Introduction

Enhanced biological phosphorus removal (EBPR) is a wastewater treatment process that protects receiving waters from eutrophication by enriching microorganisms capable of storing volatile fatty acids (VFAs) as polyhydroxyalkanoates (PHA) under anaerobic conditions (no external electron acceptors) in the first phase of the process. A distinct physiological group of microorganisms enriched in EBPR communities, known as polyphosphate-accumulating organisms (PAOs), anaerobically uptake VFAs and store them as PHA. In one PAO, known as Accumulibacter, acetate uptake is driven by active transport processes utilizing the electrochemical gradient of the proton-motive force (PMF). Generation of the PMF is hypothesized to occur via efflux of phosphate (Pi) (derived from polyphosphate (polyP) breakdown) in symport with protons across the cell membrane (Saunders et al., 2007; Burow et al., 2008). Reducing equivalents for PHA formation are generated by simultaneously catabolizing intracellular glycogen. However, some investigations have demonstrated that the budget for reducing power cannot be met exclusively by glycogen catabolism and anaerobic carbon (C) flux through the tricarboxylic acid (TCA) or glyoxylate cycles may provide this shortfall (Pereira et al., 1996; Pramanik et al., 1999; Lemos et al., 2003; Schuler and Jenkins, 2003). During the subsequent aerobic phase of EBPR, PAOs grow and reproduce using intracellular PHA and the TCA cycle. PAOs also uptake inorganic Pi and convert it to intracellular polyP during this phase. The PAOs are then gravity settled and separated from the wastewater, allowing low phosphorus (P) wastewater to be discharged to the environment (Seviour et al., 2003).

No PAOs are available in pure culture (Seviour et al., 2003; Oehmen et al., 2007), thus anaerobic central metabolic pathway models that describe C flux in these microorganisms have been developed with enrichment cultures. These models have been limited by non-existent or insufficient microbial community structure knowledge (Pereira et al., 1996; Hesselmann et al., 2000; Louie et al., 2000; Filipe et al., 2001; Lemos et al., 2003) or a lack of evidence for their physiological function (Garcia Martin et al., 2006). In situ structure-function studies of Accumulibacter (Lee et al., 2003; Kong et al., 2004) have linked phylogeny to function using fluorescence in situ hybridization-microautoradiography (FISH-MAR) and post-FISH chemical staining to clarify some aspects of PAO physiology.

To provide a more detailed model of metabolic pathways involved in anaerobic carbon metabolism in Accumulibacter, we carried out in situ structure-function analyses and enzyme inhibition investigations combined with biochemical process data acquisition. Bioinformatic analysis of the Accumulibacter genome and measurement of mRNA transcripts using relative quantitative reverse-transcriptase PCR provided further insight into the expression of genes in these pathways.

Materials and methods

Accumulibacter enrichment culture

Accumulibacter were highly enriched (see ‘Results’) from a seed sludge taken from a local full-scale EBPR wastewater treatment plant (Queensland, Australia) in an 8 l laboratory-scale bioreactor that exhibited good Pi removal. Detail of the bioreactor's operation has been previously reported (Lu et al., 2006).

Acetate uptake assays

Biomass was harvested from a parent bioreactor at the end of the aerobic period and resuspended in a 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffered salts medium (pH 7.5) containing trace metals at a final concentration of 250 μg ml−1 of total protein. Protein concentration was determined using the BCA protein assay kit using bovine serum albumin as a standard (Pierce Biotechnology, Rockford, IL, USA). The HEPES-buffered salts medium and acetate uptake assays have been previously described (Burow et al., 2008). Briefly, anaerobic uptake assays were carried out in triplicate in 100 ml batch tests in conical flasks that used oxygen-free nitrogen or helium sparging (15 min) to obtain anaerobic conditions. Batch tests were initiated by the addition of 2 mM acetate and samples were taken at T=1, 10, 20 and 30 min, with uptake rates calculated and subjected to statistical analysis using the Student's t-test. To elucidate anaerobic activity of central metabolic pathways, cells were treated with inhibitors 20 min prior to the addition of acetate. Iodoacetate was used to target glyceraldehyde-3-phosphate dehydrogenase, an enzyme in glycolysis; (mono)fluoroacetate-targeted aconitase, an enzyme common to the glyoxylate and TCA cycles; 3-nitropropionate and itaconate-targeted isocitrate lyase, an enzyme unique to the glyoxylate cycle; malonate-targeted succinate dehydrogenase, an enzyme that oxidizes succinate to fumarate and oxantel-targeted fumarate reductase, an enzyme that reduces fumarate to succinate (see Table 1 for concentrations used).

Table 1 Effect of inhibitors on anaerobic acetate uptake, Pi release and PHA production in Accumulibacter

Anaerobic:aerobic batch tests were carried out with 200 ml of enriched Accumulibacter biomass harvested from the parent bioreactor at the end of the aerobic phase. The 200 ml subsample was sparged for 15 min with oxygen-free helium to ensure anaerobic conditions. Acetate (2 mM) was then added and biomass incubated under oxygen-free conditions for 60 min, which induced intracellular PHA storage. Acetate was measured at the end of the 60 min period to verify its complete consumption. PHA stored under anaerobic conditions was the sole carbon and energy source used by the Accumulibacter enrichment culture in the subsequent aerobic phase. Aerobic conditions were applied to the batch test by sparging with air. PHA was measured at the end of both the anaerobic phase (60 min anaerobic incubation) and the aerobic phase (120 min aerobic incubation) in the enrichment culture.

Chemical analyses

Biomass samples taken during uptake assays for liquid phase (extracellular) analysis were filtered using a sterile 0.22 μm filter (Millipore, Billerica, MA, USA) and stored at 4 or −20 °C for subsequent chemical analyses. Acetate was analysed by high-pressure liquid chromatography (HPLC; Bio-Rad, Hercules, CA, USA) and Pi was determined using a flow injection analyzer (FIA, Lachat Instruments, Loveland, CO, USA). Samples taken for solid phase (intracellular) analysis of glycogen and PHA were fixed with formaldehyde. Glycogen was measured by lyophilization and digestion of biomass with HCl and the resulting supernatant liquid obtained after HCl digestion was analysed for glucose by HPLC (Bond et al., 1999). PHA analysis was determined on lyophilized biomass, acidified in a methanol solution using gas chromatography (Oehmen et al., 2005).

Fluorescence in situ hybridization and quantitative FISH

FISH was carried out with FISH probes listed in Supplementary Tables S1–S3 on fixed samples of biomass as previously described (Amann, 1995). Digital image analysis (ImageJ V1.35k; http://rsb.info.nih.gov/ij/) was used for quantitative FISH with biomass hybridized with Cy3-labelled specific probes and Cy5-labelled EUBMIX probes by calculating the percentage area of the specific probe fluorescence relative to the EUBMIX fluorescence (Burow et al., 2007).

FISH-MAR and post-FISH chemical staining

Combined FISH and MAR was carried out as described in Lee et al. (1999) on biomass samples enriched in Accumulibacter using PAOMIX and EUBMIX probes and radiolabelled acetate (14C) to determine uptake of acetate in these microorganisms in the absence of external electron acceptors (anaerobic conditions). A mixture of acetate (final concentration 2 mM) including radiolabelled acetate (specific activity 2.15 GBq mmol−1; final radioactive concentration 37 kBq ml−1) and unlabelled acetate was incubated with biomass for 2 h under anaerobic conditions prior to FISH-MAR. Assimilation of acetate and conversion to PHA was determined in Accumulibacter by post-FISH chemical staining using PAOMIX and EUBMIX probes and Nile Blue A staining according to previously described methods (Ostle and Holt, 1982; Crocetti et al., 2000).

RNA stabilization, extraction and cDNA synthesis

Expressed genes (mRNA transcripts) were measured in Accumulibacter using relative quantitative reverse-transcriptase PCR (qRT-PCR). The mRNA transcripts in 2 ml (250 μg protein ml−1) of biomass samples were stabilized by incubation with 4 ml of a 5% (v/v) phenol, 95% (v/v) ethanol solution according to methods recommended by the Institute for Food Research Microarray Facility (http://www.ifr.ac.uk). Samples were centrifuged at 21 000 g for 10 min at 4 °C and pellets were retained for storage at −80 °C.

Stabilized biomass pellets were thawed on ice and resuspended in 300 μl of RLT buffer (containing guanidine thiocyanate; RNeasy mini kit; Qiagen, Valencia, CA, USA). Floccular biomass was then dispersed by repeated pipetting through increasingly smaller pipette tips. Lysozyme was added to biomass at a final concentration of 2 mg ml−1, mixed by pipetting and incubated for 5 min at 37 °C with intermittent vortexing. Centrifugation was carried out at 21 000 g for 1 min and the resulting supernatant was transferred to a fresh tube. The manufacturer's protocol for bacterial RNA extraction was followed from this stage (RNeasy mini kit; Qiagen) except that the resulting RNA was eluted in 30 μl of RNase-free H2O containing 1 μl of RNase inhibitor (RNaseOut; Invitrogen). DNase treatment of RNA extracts using Turbo DNase (Ambion, Austin, TX, USA) was performed according to the manufacturer's protocol. RNase inhibitor (1 μl) (RNaseOut; Invitrogen) was added to the DNA-free RNA and stored at −80 °C.

RNA quality and quantity was checked by gel electrophoresis and spectrophotometric analysis (ND-1000; NanoDrop Technologies) prior to synthesis of cDNA templates. Approximately 2 μg of high-quality total RNA was reverse transcribed using the SuperScript III reverse transcriptase kit (Invitrogen) as per the manufacturer's protocol.

Primer design and qRT-PCR analysis

Primers targeting Accumulibacter metabolic genes were designed using nucleotide sequence data from strains of sequenced Accumulibacter (Garcia Martin et al., 2006). Preliminary sequence data from Accumulibacter and other bacteria was obtained from the integrated microbial genomes with microbiome samples (IMG/M) database (http://img.jgi.doe.gov/m). Metabolic gene databases containing 30–40 genes each were constructed for aconitase B (acnB), isocitrate lyase (ICL), succinate dehydrogenase B (sdhB) and cytochrome b/b6 (cyb/b6). The sequences were aligned with ClustalW (Thompson et al., 1994) and imported into ARB for further examination (Ludwig et al., 2004). Nucleotide regions conserved among strains of Accumulibacter but with mismatches to other bacteria were targeted for primer design. Primers were designed using Primer Express V1.5 (Applied Biosystems, Foster City, CA, USA) for amplification and detection of PCR products using SYBR Green chemistry.

Primer sets were evaluated after PCR amplification on the Real-Time ABI Prism 7000 detection system (Applied Biosystems) by gel electrophoresis and sequencing on 3730xl DNA Analyzer (Applied Biosystems). Each PCR volume was 25 μl and included 12.5 μl SYBR Green PCR Master Mix (Applied Biosystems), 5 μl primer (final concentration, 100–400 nM), 5 μl cDNA template and 2.5 μl high-purity pyrogen-free (clinical grade) water. Negative control PCRs included no template controls and RNA-only controls (RNA without reverse-transcriptase cDNA synthesis performed) to check for the presence of contaminating DNA. The thermal cycling parameters included an AmpliTaq Gold (Applied Biosystems) activation step at 95 °C for 10 min; 40 cycles of 95 °C for 15 s, 58 °C for 30 s and 72 °C for 30 s. Primer sets confirmed to amplify genes belonging to Accumulibacter are listed in Table 2.

Table 2 Primers used for qRT-PCR analysis of Accumulibacter

Gene expression of target (metabolic) genes was determined relative to a reference or housekeeping gene (16S rRNA) in three separate anaerobic:aerobic batch tests and the fold-relative gene expression was calculated by the method described by Pfaffl (2001). All primer sets used in qRT-PCR were developed in this study except the 16S rRNA primer set, which was previously reported by Saunders (2005).

Results

Microbial community structure of biomass used in acetate uptake assays

Biomass samples from a bioreactor demonstrating efficient and stable P-removal (reduced Pi from 20–40 mg Pi per litre to <0. 2 mg Pi or 0.645–1.29 mM to <6 μM) were harvested for use in acetate uptake assays. Further details of the process performance of this bioreactor is detailed elsewhere (Lu et al., 2006).

The microbial community structure in subsamples taken from the parent bioreactor were similar (although not identical) to those reported by Burow et al. (2008), as biomass was sampled from the same bioreactor over a similar time period during March–October 2006. Accumulibacter were the dominant microorganism across all biomass subsamples and ranged from 79(±5) to 85(±2)% of Bacteria (Supplementary Table S4). The range is statistically insignificant (P=0.12), providing evidence that the community structure was stable across each biomass sample used in acetate uptake assays.

Although biomass sampled from the bioreactor consisted of a mixed microbial community, conclusions from the use of metabolic inhibitors were made in regards to Accumulibacter central metabolism. This was possible as they were the dominant microorganism within the sludge community and because this microorganism is likely to be responsible for the majority of carbon transformations observed in these uptake assays. Results from FISH-MAR, post-FISH chemical staining and qRT-PCR assays are specific to Accumulibacter. FISH-MAR and post-FISH chemical staining showed that comparatively very few other microorganisms (<5%) were assimilating acetate and/or producing PHA.

Anaerobic acetate uptake, Pi release and PHA production in the Accumulibacter enrichment

Anaerobic assays were performed using whole cells sampled at the end of the aerobic phase from the parent bioreactor. Radiolabelled acetate was taken up by Accumulibacter as determined by FISH-MAR (Supplementary Figure S1) and assimilated to PHA as determined by post-FISH staining with Nile Blue A (Supplementary Figure S2). The Accumulibacter enrichment took up acetate (2 mM) at an average rate of 175±30 nmol min−1 mg−1 protein (Table 1). Anaerobic acetate uptake was linear over the initial 30 min (Supplementary Figure S3). Acetate was assimilated by the Accumulibacter enrichment as PHA with monomer ratios of hydroxybutyrate (HB) to hydroxyvalerate (HV) of 90:10.

Acetate uptake stimulated a concomitant Pi release rate for the Accumulibacter enrichment of 199±43 nmol min−1 mg−1 protein (Table 1) and remained linear over the initial 30 min (Supplementary Figure S4). The ratio between the average Pi release rate to average acetate uptake rate across control batch tests (n=12) was 0.57 P-mol C-mol−1. This ratio supported the finding that the sampled microbial community was highly enriched in Accumulibacter possessing the PAO phenotype (McMahon et al., 2002; Lu et al., 2006).

Anaerobic carbohydrate metabolism in the Accumulibacter enrichment

Models of carbon metabolism in PAOs predict that glycogen catabolism is an important source of energy (ATP) and reducing power (NADH2) under anaerobic conditions (Mino et al., 1998; Oehmen et al., 2007). Consumption of glycogen (stored glucose) was observed in the Accumulibacter enrichment during 60 min incubation with 2 mM acetate under anaerobic conditions. Glycogen decreased from 3.3±0.2 C-mmol glucose per g dry cell weight at the beginning of the incubation period to 1.9±0.13 C-mmol glucose g−1 dry cell weight at the end of the period. The effect of inhibition of glyceraldehyde-3-phosphate dehydrogenase (GAPDH; key enzyme in glycolysis) on glycogen catabolism was investigated by adding 1 mM iodoacetate (Bickis and Quastel, 1965; Even et al., 1999) to an anaerobic incubation prior to the addition of 2 mM acetate. After 60 min incubation with acetate under inhibited conditions the intracellular level of glycogen (3.0±0.3 C-mmol glucose g−1 dry cell weight) did not decrease and was at the same level measured at the beginning of the incubation. This suggested that glucose generated from glycogen is catabolized via the glycolytic pathway.

Anaerobic assimilation of acetate as PHA and tricarboxylic acid metabolism in the Accumulibacter enrichment

It is currently accepted that glycogen catabolism does not provide all the reducing power for generation of PHA from acetate in PAOs (Seviour et al., 2003; Oehmen et al., 2007). According to one metabolic model, PAOs generate approximately 30% of the overall reducing equivalents for anaerobic PHA production from C flux through the full oxidative TCA cycle (Pereira et al., 1996). Other metabolic models have proposed C flux through the glyoxylate cycle (Yagci et al., 2003) or the reductive and oxidative branches of the TCA pathways (split TCA cycle) (Pramanik et al., 1999).

Acetate uptake and production of PHA was investigated in the presence of the aconitase inhibitor, fluoroacetate (Lotspeich et al., 1952). The first step of the TCA and glyoxylate cycles is catalysed by citrate synthase, which converts fluoroacetate to an inhibitory fluorocitrate compound that tightly binds aconitase (Lauble et al., 1996) blocking C flow through the TCA or glyoxylate cycles. The acetate uptake rate of the Accumulibacter enrichment was strongly inhibited in the presence of fluoroacetate compared with the solvent-only (uninhibited) control. Fluoroacetate similarly inhibited concomitant Pi release rates and PHA production compared to controls (Table 1; Supplementary Figures S3 and S4).

The effect of 3-nitropropionate and itaconate, compounds known to inhibit ICL an enzyme unique to the glyoxylate cycle (Hillier and Charnetzky, 1981; Bentrup et al., 1999; Munoz-Elias and McKinney, 2005), was examined in the Accumulibacter enrichment. 3-Nitropropionate substantially inhibited acetate uptake and concomitant Pi release rates (Table 1; Supplementary Figures S5 and S6). The acetate uptake rate, Pi release rate and PHA production in the presence of itaconate were all substantially inhibited in the Accumulibacter enrichment (Table 1; Supplementary Figure S7).

Activity of succinate dehydrogenase in Accumulibacter has been predicted even in the absence of external electron acceptors due to the presence of a novel fusion protein comprising a cyb/b6 complex. It is theorized that reverse electron transport through cyb/b6 driven by the PMF (Δp) could reoxidize reduced quinones produced by succinate dehydrogenase under anaerobic conditions (Garcia Martin et al., 2006). It has been previously shown that acetate uptake in Accumulibacter is sensitive to uncoupling of the PMF using carbonyl cyanide m-chlorophenylhydrazone (CCCP) or selective inhibition of the electrical potential (Δψ) by incubation with valinomycin (Burow et al., 2008). Anaerobic PHA production in the Accumulibacter enrichment after 60 min incubation with 2 mM acetate resulted in a substantial decrease in PHA production down from 4.25±0.39 C-mmol g−1 dry cell weight in the control incubation to 0.97±0.17 C-mmol g−1 dry cell weight in the CCCP (final concentration 100 μM) inhibited incubation. In the presence of malonate, a reversible competitive inhibitor of succinate dehydrogenase (Thorn, 1953; Sumegi et al., 1990), acetate uptake, Pi release and PHA production were all significantly inhibited (Table 1; Supplementary Figure S8).

Genes encoding fumarate reductase have been identified in Accumulibacter (Garcia Martin et al., 2006). However, anaerobic PHA production in the Accumulibacter enrichment after 60 min incubation with 2 mM acetate resulted in a similar level of PHA production in control- and oxantel-inhibited incubations inferring a lack of fumarate reductase activity under these conditions (Table 1). This is consistent with a previous investigation that demonstrated no effect of oxantel on acetate uptake or Pi release rates in Accumulibacter (Burow et al., 2008).

Bioinformatic analysis of sequenced strains of Accumulibacter

Glycolysis activity in Accumulibacter has been previously suggested based exclusively on sequence data analysis that reported the presence of genes in the Embden–Meyerhof–Parnas (EMP) pathway of glucose metabolism and the absence of genes in the Entner–Doudoroff (ED) pathway of glucose metabolism (Garcia Martin et al., 2006). Glucose metabolism via the pentose phosphate (PP) pathway has also been suggested in models of anaerobic PAO metabolism (Pramanik et al., 1999). Investigation of Accumulibacter sequence data deposited in the IMG/M database was carried out to identify potential activity of the PP pathway. Glucose 6-P dehydrogenase gene (zwf), a gene unique to the PP pathway, was absent in sequenced strains of Accumulibacter.

Results from inhibition of glyoxylate cycle activity in the Accumulibacter enrichment prompted a search for genes in this cycle in the Accumulibacter metagenome data set. Genes unique to the glyoxylate cycle (ICL and malate synthase, MS) were identified in Accumulibacter (both US and OZ strains) in this study (Supplementary Figure S9).

It has been previously reported that the malic enzyme (malE) gene encoding an enzyme capable of interconverting C3 intermediates of the glycolytic pathway and C4 intermediates of the glyoxylate cycle (or TCA cycle) is present in Accumulibacter (Garcia Martin et al., 2006). Other genes that similarly catalyse C3:C4 interconversions (phosphoenolpyruvate (PEP) carboxylase, ppc and PEP carboxykinase, pck) were identified in sequenced strains of Accumulibacter in this study (Supplementary Figure S9).

Expression of tricarboxylic acid pathway genes under anaerobic:aerobic conditions in Accumulibacter

To provide molecular evidence of C flux and electron transport through central metabolic pathways in Accumulibacter, mRNA transcripts of metabolic genes were quantified relative to 16S rRNA gene transcripts using qRT-PCR. The 16S rRNA gene and metabolic genes amplified with primers listed in Table 2 using qRT-PCR were of the expected size (Supplementary Figure S10) and had greater sequence identity with the corresponding genes of sequenced strains of Accumulibacter than any other microorganisms (Table 3).

Table 3 Results of nucleotide sequencing of amplicons and BLASTN searches of the nucleotide sequences

The biomass samples used in assays for subsequent qRT-PCR analysis were taken at the end of the aerobic phase from the bioreactor enriched in Accumulibacter and subjected to a complete anaerobic:aerobic EBPR cycle in batch tests. Acetate (2 mM) was provided in the absence of external electron acceptors and PHA was produced (4.75±0.42 C-mmol g−1 dry cell weight) after 60 min anaerobic incubation. Under subsequent aerobic conditions, PHA was depleted (0.25±0.04 C-mmol g−1 dry cell weight) after 120 min incubation. Biomass was sampled from batch tests under anaerobic conditions prior to the addition of acetate to obtain a baseline transcript profile. The anaerobic transcriptional response to the addition of acetate was measured after 10 and 30 min of anaerobic incubation with acetate. The aerobic transcriptional response was measured after 10 and 30 min of aeration subsequent to the completion of a 60 min anaerobic incubation with acetate.

The expression of glyoxylate cycle genes, acnB and ICL, increased under anaerobic conditions in the presence of acetate. Expression of these genes was further enhanced under subsequent aerobic conditions compared to baseline anaerobic conditions where no C was available. The transcript profile data showed that acnB increased 2.3-fold after 10 min and 3.6-fold after 30 min under anaerobic conditions (Figure 1a) and further increased (5.2-fold after 10 min and 4.9-fold respectively after 30 min; Figure 1a) under aerobic conditions compared to baseline expression. Similarly ICL was upregulated under anaerobic conditions by 1.4-and 3.8-fold after 10 and 30 min, respectively (Figure 1b). Under aerobic conditions the profile of ICL further increased by 4.3-fold after 10 min and 6.7-fold after 30 min (Figure 1b).

Figure 1
figure 1

Expression of the genes encoding for (a) acnB and (b) ICL in the glyoxylate cycle of Accumulibacter subjected to sequential anaerobic:aerobic conditions. Transcript profiles were determined with acetate as the C source under anaerobic conditions and intracellular PHA as the C source under aerobic conditions and compared to transcripts expressed under baseline anaerobic conditions in the absence of a C source. Fold relative expression was determined with respect to the 16S rRNA gene.

sdhB and cyb/b6 transcripts increased under anaerobic conditions in the presence of acetate and were enhanced further in the presence of oxygen. An increase in sdhB (1.2-fold after 10 min and 2.5-fold after 30 min; Figure 2a) and cyb/b6 (1.2-fold after 10 min and 2.0-fold after 30 min; Figure 2b) transcripts was observed under anaerobic conditions. The transcript profile data show that sdhB and cyb/b6 transcripts were further enhanced during aeration, with sdhB increasing 3.9-fold after 10 min and 4.6-fold after 30 min (Figure 2a) and cyb/b6 increasing 2.4-fold after 10 min and 2.7-fold after 30 min (Figure 2b).

Figure 2
figure 2

Expression of the genes encoding for (a) sdhB and (b) cyb/b6 in the electron transport chain of Accumulibacter subjected to sequential anaerobic:aerobic conditions. Transcript profiles were determined with acetate as the C source under anaerobic conditions and intracellular PHA as the C source under aerobic conditions and compared to transcripts expressed under baseline anaerobic conditions in the absence of a C source. Fold-relative expression was determined with respect to the 16S rRNA gene.

Discussion

Accumulibacter was the dominant organism in the PAO sludge community (79(±5) to 85(±2)% of Bacteria) and demonstrated acetate uptake and assimilation as PHA (HB/HV) in the absence of external electron acceptors. Previous bioinformatic analyses suggest that acetate is activated via acetylcoenzyme A synthase under relatively low acetate conditions or by the action of acetate kinase and phosphotransacetylase in the presence of higher concentrations of acetate in Accumulibacter. The PHAs generated from acetate, polyhydroxybutyrate (PHB) and polyhydroxyvalerate (PHV) are likely synthesized by the same genes (phaABC) (Garcia Martin et al., 2006; Figure 3).

Figure 3
figure 3

Hypothetical model of anaerobic central carbon metabolism in Accumulibacter. Genes or pathways marked with an asterisk (*) were investigated using inhibitors or gene expression analysis. Solid lines indicate probable carbon flux, whereas broken lines represent possible carbon flux not investigated in this study. Other genes are included based on genome sequencing of Accumulibacter. Filled arrows () represent reduced electron carriers (NADH2, NADPH2 or FADH2) and unfilled arrows () represent oxidized electron carriers (NAD+, NADP+ or FAD+). Green arrows (, ) depict reactions that reduce electron carriers/generate reducing power. Reactions that oxidize electron carriers/consume reducing power are depicted by red arrows (, ). PHB, polyhydroxybutyrate; PHV, polyhydroxyvalerate.

Anaerobic consumption of glycogen (stored glucose) during acetate assimilation to PHA and the cessation of glycogen consumption in the presence of an inhibitor of GAPDH were observed. This suggests that glycolysis can generate reducing power for PHA generation in Accumulibacter (Figure 3). Glycolysis activity in Accumulibacter is consistent with previous reports that enriched Accumulibacter cultures catabolize glycogen under anaerobic conditions (Lu et al., 2006) and as reported in models of anaerobic PAO metabolism (Filipe et al., 2001). Glycolysis is likely to proceed via the EMP pathway, as genes for this pathway are all present, whereas key genes within the ED (Garcia Martin et al., 2006) and PP pathway in sequenced strains of Accumulibacter are not.

Anaerobic carbon flux through the glyoxylate cycle in Accumulibacter (Figure 3) is consistent with the significant decrease in acetate uptake rates and PHA production in the presence of metabolic inhibitors of aconitase (fluoroacetate) and ICL (3-nitropropionate and itaconate). ICL and MS genes both unique to the glyoxylate cycle were subsequently identified demonstrating the potential for activity of this cycle. Expression of acnB and ICL were upregulated in Accumulibacter in the presence of acetate under anaerobic conditions providing further evidence of anaerobic glyoxylate cycle activity. Glyoxylate cycle activity is essential for microorganisms that utilize acetate as a sole carbon source (Kornberg, 1966). In Cornyebacterium glutamicum acetate–glucose mixtures are co-utilized and the glyoxylate cycle has been shown to be essential for its optimal growth on these substrates (Wendisch et al., 2000). PHA production from acetate in PAOs requires reducing power in excess of that generated by glycogen catabolism (Seviour et al., 2003; Oehmen et al., 2007). PHA acts as an electron sink in Accumulibacter and C flux of acetate and/or glucose through the glyoxylate cycle would yield the additional reducing power predicted to be necessary for the PHA production observed in PAOs.

Interconversion of C3 carbon intermediates of the EMP pathway with C4 carbon intermediates of the glyoxylate cycle may consume reducing power in the forward flux direction and generate reducing power in the backward flux direction (Sauer and Eikmanns, 2005). The malE gene encoding for an enzyme that mediates interconversion of malate and pyruvate (concomitantly carboxylates/decarboxylates oxaloacetate) in both the forward and reverse directions has been identified within the Accumulibacter genome (Garcia Martin et al., 2006). The glyoxylate cycle intermediate malate may reenter a pyruvate pool destined for conversion to acetyl-CoA in a pyruvate–malate cycle (Petersen et al., 2000; Lu et al., 2002). Anaerobic activity of this cycle could provide a mechanism to balance the amount of reducing power generated by Accumulibacter (Figure 4). The genes ppc and pck encoding for the PEP-carboxylase and PEP-carboxykinase enzymes, respectively, were identified in this study in the genome of Accumulibacter and may provide an alternative link between central pathways in Accumulibacter (Figure 4). Application of labelled carbon techniques that enable determination of complex metabolic flux networks (Klapa et al., 2003) in highly enriched cultures of Accumulibacter would be one strategy to determine bidirectional fluxes of carbon through these cycles.

Figure 4
figure 4

Hypothesized malate–pyruvate and phosphoenolpyruvate–oxaloacetate cycling potential in Accumulibacter based on gene sequence data. Filled arrows () represent reduced electron carriers (NADH2, NADPH2 or FADH2) and unfilled arrows () represent oxidized electron carriers (NAD+, NADP+ or FAD+). Green arrows (, ) depict reactions that reduce electron carriers/generate reducing power. Reactions that oxidize electron carriers/consume reducing power are depicted by red arrows (, ).

Some metabolic models of anaerobic acetate metabolism in PAOs have overlooked the production of polyhydroxvalerate (PHV) due to the failure to assay for this type of PHA or due to detection of relatively minor amounts (Smolders et al., 1994a; Filipe et al., 2001; Zeng et al., 2003). Zeng et al. (2003) suggested that proliferation of glycogen non-polyphosphate accumulating organisms (GAOs) known to produce relatively high amounts of PHV in EBPR sludge communities performing good Pi removal could explain the presence of PHV in enrichment cultures thought to be solely dominated by PAOs. However, in this study a substantial level of PHV (10%) was measured in high enrichment cultures of Accumulibacter (79%), where GAO populations were low (4–6% of bacteria). In Accumulibacter, succinate may leave the glyoxylate cycle and be degraded to propionyl-CoA and condense with acetyl-CoA to form HV monomers, explaining anaerobic PHV production in these PAOs. Bioinformatic analysis of sequenced strains of Accumulibacter indicates that propionyl-CoA formation could proceed via methylmalonyl CoA mutase conversion of succinyl-CoA to methylmalonyl-CoA and decarboxylation of methylmalonyl-CoA by methylmalonyl decarboxylase to propionyl-CoA (Garcia Martin et al., 2006; Figure 3).

Oxidation of succinate to fumarate in the absence of external electron acceptors in EBPR sludge communities has been difficult to explain due to a requirement for a terminal electron acceptor more electropositive than that of the fumarate/succinate electron carrier (Mino et al., 1998). Despite this, evidence of the activity of succinate dehydrogenase in EBPR sludge communities performing good Pi removal has been demonstrated in the presence of acetate (Pereira et al., 1996; Louie et al., 2000) and propionate (Lemos et al., 2003). Anaerobic succinate dehydrogenase activity was observed in Accumulibacter as determined by inhibition with malonate and the upregulation of sdhB as determined by qRT-PCR assays (Figure 3). No evidence was provided to support the reduction of fumarate to succinate in the Accumulibacter enrichment due to the insignificant effect of the fumarate reductase inhibitor observed on acetate uptake rates (Burow et al., 2008) and PHA production.

Downhill transport of electrons from the fumarate/succinate couple cannot occur in the absence of external electron acceptors. However, uphill electron transport to electron carriers at the cost of PMF (Elbehti et al., 2000) could explain carbon flux mediated by succinate dehydrogenase in Accumulibacter under anaerobic conditions. Garcia Martin et al. (2006) proposed that anaerobic activity of a novel cyb/b6 fusion protein with hypothesized quinone oxidase activity could facilitate reverse electron transport driven by the PMF. The activity of quinone oxidase was inferred in Accumulibacter from the observed upregulation of cyb/b6 under anaerobic conditions in the presence of acetate. Anaerobic acetate uptake in Accumulibacter has previously been shown to require a PMF (Saunders et al., 2007; Burow et al., 2008). A substantial reduction in PHA production occurred in the Accumulibacter enrichment when the PMF was uncoupled using CCCP. Although the observations made here are consistent with quinone oxidase activity in Accumulibacter, further research efforts should investigate the quinone oxidase protein to confirm that it is functioning anaerobically.

Activity of the full oxidative TCA cycle under anaerobic conditions in PAOs is a controversial hypothesis in EBPR research (Seviour et al., 2003). Our efforts to clarify its activity under these conditions were unsuccessful (Figure 3). Attempts to target enzymes unique to the TCA cycle (isocitrate dehydrogenase, IDH or oxoglutarate dehydrogenase, OGDH) in qRT-PCR analysis failed; likely due to the intra-strain variation of gene sequences in Accumulibacter (Table 3). Several primer sets designed (based on near-complete genome sequences of two strains of Accumulibacter) to target regions of IDH (n=6) or OGDH (n=6) unique to Accumulibacter did not generate specific amplicons suitable for qRT-PCR analysis.

In PAOs, TCA pathways including the TCA cycle are thought to operate under aerobic conditions primarily for macromolecule synthesis (proteins, nucleic acids, lipids and polysaccharides) and cell growth, in contrast to their proposed primary role under anaerobic conditions to supply the cell with reducing power for PHA formation (Smolders et al., 1994b; Pereira et al., 1996). Stoichiometric data generated from PAO enrichment cultures propose that some PHA is channelled through the TCA cycle to generate macromolecules and energy via the electron transport chain, whereas another fraction is channelled through the glyoxylate cycle and gluconeogenesis to replenish glycogen (Smolders et al., 1994b; Filipe and Daigger, 1998). Aerobic upregulation of acnB and ICL in Accumulibacter with intracellular PHA as the sole C source indicates glyoxylate cycle activity may be important for flux of C derived from PHA depolymerization. The sdhB and cyb/b6 genes were also upregulated aerobically. As Accumulibacter is an aerobic heterotroph that possesses the full complement of genes in the TCA cycle (Seviour et al., 2003; Garcia Martin et al., 2006) it is likely that the TCA cycle operates aerobically together with the glyoxylate cycle. Aerobic operation of the glyoxylate cycle, which bypasses the decarboxylation steps of the TCA cycle, could be important for regulating the amount of stored carbon that is oxidized to CO2 for energy and growth and the amount used for replenishment of glycogen.

Accumulibacter are aerobic heterotrophs that proliferate under unique anaerobic:aerobic, feast:famine cycling conditions because of their capability to transform intracellular PHA. Efficient storage and utilization of PHA are of key importance to their dominance under these conditions. It is likely that flexibility in C metabolism via C flux through the glycolysis pathway, glyoxylate and TCA cycles and interconversion of intermediates between them is essential for Accumulibacter to respond to the changing electron acceptor and donor conditions in the EBPR process.