1. Introduction
Phosphorus (P) eutrophication and P mineral resource depletion are increasingly serious global problems. Wastewater treatment plants (WWTPs) are both the sink and source for P [
1], [
2]. The bioprocess known as enhanced biological phosphorus removal (EBPR) can effectively promote P removal and recovery in WWTPs. It relies on a specific microbial functional guild called polyphosphate-accumulating organisms (PAOs) [
3], [
4]. Until now, most studies of EBPR microbiology have focused on
Candidatus (
Ca.) Accumulibacter PAOs, which are commonly found in full-scale WWTPs and are widely enriched in lab-scale reactors using volatile fatty acids as carbon sources [
4]. Because of the complex wastewater compositions and luxuriant growth conditions, full-scale WWTPs can provide diverse nutritional niches to accommodate various functional guilds [
5], [
6]. PAOs other than
Ca. Accumulibacter (e.g.,
Tetrasphaera and Dechloromonas) are increasingly found to play an important but previously underappreciated role in full-scale systems [
4], [
7], [
8], which encourages continued and follow-up efforts to explore the emerging aspects of the microbiology and ecology of the EBPR process.
Tetrasphaera PAOs have gained increasing attention due to their prevalence in global WWTPs as both important contributors to P removal and potential reservoirs for P recovery [
4], [
7]. For example,
Tetrasphaera were found to be the most abundant PAOs in 18 Danish full-scale EBPR WWTPs [
7] and were further revealed by Raman-fluorescence
in situ hybridization (FISH) analysis to have a comparable or even higher P removal contribution than
Ca. Accumulibacter PAOs [
9]. In addition,
Tetrasphaera PAOs are implicated in metabolizing various carbon sources (e.g., glucose and amino acids), which may favor their growth in WWTPs with fluctuating environmental conditions [
10]. Although
Tetrasphaera PAOs play an important role in the EBPR of full-scale WWTPs, how to selectively enrich and regulate them and how to enhance biological P removal (and thus recovery) remain elusive questions. Currently,
Tetrasphaera PAOs, unlike the most intensively investigated
Ca. Accumulibacter, are still at an infancy stage of research. Our knowledge of the phylogeny, ecophysiology, and metabolism of
Tetrasphaera PAOs, especially their capacity for wastewater P uptake and removal, is limited [
4], [
9], [
10]. The successful isolation of six
Tetrasphaera strains partially filled these knowledge gaps, but the physiologies and metabolic characteristics of these isolates in pure culture cannot explain the interactive behaviors of
Tetrasphaera PAOs in WWTPs [
11]. In addition, the high diversity and relative abundance of
Tetrasphaera PAOs in EBPR plants generally imply the existence of yet-to-be-discovered PAOs. Therefore, it is essential to explore the enrichment method and microbial interactions of
Tetrasphaera PAOs in the context of the EBPR microbiome.
Until now, only a few studies have reported
Tetrasphaera-enriched microbiomes that were fed amino acids as a sole carbon source [
10], [
12], [
13], [
14]. The enrichment strategy, process, and key biotic and abiotic factors are unclear, hindering the in-depth study of the ecophysiology and metabolism of
Tetrasphaera PAOs. Notably, allylthiourea (C
4H
8N
2S), a specific nitrification inhibitor, is empirically supplemented to facilitate the enrichment of model PAOs (mainly
Ca. Accumulibacter) in lab-scale EBPR reactors by suppressing nitrite production by ammonia-oxidizing bacteria (AOB) to alleviate adverse competition from denitrifying bacteria [
15], [
16]. Because
Tetrasphaera PAOs possess versatile metabolism, including the denitrifying capability, the effect of allylthiourea on enriching
Tetrasphaera PAOs needs to be explored. Recently, a top-down approach has been proposed that uses intricately designed operation conditions (e.g., substrate composition, loading rate, and redox conditions) to manipulate an existing microbiome through ecological selection to perform the desired biological processes [
17]. While this new methodological framework provides an ideal theoretical guide for the establishment of the
Tetrasphaera-enriched microbiome, a practical strategy is lacking and needs to be established. Additionally, understanding of the microbiome dynamics and treatment performance during the
Tetrasphaera-enriched microbiome establishment process is lacking, especially regarding the influence of biotic and abiotic factors and the link to EBPR performance (i.e., P removal and recovery). This knowledge is the key to the reproducible establishment and precise regulation of a
Tetrasphaera-enriched microbiome to improve EBPR performance.
Given the abovementioned knowledge gaps, our study aims to ① establish and explore a top-down strategy for Tetrasphaera-enriched microbiome establishment and characterize its P removal and recovery potentials, ② investigate the microbiome temporal dynamics and interaction patterns of the Tetrasphaera-enriched microbiome establishment process, and ③ quantify the contribution of biotic and abiotic factors to microbiome dynamics and treatment performance. In addition, the nexus between microbial composition, environmental (e.g., operational) conditions, and treatment performance of the EBPR microbiome as well as the effect of allylthiourea on the enrichment of Tetrasphaera PAOs and EBPR performance have been investigated. To the best of our knowledge, this is the first proof-of-concept trial study to investigate the approaches and mechanisms of the establishment of the Tetrasphaera-enriched microbiome from its strategy design to the potential evaluation of P recovery.
2. Materials and methods
2.1. Bioreactor setup and operation
To enrich Tetrasphaera PAOs, a lab-scale sequencing batch reactor with a working volume of 10 L was established and inoculated with activated sludge collected from a local municipal WWTP in Hangzhou, China. The WWTP employs the anaerobic-anoxic-oxic (A/A/O) process with AlCl3 coagulating sedimentation in the primary setting tank, treating 50 000 m3·d−1 of domestic wastewater. The 8 h cycle of reactor operation consisted of four phases: 150 min of anaerobic treatment, 320 min of aerobic treatment, 5 min of settling, and 5 min of water discharge and feeding. Operation details are provided in Text S1 in Appendix A.
2.2. Synthetic wastewater preparation and enrichment stages
The synthetic wastewater was prepared by mixing different volumes of stock solutions (solutions A, B, and C, and trace element solution; Table S1 in Appendix A) to maintain a total organic carbon (TOC) concentration of (394.8 ± 34.7) mg·L
−1, PO
43−-P concentration of (24.8 ± 3.9) mg·L
−1, and NH
4+-N concentration of (113.6 ± 36.9) mg·L
−1. The characteristics of the prepared synthetic wastewater are similar to those of high-concentration wastewater (e.g., livestock wastewater [
18]). Specifically, solution A (multicarbon sources) contained amicase (CAS No. 65072-00-6; Sigma-Aldrich, USA), glucose (CAS No. 50-99-7; HuShi, China), and sodium acetate (CAS No. 127-09-3; HuShi). Solution B (P sources) contained K
2HPO
4 and KH
2PO
4. Solution C (mineral medium) contained NH
4Cl, MgSO
4·7H
2O, and CaCl
2·2H
2O. The trace element solution was prepared according to a previous study [
15] and added to synthetic wastewater (1 mg·L
−1). To investigate the effect of allylthiourea on
Tetrasphaera PAOs enrichment and EBPR function, the entire reactor operation was divided into the following four stages with different dosages of allylthiourea:
Stage I (days 0-73): In the natural enrichment phase, no allylthiourea (0 mg·L−1) was supplemented as a nitrification inhibitor.
Stage II (days 74-120): In the reinforcement enrichment phase, allylthiourea was supplied at 1 mg·L−1 to enhance the Tetrasphaera-related PAOs enrichment by partially inhibiting nitrification activity.
Stage III (days 121-127): In the transient shock phase, the allylthiourea was added to 5 mg·L−1 to strongly inhibit or suppress nitrification activity.
Stage IV (days 128-170): In the recovery phase, the allylthiourea concentration was adjusted back to 1 mg·L−1.
2.3. Batch experiments
Three batch tests were performed following experimental conditions similar to the mother reactor. Each test was seeded with the biomass collected from the EBPR reactor on day 143 when the relative abundance of Tetrasphaera sequences reached 30.3%. The composition of synthetic wastewater was consistent with Section 2.2 but only amended with three dosage levels of allylthiourea: 0, 1, and 5 mg·L−1. The pH and dissolved oxygen (DO) concentration were monitored using a portable multiparameter analyzer (HQ40d; Hach, USA). Seven samples were taken at three time points of the anaerobic phase (13, 45, and 135 min) and four time points of the aerobic phase (30, 90, 210, and 320 min). The sludge samples were centrifuged at 3000 r·min−1 for 3 min, and then the supernatant was collected for further physicochemical analysis.
2.4. Chemical analysis
The collected water samples were pretreated using 0.22 μm Millex GP syringe-driven filters to eliminate the potential solids before analysis. PO
43−-P, TN, and NH
4+-N were analyzed using a segmented flow analyzer (San
++; SKALAR, the Netherlands). The TOC was assessed by a TOC-L total organic carbon analyzer (Shimadzu, Japan). Total solid (TS) and volatile solid (VS) were analyzed according to standard methods [
19]. To evaluate the P recovery potential of the
Tetrasphaera-enriched microbiome, the P weight percentage content in sludge ash was assessed and compared with the WWTP. Briefly, the sludge samples obtained from the
Tetrasphaera-enriched reactor and local WWTP (the sampling site was the same as seeding sludge) were adjusted to the same TS content using pure water. Samples with a 50 mL volume were taken and put into a crucible for incineration (600 °C, 8 h). Then, the ash was collected and mixed using a mortar. The P weight percentage content in the sludge ash was assessed by a scanning electron microscope equipped with an energy dispersive spectrometer (SEM-EDS; Hitachi, Japan). Every experiment was conducted with the triple treatments.
2.5. FISH analysis of polyphosphate-accumulating organisms
FISH analysis was performed to investigate the spatial organization of targeted PAOs in microbiome samples according to a previous protocol with some modifications [
20]. The detailed experimental flow and the information of the FISH probes can be found in Text S2 and Table S2 in Appendix A. FISH samples were observed using a laser scanning confocal microscope (LSM800; Zeiss, Germany).
2.6. DNA extraction, PCR amplification, and sequencing
Sludge samples were collected regularly at the end of the aerobic phase for DNA extraction. A total of 16 samples were collected over 170 days. DNA extraction was performed in duplicate from 1.5 mL collected reactor biomass using the Fast DNA SPIN Kit for Soil (MP Biomedicals, USA). The duplicate DNA extracts were mixed equally for bacterial 16S ribosomal RNA (rRNA) gene high-throughput sequencing targeting the V1-V3 hypervariable regions (27F: 5′-AGAGTTTGATCCTGGCTCAG-3′ and 534R: 5′-ATTACCGCGGCTGCTGG-3′) [
21] using Illumina’s MiSeq platform with a paired-end (2 × 300) sequencing strategy at Guangdong Magigene Biotechnology Co., Ltd., China. The thermocycler (Bio-Rad S1000; Bio-Rad Laboratory, USA) settings for polymerase chain reaction (PCR) were initial denaturation at 95 °C for 2 min, 30 cycles of 95 °C for 20 s, 56 °C for 30 s, and 72 °C for 60 s, and final elongation at 72 °C for 5 min [
21].
2.7. Bioinformatics analysis
Fastp was used to check the quality of the raw data with the sliding window parameter of -W 4 -M 20. The primers were removed by using cutadapt software (Germany) according to the primer information at the beginning and end of the sequence [
22]. The obtained clean data were merged using QIIME2 (v2020.6; USA) pipeline [
23] with the DADA2 algorithm [
24]. Briefly, the following steps were performed: ① filter out noisy sequences, ② correct errors in marginal sequences, ③ remove chimeras sequence and singletons, ④ join denoised paired-end reads, and ⑤ conduct sequence dereplication using a command of qiime dada2 denoise-paired with the parameter of --p-trunc-len-f 250 and --p-trunc-len-r 247 to maintain at least 12 bp overlap. The amplicon sequence variant (ASV) table was generated based on 100% sequence similarity for 16S rRNA genes, which contained 1389 ASVs in total. To exclude spurious sequences from the downstream analysis, the rare feature (i.e., ASV) with a total sequence count of less than three was removed using a command of qiime feature-table filter-features with the parameter of --p-min-frequency 3. To determine if the richness of the samples was fully sequenced, alpha rarefaction was analyzed using a command of qiime diversity alpha-rarefaction and visualized by the qiime 2 view website
†. A specific taxonomic classifier was trained based on the Silva_138.1_SSURef_NR99 database [
25] and used for taxonomic analysis.
2.8. Phylogenetic analysis
The most abundant ASVs assigned to
Tetrasphaera were selected to construct the phylogenetic tree using MEGA X [
26]. Multiple sequence alignment was conducted by MUSCLE [
27] with the default parameter. The evolutionary history was inferred by the maximum likelihood method and Kimura two-parameter model. Two Nitrosomonas europaea sequences were used as the outgroup.
2.9. Statistical and network analysis
The statistical analyses, including redundancy analysis (RDA), principal coordinate analysis (PCoA), Mantel test, and PERMANOVA test, were performed using the R package vegan [
28]. Co-occurrence networks were constructed using the ‘Co-occurrence_network.R’ script of package MbioAssy1.0
‡ [
6], [
29]. Detailed information on the co-occurrence network analysis is provided in Text S3 in Appendix A. The network was visualized in Gephi 0.9.2
††.
3. Results and discussion
3.1. The establishment strategy and performance of the Tetrasphaera-enriched microbiome
In this study, we established a strategy including four stages to selectively enrich
Tetrasphaera PAOs based on the top-down design approach [
17]. In stage I, the multicarbon sources (containing amicase, glucose, and sodium acetate) and periodic operational conditions (anaerobic/aerobic alternations) were designed as the selective tactics to conduct the initial enrichment. In stage II, 1 mg·L
−1 allylthiourea was introduced to facilitate enrichment by suppressing nitrification. In the following stages (stages III and IV), a transient shock with a high dosage of allylthiourea (5 mg·L
−1) was applied to investigate the effect of allylthiourea dosage on the
Tetrasphaera-enriched microbiome.
The treatment performance of nutrients (i.e., NH
4+-N, total nitrogen (TN), and PO
43−-P) and organic carbon (i.e., TOC) in wastewater were followed over nearly six months (
Fig. 1). In stage I, the reactor showed an overall decline in P removal efficiency (average 35%), but NH
4+-N removal increased with prolonged operation (the highest NH
4+-N removal was up to 92% on day 49). With the addition of allylthiourea (stage II), the efficiency of P removal significantly increased from 27.6% (day 73) to 86.2% (day 94) and then remained stable between 78.1% and 87.4% during days 105-121. The NH
4+-N removal decreased significantly during days 73-90 (the lowest percentage was 3.9% on day 87) and then recovered during days 105-121 (the average removal percentage between 89.7% and 98.8%). When the concentration of allylthiourea was increased to 5 mg·L
−1 in stage III (transient shock to suppress nitrification), the efficiency of P removal presented a slightly decreasing trend (from 76.8% of day 122 to 58.8% of day 125,
Fig. 1(a)), accompanying an obvious decrease and a final failure in NH
4+-N removal. With the downregulation of the allylthiourea level back to 1 mg·L
−1 on day 128 (stage IV), the efficiency of P removal was recovered to the original level (67.4% to 89.3%), and the efficiency of TN removal was stabilized at approximately 80% on day 148-170. In addition, the utilization and degradation of organic carbon during the 170-day operation remained stable, and approximately 95% of the carbon source could be absorbed and utilized (
Fig. 1(b)).
In stage I, the noticeable increase in NH
4+-N removal and general decrease in P removal (between day 23 and day 46, see
Fig. 1(c) against
Fig. 1(a)) suggested an overall negative relationship between nitrification and the P removal process. Considering the high TN concentration in the effluent (
Fig. 1(d)), the nitrification products (i.e., NO
2− and NO
3−) may be accumulated, which could trigger the competition of denitrification for the organic carbon source. Therefore, the nitrification process should be controlled to promote PAO enrichment and enhance P removal. High P removal (85%) was achieved in only 17 days upon adding 1 mg·L
−1 allylthiourea in stage II, indicating that the growth and activity of PAOs were enhanced with the regulation of nitrification inhibitor. Although the possibility could not be excluded that the increase in N removal in the late phase of stage II could be partially attributed to the increase in the TOC/TN ratio (from 1.31 to 2.84); in stage IV, the TOC/TN ratio (2.90 ± 0.3) was stable, and both TN and NH
4+-N removal increased after the transient shock and were maintained in the late phase. These results imply that the nitrification process of the microbial community recovered, contributing to the increase in nitrogen (N) removal. After the long-term selection pressure of low dosage allylthiourea, the composition of the nitrifying bacteria in the EBPR microbiome could be shifted, which agreed with a previous study in that allylthiourea had an obvious effect on the structure of the nitrifying bacterial community [
30].
3.2. The effect of allylthiourea on EBPR performance
To further investigate the effect of allylthiourea on the EBPR microbiome, batch experiments were conducted using 143-day microbiome biomass under different allylthiourea dosages (0, 1, and 5 mg·L
−1). Based on 16S rRNA gene amplicon sequencing analysis, the relative abundance of
Tetrasphaera PAOs in the sampled 143-day EBPR microbiome was 30.3%. In addition, the corresponding removal efficiencies of PO
43−-P, TOC, NH
4+-N, and TN in the mother EBPR reactor at 143 days were 91.9%, 97.9%, 92.3%, and 64.4%, respectively (
Fig. 1). The P uptake rate with 1 mg·L
−1 allylthiourea addition was highest (0.07 mg·L
−1·min
−1), followed by 0 and 5 mg·L
−1 (0.04 and 0.05 mg·L
−1·min
−1, respectively) (
Fig. 2(a)). In addition, the P removal efficiency in the 5 mg·L
−1 group was 67.9%, which was the lowest among the three groups (
Fig. 2(b)). The NH
4+-N profiles of the three groups were different, especially in the aerobic phase. NH
4+-N removal did not occur under 5 mg·L
−1 allylthiourea, while a remarkable decline in NH
4+-N concentration was observed with 0 and 1 mg·L
−1 allylthiourea. For example, the NH
4+-N concentration decreased from 24.64 to 1.48 mg·L
−1 when 1 mg·L
−1 allylthiourea was added (
Fig. 2(c)). In addition, the dynamics of DO showed no visible difference in the DO concentration among the three experimental groups in the anaerobic phase (< 0.2 mg·L
−1), while in the following aerobic phase, the DO concentration increased to 5.21 mg·L
−1 at 53 min when exposed to 5 mg·L
−1 allylthiourea (
Fig. 2(d)). In comparison, the DO concentrations were 1.09 and 1.04 mg·L
−1 with allylthiourea concentrations of 0 and 1 mg·L
−1, respectively.
With the dosages of 0 and 1 mg·L
−1 allylthiourea, the enriched microbial community achieved stable N and P removal. In contrast, the high allylthiourea dosage (5 mg·L
−1) showed a negative effect on the
Tetrasphaera-enriched microbiome, consistent with the impact of allylthiourea addition on the nutrient removal performance previously observed in the EBPR reactor at stage III. The increased allylthiourea dosage (5 mg·L
−1) inhibited the nitrification process, which could cause ammonia accumulation and disturbance in the microbial community [
31]. In addition, when nitrification was completely inhibited, the production of NO
3− and NO
2− decreased, which may affect the activity and function of denitrifying PAOs. A previous study reported that although allylthiourea is soluble in water, it can be adsorbed by soil [
30], indicating that allylthiourea may also be distributed in the sludge in addition to the aqueous phase. In addition, allylthiourea can be degraded by the microorganism; for example, a previous study indicated that 20 mg·L
−1 allylthiourea was degraded to 0.5-1.0 mg·L
−1 by activated sludge under aerobic conditions in four days [
32], implying its low-security risk when applied for short-term biostimulation of wastewater treatment systems.
3.3. FISH analysis of the Tetrasphaera-enriched microbiome and evaluation of P recovery potential
The
Tetrasphaera-enriched microbiome on day 143 was visualized by FISH. The targeted
Ca. Accumulibacter and
Tetrasphaera PAOs were visually detected in both the activated sludge sample collected from the same WWTP as the seeding sludge (on December 2, 2020;
Fig. 3(a)) and the
Tetrasphaera-enriched microbiome (
Fig. 3(b)) in our EBPR reactor. Overall,
Tetrasphaera PAOs were morphologically dominant in the EBPR microbiome, indicating that they were successfully enriched by the top-down design approach employed. Furthermore, the P storage ability and recovery potential of the
Tetrasphaera-enriched microbiome were evaluated by determining the P content in the sludge ash using SEM-EDS. The P content (counted as the weight percentage) in
Tetrasphaera-enriched sludge ash and secondary sink sludge ash was 32.5% and 1.4%, respectively. Therefore, the percentage of P in
Tetrasphaera-enriched microbiome sludge ash dramatically increased by 23.2 times compared with the normal full-scale WWTP sludge, indicating that the P-storage ability of the
Tetrasphaera-enriched microbiome was largely enhanced. A strong signal intensity of Na and Mg elements was detected in the
Tetrasphaera-enriched sludge, while the Al and Si signal intensities were higher in WWTP sludge (
Fig. 3(c)). These results corresponded with the surface morphology of these two kinds of sludge (
Figs. 3(a) and (b)). The WWTP sludge ash showed a red color with a loose structure because the WWTP influent contained sand, and the WWTP used Al-based coagulants to aid nutrient removal from wastewater. In comparison, the
Tetrasphaera-enriched sludge ash showed a white color and compact morphology, which resulted from the designed operation process and influent compositions (e.g., MgSO
4 and CaCl
2). Theoretically,
Tetrasphaera-enriched sludge ash is attractive for further use in white P recovery and is promising to replace the depleting phosphate rock, in which Ca
3(PO
4)
2 is the main component [
33]. Therefore, three essential conditions to promote P recovery from wastewater are achieved: ① high P removal efficiency, ② microbiome with enhanced P storage ability, and ③ P-rich ashes and compounds for further value-added P recovery. The establishment of the
Tetrasphaera-enriched microbiome holds promise for exploiting
Tetrasphaera microbiology and EBPR biotechnology for the targeted regulation and promotion of biological P recovery in full-scale WWTPs.
3.4. The microbiome composition and dynamics during the establishment process of the Tetrasphaera-enriched microbiome
The microbiome composition and temporal dynamics during the enrichment process of
Tetrasphaera PAOs were investigated based on high-throughput 16S rRNA gene amplicon sequencing (
Fig. 4). In general, the microbiome was dominated by
Tetrasphaera PAOs in coexistence with
Microlunatus PAOs and glycogen accumulating organisms (GAOs) (i.e.,
Ca. Competibacter). In seed sludge,
Tetrasphaera PAOs,
Microlunatus PAOs, Brevundimonas, and
Ca. Competibacter GAOs showed relative abundance of 1.5%, 0.1%, 0.3%, and 12.4%, respectively. For the initial establishment of the EBPR microbiome, their relative abundance on day 73 shifted to 13.5%, 2.3%, 15.1%, and 0.7%, respectively. In addition, the relative abundance of
Tetrasphaera PAOs reached approximately 40% on day 113, approximately equivalent to a cell or biovolume percentage of 62% to 67% by assuming an average 16S rRNA gene copy number of 1.0 for
Tetrasphaera and between 2.5 and 3.0 for activated sludge bacteria [
34]. This result implied that allylthiourea promoted significant enrichment of
Tetrasphaera PAOs (
P value = 0.017, Kruskal-Wallis test, Fig. S1(a) in Appendix A). Notably, the
Tetrasphaera PAOs were the most abundant 16S rRNA gene ASV in the EBPR microbiome (i.e., EBPR-ASV0001). Phylogenetic analysis showed that
Tetrasphaera sp
. EBPR-ASV0001 was phylogenetically different from the six previously reported isolates (
Fig. 4(b) and Table S3 in Appendix A). The global similarity of 16S rRNA gene sequences between EBPR-ASV0001 and
Tetrasphaera japonica (
T. japonica) was 96.8%, indicating that EBPR-ASV0001 is most likely a denitrifying PAO similar to
T. japonica.
The relative abundance fluctuations of
Ca. Competibacter GAOs,
Brevundimonas, and
Tetrasphaera PAOs during stage I (days 0-73;
Fig. 4(a)) were consistent with unstable biological phosphorus and nitrogen removal. Specifically,
Ca. Competibacter GAOs decreased significantly in the first 33 days (
P value = 0.05, Kruskal-Wallis test), while
Brevundimonas and
Paracoccus were markedly increased during days 49-73 (
P value = 0.05, Kruskal-Wallis test) (Figs. S1(b)-(d) in Appendix A). In addition, the decrease in the relative abundance of
Tetrasphaera PAOs and the P removal efficiency coincided with the prosperity of
Brevundimonas and
Paracoccus. Previous reports indicated that the occurrence of the denitrification process in the anaerobic phase could cause EBPR failure because some denitrifiers could compete for organics with PAOs [
16].
Brevundimonas and
Paracoccus are well known not only as denitrifiers but also as heterotrophic nitrification and aerobic denitrification (HN-AD) bacteria, which can simultaneously perform nitrification and denitrification under aerobic conditions [
35], [
36], [
37]. Therefore, their relative abundance increase could lead to potential competition for organic molecules and/other resources with
Tetrasphaera PAOs, especially in full-scale WWTPs where the dissolved oxygen is heterogeneous.
With the addition of 1 mg·L
−1 allylthiourea, the following gradual increase in the relative abundance of
Tetrasphaera PAOs (13.0% to 37.7%) was coupled with the sharp decrease in the relative abundance of Nitrosomonas AOBs (0.35% to 0.02%) as well as the denitrifiers
Brevundimonas (19.9% to 6.9%) and
Paracoccus (5.8% to 1.1%) from day 66 (stage I) to day 113 (stage II). This result can be explained by the chain effects associated with the allylthiourea inhibition of nitrifying
Nitrosomonas, that is, the inhibition of autotrophic ammonia oxidation largely reduced denitrifying substrates available to the dentritrifers, thus greatly alleviating their competition against PAOs for organic resources (
Fig. 4(a)). Interestingly, the ammonia oxidizer
Nitrosomonas decreased by 13.7 times from day 73 to day 113 and disappeared on day 121, while high ammonia removal was still observed (e.g., 98.7% on day 121;
Fig. 1(c)), implying the existence of an alternative nitrification process. We speculated that some heterotrophic nitrifiers should participate in the nitrification process when autotrophic nitrifiers (e.g.,
Nitrosomonas) are evidently inhibited under a low dosage of allylthiourea (i.e., 1 mg·L
−1), facilitating the nitrogen removal. In stage III, when the transient shock of 5 mg·L
−1 allylthiourea was applied, PAOs fluctuated in their relative abundance, which first decreased to 12.6% on day 147 and then increased to 28.2% on day 157. We hypothesized that a high allylthiourea dosage would adversely affect the proliferation and function of denitrifying PAOs that can couple nitrate and/or nitrite reduction to P uptake due to the heavy suppression of nitrification. Supporting this hypothesis, ammonia removal was completely suppressed in stage III, accompanying a dramatic decline in the removal efficiency of TN and PO
43−-P to 19.5% and 58.4%, respectively (
Fig. 1(a)). Furthermore, a high allylthiourea dosage restrained the propagation of the most abundant PAOs,
Tetrasphaera sp. EBPR-ASV0001, which showed a marked decline on day 142 (22.64%,
Fig. 4(b)). Further genome-based study and laboratory isolation of
Tetrasphaera sp. EBPR-ASV0001 to reveal its metabolic pathways, physiology, and application potential in full-scale EBPR WWTPs are warranted.
3.5. The microbial co-occurrence patterns during the establishment process of the Tetrasphaera-enriched microbiome
To predict microbial interactions in the
Tetrasphaera-enriched microbiome, microbial co-occurrence patterns over the 170-day reactor operation were analyzed based on network analysis (
Fig. 5). The positive bacterial association network had 85 nodes (ASVs) and 167 edges (correlations, Table S4 in Appendix A), which were topologically partitioned into nine modules presumably representing discrete and interactive ecological niches over time (
Fig. 5(a)). The topological properties of the observed network were calculated and compared with those of identically sized Erdös-Réyni random networks (Table S4). The clustering coefficient (CC) and modularity index of the observed
Tetrasphaera-enriched co-occurrence network were 0.449 and 0.687 and higher than those of the corresponding Erdös-Réyni random networks (0.046 and 0.418), suggesting that the microbiome assembly process is nonrandom (or deterministic). Moreover, the network of the
Tetrasphaera-enriched microbiome showed small-world properties with a coefficient (
, where
σ is small-coefficient and APL is average path length) [
38] of 8.97, suggesting that the enrichment strategy adjusts the overall microbial interconnections and interior interaction relationship.
To link ecological niches of co-occurring taxa to their functionality, each network node was taxonomically assigned to three functional groups (
Fig. 5(b)), according to previous reports [
7], [
29], [
36], [
39], including PAOs, GAOs, and nitrogen-removing bacteria (e.g., denitrifiers and HN-AD bacteria). Overall,
Tetrasphaera was distributed in four modules of the network (i.e., modules II, IV, V, and VIII), implying niche differences between members of the genus
Tetrasphaera. This observation supported the previous conclusion [
4], [
10], [
12], [
13], [
14], namely, the different species of
Tetrasphaera could have a specific metabolic preference; therefore, the genus
Tetrasphaera was extremely versatile and could widely interact with coexisting bacteria in EBPR. In particular,
Tetrasphaera PAOs significantly co-occurred with
Microlunatus PAOs in module II (Spearman’s
ρ = 0.906, false discovery rate (FDR) adjusted
P < 0.01;
Fig. 5), indicating their potential cooperative relationships. In addition, known denitrifiers (e.g., Hyphomicrobium,
Bosea, and Xanthobacter) occurred with GAOs (e.g.,
Ca. Competibacter,
Kineosphaera, and
Micropruina) in modules I, II, and IV, implying a shared niche preference, such as taking up organic matter for growth. The wide within-module co-occurrence between denitrifiers and organic-degrading bacteria (e.g.,
Hyphomicrobium and
Shinella in module I,
Bosea and
TM7a in module II, and
Xanthobacter and
Amaricoccus in module IV) supports the classical presumption of their cooperative interactions, such as cross-feeding, that is, the former provides small organic molecules for the latter as a carbon source.
The microbial interaction and niche patterns revealed in the co-occurrence network of the
Tetrasphaera-enriched microbiome (
Fig. 5) were associated with the dynamics of the removal efficiency of nutrients (Figs. 1(a), (c) and (d)) rather than organic carbon (
Fig. 1(b)). First, the wide co-occurrence and prevalence of denitrifiers and organic-degrading bacteria in the same modules, together with the near-perfect TOC removal (95.0% ± 1.2%) and mostly favorable TN removal (77.0% ± 2.9%) when no inhibition occurred), implies that chemoorganotrophic respiration and heterotrophic denitrification could be active during the microbiome assembly process. Moreover,
Micropruina was a fermentative GAO [
4], and its positive correlations with
Brevundimonas in module V (Spearman’s
ρ = 0.816) and
Paracoccus in module IV (Spearman’s
ρ = 0.818) indicated the fermentative bacteria may cross-feed co-occurring HN-AD bacteria with fermentation products. Likewise, a positive correlation between
Tetrasphaera PAOs and
Bosea denitrifiers (Spearman’s
ρ = 0.910,
Fig. 5) suggests that denitrifying phosphorus removal should have occurred in the reactor, promoting simultaneous biological N and P removal. A prior genomic survey indicated that
Tetrasphaera can utilize nitrite for poly-P production [
11], which could explain the abovementioned enrichment scenarios of
Tetrasphaera PAOs facilitated by cross-feeding partnerships with nitrate denitrifiers. Therefore, the enrichment strategy with intricately designed multicarbon sources and low-dose allylthiourea creates microbial niches and fosters diverse cooperative interactions that affect the microbiome composition and establishment.
3.6. The nexus between microbiome, environment, and EBPR performance
In addition to microbiome dynamics and co-occurrence patterns, the roles of abiotic factors (i.e., environmental condition and influent composition) and biotic factors (i.e., microbiome composition and interaction) in driving
Tetrasphaera-enriched microbiome establishment were explored based on RDA and PCoA analyses. The microbial communities shifted away upon reactor startup and quickly assembled with the continuous feedings of synthetic wastewater containing multicarbon sources (stage I in
Fig. 6; and Fig. S2 in Appendix A). When the nitrification inhibitor (i.e., allylthiourea) was introduced, the microbial community was significantly selected and restructured (stages II-IV in
Fig. 6 and Table S5 in Appendix A). The nexus between the environmental condition and microbial community was visualized with the first two RDA axes (
Fig. 6(a)), which jointly explained 52.2% of the total microbiome variances and indicated that the designed environmental parameters, especially the inhibitor (adjusted
R2 = 0.23,
P value < 0.01; Table S6 in Appendix A), were among the major drivers of the microbiome assembly process. In addition, the
Tetrasphaera PAOs and
Microlunatus PAOs positively correlated with allylthiourea and PO
43−-P, indicating that
Tetrasphaera PAOs and
Microlunatus PAOs have growth advantages under selective conditions. The association analysis between the microbial community and EBPR performance showed that
Tetrasphaera PAOs,
Microlunatus PAOs, and
Saccharimonadales positively correlated with PO
43−-P, TN, and TOC removal. Meanwhile,
Brevundimonas and
Kineosphaera were positively correlated with NH
4+-N removal (
Fig. 6(b)). Consistent with the RDA results, Mantel test analysis revealed that allylthiourea significantly affected the microbiome dynamics and PO
43−-P removal (
P value < 0.005; Table S7 in Appendix A).
Overall, the correlation nexus between microbial (M) community, environmental (E) variables, and EBPR performance (P) (defined here as the “MEP nexus”) indicated that 1 mg·L
−1 allylthiourea, as a key abiotic factor, promoted the selection and proliferation of
Tetrasphaera PAOs,
Microlunatus PAOs, and organic degrading bacteria
Saccharimonadales while inhibiting
Nitrosomonas AOBs and indirectly affecting potential competitors such as denitrifying
Brevundimonas. Furthermore, the removal of carbon and nutrients (i.e., TOC, TN, and PO
43−-P) was enhanced. Notably,
Tetrasphaera and
Microlunatus PAOs showed no significant negative correlation in their relative abundance in the whole process (Fig. S3 in Appendix A), and their positive co-occurrence patterns (module II,
Fig. 5(a)) made their coexistence feasible in the
Tetrasphaera-enriched EBPR microbiome. In addition,
Saccharimonadales is reported to be a key degrader of organic molecules (e.g., glucose and sugars) in aerobic sludge systems [
40]. Based on their temporal dynamics,
Microlunatus PAOs and
Saccharimonadales should be the key biotic factors that benefit from the designed multicarbon sources and positively contribute to building a
Tetrasphaera-enriched microbiome and enhancing the reactor performance for organics and P removal. In summary, the MEP nexus indicates that the enrichment strategy designed here has realized the selective enrichment of
Tetrasphaera PAOs as an outcome of co-actions from both abiotic and biotic factors.
4. Conclusions
In this study, an enrichment strategy featuring intricately designed multicarbon sources amended with low dosage allylthiourea was proposed and applied to build a Tetrasphaera-enriched EBPR microbiome. Based on 16S rRNA gene amplicon sequencing analysis, a novel putative Tetrasphaera PAO species, EBPR-ASV0001, was identified and enriched at a relative abundance of 40%. The established microbiome showed enhanced P removal (∼85%) and N removal (∼80%) and a 23.2 times higher P content in the sludge ash than in the normal full-scale WWTP sludge. The addition of low-dosage allylthiourea (e.g., 1 mg·L−1 demonstrated in this study) regulated the community structure and affected the microbial co-occurrence patterns, which can be advised as a short-term biostimulation approach for the quick selection and enrichment of Tetrasphaera PAOs to strengthen biological P removal. The findings of this study expand our knowledge of the temporal dynamics and interaction patterns of a novel putative PAO member of Tetrasphaera. Future studies with metagenomic analysis or full-length 16S rRNA sequencing of the enriched Tetrasphaera species are needed to verify their phylogeny and putative function more reliably as new PAOs. Nonetheless, from an engineering perspective, this outcome provides theoretical guidance for exploiting and optimizing EBPR application for sustainable P removal and recovery coupled with biological nitrogen removal from high-concentration wastewater.
Compliance with ethical guidelines
Hui Wang, Yubo Wang, Guoqing Zhang, Ze Zhao, and Feng Ju declare that they have no conflicts of interest or financial conflicts to disclose.
the Key Research and Development Program of Zhejiang(2022C03075)
National Natural Science Foundation of China(22241603)
Zhejiang Provincial Natural Science Foundation of China(LR22D010001)
PDF
(4474KB)
