate concentration, as shown in Table 1. However, to ensure stable EBPR operation, complete elimination of the oxic phase was not introduced.

Fig 1.

Concentration profiles of PO₄³⁻ P (●), NO₃⁻ N (○), NO₂⁻ N (▼), and pH (×) during a typical SBR cycle of each acclimation stage for denitrifying PAO enrichment. These profiles were obtained from acclimation stages 1 (A, day 91), 3 (B, day 164), 4 (C, day 213), and 5 (D, day 241) shown in Table 1. F, AN, ANO, O, and SD, feeding, anaerobic, anoxic, oxic, and setting and decanting phases, respectively.

Figure 1 shows the profiles of ortho-P, nitrate, nitrite, and pH determined during a typical cycle of each acclimation stage for denitrifying “Ca. Accumulibacter” enrichment. Dissolved oxygen was nearly 0 mg liter⁻¹ during the anaerobic and anoxic periods, but it was maintained at more than 3 mg liter⁻¹ during the aerobic period (data not shown). P uptake efficiencies increased with the increase of nitrate additions during the anoxic phase, and eventually, efficient P uptake was accomplished when 50 mg liter⁻¹ of NO₃⁻ N was supplied. Rapid P uptake was relatively well coupled with the decrease in nitrate concentration during the early anoxic phase, but concurrently, the nitrite concentration in the bulk solution increased. When nitrate and nitrite were depleted, P uptake also stopped and the P concentration increased slightly. A rapid increase of pH values occurred at the early anoxic stage (primarily by the reduction of nitrate) and subsequent pH values decreased in the aerobic stage (mainly due to nitrification), which was the typical pH profile of denitrifying EBPR processes in the SBR (12, 44). These results indicate the successful enrichment of “Ca. Accumulibacter” populations performing denitrifying EBPR.

Population structures of “Ca. Accumulibacter” at the AO and A2O stages.

To compare the populations of “Ca. Accumulibacter” under AO and A2O conditions, bacterial 16S rRNA and “Ca. Accumulibacter” ppk1 gene clone libraries were constructed using sludge samples from stages 1 and 5, respectively. To compare “Ca. Accumulibacter” populations on the basis of 16S rRNA gene sequences, approximately 400 colonies of the two 16S rRNA gene clone libraries were randomly selected and evaluated by RFLP analysis. Representative clones with unique RFLP patterns were partially sequenced (approximately 650 nucleotides), and the resulting sequences were submitted to GenBank for BLASTN searches. The results of the BLASTN searches showed that clones affiliated with “Ca. Accumulibacter,” “Candidatus Competibacter phosphatis,” and Dechloromonas were identified as the major bacterial groups in the AO and A2O sludge (data not shown). Other bacterial genera, including Thauera, Azospira, Azoarcus, Rhodobacter, and Cytophaga, were also identified as minor populations in the AO and A2O sludge, and these communities did not show clear differences with operational changes in the SBR (data not shown). Approximately 50 “Ca. Accumulibacter”-related 16S rRNA gene clones from each clone library were selected. Representative clones showing unique RFLP patterns were sequenced more completely (approximately 1450 nucleotides) and analyzed phylogenetically. “Ca. Accumulibacter” 16S rRNA gene sequences fell into four major “Ca. Accumulibacter” subgroups (Fig. 2), as described previously (15). In the AO sludge, the most dominant “Ca. Accumulibacter” clones were affiliated with members of Acc-SG4 (30 of 50 “Ca. Accumulibacter” clones), and those belonging to members of Acc-SG2 (18 of 50 “Ca. Accumulibacter” clones) were the next most dominant. However, members of Acc-SG1, which contained the type 16S rRNA gene sequence, clone R6, of “Candidatus Accumulibacter phosphatis,” comprised minor proportions (2 of 50 “Ca. Accumulibacter” clones). On the other hand, in the A2O sludge, members of Acc-SG1 became the most dominant “Ca. Accumulibacter” clones (31 of 51 “Ca. Accumulibacter” clones), followed by clones of Acc-SG3 (11 of 51 “Ca. Accumulibacter” clones) and Acc-SG2 (9 of 51 “Ca. Accumulibacter” clones). Interestingly, clones belonging to Acc-SG4, the most dominant “Ca. Accumulibacter” subgroup in the AO sludge, were not detected from the clone library of the A2O sludge, and clones of Acc-SG3 were found only in the A2O sludge. Comparisons of the “Ca. Accumulibacter” clones showed that members of Acc-SG1 and Acc-SG2 clearly shifted phylogenetically within their subgroups (Fig. 2). In summary, the comparative analyses based on 16S rRNA gene sequences demonstrated that the operational change from AO to A2O conditions led to a clear shift in the “Ca. Accumulibacter” communities.

Fig 2.

Phylogenetic analysis of “Ca. Accumulibacter”-affiliated 16S rRNA gene sequences. Sequences found in this study were compared with reference sequences in the GenBank database. AO and A2O clones in bold characters represent clones derived from the AO (stage 1) and A2O (stage 5) sludge samples, respectively. “Ca. Accumulibacter” 16S rRNA gene sequences were classified as Acc-SG1, Acc-SG2, Acc-SG3, and Acc-SG4 on the basis of a previous description (15). The numbers in parentheses indicate the frequencies of clones exhibiting the same fragment patterns as those from the RFLP analysis. Dechloromonas agitata CKB was used as an outgroup. Bootstrap values are shown as percentages of 1,000 replicates when greater than 70%. The scale bar indicates the number of changes per nucleotide position.

The ppk1 gene has been suggested to be an alternative good phylogenetic marker for the classification of “Ca. Accumulibacter” clades because their ppk1 gene sequences allow enough resolution of closely related “Ca. Accumulibacter” populations (19, 20, 45). Therefore, “Ca. Accumulibacter” ppk1 gene homologs were also analyzed to facilitate comparisons between the “Ca. Accumulibacter” communities in the AO and A2O sludge. A phylogenetic analysis was performed using ppk1 gene sequences showing unique RFLP patterns, and the relations between ppk1 gene homologs and 16S rRNA gene sequences of “Ca. Accumulibacter” were linked on the basis of previous results (15). The phylogenetic analysis demonstrated that the ppk1 gene homologs retrieved from the AO and A2O sludge were differentiated into four distinct “Ca. Accumulibacter” clades (Fig. 3), which is consistent with the results of the analysis of the 16S rRNA gene sequences shown in Fig. 2. In the AO sludge, the most dominant ppk1 gene homologs were affiliated with clade IIA (50 of 86 clones), corresponding to Acc-SG4 members of “Ca. Accumulibacter.” “Ca. Accumulibacter” ppk1 gene homologs belonging to clade IA, corresponding to Acc-SG1, were relatively abundant (24 of 86 clones) in the AO sludge, although members of Acc-SG1 were considered minor on the basis of the analysis of 16S rRNA gene sequences (this discrepancy might be the result of PCR bias). “Ca. Accumulibacter” ppk1 gene homologs of clade IIF, corresponding to Acc-SG2, were the next most abundant (11 of 86 clones). Only one clone of the clade IIC ppk1 gene homologs, corresponding to Acc-SG3, was identified (1/86), which was in accordance with the above-presented result that no Acc-SG3 member was detected from the AO sludge. On the other hand, ppk1 gene homologs of clade IIC (42 of 85 clones, Acc-SG4) became predominant in the A2O sludge, and ppk1 gene homologs of clades IA (29 of 85 clones, Acc-SG1) and IIF (14 of 85 clones, Acc-SG2) were the next most abundant. ppk1 gene homologs of clade IIA (Acc-SG4) were not detected in the A2O sludge; this observation is consistent with the corresponding absence of Acc-SG4 members from the 16S rRNA gene library. The analysis of “Ca. Accumulibacter” ppk1 gene sequences showed that the “Ca. Accumulibacter” communities changed during the acclimation process and dominant members of clade IA, which were present in both AO and A2O sludge, clearly shifted within this clade, as was also shown in the 16S rRNA gene sequencing analysis (Fig. 2 and 3).

Fig 3.

Phylogram indicating inferred relatedness of ppk1 gene homologs from “Ca. Accumulibacter” clades on the basis of DNA sequences. AO and A2O clones in bold characters represent clones derived from the AO (stage 1) and A2O (stage 5) sludge samples, respectively. The numbers in parentheses indicate frequencies of clones exhibiting the same fragment patterns as those from the RFLP analysis. Clade names of ppk1 gene homologs were based on previously assigned clade names (19). The relations between ppk1 gene homologs and 16S rRNA gene sequences of “Ca. Accumulibacter” shown in parentheses were linked on the basis of previous results (15). Bootstrap values are shown as percentages of 1,000 replicates when greater than 70%. The scale bar indicates the number of changes per nucleotide position.

FISH probe design and quantitative FISH analysis.

The sequences of previously designed oligonucleotide FISH probes Acc72, DECH454, and DECH472, targeting the Acc-SG3 subgroup and the genus Dechloromonas, did not completely match the 16S rRNA gene sequences of Acc-SG3 and Dechloromonas members retrieved from the AO and A2O sludge of this study (15, 46). Therefore, new FISH probes Acc623 and Dech453, targeting all Acc-SG3 and Dechloromonas members of this study, respectively, were designed, and their specificities were confirmed using the Probe Match tool of the Ribosomal Database Project (37). Optimal FA concentrations for the two newly designed FISH probes were determined by increasing the FA concentration in PFA-fixed sludge since no pure culture was available and their optimal FA concentrations were approximately 35%.

To enumerate the populations of the four “Ca. Accumulibacter” subgroups and other associated major bacteria, “Ca. Competibacter” and Dechloromonas, in the AO and A2O sludge, quantitative FISH approaches using the FISH probes listed in Table 2 were applied, and the relative abundance of these organisms was quantified as the percentage of respective probe-binding cells against the total number of EUBmix-binding cells. Quantitative FISH analysis also demonstrated that the operational change from the AO to A2O conditions led to clear shifts in “Ca. Accumulibacter” communities as well as in “Ca. Competibacter” and Dechloromonas populations (Table 3). During the acclimation process, the shifts in the “Ca. Accumulibacter” communities coincided relatively well with the shifts shown in the analyses based on “Ca. Accumulibacter” 16S rRNA and ppk1 gene sequences, but the relative abundances of the individual “Ca. Accumulibacter” subgroups were slightly different, likely due to PCR bias. Bacterial cells responding to probe Acc119 (members of Acc-SG4) were the most dominant in the AO sludge, but they were not detected in the A2O sludge. On the other hand, bacterial cells binding to probe Acc623 (members of Acc-SG3) were not detected in the AO sludge, but they became one of the dominant “Ca. Accumulibacter” subgroups in the A2O sludge. The relative abundance of total “Ca. Accumulibacter” clades decreased from approximately 55% to 29% during the acclimation process, while that of “Ca. Competibacter” and Dechloromonas increased. The relative abundance of Dechloromonas populations increased especially sharply from approximately 1.2% to 19%.

Table 3.

Distribution of major bacteria in sludge samples during the acclimation procedure

Probe	Target organism (clade)	Relative abundance (%)a
		Stage 1 (day 91)	Stage 2 (day 136)	Stage 3 (day 164)	Stage 4 (day 213)	Stage 5 (day 241)
Acc444	Acc-SG1 (IA)	11.0	28.6	23.7	13.9	9.5
Acc184	Acc-SG2 (IIF)	8.3	5.2	8.6	9.3	8.9
Acc623	Acc-SG3 (IIC)	ND	1.2	6.7	6.2	10.8
Acc119	Acc-SG4 (IIA)	35.8	18.2	4.3	0.9	ND
Dech453	Dechloromonas	1.2	NA	NA	NA	19.2
GBmix	“Ca. Competibacter”	16.4	NA	NA	NA	20.0

^aMean percentages of each probe-targeted cell against the total EUBmix-binding cells using more than 20 FISH images per sample. The days in parentheses indicate the sludge sampling time for the abundance analysis using FISH. ND, not detected; NA, not analyzed.

Interestingly, the FISH analysis showed that the “Ca. Accumulibacter” populations in the AO and A2O sludge had different morphologies and sizes, although the same FISH probes were applied (Fig. 4). “Ca. Accumulibacter” cells hybridized with probes Acc444 and Acc184 targeting clades Acc-SG1 and Acc-SG2, respectively, had a small (approximately 1-μm) coccus-shaped morphology in the AO sludge (Fig. 4A and B), while “Ca. Accumulibacter” cells from the A2O sludge that hybridized with the same FISH probes had quite different morphologies (Fig. 4D and E). That is, probe Acc444-positive cells in the A2O sludge were long (approximately 2 to 3 μm) and rod shaped (Fig. 4D), and probe Acc184-targeted cells in the A2O sludge also had rod morphologies and were approximately 1 to 2 μm long (Fig. 4E). However, the 16S rRNA and ppk1 gene sequence analyses showed that the “Ca. Accumulibacter” community shifted within the Acc-SG1 and Acc-SG2 subgroups with the operational change in the SBR (Fig. 2). Therefore, the cells of the AO and A2O sludge might be different phylogenetic clades with distinct morphologies, although they were hybridized with the same FISH probes. Acc-SG4 from the AO sludge that bound probe Acc119 had a small (approximately 1-μm) coccoid morphotype, while Acc-SG3 cells from the A2O sludge that hybridized with probe Acc623 had large (approximately 2- to 3-μm) coccobacillus cell types, consistent with previous results (15).

Fig 4.

FISH images of “Ca. Accumulibacter” subgroups showing different morphologies and sizes in the AO (A to C) and A2O (D to F) sludge. The FISH images show “Ca. Accumulibacter” cells hybridized with the PAOmix (fluorescein, green) and “Ca. Accumulibacter” (Cy3, red) subgroup-specific probes Acc444, Acc184, Acc119, and Acc623, which targeted members of clades Acc-SG1 (A and D), Acc-SG2 (B and E), Acc-SG4 (C), and Acc-SG3 (F), respectively. Members of clades Acc-SG1 (A and D) and Acc-SG2 (B and E) had different morphologies and sizes in the AO and A2O sludge, although they hybridized with the same FISH probes.

Characterization of denitrifying P uptake properties of “Ca. Accumulibacter” clades using FISH/MAR analysis.

To characterize the denitrifying P uptake properties of respective “Ca. Accumulibacter” clades, a series of batch experiments combined with FISH/MAR analysis was carried out under different electron acceptor conditions. First, a FISH/MAR experiment was performed using the slightly diluted sludge sample (approximately 0.5 g [dry weight] sludge per liter) in the presence of nitrate. The FISH/MAR analysis clearly showed that all “Ca. Accumulibacter” clades responding to PAOmix (which covered all “Ca. Accumulibacter” clades that hybridized with Acc444, Acc184, and Acc623 FISH probes) took up ³³P_i when nitrate was present (Fig. 5A and D). Formation of nitrite was confirmed using the FIAstar 5000 (FOSS, Sweden) analyzer, meaning that nitrate reduction to nitrite occurred in the FISH/MAR solution, which coincided with the increase in the nitrite concentration during the early anoxic phase under A2O conditions (Fig. 1B to D). Next, the sludge samples were highly diluted (approximately 0.005 g [dry weight] sludge per liter) to exclude the possibility of nitrite formation from nitrate by some “Ca. Accumulibacter” clades or other associated nitrate-reducing bacteria, and a FISH/MAR experiment was performed in the presence of nitrate. The concentration of nitrite in the FISH/MAR solution was below the detection limit during the batch test. Interestingly, the FISH/MAR experiment using the highly diluted sludge showed that all “Ca. Accumulibacter” clades responding to PAOmix probes did not take up ³³P_i in the nitrate treatment (Fig. 5B and E). However, another FISH/MAR experiment using the same sludge providing nitrite instead of nitrate as the final electron acceptor showed that all “Ca. Accumulibacter” clades successfully took up ³³P_i (Fig. 5C and F). These results indicate that “Ca. Accumulibacter” clades were not capable of reducing nitrate but they were able to take up phosphorus using nitrite that might be provided by associated bacteria capable of nitrate reduction.

Fig 5.

FISH and MAR images of the A2O sludge under nitrate and nitrite conditions. The FISH images show bacteria hybridized with PAOmix (PAO462, PAO651, and PAO844; Cy3, red) and EUBmix (EUB338I, EUB338II, and EUB338III; FITC, green) probes. FISH images were obtained under nitrate (A and B) and nitrite (C) conditions. Nomarski images show MAR-positive cells taking up ³³P_i under nitrate (D and E) and nitrite (F) conditions. The FISH/MAR experiments were performed using slightly diluted (0.5 g [dry weight] sludge per liter; A and D) and highly diluted (0.005 g [dry weight] sludge per liter; B, C, E, and F) sludge samples. Arrows, PAOmix-positive cells and their corresponding MAR signals.

DISCUSSION

Denitrifying PAOs have generally been defined to be specific PAOs capable of using oxygen or nitrite as well as nitrate for P uptake, and they are differentiated from PAOs with the ability to use only oxygen and nitrite for P uptake (8, 11, 12, 21, 44, 47–50). In this study, the denitrifying P uptake properties of “Ca. Accumulibacter” clades were characterized through a community analysis of “Ca. Accumulibacter” clades performing denitrifying P uptake and a series of FISH/MAR analyses. For the successful acclimation of the denitrifying EBPR sludge, SBR operation was gradually changed from anaerobic-oxic (AO) to anaerobic-anoxic-oxic (A2O) conditions by stepwise increases in nitrate concentrations and anoxic time as described by Carvalho et al. (12). Eventually, P uptake occurred efficiently, with the decrease of nitrate occurring during the anoxic phase, meaning that the enrichment of “Ca. Accumulibacter” populations performing denitrifying P uptake was successful. Moreover, the anoxic P uptake continued until nitrate was completely depleted (Fig. 1B to D). This observation was slightly different from results reported in previous studies suggesting the presence of two PAO types, denitrifying PAOs and PAOs. These studies showed that denitrifying P uptake ceased, despite the presence of available nitrate due to internal PHA depletion of denitrifying PAOs, and P uptake by other PAOs continued again when oxygen was supplied (12, 23, 51, 52). In the present study, our SBR operation showed that nitrite concentrations in the bulk solution increased concurrently with the decrease of nitrate (Fig. 1B to D), which suggested that “Ca. Accumulibacter” clades could not use nitrate directly for P uptake.

A metagenomic analysis of EBPR sludge revealed the absence of a respiratory nitrate reductase homologue in sludge enriched with “Ca. Accumulibacter” clade IIA (13). Flowers and colleagues (27) analyzed the denitrifying P uptake properties of “Ca. Accumulibacter” clades using specific FISH probes targeting “Ca. Accumulibacter” type I (clade IA) and type II (clades IIA, IIC, and IID), and they suggested that clade IA was able to take up phosphorus using nitrate as the final electron acceptor, but clade IIA was not. Previous studies also proposed that “Ca. Accumulibacter” cells with rod morphotypes, including clade IA, had denitrifying P uptake properties (12, 28, 53). Our study also showed that members of clade IIA (Acc-SG4) with coccus morphotypes were dominant in the AO sludge, but they were not detected in the A2O sludge, in which good denitrifying EBPR was observed (Fig. 2 to 4 and Table 2). However, members of “Ca. Accumulibacter” clade IA (Acc-SG1) were detected as one of the dominant “Ca. Accumulibacter” clades in the AO sludge as well as the A2O sludge. Interestingly, all “Ca. Accumulibacter” cells in the AO sludge had coccus morphotypes, whereas all “Ca. Accumulibacter” cells in the A2O sludge had rod morphotypes, although cells of clades IA (Acc-SG1) and IIF (Acc-SG2) were commonly identified in the AO and A2O sludge (Fig. 4). However, it was inferred from the phylogenetic analyses based on 16S rRNA and ppk1 gene sequences that these morphological differences of the same “Ca. Accumulibacter” clades were not induced by phenotypic changes of the same “Ca. Accumulibacter” organisms but might result from subpopulation changes within the same phylogenetic clades (Fig. 2 and 3). Our observation of different morphologies in the AO and A2O sludge is consistent with previous results that sludge with coccus morphotypes as the major “Ca. Accumulibacter” clade was unable to take up phosphorus under nitrate conditions, while sludge with rod morphotype cells as the major “Ca. Accumulibacter” clade had a P uptake capability using nitrate (12, 26).

Besides the shifts in “Ca. Accumulibacter” clades, the FISH analysis showed that the relative abundance of the total “Ca. Accumulibacter” cells decreased in the A2O sludge, while that of “Ca. Competibacter” and Dechloromonas increased in the A2O sludge (Table 3). The Dechloromonas population increased especially sharply from approximately 1.2% to 19.2%. Previous studies have demonstrated that members of the genera Dechloromonas and “Ca. Competibacter” have nitrate reduction abilities and can accumulate internal polyhydroxyalkanoate (PHA) throughout the SBR cycle (46, 54–59). These results suggested that the nitrite increase observed during the early anoxic phase (Fig. 1) might result from nitrate reduction using internal PHA by Dechloromonas and “Ca. Competibacter” members.

Based on the results presented above, we hypothesized that “Ca. Accumulibacter” clades took up phosphorus using nitrite produced by associated nitrate-reducing bacteria such as Dechloromonas and “Ca. Competibacter” under nitrate conditions. Our FISH/MAR analyses showed that all “Ca. Accumulibacter” clades with rod morphotypes, including clades IIC, IA, and IIF, performed successful P uptake using nitrate in the slightly diluted sludge with sufficient nitrate reduction activity (Fig. 5A and D). However, all “Ca. Accumulibacter” clades failed to take up phosphorus using nitrate in the highly diluted sludge with a lack of sufficient nitrate reduction capability (Fig. 5B and E), while “Ca. Accumulibacter” clades successfully took up phosphorus using nitrite as the final electron acceptor (Fig. 5C and F). These results were in accordance with previous results showing the absence of nitrate reductase and the presence of nitrite reductase in the “Ca. Accumulibacter” metagenome, although the metagenomic analysis was primarily limited to clade IIA genomes (13). Our results indicate that “Ca. Accumulibacter” cells were not able to reduce nitrate themselves but could take up phosphorus using nitrite reduced from nitrate by other associated nitrate-reducing bacteria, such as Dechloromonas and “Ca. Competibacter,” under nitrate conditions. These results are contrary to previous results that suggested that only some fraction of PAOmix-binding cells could take up phosphorus under nitrate conditions (8) and showed that cells in clade IA successfully took up P using nitrate as the final electron acceptor (12, 27, 28). However, we could not be sure that the “Ca. Accumulibacter” populations in our sludge were the same as those in the previous studies, even though they belonged to the same “Ca. Accumulibacter” clades, because “Ca. Accumulibacter” morphologies and phylogenies may be different even within the same “Ca. Accumulibacter” clade (Fig. 2 to 4). Additional studies using various “Ca. Accumulibacter” populations will be required for clearer characterization of denitrifying P uptake properties because “Ca. Accumulibacter” populations are highly diverse, both functionally and phylogenetically (2, 14, 15, 20).