Genomic insights into metabolic potentials of two simultaneous aerobic denitrification and phosphorus removal bacteria, Achromobacter sp. GAD3 and Agrobacterium sp. LAD9

Abstract

Bacteria capable of simultaneous aerobic denitrification and phosphorus removal (SADPR) are promising for the establishment of novel one-stage wastewater treatment systems. Nevertheless, insights into the metabolic potential of SADPR-related bacteria are limited. Here, comprehensive metabolic models of two efficient SADPR bacteria, Achromobacter sp. GAD3 and Agrobacterium sp. LAD9, were obtained for the first time by high-throughput genome sequencing. With succinate as the preferred carbon source, both strains employed a complete TCA cycle as the major carbon metabolism for potentials of various organic acids and complex carbon oxidation. Complete and truncated aerobic denitrification routes were confirmed in GAD3 and LAD9, respectively, facilitated by all the major components of the electron transfer chain via oxidative phosphorylation. Comparative genome analysis revealed distinctive ecological niches involved in denitrification among different phylogenetic clades within Achromobacter and Agrobacterium. Excellent phosphorus removal capacities were contributed by inorganic phosphate uptake, polyphosphate synthesis and phosphonate metabolism. Additionally, the physiology of GAD3/LAD9 is different from that displayed by most available polyphosphate accumulating organisms, and reveals both strains to be more versatile, carrying out potentials for diverse organics degradation and outstanding SADPR capacity within a single organism. The functional exploration of SADPR bacteria broadens their significant prospects for application in concurrent aerobic carbon and nutrient removal.

simultaneous aerobic denitrification and phosphorus removal (SADPR), metabolic potentials, genome assembly, carbon metabolism, nitrogen metabolism, phosphorus metabolism

INTRODUCTION

Excessive discharge of nitrogen (N) and phosphorus (P) into water bodies has been identified as one of the major reasons for eutrophication. Biological nutrient removal processes have been verified to be the most effective for N and P elimination from municipal wastewater. Conventional biological N removal technology commonly consists of nitrification facilitated by autotrophic nitrifiers in aerobic conditions and denitrification conducted by heterotrophic denitrifiers in anaerobic or anoxic conditions (Yao et al.2013). In terms of widely applied enhanced biological P removal (EBPR) systems, typical polyphosphate accumulating organisms (PAOs) have a vital function. During the anaerobic phase, PAOs can uptake volatile fatty acids to synthesize intracellular polyhydroxyalkanoates (PHAs), hydrolyze polyphosphate (poly-P) and glycogen to generate energy, and release orthophosphate (ortho-P) extracellularly. In the subsequent aerobic stage, the PAOs are able to incorporate more ortho-P into the cell with formation of poly-P than has been released during the anaerobic period, and thereby achieve efficient P removal from wastewater (Mino, Van Loosdrecht and Heijnen 1998). However, aerobic and anaerobic conditions are alternately needed in biological nutrient removal processes due to the different requirements for nitrification, denitrification, and P release and uptake, commonly resulting in complex and large-size reactors and increased cost, which has become a primary obstacle in the development and innovation of biological nutrient removal technologies.

Novel biological N removal processes such as simultaneous nitrification and denitrification (SND), partial nitrification and denitrification (PND), anaerobic ammonia oxidation (Anammox) and aerobic denitrification have received widespread attention around the world. Although both SND and PND are economically favorable and technically feasible in combination with PAOs for simultaneous N and P removal (Meyer et al.2005; Ji and Chen 2010), the competition for limited organic carbon sources between denitrifiers and PAOs sometimes causes disruption (Lu, Chandran and Stensel 2014). Anammox, shortening the ammonia removal cycle, has become a popular energy-saving and cost-effective biological method; however, it is commonly inhibited by the high concentration of organic carbon and competition for nitrite supply by nitrite oxidation and denitrifying bacteria in wastewater (Zhang et al.2017), and rarely applied in P removal systems. Aerobic denitrification was first reported in the bacterium Thiosphaera pantotropha (Robertson and Kuenen 1984). Since then, diverse aerobic denitrifiers including Alcaligenes faecalis (Joo, Hirai and Shoda 2005), Bacillus methylotrophicus (Zhang et al.2012), Acinetobacter calcoaceticus (Zhao et al.2010) and Psychrobacter sp. (Zheng et al.2011) have been isolated from various wastewater treatment systems. These microorganisms, due to their high growth rate and excellent ability to reduce nitrate to nitrogenous gas aerobically, are of great significance for biological N removal systems where concomitant nitrification and denitrification is much desired (Chen et al.2015). More importantly, a small number of aerobic denitrifiers, such as Pseudomonas stutzeri YG-24 (Li et al.2015) and Agrobacterium sp. LAD9 (Ma et al.2016), have been proven to have the ability to carry out simultaneous aerobic denitrification and phosphorus removal (SADPR) coupled with the capacity for heterotrophic organic carbon degradation, which is believed to have great application potential in a one-stage nutrients treatment process. Compared with conventional processes, the implementation of these SADPR bacteria could not only simplify procedures for a low energy cost, but also increase the nutrient removal rate to meet stringent environmental regulations.

Generally, in addition to the measured characteristics of a single strain, the corresponding functional genes for N and P removal in these bacteria have been inferred. For instance, the coordinated expression of denitrification genes of a novel highly efficient aerobic denitrifier, Pseudomonas stutzeri PCN-1, was confirmed with low NO and N₂O accumulation (Zheng et al.2014). Another good aerobic denitrifier, Klebsiella pneumoniae CF-S9, was detected expressing periplasmic nitrate reductase and nitrite reductase during an aerobic denitrification process (Padhi et al.2013). Interestingly, the poly-P kinase encoded by the ppk gene, catalyzing the formation of poly-P and commonly existing in PAOs, was also found in some denitrifying bacteria (Zeng et al.2016). It was speculated that the occurrence of ppk was possibly related to the simultaneous P removal by denitrifiers. However, these results for N and P removal were all obtained merely by the determination of one or several kinds of functional genes targeted by polymerase chain reaction (PCR) amplification, which could not help in deciphering the comprehensive metabolic potential of single strains exactly and rapidly, due to the low throughput resolution and PCR bias (Chao et al.2016).

Recently, next-generation sequencing technologies (NGST) have been widely utilized to identify both the taxonomic and the functional content of many natural communities, including activated sludge (Yu and Zhang 2012), an acid mine drainage ecosystem (Hua et al.2015) and global soils (Nelson, Martiny and Martiny 2016). With the aid of NGST, high-quality microbial draft genomes have been successfully reconstructed for thousands of microbial organisms, in which the taxonomic status and biotechnological potential could be especially explored to expand the phylogenetic tree of life (Hug et al.2016). The acquired near-complete genomes enabled researchers to deeply understand the metabolic functions and diverse ecological roles of cultured or uncultured species. To date, most studies have only concentrated on the nutrient removal features of single aerobic denitrifying strains; there has been a lack of reconstruction of the draft genomes of SADPR bacteria to outline their functional traits in carbon, nitrogen and phosphorus metabolism simultaneously.

Accordingly in this study, the SADPR performance of two previously isolated bacteria, named Achromobacter sp. GAD3 and Agrobacterium sp. LAD9 (Chen and Ni 2011), were characterized and compared. Their taxonomic and functional profiles were identified through Illumina shotgun sequencing, followed by a series of bioinformatics procedures. Based on the performance of simultaneous N and P removal, genes encoding carbon, nitrogen, phosphorus and extracellular polysaccharide (EPS) related metabolism, and stress response and defense mechanisms were thoroughly analyzed to provide genomic evidence for microbial ecological niches. Comparative genomics discussions were also conducted to disclose the functional difference among these SADPR bacteria with their closely related genomes. The exploration of SADPR bacterial functionality enhances their application value in actual wastewater treatment.

MATERIALS AND METHODS

Evaluation of simultaneous aerobic denitrification and phosphorus removal by two strains

The strains LAD9 (CGMCC No. 2962) and GAD3 (CGMCC No. 2964) were isolated from the activated sludge in landfill leachate treatment systems and identified as Agrobacterium sp. and Achromobacter sp. based on 16S rRNA gene analysis (Chen and Ni 2011). They were stored in 25% glycerol solution and frozen at −80°C.

Two isolated strains were first pre-cultured in 100 mL Luria-Bertani (LB) medium that contained (L⁻¹) 5 g yeast extract, 10 g peptone and 10 g NaCl for 24 h. Enriched cultures were centrifuged (4500× g, 10 min) and washed twice with DD H₂O. The pellets were resuspended in basal medium (BM) for carbon source and SADPR performance evaluation. The ingredients of BM were as follows (L⁻¹): 0.044 g KH₂PO₄, 0.50 g MgSO₄·7H₂O, 0.10 g CaCl₂, 0.006 g FeSO₄·7H₂O, 4.5 g sodium succinate hexahydrate, and 1 mL trace element solution (Zheng et al.2014). Another two types of carbon compound, glucose and sodium acetate, were used as carbon sources instead of sodium succinate in the BM. The amount of each carbon source was determined by the fixed C/N ratio (w/w) of 8. Sodium succinate was the best carbon source for both strains, giving a more rapid growth rate and a higher maximal OD₆₀₀ value than the other two carbon compounds (Supplementary Fig. S1).

To assess the SADPR performance of the two strains, the indicated amount of ¹⁵N-labeled NaNO₃ was added into the BM as the sole N source (nitrate N concentration: 100 mg L⁻¹), with the appropriate amount of sodium succinate maintaining the C/N ratio (w/w) of 8. Isotopic labeled ¹⁵N was employed to avoid contamination by atmospheric N₂ (Zheng et al.2014; Ma et al.2016; Zheng et al.2016). All media were adjusted to pH 7.0 and sterilized for 20 min at 121°C. Cell suspension (1 mL) was inoculated in 70 mL of the above medium stored in 300 mL sealed serum bottles with the headspace filled with air. The culture was cultivated with a shaking speed of 150 rpm at 30°C. All SADPR tests were performed in triplicate. During incubation, periodic measurements of nitrate, nitrite and ortho-P in supernatants, cell OD₆₀₀ and gas (O₂, N₂O and ¹⁵N₂) were conducted based on the methods of our previous research (Ma et al.2016).

Bacterial DNA extraction and genome sequencing

The cultures of strains LAD9 and GAD3 were washed twice with phosphate-buffered saline and then centrifuged at 3420× g for 3 min at 4°C. The pellets were then used for DNA extraction. Total bacterial DNA was extracted using a genomic DNA extraction kit (TianGen, China) according to the manufacturer's protocols and stored at −80°C before use. DNA Library construction and genome sequencing were performed in the Beijing Novogene Bioinformatics Technology Co. Ltd (Beijing, China), by following the manufacturer's instructions.

Shotgun sequencing was conducted with two different constructed libraries: (a) the Illumina Hiseq 2000 platform paired-end (2 × 100 bp) with a library of 5 kbp insert size plus Miseq library paired-end (2 × 250 bp) with insert size of 500 bp for LAD9 genome sequencing; and (b) the Illumina Hiseq 2000 paired-end (2 × 100 bp) with a library of 5 kbp insert size plus Miseq library paired-end (250 and 150 bp) with insert size of 436 bp for GAD3 genome sequencing. Raw reads were removed when the ambiguous nucleotides were more than 10 bp, or more than 40 nucleotides had quality scores of less than 38. A total of more than 2.6 Gb high-quality clean data were generated for the two strains after data filtering. The Illumina sequencing raw data have been submitted to the NCBI Sequence Read Archive (SRA) database with the accession number SRP097652.

Post-run bioinformatics analyses

De novo assembly was conducted using SOAPdenovo2 with optimum k-mer size and parameters (Luo et al.2012). Contigs with a length of more than 500 bp were retained for downstream analysis. Local inner gaps were then closed and the single base errors were corrected using the software Gapcloser (http://sourceforge.net/projects/soapdenovo2/files/GapCloser/). The assembled sequences of both draft genomes were deposited in the GenBank database under the accession numbers PHGT00000000 (GAD3) and PHGS00000000 (LAD9). A series of indicators including genome size, average coverage, G+C content, and so on were summarized to elucidate the assembly features of the two genomes. CheckM (v 1.0.7) was employed to assess the completeness and contamination of the two assembled genomes by considering the universal ribosomal proteins and single-copy genes (Parks et al.2015). The tRNA genes were found using tRNAscan-SE software (v 2.0) (Lowe and Eddy 1997), while ribosomal RNAs were predicted by RNAmmer server (v 1.2) (Lagesen et al.2007). The closely related genomes of GAD3 and LAD9 were downloaded from the NCBI genome repository according to the Phylosift (v 1.0.1) standard (Darling et al.2014), and the average nucleotide identity (ANI) between GAD3/LAD9 and their closely related genomes was calculated using the OrthoANI tool (v 0.93) (Lee et al.2016), as shown in Supplementary Tables S1 and S2. The open reading frames (ORFs) were predicted using GeneMarkS (http://topaz.gatech.edu/) with the default parameters (Besemer, Lomsadze and Borodovsky 2001), and the presumed protein sequences were deduced. The amino-acid predicted ORFs were queried against NCBI non-redundant (NR) (accessed May 2017) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (accessed May 2017) database using DIAMOND (v 0.8.38) with an e-value <10⁻⁵ (Buchfink, Xie and Huson 2015). The matched genes were then assigned to KEGG orthologues (KOs), pathways and categories. Based on the gene annotation, the carbon, nitrogen, phosphorus, EPS and stress response-related metabolisms of the two strains could be inferred.

RESULTS AND DISCUSSION

Comparative performance of simultaneous aerobic denitrification and phosphorus removal by the two strains

The SADPR performance of strain GAD3 is shown in Fig. 1, where nitrate was used as the sole nitrogen source and the O₂ percentage in the headspace of the bottles was set at about 21%. Performance comparison was conducted with strain LAD9, which had been previously measured and reported in Ma et al. (2016).

Figure 1.

Cell growth (a), performance of simultaneous aerobic denitrification (b) and phosphorus removal (c) by GAD3. Open squares, OD₆₀₀; closed squares, O₂; closed triangles, nitrate; open triangles, nitrite; closed inverted triangles, ¹⁵N₂; open inverted triangles, N₂O-N; circles, ortho-P. The data are shown as mean value ± standard deviation from three independent measurements.

As depicted in Fig. 1a, after nearly 6 h time lag for growth, the strain GAD3 entered into logarithmic growth phase and the OD₆₀₀ value reached 1.23 ± 0.02 at 30 h. The lag phase possibly resulted from a sudden environmental change, such as the medium shift of strain GAD3 from LB medium to BM (Swinnen et al.2004). No obvious lag phase was observed in the growth of LAD9, and its OD₆₀₀ finally reached around 0.78, which was far less than that of GAD3. For GAD3, nitrate was degraded gradually after the lag phase (Fig. 1b). One hundred percent of the nitrate was removed within 24 h, and the averaged removal rate reached approximately 4.29 mg L⁻¹ h⁻¹, which was equivalent to LAD9 of 4.20 mg L⁻¹ h⁻¹. The nitrate removal rates of these two strains were much higher than those of a series of other aerobic denitrifiers (Supplementary Table S3) (Patureau et al.2000; Zheng et al.2011; Chen et al.2012; Padhi et al.2013; Liu et al.2016). Accompanying the nitrate reduction in strain GAD3, nitrite started to accumulate at about 12 h and reached a peak with the concentration of 27.57 ± 1.98 mg L⁻¹ at 18 h, but afterwards nitrite was non-detectable any more (Fig. 1b), which differentiated it from LAD9 in which there was hardly nitrite accumulation during nitrate reduction (Ma et al.2016).

The denitrifying intermediate N₂O and final product N₂ were monitored to investigate the pathway for nitrogen removal (Fig. 1b). For GAD3, the denitrification ultimate product, N₂, was detected at 18 h, and achieved a peak of 22.57 ± 2.05 mg L⁻¹ (nearly 25% of removed nitrate) at 30 h, which demonstrated that strain GAD3 could carry out a complete aerobic denitrification pathway with N₂ as the terminal product, even though a small amount of denitrifying intermediate gas, N₂O, occurred at 18 and 21 h. On the contrary, N₂O rather than N₂ was the final nitrogenous product for LAD9 with 55.1% of removed nitrate being transformed into N₂O, which possibly indicated that LAD9 carried out a truncated denitrification pathway. Previous studies demonstrated that the final product of aerobic denitrification could be N₂O or N₂. Marinobacter strain NNA5 (Liu et al.2016), Pseudomonas stutzeri PCN-1 (Zheng et al.2014), Pseudomonas versutus LYM (Shi et al.2013) and Acinetobacter sp. HA2 (Yao et al.2013) were shown to generate N₂ as the single aerobic denitrification product, while Alcaligenes faecalis TUD and Paracoccus denitrificans produce large amounts of N₂O rather than N₂ during aerobic denitrification (Liu et al.2016). This suggests that different species exhibit distinct patterns for aerobic nitrate reduction. Besides, it was notable that the percentage of O₂ decreased to a low level for both GAD3 (2.37%) and LAD9 (4.40%) during the nitrate reduction, suggesting that both strains could perform co-respiration of nitrate and oxygen during the inoculation (Gao et al.2010; Zheng et al.2014).

P removal by GAD3 is shown in Fig. 1c. Ortho-P declined from 9.99 ± 0.19 to 0.75 ± 0.13 mg L⁻¹ within 30 h (with a removal efficiency of 92.5%), illustrating a stronger capability for phosphorus removal than LAD9 has (with a removal efficiency of 76.5%). To date, aerobic denitrifiers capable of simultaneous P removal have been rarely reported. Li et al. (2015) demonstrated Pseudomonas stutzeri YG-24 could simultaneously remove N and P with a P removal efficiency of 51.21% (original P: 2.07 mg L⁻¹), but the removal efficiency was much lower than with strains GAD3 and LAD9 in this study. Compared with most other reported aerobic denitrifiers, strains GAD3 and LAD9 showed a high SADPR capability. To our knowledge, this is the first report of SADPR performance for both Achromobacter and Agrobacterium phylogenies. Considering the potential for application, it is important to deeply understand the comprehensive metabolic characteristics of these SADPR bacteria.

Genome features of two strains

Draft genomes of both strains were successfully reconstructed with around 110- to 130-fold average coverage. GAD3 and LAD9 possessed a total of 28 and 22 scaffolds, with the maximum scaffold size of 1547 and 2599 kbp, respectively. CheckM confirmed that the completeness of both genomes was as high as 99% and the contamination was extremely low (<1%), suggesting that the two acquired draft genomes were of high quality. The genome of GAD3 was estimated to be 6 475 484 bp in total sequence, with a 67.79% G+C content, 5862 ORFs, 6 rRNA and 54 tRNA operons. Complete sequences of 5S, 16S (1519 bp) and 23S rRNA (2819 bp) were recovered from the assembly successfully. GAD3 had an ANI of >98% with 11 draft genomes of Achromobacter xylosoxidans phylogenetic clade II, much higher than those (ANI: 82–92%) with any other Achromobacter clades (Supplementary Table S1). For LAD9, the genome was estimated to be 5 931 195 bp in total length, with G+C content of 59.12%, 5729 ORFs, 45 tRNA and no successfully recovered rRNA operons, sharing a much higher ANI (>97%) with six strains belonging to Agrobacterium tumefaciens phylogenetic clade II than those (ANI: 72–92%) with any other Agrobacterium clades (Supplementary Table S2). Based on the ANI cutoff of 95–96% (Richter and Rosselló-Móra 2009), both GAD3 and LAD9 would likely share the same species level phylogeny with their closely related genomes. The gene annotation results of strains showed that 47–53% of total ORFs could be annotated by KEGG KO, and more than 90% of total ORFs could be annotated by the NCBI NR database. Approximately 18–24% of total ORFs were annotated as hypothetical proteins, revealing that more functionality is yet to be discovered for these two strains.

Carbon metabolism

The gene sets of GAD3 and LAD9 were dominated by those involved in the KEGG categories of membrane transport, amino acid metabolism, carbohydrate metabolism, energy metabolism and cellular community (Table 1). The utilization of multiple carbon sources is of great significance in relation to N and P removal, as the carbon materials not only serve for cellular macromolecule biosynthesis, but also as the electron and energy sources for aerobic denitrification and phosphorus metabolism (Kristiansen et al.2013; Liu et al.2016). In an actual wastewater treatment system, the carbon sources potentially have a stronger influence on the denitrifying communities than most other factors, as organic carbon metabolism with diverse pathways is the foundation of heterotrophic growth (Lu, Chandran and Stensel 2014). Thus, looking into carbon consumption routes within both SADPR strains is necessary.

Table 1.

The amount of genes assigned to KEGG categories (second level) and identification of central metabolic pathways by two strains.

KEGG categories	GAD3	LAD9	Central metabolic pathways or functions	GAD3	LAD9
Xenobiotics biodegradation and metabolism	98	80	Carbon fixation	−	−
Nucleotide metabolism	84	112	Glycolysis EMP	(+)	+
Metabolism of terpenoids and polyketides	46	43	Glycolysis ED	−	+
Metabolism of other amino acids	103	65	Pyruvate oxidation to acetyl-CoA through pyruvate dehydrogenase	+	+
Metabolism of cofactors and vitamins	176	182	Citrate cycle (TCA cycle)	+	+
Lipid metabolism	109	90	Pentose phosphate cycle	(+)	+
Glycan biosynthesis and metabolism	82	68	Starch and sucrose hydrolysis	(+)	+
Enzyme families	95	78	Glyoxylate cycle	+	+
Energy metabolism	212	182	Nucleotide sugar biosynthesis	+	+
Carbohydrate metabolism	281	276	Organic acids utilization	+	+
Biosynthesis of other secondary metabolites	54	46	PHA transformation	+	+
Amino acid metabolism	384	306	β-Oxidation (fatty acids)	+	+
Translation	87	87	Nitrogen fixation (N2→ammonia)	−	−
Transcription	4	5	Dissimilatory nitrite reduction to ammonia	+	+
Replication and repair	52	53	Complete denitrification (nitrate→N2)	+	(+)
Folding, sorting and degradation	45	40	Assimilatory sulfate reduction	+	+
Signaling transduction	159	129	Low-affinity phosphate transporters	+	+
Membrane transport	364	374	High-affinity phosphate transport system	+	+
Transport and catabolism	20	13	Polyphosphate accumulation and degradation	+	+
Cell motility	58	70	Phosphonate metabolism	+	+
Cell growth and death	28	49	F-type ATPase	+	+
Cellular community–prokaryotes	315	239	V-type ATPase	−	−
Drug resistance	65	60	NADH:quinone oxidoreductase	+	+
			Succinate dehydrogenase	+	+
			Cytochrome bc1 complex	+	+
			aa3-type cytochrome c oxidase	+	+
			cbb3-type cytochrome c oxidase	+	+
			Cytochrome bd complex	+	+

+, a complete pathway or function is present; (+), an incomplete pathway with the absence of a few functions; −, a pathway or function is absent or nearly absent. ED, Entner–Doudoroff; EMP, Embden–Meyerhof–Parnas; PHA, polyhydroxyalkanoate.

The potential for a heterotrophic lifestyle of both SADPR strains was inferred from a lack of key genes for carbon fixation and the existence of genes for organic carbon degradation. Specifically, as predicted from annotation results, all the CO₂ fixation routes, including the Wood–Ljungdahl pathway, Calvin cycle, reverse TCA cycle, 3-hydroxypropionate bi-cycle, dicarboxylate–hydroxybutyrate cycle and hydroxypropionate–hydroxybutyrate cycle, were incomplete or nearly absent for both strains (Table 1). The glucose degradation pathway and TCA cycle were of particular concern, as they are the central metabolic routes involved in carbohydrate oxidation and energy production for most heterotrophs. GAD3 and LAD9 were observed to harbor a complete TCA cycle, including a total of 28 and 18 separately involved genes (Fig. 2; Supplementary Tables S4 and S5). For strain GAD3, the glycolytic Embden–Meyerhof–Parnas (EMP) pathway, where α-D-glucose is irreversibly transformed into pyruvate to produce ATP and NADH, was putatively partial due to the lack of two critical enzymes (phosphorylating glucokinase and 6-phosphofructokinase). The Entner–Doudoroff (ED) pathway also appeared to be incomplete for GAD3, as a result of the absence of glucose-6-phosphate 1-dehydrogenase and phosphogluconate dehydratase. These were possibly responsible for the low utilization efficiency of glucose (Supplementary Fig. S1a). Definitely, the glycolytic EMP and ED pathways were also incomplete for all the selected closely related strains of GAD3 based on the KEGG annotation. With respect to LAD9, the glycolytic EMP and ED pathways and the pentose phosphate pathway were all observed completely in the genome, and many genes did not share the same orthologues with GAD3. Taking the carbon source effectiveness evaluation into account (Supplementary Fig. S1a–c), glucose was actually not the most ideal carbon source for both strains, as the OD₆₀₀ only reached as low as 0.2 (GAD3) and 0.3 (LAD9) after 24 h cultivation, which demonstrated that the glycolytic EMP pathway route (with requirements of multiple-step reactions from glucose to pyruvate) was inefficient in a short period for both strains’ metabolic activities. On the other hand, succinate was the optimal carbon source for GAD3 and LAD9 observed in this study. Previous research has reported that succinate was the most preferred carbon source for a variety of aerobic denitrifiers, as succinate yields energy via the TCA cycle more easily and directly than glucose and acetate (Ji et al.2015). As the TCA cycle appeared to be complete in both strains, most organic carbon sources could be metabolized via the TCA cycle to produce a large amount of ATP and reducing power for the SADPR process. It could also produce precursors such as citrate, oxoglutarate and oxaloacetate for biosynthesis of amino acids, fatty acids and cellular building blocks. The co-existence of UTP-glucose-1-phosphate uridylyltransferase (galF/galU) and those enzymes catalyzing oxoglutarate to α-glucose-6P could facilitate the unidirectional generation of UDP-glucose, which is the crucial precursor for exopolysaccharide biosynthesis and other anabolic pathways in both strains (Fig. 2; Supplementary Tables S4 and S5).

Figure 2.

Proposed metabolic potential of GAD3 and LAD9 inferred from their draft genomes. Predicted genes or functions indicated by numbers in the figure can be found in Supplementary data.

Moreover, GAD3 and LAD9 were predicted to utilize ethanol and a series of other organic acids (including formic acid, acetic acid, acetoacetic acid, propionic acid, lactic acid and malic acid) as electron donors to derive energy for the SADPR process (Fig. 2; Supplementary Tables S4 and S5). There was an acetyl-CoA synthetase (acs) encoded in draft genomes of GAD3 and LAD9, which might be the reason why these bacteria could also efficiently utilize acetate (Supplementary Fig. S1b). GAD3 also contained phosphate acetyltransferase (pta) and acetate kinase (ackA), reversibly catalyzing the transformation of acetate to acetyl-CoA with consumption of ATP. However, the pta–ack reaction forms a lower affinity pathway for acetate assimilation compared with the high-affinity acetyl-CoA synthetase pathway (Wolfe 2005). In the meantime, the propionic acid could be metabolized and consumed via the 2-methylcitrate pathway in GAD3 and via the (S)-methylmalonyl-CoA degradation route in LAD9. The complete PHA transformation routes were also identified in the genomes of GAD3/LAD9, facilitated by a series of enzymes involved in the pathways from acetyl-CoA to poly-β-hydroxybutyrate or polyhydroxyvalerate/polyhydroxy-2-methylbutyrate (Mao et al.2014). The PHA accumulation is usually a natural process for bacteria to store carbon and energy when growth is limited by depletion of N, P or O₂ and an excessive amount of carbon source is still present (Verlinden et al.2007). In typical EBPR systems, the biosynthesis of PHAs in the anaerobic period is vital for both the proliferation of cells and the accumulation of poly-P for PAOs in the aerobic stage. In the present study, the SADPR process was performed with a relatively low C/N and C/P ratio, so it was unlikely that cells of either strain accumulated a considerable amount of PHAs during the incubation. However, as the pathways for PHA biodegradation and synthesis were certainly complete, it could be expected that GAD3/LAD9 could perform the process to some extent.

The potential for complex carbon degradation was further investigated in both heterotrophic strains (Fig. 2; Supplementary Tables S4 and S5). The strain LAD9 employed much more varied types of oligosaccharide and monosaccharide transport system ATP-binding cassette (ABC) proteins than GAD3. Once transported into the cytoplasm, the intracellular saccharides might undergo breakdown to glucose-1P or fructose-6P via starch, sucrose, galactose or mannose, then enter the glycolytic pathway or pentose phosphate cycle for biomass synthesis. Starch, sucrose, cellodextrin, cellobiose, chitobiose, xylose, galactose, raffinose, fructose and mannose degrading enzymes and related pathways were particularly identified in the genome of LAD9, indicating an extensive capacity to degrade complex carbon compounds. Indeed, the functional traits of some plant-derived metabolites for decomposition and core energy metabolism (such as the EMP and ED pathways and the TCA cycle) were also detected in some species of Agrobacterium tumefaciens previously (Bouzar, Jones and Hodge 1993; Wood et al.2001), which was in agreement with our isolate LAD9. The GAD3 genome had only starch hydrolysis and chitobiose degrading enzymes, rendering it a lesser degrader of plant-derived organics compared with LAD9. Additionally, complete β-oxidation of long chain fatty acids was detected in both strains, the enzymes involved in the degradation of benzoate, naphthalene, salicylate, 2-amino-benzoate, nitrobenzene and catechol resided in GAD3, and pathways for 4-hydroxy-benzoate and 3,4-dihydroxy-benzoate degradation to succinyl-CoA were conserved in the LAD9 genome. Notably, GAD3 shared a high ANI (>98%) with the well-known aromatic compound-degrading organism Achromobacter xylosoxidans, which could maintain survival with a high amount of p-nitrophenol, chrysene and bisphenol A based on previous research (Wan, Gu and Yan 2007; Zhang et al.2007; Ghevariya, Bhatt and Dave 2011). However, given the lack of data here, further tests need to be done to reveal the breakdown potentials and mechanisms of recalcitrant organic materials for GAD3. As the carbon pool of combined municipal and industrial wastewater is continually complicated and rich in types, multiple heterotrophic carbon consumption routines within a single SADPR organism are normally required for sustainable performance. Both GAD3 and LAD9 might be key denitrifying and phosphorus removal populations through potentially utilizing more diverse carbon sources.

Nitrogen metabolism

Microbial denitrification is primarily dependent on the activities of four important denitrifying enzymes, nitrate reductase (NAR), nitrite reductase (NIR), nitric oxide reductase (NOR) and nitrous oxide reductase (N₂OR). GAD3 maintained many key genes for the complete aerobic denitrifying pathways (NO₃⁻→NO₂⁻→NO→N₂O→N₂), including both heme-containing membrane respiratory NAR clusters (narGHJI) and periplasmic NAP subunits (napABCDE), copper-type NIR (nirK), cytochrome c-dependent NOR (norB) and N₂OR clusters (nosXLYFRZD) (Fig. 3a). The unabridged denitrifying gene clusters would well explain the experimental data that N₂ rather than N₂O was the final product of aerobic denitrification in GAD3. It was notable that GAD3 possessed two types of NAR, namely the nar operons responsible for anaerobic nitrate reduction and the periplasmic nap operons that reduce nitrate in the presence and absence of O₂ (Richardson et al.2001), showing the capacity to undergo denitrification in oxic, micro-oxic and anoxic environments. narX–narL, a two-component system acting as a sensor for changes in nitrate/nitrite concentrations, was located not far from narGHJI, even though they were blocked by genes coding for molybdopterin-guanine dinucleotide biosynthesis protein (mobB) and molybdenum cofactor biosynthesis protein, which are quite important for the formation of nitrate reductase delta subunit (narJ) (molybdenum cofactor assembly chaperone). Two nitrate/nitrite transporters (narK) were closely linked to nar clusters. Various types of genes in response and signaling to NO, such as nnrS, nnrR and dnrN, were present in GAD3, indicating a sensitive unique mechanism of NO detoxification in the cells. For LAD9, the periplasmic NAR clusters (napABCDEF), copper-type NIR (nirK) and NOR clusters (norBCDEFQ) all exist close to one another in the genome, but the respiratory nar operons and the nos operons for N₂O reduction were both lacking (Fig. 3a), which was consistent with our present culture-based finding that LAD9 carried out a truncated denitrifying pathway (NO₃⁻→NO₂⁻→NO→N₂O). One NO sensing protein (nnrS) and one denitrification regulatory protein (nnrU) were adjacent to the nor operons and nirKV. Both strains harbored compact aerobic denitrification gene arrangements with a relatively high alignment similarity (Supplementary Tables S4 and S5), suggesting that GAD3 and LAD9 are both ideal aerobic denitrifiers. Besides, both strains had nitrate/nitrite transport system substrate-binding proteins encoded by genes nrtABC and dissimilatory nitrite reduction to ammonia pathways encoded by NADH-nitrite oxidoreductase (nirBD); these two functions are crucial for microbes to take up nitrate/nitrite into the cytoplasm and then produce ammonia for L-glutamate biosynthesis with nitrate or nitrite as the sole nitrogen source.

Figure 3.

(a) Nitrate reduction and denitrifying related gene arrangements located on draft genomes of GAD3 and LAD9. Parallel double lines indicate a break in locus organization among scaffolds, and dotted black lines indicate where unrelated continuity loci are not shown. Numbers below the line symbolize the locations. Genes and non-coding regions are nearly drawn to scale. (b) The substance transformation and electron transfer chains during the aerobic denitrification process were identified from draft genomes of both strains.

Our present study relied on genome sequencing for a more complete understanding of the nitrogen metabolism within these two SADPR bacteria, compared with quantitative PCR results of key aerobic denitrifying genes (nap, nir, nor and nos) in a series of other strains (Padhi et al.2013; Zheng et al.2014; Zheng et al.2016). A few Achromobacter strains, such as Achromobacter sp. DBTN3 (Kathiravan and Krishnani 2014) and Achromobacter xylosoxidans NH44784-1996 (Jakobsen et al.2013), were reported previously to be involved in anaerobic denitrification processes, and thus our findings that GAD3 could carry out a complete aerobic denitrification might offer a new perspective on the nitrogen metabolism in this phylogenic group. Interestingly, the complete aerobic denitrification enzymes were mostly found in 91% of strains (except str. X02736) belonging to A. xylosoxidans clade II, strain A8 of clade III and representative genomes of three other species (A. denitrificans, A. arsenitoxydans and A. insuavis), while 80% of strains (excluded str. FDAARGOS_147) within A. xylosoxidans clade I and four other species (A. piechaudii, A. insolitus, A. ruhlandii and A. spanius) might not possess the denitrification pathway. This might first reveal the distinctive ecological niches involved in the denitrification process for different phylogenetic clades of the genus Achromobacter. As a closely related species to LAD9, Agrobacterium tumefaciens C58 cells have been confirmed to express both nitrite and nitric oxide reductases (Baek and Shapleigh 2005), and not harbor nos clusters in the genome (Wood et al.2001), which made strain C58 act as a sink for the N₂O production in soil (Jones et al.2014). Comparative genome analysis demonstrated that all the reconstructed genomes affiliated with A. tumefaciens clade I, II and III employed the truncated aerobic denitrification routes, and six representative genomes of other Agrobacterium species lacked all the key denitrification enzymes. For now, the truncated denitrifying potential of Agrobacterium was predicted to be associated with specific clades. We assume the isolation and application of our SADPR organisms with complete or truncated aerobic denitrification routes could greatly help to enhance total N and P scavenging efficiency and facilitate N and P cycling in complex ecosystems.

Substance transformation and electron transfer under aerobic denitrification are facilitated by sophisticated genes encoding oxidative phosphorylation pathways obtained from draft genomes of GAD3/LAD9 (Fig. 3b; Supplementary Tables S4 and S5). First of all, the NADH dehydrogenase (respiratory complex I encoded by nuo subunits) and succinate dehydrogenase (complex II encoded by sdhABCD, which connects the TCA cycle to the quinone pool) provide reducing power through the conversions of NADH/H⁺ to NAD⁺ and succinate to fumarate, respectively. The generated electrons are then transferred to the ubiquinone (UQ), cytochrome bc₁ complex (complex III) and finally cytochrome c (cyt c). Only nap clusters can compete with O₂ for electrons transferred from UQH₂ and reduce nitrate to nitrite under aerobic condition, while nar operons are limited when O₂ is present (Richardson et al.2001). Cyt c can transfer electrons to the key denitrification enzymes including nirK, norB and nosZ, and subsequently nitrite can be reduced to NO, N₂O or N₂ at the end. In addition, with respect to cytochrome oxidases (complex IV), both strains also harbored a low-affinity aa₃-type, cbb₃-type cyt c oxidase and a high-affinity quinol oxidase bd complex for O₂, which might be associated with the oxidation of alternative electron donors (e.g. organic carbon) during aerobic nitrate respiration. Their different affinities to O₂ could protect cells against oxidative stress in a fluctuating environment (Mardanov et al.2016). Evidence showed that cbb₃-type cyt c was highly expressed under growth conditions where nap was also expressed at high levels (Van Spanning et al.1997).

In addition to the aerobic denitrification, GAD3 and LAD9 were predicted to take up and utilize urea due to the presence of urtABCDE coding for a high-affinity urea ABC transporter in both genomes, and urea carboxylase and urease present in GAD3 and LAD9 respectively. On the other hand, GAD3 was likely to degrade nitroalkane, cyanate, formamide and nitrile in the presence of nitronate monooxygenase, cyanate lyase, formamidase and nitrilase, while LAD9 might utilize nitroalkane and nitrile only. This might confer a benefit for GAD3/LAD9 in environments with the presence of other nitrogen sources coupled with low nitrate and ammonium concentrations.

Phosphorus metabolism

The most intensively studied processes of bacterial phosphorus metabolism in EBPR technology are the uptake and transport of inorganic phosphate (P_i) and poly-P synthesis (Mao et al.2014). P_i transport is the first important step for EBPR metabolism and has a direct effect on the performance of P removal. For both GAD3 and LAD9, several types of P_i transport systems, such as low affinity inorganic phosphate transport encoded by pit and phosphate-specific transport permease proteins encoded by pst clusters and phoU, were identified on the genomes (Fig. 4; Supplementary Tables S4 and S5). pit encodes a P_i-binding protein located in the periplasm and functions in mediating P_i translocation and energizing the transport, while pst clusters (comprising pstSCAB) belong to the superfamily of ABC transporters functioning in high-affinity P_i transport (Aguena, Yagil and Spira 2002). Both pit and pst are most likely induced when cells undergo P_i starvation during aerobic growth. After being transported into the cells, P_i is considered to combine with ADP to form ATP through the F-type H⁺-transporting ATPase (Martín et al.2006) encoded by atp clusters identified on genomes of both strains, and the arrangement of atp clusters was quite similar between the two strains as atpCDGAHFEBI. Due to the limitations of genome assembly from clean reads, the atp clusters on LAD9 were arranged on two separate scaffolds. The other V-type H⁺-transporting ATPase, which could also combine P_i with ADP to generate ATP, was totally missing from both strains’ draft genomes and any other closely related organisms applied in the present study. GAD3/LAD9 shared common genes of P_i transport with all the Candidatus Accumulibacter and Tetrasphaera PAOs (Martín et al.2006; Kristiansen et al.2013; Mao et al.2014; Skennerton et al.2015; Mao et al.2016), which could demonstrate an excellent P_i uptake capacity for the two strains during the SADPR process.

Figure 4.

The organization of key phosphorus metabolism genes in GAD3 and LAD9 draft genomes (phosphonate operons; see Supplementary Tables S4 and S5). Parallel double lines indicate a break in locus organization among scaffolds, and dotted black lines indicate where unrelated continuity loci are not shown. Numbers below the line symbolize the locations. Genes and noncoding regions are drawn to nearly scale.

Poly-P metabolism has been studied in detail for many bacteria including Escherichia coli (Kristiansen et al.2013). Similar to all the available PAOs, both GAD3 and LAD9 also harbored the poly-P kinase (ppk) gene and exopolyphosphatase (ppx) gene, which catalyze the biosynthesis and hydrolysis of poly-P respectively (Fig. 4; Supplementary Tables S4 and S5). Phylogenetic analysis showed that ppk genes from GAD3/LAD9 were rather distant from any clades of available PAOs, clustering with their closely related organisms (Supplementary Fig. S2). In the GAD3 draft genome, the ppk gene was identified upstream of the ppx gene and both ppk and ppx genes were linked closely with the pst clusters, which constituted a complete phosphate transport, poly-P accumulation and decomposition pathway for GAD3. In the draft genome of LAD9, the pair of ppk and ppx genes, atp clusters, pst clusters and phoBU were all included on the same reconstructed scaffold, but appeared to be disconnected with each other. The ppk gene was absent in four strains (str. CCNWGS0286, B140_95, B6 and Kerr 14) of A. tumefaciens clade II and all strains of clade I, which revealed the poly-P might not be accumulated by these organisms. Besides, the polyphosphate AMP phosphotransferase encoded by the pap gene, which directly catalyzes the reverse conversions of AMP to ADP at the expense of poly-P, was absent from GAD3/LAD9, indicating the pap/adk reaction might not serve as an additional mechanism for poly-P synthesis and degradation (Kristiansen et al.2013). All the closely related strains of LAD9 did not contain the pap/adk reaction either. Therefore, during the SADPR process of GAD3/LAD9, it could be hypothesized that a considerable amount of P_i is first transported into the intracellular cytoplasm through both the low- and high-affinity P_i transport system, and then some of the P_i is thought to be synthesized into poly-P by the ppk activity, which transfers high-energy phosphate groups from ATP to a growing poly-P linear polymer, accompanied by carbon source consumption to produce reducing power and energy. Intracellular stored poly-P granules are dynamic and could be depleted in reactions catalyzed by the ppx enzymatic activity under phosphorus starvation (Ma et al.2016). As previously shown, ∼73% of intracellular P of strain LAD9 was dominated by organic P including ortho-P monoesters, phospholipids and DNA, which also demonstrated that anabolic P routes could make major contributions to aerobic P removal (Ma et al.2016). Further transcriptomic analysis has been taken into consideration to reveal the active P transformation pathways for GAD3/LAD9.

In addition, a two-component system encoded by phoBR, which acts as the phosphate starvation response regulator, and phosphate starvation-inducible proteins phoH/phoL were found in both GAD3 and LAD9 genomes. They might be expressed at a high level when bacterial growth encounters phosphate limitation (Wanner and Chang 1987). However, the phoBR operons were absent on Candidatus Accumulibacter clade IIC HKU-2, IIF SK-11 and all four Tetrasphaera species based on KEGG or NCBI genome searching. Both strains also contained the phosphonate ABC transporter system and phosphonate hydrolase (phn operons) for the phosphonate uptake and hydrolysis, similar to previously proposed in activated sludge bacteria Zoogloea resiniphila (An et al.2016). But the phn operons were observed to be nearly absent for any PAOs, suggesting the lack of an alternative dissolved organic P accumulation route. In total, both strains are promising organisms with the capabilities of aerobic denitrification and P accumulation, as demonstrated by cultured data and genome annotation.

Extracellular polysaccharide-related gene clusters of the two strains

The EPS produced by activated sludge bacteria plays an important role in the formation of stable multicellular conglomerates of microbial clusters. The bacterial flocs that are formed make it possible to improve activated sludge settleability by gravitational precipitation in a settlement tank and release the settled sludge back to an aeration tank with high cell densities to maintain the wastewater N and P removal process (An et al.2016). Here, the various putative EPS biosynthesis and export gene clusters identified in the genomes could also make both GAD3 and LAD9 promising SADPR performers in activated sludge.

Complete capsular polysaccharide biosynthesis, transport and export gene clusters were identified on GAD3 scaffold_1 (Supplementary Table S6). The bacterial capsular polysaccharides enable surface adhesion and resistance to environmental toxins and desiccation, and are also key viral factors of human disease (Nielsen et al.2016). The presence of pgaABCD operons encoding biofilm production, which is involved in cell-to-cell and cell-to-surface adhesion, was not surprising, as they were also observed in the genome of Achromobacter xylosoxidans strain NH44784-1996 (Jakobsen et al.2013). Additionally, we have also obtained genes coding for the lipoprotein releasing system (lolCD), lipopolysaccharide biosynthesis and export system (lptABC), and polysaccharide and colanic acid biosynthesis in the genome of GAD3 with a high identity (mostly >99%), coupled with the EPS chain length determining protein Etk/Wzc (Supplementary Table S6). Previous study has demonstrated that the Achromobacter along with Zoogloea, Alcaligenes, Pseudomonas and Flavobacterium genus constituted major organisms responsible for the formation of flocs in activated sludge (Maurines-Carboneill et al.1998).

The EPS composition and structure of LAD9 might be different from those of GAD3. A large EPS biosynthesis gene cluster including 13 compactly connected genes that participate in succinoglycan biosynthesis, polymerization and export was employed on LAD9 scaffold_6 (Supplementary Table S7). Previous studies have demonstrated that the exopolysaccharide succinoglycan is required for biofilm formation and motility, and is usually synthesized by a variety of bacteria belonging to the family Rhizobiaceae (Rhizobium, Agrobacterium) (Tomlinson et al.2010). Also, as revealed by the comparative genome analysis, most of the Agrobacterium related species (except A. albertimagni AOL15) or strains shared very similar EPS biosynthesis genes with LAD9, which could indicate a lower EPS composition and structure variation within the genus Agrobacterium. Besides, our previous research concluded that the LAD9 EPS harbored 16.5% of removed total phosphorus and acted as a reservoir of phosphorus removal (Ma et al.2016), which was comparable to EBPR with 5–9% of triose phosphate removal contributed by EPS (Zhang et al.2013). Similarly with GAD3, the lipoprotein releasing system, lipopolysaccharide biosynthesis and export proteins were recognized in LAD9, even though they were somewhat disconnected in the genome. Further chemical analysis on EPS of GAD3 and LAD9 is still ongoing and more trials need to be made to reveal the EPS biosynthesis pathways and regulatory mechanisms among different genera, as the available achievements in EPS production are just the tip of the iceberg.

Stress response, defense and resistance

For practical applications, both SADPR bacteria are usually exposed to dissolved oxygen, and thus should harbor multiple oxidative stress defense mechanisms (Supplementary Tables S8 and S9). Genes coding for catalase, superoxide dismutase and peroxidase, providing bacteria with protection against reactive oxygen species, were detected on both genomes. GAD3 employed diverse genes related to reactive oxygen species protection including two kinds of catalase and superoxide dismutase and three types of peroxidase (cytochrome c, glutathione and putative iron-dependent peroxidase), while LAD9 possessed one type of catalase and superoxide dismutase, and two types of peroxidase (cytochrome c and non-heme chloroperoxidase), indicating that they were associated with the oxidative stress resistance and excess oxygen scavenging.

Furthermore, diversiform mechanisms that allow microbes to respond to the changing environment are described here (Supplementary Tables S8 and S9). A series of two-component systems that are involved in sensing and responding to changes in the surrounding environment (for instance, osmolarity, nutrient availability, low pH tolerance and chemotaxis) appeared to be present in both strains. In addition to signal transduction systems (methyl-accepting chemotaxis protein and che operons) of chemotaxis, both genomes had a set of genes involved in flagellin and flagella assembly and related motor rotation proteins. Hence, based on these genome characteristics, both strains might be capable of a chemotactic response to the changing environmental conditions with great motility, moving towards a favorable location in a natural environment. Also, more than five copies of cold shock protein cspA, one copy of membrane-bound heat shock protein htpX and molecular chaperone dnaKJ were definitely obtained for GAD3/LAD9, and these genes were putatively induced to regulate the adaption to temperature variation after a rapid temperature downshift or upshift (Arsène, Tomoyasu and Bukau 2000; Rohlin et al.2005; Horn et al.2007). Some psychrotrophic aerobic denitrifying bacterial strains have recently been isolated and characterized with highly efficient nitrate and nitrite removal capacity below 10°C (Huang et al.2013; Yao et al.2013), implying the potential for promising applications in typical wastewater treatment plants of northern China during the winter time. It was further revealed that GAD3/LAD9 possessed various types of copper, arsenic and chromate resistance proteins that conferred tolerance to the overload of such bio-toxic metals in wastewater, which is likely to expand the applications of both strains in industrial wastewater treatment.

Engineering implications

Taken together, the present study determined the SADPR performance and constructed metabolic models for two isolates, Achromobacter xylosoxidans strain GAD3 and Agrobacterium tumefaciens strain LAD9, and is the first report of SADPR based on both culture-based test and genome insights for Achromobacter and Agrobacterium phylogenies. Both strains were characterized as obligate heterotrophic organic carbon transformers, accompanied by sets of genes for aerobic denitrification pathways, phosphorus/phosphonate uptake, poly-P/phosphonate accumulation, biofilm formation and multiple stress defense and resistance. For all the PAOs with draft genomes available (Martín et al.2006; Kristiansen et al.2013; Mao et al.2014; Skennerton et al.2015; Mao et al.2016), we found that the glycolytic EMP pathway, TCA cycle, transport of orthophosphate and polyphosphate metabolism existed in all Candidatus Accumulibacter and four members of the genus Tetrasphaera; however, PHA transformation routes were absent for the Tetrasphaera isolates. Unlike our two SADPR bacteria, all the Tetrasphaera isolates and Accumulibacter clade IIC strains possessed the potential to use nitrate/nitrite as the terminal electron acceptors in the anaerobic denitrification process, and the strains SK-11, SK-12 and BA-94 of clade IIF and HKU-1 of clade IB all lacked the key nitric oxide reductase (nor operons) in their draft genomes. Moreover, the oxidative phosphorylation pathways of all PAOs were investigated through KEGG and NCBI database searches, for which there have been very few previous reports in the literature. Compared with all major components of the electron transfer chain found on GAD3/LAD9, four Tetrasphaera PAOs did not encode any subunits of cyt bc₁ complex, cbb₃-type and cyt o ubiquinol oxidase, and cyt bd-type quinol oxidase deficiency was observed for most Accumulibacter species except BA-93. Cyt bd can play an important role in protection against different types of oxidative stress (Lu et al.2015). Based on the above, the physiology of our SADPR bacteria is different from that displayed by most available PAOs, and reveals GAD3/LAD9 to be more versatile organisms. This is greatly helpful in overcoming the limitation of conventional N and P removal systems such as EBPR where denitrifiers and PAOs commonly compete for the organic carbon sources.

In the past 30 years, a mass of bacterial strains, mainly belonging to Proteobacteria, have been found and characterized with the huge capability for aerobic denitrification, and relatively intense applications for heterotrophic nitrogen removal have been substantially conducted in sequencing batch reactors, continuous wastewater treatment and membrane processes (Ji et al.2015). These two SADPR bacteria were named as PCN bacteria before, owing to the co-metabolism of chemical oxygen demand and nutrients within a high-stability system. Bioaugmentation treatment of municipal wastewater with the consortium including our isolates GAD3 and LAD9 was realized in an aerobic pilot-scale sequencing batch reactor, exhibiting stable carbon, total nitrogen and phosphorus removal efficiency to meet the first class level of China National Municipal Wastewater Discharge Standards (Chen et al.2015). The on-time addition of these PCN bacteria showed excellent accommodation and compatibility with the indigenous microorganisms, and the further dynamic of metabolic interactions among them should be revealed. On the basis of the observed set of genes, it is likely that these strains will display excellent SADPR performance in the multiple carbon source-, N- and P-rich industrial and municipal wastewater treatment systems under a normal or low temperature, with novel operation conditions and process design needed. For the other aspect, genome-based analysis has revealed the close metabolic interactions between Anammox bacteria and heterotrophic partial or full denitrifying bacteria for enhanced nitrogen removal performance, and the niche differentiation of aerobic and anaerobic communities within Anammox granules (Speth et al.2016). Even though the complete denitrifying organisms may potentially be detrimental to Anammox bacteria as they compete with Anammox bacteria in the utilization of nitrite, we still speculate that a possible combination of SADPR bacteria with the Anammox process in an appropriate method. If this is the case, it will prompt a possible re-evaluation of the limited knowledge of heterotrophic biological nitrogen removal processes, upgrading and retrofitting the conventional systems in existence.

SUPPLEMENTARY DATA

Supplementary data are available at FEMSEC online.

Acknowledgements

This work was performed using the equipment of Department of Environmental Engineering from Peking University. We thank Beijing Novogene Bioinformatics Technology Co. Ltd for providing genome sequencing and bioinformatics support. We also thank the editor and anonymous reviewers for their comments, which helps to improve the manuscript.

FUNDING

This work was supported by National Natural Science Foundation of China [Grant No. 51539001, 51208007] and Yellow River Institute of Hydraulic Research [Grant No. HKY-JBYW-2016-01].

Conflict of interest. None declared.