Using DNA-based stable isotope probing to reveal novel propionate- and acetate-oxidizing bacteria in propionate-fed mesophilic anaerobic chemostats

Abstract

Propionate is one of the most important intermediates of anaerobic fermentation. Its oxidation performed by syntrophic propionate-oxidizing bacteria coupled with hydrogenotrophic methanogens is considered to be a rate-limiting step for methane production. However, the current understanding of SPOB is limited due to the difficulty of pure culture isolation. In the present study, two anaerobic chemostats fed with propionate as the sole carbon source were operated at different dilution rates (0.05 d⁻¹ and 0.15 d⁻¹). The propionate- and acetate-oxidizing bacteria in the two methanogenic chemostats were investigated combining DNA-stable isotope probing (DNA-SIP) and 16S rRNA gene high-throughput sequencing. The results of DNA-SIP with ¹³C-propionate/acetate suggested that, Smithella, Syntrophobacter, Cryptanaerobacter, and unclassified Rhodospirillaceae may be putative propionate-oxidizing bacteria; unclassified Spirochaetaceae, unclassified Synergistaceae, unclassified Elusimicrobia, Mesotoga, and Gracilibacter may contribute to acetate oxidation; unclassified Syntrophaceae and Syntrophomonas may be butyrate oxidizers. By DNA-SIP, unclassified OTUs with 16S rRNA gene abundance higher than 62% of total Bacteria in the PL chemostat and 38% in the PH chemostat were revealed to be related to the degradation of propionate. These results suggest that a variety of uncultured bacteria contribute to propionate degradation during anaerobic digestion. The functions and metabolic characteristics of these bacteria require further investigation.

Introduction

Anaerobic digestion is a key technology in the development of a future bio-based economy because of the mineralization of organic pollutants in waste/wastewater and simultaneous production of methane as clean energy. Converting organic matter into biogas relies upon an intricate balance of multiple trophic microbial groups, including hydrolytic and fermentative bacteria, syntrophic acetogenic bacteria, and methanogenic archaea¹. Volatile fatty acids (VFAs) are the major intermediate metabolites during anaerobic digestion. In thermodynamics, propionate is more difficult to anaerobically oxidize than other VFAs such as butyrate and valerate. Propionate oxidation is energetically unfavorable with a standard change in Gibbs free energy (ΔG°) of +76 kJ per mol reaction. The propionate oxidation under methanogenic condition can be realized only by syntrophic association of propionate-oxidizing bacteria and hydrogenotrophic methanogens. Therefore, propionate often accumulates when anaerobic digestion becomes unstable². The oxidation of intermediate propionate to acetate and hydrogen is regarded as the rate-limiting step in anaerobic digestion.

Owing to their symbiotic relationships with methanogens, isolating syntrophic propionate-oxidizing bacteria (SPOB) is challenging. Until now, only eleven SPOB species affiliated with the genera Syntrophobacter, Desulfotomaculum, Pelotomaculum, Smithella, and candidatus Desulfonatronobulbus have been isolated^{3,4,5,6,7,8,9,10,11,12}, and a SPOB named Candidatus Syntrophosphaera thermopropionivorans is identified in a thermophilic propionate-degrading enriched culture¹³. The limited knowledge of SPOB leads to a gap in knowledge of syntrophic propionate degradation during anaerobic digestion. DNA-based molecular techniques permit the comprehensive determination of microbial diversity but do not generally reveal the relationship between the taxonomy and function of microorganisms. Stable isotope probing (SIP) provides a link between phylogeny and function, and in recent years, this method has been widely used to identify the functions of uncultured microorganisms in various environments^14,15. Several reports demonstrated that SIP is an effective tool for identifying potential organic acid-degraders in different methanogenic habitats^14,16. Species other than Syntrophomonas have been found to be responsible for syntrophic butyrate-oxidation, such as the genera Tepidanaerobacter, Clostridium, the families Syntrophospora, Syntrophomonadaceae, Syntrophaceae and the phylum Actinobacteria¹⁷. Some species related to genera Thermacetogenium and unclassified Thermoanaerobacteraceae were reported responsible for syntrophic acetate oxidation¹⁸. In addition to the known SPOB genera Syntrophobacter, Pelotomaculum, and Smithella, Geobacter and some other anaerobic bacteria affiliated with family Rhodocyclaceae, class Thermomicrobia, and phyla Acidobacteria and Actinobacteria were labeled by ¹³C₃-propionate, which might utilize propionate or metabolic intermediates of propionate oxidation¹⁴. To the best of our knowledge, no study investigated the SPOB in methanogenic reactor by DNA-SIP up to now. To link the uncultured bacteria in methanogenic reactors with the propionate degradation would be essential for revealing the microbial mechanism of propionate degradation and therefore would contribute to the regulation of anaerobic digestion processes.

In anaerobic reactors treating actual waste/wastewaters, microbial communities are very complex and SPOB abundance is generally low. Therefore, it is difficult to target SPOB using ¹³C₃-propionate. In this study, we established two mesophilic methanogenic chemostats fed with propionate as the sole carbon source. Stable propionate-degrading microbial communities were obtained through long-term cultivation and enrichment, which is conducive to the effective identification of propionate- and acetate-oxidizing bacteria. Considering that dilution rate (the reciprocal of hydraulic retention time) could seriously affect community structure¹⁹, two chemostats were operated at different dilution rates (0.05 and 0.15 d⁻¹). The potential SPOB in these two chemostats were investigated using DNA-SIP combined 16S rRNA gene high-throughput sequencing. Acetate-oxidizing bacteria were investigated simultaneously, as acetate was the intermediate metabolite of propionate oxidation.

Results

Microbial communities in propionate-fed chemostats operated at different dilution rates

The operation performance of each chemostat is displayed in Fig. S1. Both PL (0.05 d⁻¹) and PH (0.15 d⁻¹) chemostats maintained steady states during the 1000-day and 130-day operation periods, respectively. The biogas production of the PL and PH chemostat was stable at 300 and 1800 mL L⁻¹ d⁻¹ during this period, respectively. The methane content in biogas was 65–68% for the PL chemostat and 64–66% for the PH chemostat. The organic acid concentration was below 100 mg L⁻¹ in both chemostats. The average VSS concentration was 0.9 g L⁻¹ for the PL chemostat and 1.0 g L⁻¹ for the PH chemostat. The sludge from each chemostat was used to analyze the respective microbial communities.

The ratio of bacteria to archaea in the PL and PH chemostats was 0.89 and 1.21, respectively. The bacterial community structure in each chemostat was obviously distinct (Fig. 1). The total abundance of the top 20 bacterial genera accounted for 93.36% and 95.59% of total bacteria in PL and PH, respectively (Fig. 1). Of the bacteria in the PL chemostat, 88.70% were assigned to six phyla: Spirochaetes (45.49%), Synergistetes (14.34%), Ignavibacteriae (9.64%), Proteobacteria (8.72%), Bacteroidetes (6.07%), and Firmicutes (4.44%) (Fig. 1a). Of the bacteria in the PH chemostat, most were assigned to the phylum Proteobacteria (47.90%), followed by the phyla Synergistetes (23.64%), Spirochaetes (7.93%), Ignavibacteriae (6.50%), Firmicutes (4.61%), and Bacteroidetes (1.98%) (Fig. 1b). Unclassified Spirochaetaceae_1, unclassified Synergistaceae, unclassified Ignavibacteriales, and Smithella were the dominant genera at the dilution rate of 0.05 d⁻¹ (PL chemostat), accounting for 44.34%, 14.34%, 9.64%, and 7.48% of the total bacterial reads, respectively. These genera were also detected in the PH chemostat, but their levels of abundance differed. Among them, Smithella is known as SPOB⁴. It was the predominant genus in the PH chemostat and accounted for 30.23%. The other pure cultured SPOB genus Syntrophobacter³ was also detected in both chemostats, and accounted for 0.59% and 13.00% in PL and PH, respectively. It is worth noting that 76.60% and 48.13% of total bacteria in PL and PH, respectively, was unable to be categorized to known genera, suggesting that a variety of unknown bacteria propagate in propionate-degrading bacterial communities.

Figure 1

Composition of bacterial community in two mesophilic propionate-fed chemostats. (a) PL chemostat: operated at a dilution of 0.05 d⁻¹; (b) PH chemostat: operated at a dilution of 0.15 d⁻¹. *Unclassified Spirochaetaceae_1 and unclassified Spirochaetaceae_2* were two different genera affiliated to the family Spirochaetaceae.

The archaeal community of the two chemostats was listed in Table S1. In PL, hydrogenotrophic Methanoculleus (92.72%) and acetoclastic Methanothrix (6.18%) were the primary genera. In PH, Methanothrix (45.58%) became the most dominant methanogen, followed by Methanospirillum (26.92%), Methanoculleus (17.65%), and Methanolinea (7.72%). The abundance of acetoclastic methanogens and the diversity of hydrogenotrophic methanogens were higher than that of PL chemostat.

DNA-SIP analysis

Sludges from the PL and PH chemostats were incubated with ¹³C- and ¹²C-substrates for 7 d/14 d and 5 d/10 d, respectively (Table 1). The VFAs consumption and biogas yield in each microcosm were measured. Three biological replicates of each treatment exhibited almost the same biogas yield (Fig. 2) and VFA consumption. The sludge obtained from the PL chemostat consumed over 96% propionate and acetate, and the biogas yields were approximately 62–72% of the theoretical value. For the PH chemostat-derived sludge, over 94% added propionate and acetate were consumed, and the biogas yields were approximately 70–87% of the theoretical value.

Table 1 Set up of DNA-SIP incubation.

Figure 2

Biogas yields of DNA-SIP incubations with PL (a,b) and PH (c,d) sludge. Error bars represent the standard deviation of three replicates.

DNA samples from each microcosm were used for density-gradient centrifugation and fractionation. The DNA distribution profiles in different density fractions were illustrated using relative copies of bacterial and archaeal 16S rRNA genes (Figs 3 and S2). DNA from ¹²C-substrate treatments was mainly distributed between buoyant density values of 1.68–1.72 g mL⁻¹. However, the peak DNA moved to a heavier buoyant density for DNA obtained from ¹³C-substrate treatments. As shown in Fig. 3a, the heavy density fractions from 1.72 to 1.73 g mL⁻¹ (fraction H) contained more bacterial 16S rRNA genes in the ¹³C₃-propionate treatments (10.81–13.83%) than that in the ¹²C₃-propionate treatments (0.59–0.90%) with PL sludge after a 7-day incubation period. Fig. S2a showed that the abundance of archaeal 16S rRNA genes increased from 0.75–1.28% (¹²C₃-propionate treatments) to 26.34–30.76% (¹³C₃-propionate treatments) in the heavy density fractions. For the ¹³C₂-acetate treatments with PL sludge after a 7-day incubation period, no obvious enriched DNA was found in heavy fractions (Figs 3b and S2b). After the 14-day incubation period, bacterial and archaeal 16S rRNA genes were enriched more significantly in heavy fractions of ¹³C-substrate treatments compared with those after the 7-day incubation period (Figs 3c,d and S2c,d). Particularly, 12.47–18.84% bacterial 16S rRNA genes were enriched at the density fractions from 1.73 to 1.74 g mL⁻¹ (fraction HH) in the ¹³C₃-propionate treatments (Fig. 3c).

Figure 3

Relative abundance of bacterial 16S rRNA genes in the gradient fractions. (a) PL sludge (0.05 d⁻¹), propionate treatments, 7 d-incubation; (b) PL sludge, acetate treatments, 7 d-incubation; (c) PL sludge, propionate treatments, 14 d-incubation; (d) PL sludge, acetate treatments, 14 d-incubation; (e) PH sludge (0.15 d⁻¹), propionate treatments, 5 d-incubation; (f) PH sludge, acetate treatments, 5 d-incubation; (g) PH sludge, propionate treatments, 10 d-incubation; (h) PH sludge, acetate treatments, 10 d-incubation. (1), (2) and (3) in each figure means three replicates. The fractions labeled with filled dot were used for 16S rRNA gene sequencing. The filled dot in (a) named LL, L, M and H from buoyant density 1.68–1.73 g mL⁻¹; the filled dot in (b,d,e, f,h) named L, M and H from buoyant density 1.69–1.73 g mL⁻¹; the filled dot in (c,g) named L, M, H and HH from buoyant density 1.69–1.74 g mL⁻¹.

For the ¹³C₃-propionate treatments with PH sludge, a small amount DNA (4.47–7.64%) was enriched in heavy fractions (1.72–1.73 g mL⁻¹) after a 5-day incubation period (Fig. 3e). The abundance of bacterial 16S rRNA genes increased from 10.88–16.19% (¹²C₂-acetate treatments) to 18.78–23.47% in middle fractions (1.71–1.72 g mL⁻¹) for the ¹³C₂-acetate treatments (Fig. 3f). Archaeal 16S rRNA genes were enriched in heavy fractions for ¹³C₃-propionate and ¹³C₂-acetate treatments (Fig. S2e,f). After 10-d incubation, both bacterial and archaeal 16S rRNA genes were enriched in heavy fractions (1.72–1.73 mg L⁻¹) for ¹³C₃-propionate and ¹³C₂-acetate treatments, and 4.43–7.81% bacterial 16S rRNA genes were enriched at the density fractions from 1.73 to 1.74 g mL⁻¹ (fraction HH) in the ¹³C₃-propionate treatments (Fig. 3g). These results suggest that bacterial and archaeal 16S rRNA genes were labeled successfully by ¹³C-acetate and propionate. In order to more closely reflect potential propionate/acetate-oxidizing bacteria and cooperators, recovered DNA of the fractions labeled with filled dots in Fig. 3 and 2S were used for high-throughput sequencing of the 16S rRNA gene.

Phylogenetic identification of the labeled bacterial species

Bacterial community distribution in different buoyant density fractions is shown in Figs S3–S6. For ¹²C-treatments, the amount of bacterial 16S rRNA genes in each fraction of the same treatment was vastly different, but the structure of bacterial community was similar. For ¹³C-treatments, the structure of the bacterial community in each fraction of the same treatment exhibited distinct differences, which suggested that ¹³C-substrate enriched some species in the heavy fractions. After a 7-day incubation period with PL sludge, in fraction H of ¹³C₃-propionate treatments and ¹³C₂-acetate treatments, the abundances of bacterial 16S rRNA genes increased 15.8-fold and 2.2-fold, respectively, compared with the ¹²C-treatment controls (Fig. S3e). Unclassified Synergistaceae increased from 0.17% in ¹²C₃-propionate treatments to 9.06% in ¹³C₃-propionate treatments (53.6-fold increase) (Fig. 4a). Unclassified Spirochaetaceae_1 and unclassified Spirochaetaceae_2 increased 8.1- and 6.2-fold, respectively. The two known SPOB Smithella and Syntrophobacter also increased 7.0- and 13.2-fold, respectively. For the ¹³C₂-acetate treatments with PL sludge, only unclassified Synergistaceae was obviously enriched in fraction H suggesting its role in acetate oxidation, and its abundance increased from 0.30% in ¹²C₂-acetate treatments to 1.56% in ¹³C₂-acetate treatments. The low enrichment of unclassified Synergistaceae in ¹³C₂-acetate treatments suggested that high concentrations of acetate in acetate treatments was more easily utilized directly by acetoclastic methanogens. Unclassified Synergistaceae might have higher acetate affinity and metabolic activity under low acetate concentrations in propionate treatments. After a 14-day incubation period, bacterial 16S rRNA genes were enriched more obviously in ¹³C₃-propionate treatments, and reached 25.83% and 25.10% in fraction H and HH, which was 25.5- and 104.2-fold of that in ¹²C₃-propionate treatments, respectively (Fig. S4e). In addition to the five genera labeled by ¹³C₃-propionate after a 7-day incubation period, three more genera (Soehngenia, Mesotoga, and unclassified Syntrophaceae) were enriched in fraction H (Fig. 4a). The abundance of them increased 49.6-, 32.6-, and 13.2-fold more than those in ¹²C₃-propionate treatments. In fraction HH, the labeled genera were similar with ¹³C₃-propionate treatments after 7-day incubation period. Unclassified Synergistaceae was still the predominant bacteria in ¹³C₃-propionate treatments, and the abundance increased 408.3-fold more than that in ¹²C₃-propionate treatments. For ¹³C₂-acetate treatments, the abundance of bacterial 16S rRNA genes in fraction H after a 14-day incubation period was much higher than that after a 7-day incubation period, which suggested that more bacteria were labeled by ¹³C₂-acetate. However, only unclassified Synergistaceae was notably enriched (25.3-fold) in fraction H, which was consistent with the result after the 7-day incubation period. In addition, unclassified Spirochaetaceae_1 was lightly enriched (2.4-fold) in fraction H of ¹³C₂-acetate treatments.

Figure 4

Relative abundance of main bacterial genera (top20 in chemostat) in the heavy fractions of ¹³C- and ¹²C-substrate treatments. (a) PL sludge; (b) PH sludge.

After incubating for 5 days, the abundance of bacterial 16S rRNA genes in ¹³C₃-propionate treatments with PH sludge increased 7.5-fold in fraction H and 1.9-fold in ¹³C₂-acetate treatments, compared with their ¹²C-controls (Fig. S5e). Compared with ¹²C₃-propionate treatments, the abundance of the two known SPOB Smithella and Syntrophobacter increased 5.0-fold and 13.7-fold in the fraction H of ¹³C₃-propionate treatments, respectively (Fig. 4b). Unclassified Spirochaetaceae_1 increased from 0.13% in ¹²C₃-propionate treatments to 2.42% in ¹³C₃-propionate treatments (19.2-fold increase in fraction H). In addition, the abundance of unclassified Rhodospirillaceae, Syntrophomonas, and Gracilibacter increased approximately ten-fold in fraction H of the ¹³C₃-propionate treatments (Fig. 4b). For ¹³C₂-acetate treatments, the enrichment of the bacterial 16S rRNA gene was not obvious, and only the abundance of unclassified Spirochaetaceae_1 and Gracilibacter in fraction H were enriched 3.5-fold and 2.5-fold of that in ¹²C₂-acetate treatments, respectively (Fig. 4b). After a 10-day incubation period, bacterial 16S rRNA genes were enriched more obviously in heavy fractions in ¹³C-substate treatments (Figs 4b and S7). In addition to the six genera labeled by ¹³C₃-propionate after 5-day incubation period, seven more genera (unclassified Synergistaceae, Cryptanaerobacter, unclassified Chloroflexi, unclassified Elusimicrobia, unclassified Parcubacteria, Mesotoga, and unclassified Syntrophaceae) were enriched 7.0~48.1-fold in fraction H by ¹³C₃-propionate (Fig. 4b). The abundance of bacteria from these genera increased in fraction HH of ¹³C₃-propionate treatments compared with ¹²C₃-propionate treatments (Fig. 4b). In addition, unclassified Synergistaceae and unclassified Spirochaetaceae_1 increased from 0.17% and 0.09% in ¹²C₂-acetate treatments to 2.92% and 1.29% in ¹³C₂-acetate treatments in fraction H, respectively (Fig. 4b).

Phylogenetic identification of labeled archaeal species

The distribution of archaeal community in different buoyant density fractions is presented in Figs S7–S10. For ¹³C₃-propionate treatments with PL sludge, the abundance of main methanogen genera Methanoculleus and Methanothrix was observed with significant enrichment in heavy fractions. However, only acetoclastic Methanothrix was labeled by ¹³C₂-acetate. Similar results were found with PH sludge. Main methanogen genera including Methanothrix, Methanospirillum,Methanoculleus, and Methanolinea were all labeled by ¹³C₃-propionate, but only acetoclastic Methanothrix was labeled by ¹³C₂-acetate. These results suggested that an acetate decarboxylation pathway was used for methane formation of the propionate-degrading microbial community when acetate was the sole carbon source.

Discussion

In this study, two mesophilic anaerobic chemostats fed with propionate as the sole carbon source were constructed and operated at different dilution rates. The potential propionate- and acetate-oxidizing bacteria during anaerobic propionate digestion were investigated using DNA-SIP and 16S rRNA gene high-throughput sequencing. The results indicate that the dilution rate may influence the structure of putative mesophilic propionate- and acetate-oxidizing bacterial communities.

The phylogenetic relationship of bacterial genera labeled by ¹³C-substrate is shown in Fig. 5. In total, 15 bacterial genera were labeled by ¹³C₃-propionate or ¹³C₂-acetate, and 8 of them could not be classified to a genus. These labeled genera exhibited extensive phylogenetic diversity and distributed in 8 phyla (Fig. 5). Four genera (Smithella, Syntrophobacter, unclassified Syntrophaceae, and Mesotoga) were labeled by ¹³C₃-propionate but not ¹³C₂-acetate with both PL and PH sludge. The label of Smithella and Syntrophobacter, which are known SPOB, demonstrated the effectiveness of DNA-SIP used in methanogenic community development. Smithella and Syntrophobacter accounted for 43.23% in PH chemostat, which suggest that they were the primary SPOB under high dilution rate conditions (0.15 d⁻¹). The OTUs of unclassified Syntrophaceae (PL-OTU204 and PH-OTU630) exhibited a 94% sequence (16S rRNA gene) similarity with SPOB Smithella propionica (NR_024989) and a 93% sequence similarity with aromatic compound degrader Syntrophus aciditrophicus (NR_117565). Nobu et al. reported a novel Syntrophaceae member in a methanogenic bioreactor degrading terephthalate (TA) likely performing syntrophic degradation of butyrate and branched-chain fatty acids based on metagenomic and metatranscriptomic analysis²⁰, but unclassified Syntrophaceae (PL-OTU204 and PH-OTU630) only had a 94% sequence similarity with this novel Syntrophaceae member. However, unclassified Syntrophaceae (PL-OTU204 and PH-OTU630) were found having a 100% sequence similarity with unclassified Syntrophaceae OTUs enriched in two mesophilic isovalerate-fed methanogenic chemostats in our lab, which were verified having butyrate degradation pathway and high expression activity based on metagenomic and metatranscriptomic analysis (unpublished data). Therefore, unclassified Syntrophaceae (PL-OTU204 and PH-OTU630) may be related to butyrate oxidation in both PL and PH chemostats. Syntrophobacter degrade propionate to produce acetate, hydrogen, and carbon dioxide (propionate + H₂O → acetate + 3H₂ + H⁺ + HCO₃⁻) using the randomizing methyl-malonyl-coenzyme A (CoA) (MMC) pathway, whereas the genus Smithella uses a different pathway that results in the formation of acetate and butyrate, followed by the syntrophic β-oxidation of butyrate to acetate (2propionate + 2H₂O → 3acetate + 2H₂ + H⁺)^21,22,23. Unclassified Syntrophaceae (PL-OTU204 and PH-OTU630) should be labeled by ¹³C₄-butyrate, the intermediate metabolite of ¹³C₃-propionate. Mesotoga PL-OTU64 and PH-OTU361 were affiliated to Mesotoga infera VNs100 (NR_117646, 99% similarity), which is a member of phylum Thermotogae. According to the results of the metagenomic analysis, Nobu et al. found that Mesotoga may syntrophically oxidize acetate though a previously uncharacterized pathway²⁰. However, Mesotoga was more easily labeled by ¹³C₃-propionate than ¹³C₂-acetate, which suggest that Mesotoga may prefer to utilize acetate in very low acetate concentrations.

Figure 5

Neighbor-joining phylogenetic tree of ¹³C-substrate labeled representative bacteria.

Unclassified Spirochaetaceae_2 and Soehngenia were only labeled by ¹³C₃-propionate with PL sludge suggesting that they might be related to propionate degradation in PL chemostat. Unclassified Spirochaetaceae_2 PL-OTU212 exhibited a 98% sequence similarity with clone M235 (JN688037) obtained from an anaerobic cellulolytic microbial consortium²⁴. Soehngenia PL-OTU266 exhibited 95% sequence similarity with a benzaldehyde-converting bacterium Soehngenia saccharolytica (NR_025761)²⁵. Six genera (unclassified Rhodospirillaceae, Syntrophomonas, Cryptanaerobacter, unclassified Chloroflexi, unclassified Elusimicrobia, and unclassified Parcubacteria) were labeled by ¹³C₃-propionate but not ¹³C₂-acetate with PH sludge. Unclassified Rhodospirillaceae PH-OTU669 only exhibited a 91% sequence similarity with a nitrogen-fixing bacterium Azospirillum soli (NR_145884), but had a 99% sequence similarity with clone PMB07 (AB232813, 99%) from an enriched mesophilic propionate-degrading methanogenic consortium¹⁹. Unclassified Rhodospirillaceae may be a potential propionate-oxidizing bacterium in the PH chemostat. Syntrophomonas PH-OTU640 had a 97% sequence similarity with a syntrophic butyrate-oxidizing bacterium Syntrophomonas cellicola (NR_125510)²⁶. Syntrophomonas should be labeled by ¹³C₄-butyrate, the intermediate metabolite of ¹³C₃-propionate. Cryptanaerobacter PH-OTU231 exhibited 99% sequence similarity with the Cryptanaerobacter phenolicus strain LR7.2 (NR_025757), which can transform phenol and 4-hydroxybenzoate into benzoate²⁷. Cryptanaerobacter PH-OTU231 also shared a 97% sequence similarity with Pelotomaculum propionicicum strain MGP (NR_041000), an obligately syntrophic propionate-oxidizing bacterium⁵. Ahlert et al. reported that Cryptanaerobacter sp./Pelotomaculum sp., whose sequences are related to both genera Cryptanaerobacter and Pelotomaculum, were the potentially propionate-oxidizing key species²⁸. Therefore, Cryptanaerobacter PH-OTU231 may participate in syntrophic propionate oxidation. Unclassified Chloroflexi PH-OTU54, unclassified Parcubacteria PH-OTU639, and unclassified Elusimicrobia PH-OTU745 had low sequence similarities (<90%) with pure cultured bacterial species. Unclassified Chloroflexi PH-OTU54 exhibited 99% sequence similarity with clone QEDR1AF11 (CU922883) obtained from anaerobic sludge digesters and defined as a core microorganism involved in anaerobic digestion²⁹. Nobu et al. found an uncultivated Chloroflexi subphylum I member which may be capable of H₂-oxidizing homoacetogenesis in a methanogenic bioreactor degrading terephthalate (TA)²⁰. However, unclassified Chloroflexi PH-OTU54 only had an 83% sequence similarity with this potential homoacetogen. Unclassified Parcubacteria PH-OTU639 was most closely related to clone B16 (JX100399, 99% similarity), which was obtained from an upflow anaerobic sludge blanket reactor with propionate as sole carbon source³⁰. PH-OTU54 and PH-OTU639 may be related to propionate oxidation in the PH chemostat as they were labeled only by ¹³C₃-propionate and no reports on their capacity of acetate oxidation or acetogenesis could be found. Unclassified Elusimicrobia PH-OTU745 exhibited 99% sequence similarity with clone Ace13(8–20) (KU667238) labeled by ¹³C₂-acetate under methanogenic conditions³¹. Considering these findings and the results of the present study, it is suggested that unclassified Elusimicrobia PH-OTU745 may be related to the degradation of the intermediate metabolite acetate.

Three genera were labeled by both ¹³C₃-propionate and ¹³C₂-acetate, including unclassified Spirochaetaceae_1, unclassified Synergistaceae, and Gracilibacter. The OTUs of unclassified Spirochaetaceae_1 (PL-OTU229 and PH-OTU249), predominant bacteria in both PL and PH chemostats, both exhibited 99% similarity with clone NBLE38B (GU389895) from an anaerobic digester treating food-processing wastes³². Similar unclassified Spirochaetaceae OTUs in an acetate-fed and a butyrate-fed mesophilic methanogenic chemostat were also labeled by ¹³C-acetate³³. Acetate stimulated the activity of Spirochaetes the most compared with other VFAs during anaerobic digestion³⁴, and the increase in the cluster II Spirochaetes was found to be positively correlated with increase in hydrogenotrophic methanogens in batch reactors seeded with five different anaerobic sludge samples supplemented with acetate as the sole carbon source, suggesting the possible role of Spirochaetes in syntrophic acetate oxidation³⁵. The OTUs of unclassified Synergistaceae (PL-OTU209 and PH-OTU307), predominant bacteria in both PL and PH chemostats, exhibited a 95% similarity with an amino-acid-degrading bacterium Thermovirga lienii (NR_043522)³⁶. Similar unclassified Synergistaceae OTUs in an acetate-fed and a butyrate-fed mesophilic methanogenic chemostat were also labelled by ¹³C-acetate^31,33. Xu et al. reported that Thermovirga belonging to Synergistaceae accounted for nearly half of the total OTUs in an anaerobic chemostat fed with acetate³⁷. In other studies conducted in our lab, a high abundance of this OTU also detected in acetate-fed anaerobic chemostat (data not shown). The syntrophic acetate oxidation ability of Synergistaceae spp. were revealed by genome-centric metagenomics³⁸ and SIP-based functional study³⁹. Together, unclassified Synergistaceae (PL-OTU209 and PH-OTU307) may be syntrophic acetate oxidizer in the PL and PH chemostats. Gracilibacter PH-OTU760 exhibited 98% sequence similarity with Gracilibacter thermotolerans strain JW/YJL-S1 (NR_115693) which can ferment glucose to produce acetate⁴⁰. These species might be able to utilize acetate based on our DNA-SIP results.

Of the archaea, one genus of hydrogenotrophic methanogen Methanoculleus was detected at the dilution rate of 0.05 d⁻¹, whereas the Methanoculleus, Methanospirillum, Methanolinea, and Methanobacterium genera were detected at a higher dilution rate of 0.15 d⁻¹. The minimal threshold for hydrogen partial pressures of Methanoculleus is much lower than other hydrogenotrophic methanogens^41,42,43, which may explain the dominance of Methanoculleus among hydrogenotrophic methanogens in the chemostat under the two dilution rates. The higher propionate degradation and hydrogen production rates at a high dilution rate (0.15 d⁻¹) allowed the abundance of other hydrogenotrophic methanogens with lower hydrogen affinities to increase. Aceticlastic methanogen Methanothrix, which exhibits stronger acetate affinity than Methanosarcina⁴⁴, was detected in both chemostats. Methanothrix was also found to be the dominant aceticlastic methanogen in a variety of anaerobic reactors at low acetate concentrations⁴⁵, which is consistent with the findings of this study. In the ¹³C₃-propionate treatments with either PL or PH sludge, all main methanogens in original chemostats were concentrated in the heavy fractions, which suggested that these methanogens participated in propionate-related methanogenesis processes. In the ¹³C₂-acetate treatments, only the aceticlastic methanogen Methanothrix was concentrated in the heavy fractions, which suggest that aceticlastic methanogenesis pathway is dominant in acetate treatments.

In SIP experiment, it is difficult to eliminate cross-feeding, but it is possible to reduce it by optimizing parameters including incubation time, substrate concentration, and substrate adding frequency. Many preliminary SIP experiments had been down to optimize these parameters before the formal SIP experiments carried out in this study. In addition, attention was not payed to those bacteria that were identified in the heavy fraction but having fairly low abundance. Compared with those OTUs having high abundance in the heavy fraction, the bacteria having fairly low abundance might have higher possibility of cross-feeding. However, the functions of the OTUs labeled in this study should be further studied and direct evidences of their functions should be obtained.

In conclusion, DNA-SIP successfully identified the propionate and acetate oxidation players in propionate-fed chemostats. Different species were retrieved from the two chemostats, which were operated at different dilution rates suggesting that, in addition to known SPOB, more uncultured bacteria may be involved in propionate degradation during anaerobic digestion. However, to reveal the propionate and acetate oxidation functions and metabolic pathways of these labeled unknown species, further study employing metagenomics and culture-dependent techniques is required.

Materials and Methods

Construction and operation of chemostats fed with propionate as the sole carbon source

Two anaerobic chemostats were constructed using two continuous stirred tank reactors (CSTRs) each with a working volume of 1.8 L as described previously¹⁹. The seed sludge was obtained from an anaerobic digester treating distillation wastewater of an alcohol plant in Sichuan Province, China. The chemostats were fed with synthetic wastewater containing propionate as the sole carbon source (TOC 8000 mg L⁻¹) at dilution rates of 0.05 d⁻¹ and 0.15 d⁻¹, respectively (designated PL and PH chemostats, respectively). The components of the synthetic wastewater were the same as described in a previous study¹⁹. The wastewater comprised (grams per liter): sodium propionate, 4.27; propionic acid, 13.16; KH₂PO₄, 0.3; KHCO₃, 4.0; NH₄Cl, 1.0; NaCl, 0.6; MgCl₂·6H₂O, 0.82; CaCl₂·2H₂O, 0.08; cysteine–HCl·H₂O, 0.1; 10 ml of a trace element solution of Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH medium 318 containing 21.3 mg L⁻¹ of NiCl₂·6H₂O and 24.7 mg L⁻¹ of CoCl₂·6H₂O; and 10 ml of vitamin solution of DSMZ medium 318 without B₁₂. The two chemostats operated stably at 37 °C for approximately 1500 and 400 days, respectively. During the operation, several parameters including pH, SS, VSS, TOC, and VFAs were measured regularly as described previously⁴⁶. The sludge from each chemostat was used for microbial community analysis and DNA-SIP experiments.

SIP incubation with ¹³C₃-propionate and ¹³C₂-acetate

Sludge was sampled directly from the PL and PH chemostats on day 1467 and 351, respectively. Microcosms were individually constructed using 50-mL serum bottles with 30 mL sludge from PL or PH digester. The headspaces were flushed with pure N₂ for 3 min, and the bottles were sealed with a butyl rubber stopper and an aluminum cap. Cysteine-HCl was added as reducing agent and resazurin was added as redox indicator. Considering the co-existence of acetate and propionate degraders during the propionate metabolism, the sludge was incubated separately with ¹³C₃-propionate and ¹³C₂-acetate to aid in identifying of SPOB. Incubations with ¹²C₃-propionate and ¹²C₂-acetate were used as controls. Propionate or acetate were added to the bottles using a gas-tight syringe every day during the incubation period, as shown in Table 1. Sixteen different treatments were established and conducted in triplicate. They are summarized in Table 1. All microcosms were incubated at 37 °C and 150 rpm on a shaker. Biogas production was measured using a syringe every day. After incubation, remaining VFAs were detected and sludge from each microcosm was collected and used for DNA extraction. ¹³C₃-propionate (99 atom% in ¹³C) and ¹³C₂-acetate (99 atom% in ¹³C) were purchased from Cambridge Isotope Laboratories (USA).

DNA extraction, density-gradient centrifugation and fractionation

Total genomic DNA of each sludge sample was extracted by using CTAB method according to the method outlined by Griffiths et al.⁴⁷. Purified DNA was prepared for density-gradient centrifugation and fractionation as described by Lueders et al.⁴⁸. Briefly, total DNA (~2.5 μg) was added to Quick-Seal polyallomer tubes (6.3 mL, Beckman Coulter, Australia), along with 1.2 mL of the gradient buffer (GB) containing 0.1 M Tris-HCl with pH of 8.0, 0.1 M KCl, and 1 mM EDTA, and 4.8 mL CsCl solution (final buoyant density of 1.90 g mL⁻¹). Then, the tubes were sealed, transferred to a Beckman ultracentrifuge with Ti90 fixed angle rotor (Beckman, USA), and centrifuged at 177,000 × g for 40 h at 20 °C. Following centrifugation, 15 density fractions (400 μL of each fraction) were collected from each tube using a fraction recovery system (Beckman Coulter, USA). The buoyant density of each fraction was determined using a digital refractometer (AR200, Reichert, USA)⁴⁸. DNA was recovered from each fraction by PEG6000 precipitation with glycogen.

In order to profile the DNA gradient distribution, bacterial and archaeal 16S rRNA genes in each fraction were quantified by qPCR using the EcoTM real-time PCR system (Illumina, USA) with primer sets Eu27f/Eu518r⁴⁹ and Arch349f/Arch806r⁵⁰, respectively. According to the results of qPCR, the DNA samples from fractions with buoyant density during 1.69–1.74 g mL⁻¹ were selected for high-throughput sequencing of the 16S rRNA gene.

High-throughput sequencing of 16S rRNA gene and phylogenetic analysis

DNA samples were amplified with primer 515 F (5′-GTGCCAGCMGCCGCGGTAA-3′) and 909 R (5′- CCCCGYCAATTCMTTTRAGT -3′) to amplify the V4-V5 region of 16S rRNA gene in both bacteria and archaea. Sequencing was performed with an Illumina MiSeq platform (Illumina, San Diego, USA) according to the standard protocols described by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Raw fastq files were quality-filtered by Trimmomatic and merged by FLASH with the following criteria: (i) the reads were truncated at any site receiving an average quality score <20 over a 50 bp sliding window. (ii) Sequences whose overlap exceeded 10 bp were merged according to their overlap with mismatch no more than 2 bp. (iii) Sequences of each sample were separated according to barcodes (exactly matching) and primers (allowing 2 nucleotide mismatching), and reads containing ambiguous bases were removed. Operational taxonomic units (OTUs) were clustered at a 97% similarity cutoff using UPARSE-OTU (version 7.1 http://drive5.com/uparse/) with a novel “greedy” algorithm that simultaneously performs chimera filtering and OTU clustering⁵¹. Final OTUs were taxonomically classified using Ribosomal Database Project classifiers and NCBI BLAST⁵². Sequences of 21 target OTUs and thirty-seven reference sequences were used to constructed phylogenetic tree. Multiple alignments were run using the Clustalx1.8 (http://www.clustal.org/). Distance matrix trees were constructed using the MEGA4 software package (http://megasoftware.net/mega4/) by using the neighbor-joining method with the Kimura two parameter model and complete-deletion option. Bootstrap resampling analyses for 1000 replicates were performed to estimate the confidence of tree topologies.

Nucleotide sequence accession numbers

The original sequencing data is available at the National Center for Biotechnology Information (Accession No. PRJNA505222).