Bacteria Responsible for Nitrate-dependent Antimonite Oxidation in Antimony-contaminated Paddy Soil Revealed by the Combination of DNA-SIP and Metagenomics

Background: Antimonite (Sb(III)) oxidation (SbO) can decrease the toxicity of antimony (Sb) and its uptake into plants (e.g., rice), thus serving an ecological role in bioremediation of Sb contamination. In some anoxic environments, Sb(III) can be oxidized coupled with nitrate as the electron acceptor. Here we investigate the potential for nitrate-dependent SbO in Sb contaminated rice paddies and identify nitrate-dependent Sb(III)-oxidizing bacteria (SbOB) using stable isotope probing (SIP) coupled with amplicon and shotgun metagenomic sequencing. Results: Anaerobic SbO was exclusively observed in the paddy soil amended with both Sb(III) and NO 3- , whereas no apparent SbO was detected in the soil amended with Sb(III) only. The increasing abundance of the arsenite oxidase (aioA) gene suggests that nitrate-dependent SbO was catalysed by microorganisms harbouring the aioA gene. After 60-day DNA-SIP incubation, obvious shift in the aioA gene to heavy DNA fractions only in the treatment amended with 13 C-NaHCO 3 , Sb(III) and NO 3 - suggested the incorporation of 13 C by nitrate-dependent SbOB. Accordingly, DNA-SIP identied a number of putative nitrate-dependent SbOB in the paddy soil, including Azoarcus, Azospira and Chelativorans. Metagenomic analysis further revealed that they contained aioA gene and genes involved in denitrication and carbon xation, supporting their capability for nitrate-dependent SbO. Conclusions: These observations in this study suggested the occurrence of nitrate-dependent SbO in paddy soils. A number of putative nitrate-dependent SbOB (i.e., Azoarcus, Azospira and Chelativorans) were reported here, which expands our current knowledge regarding the diversity of nitrate-dependent SbOB. In addition, this study provides a proof of concept using DNA-SIP to identify nitrate-dependent SbOB.

However, Stibiobacter senarmontii was reported to be capable of growing chemoautotrophically using the energy obtained from SbO [25]. Currently, over 60 bacterial strains have been reported to oxidize Sb(III) via arsenite oxidase AioAB or antimonite oxidase AnoA [26]. Southwest China is the major rice producing area and also a key Sb mining area. The Sb mining activities have caused frequent contamination of rice elds. For instance, it has been reported that rice is a major route for Sb exposure near Xikuangshan Sb mining area, contributing to over 30% of the daily intake of the Sb to exposed population [25]. Therefore, Sb contamination in rice paddies n + ear Sb mining area is an important environmental issue. In addition, rice is found to be more e cient in uptake of Sb(III) than of Sb(V) [25,26]. Hence, SbO may be an important biogeochemical process that attenuates Sb toxicity in rice paddies and decreases the translocation and accumulation of Sb in the rice. However, the anoxic conditions in some habitats such as river sediment and ooded rice paddies may hinder SbO because of the lack of O 2 . Notably, nitrate is an alternative to O 2 as an oxidant that can support SbO under anoxic conditions. Indeed, three anaerobic Sb(III)-oxidizing bacteria (SbOB) (i.e., Hydrogenophaga taeniospiralis strain IDSBO-1, Ensifer sp. NLS4 and Sinorhizobium sp. GW3) have been isolated using nitrate as the electron acceptor [13,18,27]. The anoxic conditions and high concentrations of nitrate caused by N fertilization would allow rice paddies to facilitate nitrate-dependent SbO. However, nitrate-dependent SbO has never been reported in rice paddies so far.
Xikuangshan Sb mine, designated as "the World Antimony Capital", located in Lengshuijiang City in Hunan Province of China, is the world largest Sb deposit with over 120 years of mining history and environmental contamination. In order to investigate nitrate-dependent SbO in rice paddy soils and identify the bacteria responsible for this process, Sb-contaminated paddy samples were collected near the Xikuangshan mining area [28,29]. It is proposed that the long-term Sb contaminated may enrich Sbmetabolising bacteria including nitrate-dependent SbOB and makes these Sb-contaminated rice paddy soils ideal for this study. Although three nitrate-dependent SbOB have been isolated and identi ed so far [13,18,27], their overall diversity and distribution in soil, especially in rice paddy soil, is not known. Culture-independent DNA-stable isotope probing (SIP) incubations can link microbial identity with function in environment and enable a greater understanding of active microbial communities involved in the process under study [30]. Indeed, SIP has been used to identify chemolithotrophs in various habitats such as rice elds [31], Karst caves [29], and marine sediments [32]. More specially, DNA-SIP has been used to identify microorganisms responsible for nitrate-dependent Fe(II) and As(III) oxidation [33,34]. Therefore, DNA-SIP may be capable of identifying nitrate-dependent SbOB as well. Combining DNA-SIP with amplicon and shotgun metagenomics herein, we aimed to (i) investigate the potential of nitratedependent SbO in Sb-contaminated rice paddy soil; (ii) identify bacteria responsible for nitrate-dependent SbO; and (iii) explore the metabolic potentials of the putative nitrate-dependent SbOB.

Methods
The analyses carried out in this study included chemical characterization of the soil, microcosms incubation, DNA-SIP, and amplicon and shotgun metagenomic sequencings, which are described in detail Page 5/24 below.

Soil collection and geochemical analyses
Rice soils were sampled at 30-40 cm depth from the surface in various ooded rice paddy elds (27°44′38″N, 111°27′46″E) near Xikuangshan mining area, which were immediately sealed and transported at 4 °C to the laboratory. In addition, soils from Sb-contaminated arid elds were also collected near the mining area. A preliminary set of experiments were performed to monitor nitratedependent SbO in all of the soils collected from various rice elds and arid elds. The rice paddy soil showing the most rapid rate of nitrate-dependent SbO was selected for further experiments while no soils from arid elds showing nitrate-dependent SbO. The selected rice paddy soil was subsequently subsampled for geochemical analyses and for microcosm setup. The soil contained over 750 mg kg − 1 Sb, 200 mg kg − 1 As, and a pH of 6.89 (Table S1).

Soil microcosms incubations
Three sets of microcosms were performed: (i) nitrate-dependent SbO activity incubation -to examine the potential of nitrate-dependent SbO in the rice paddy; (ii) nitrate-dependent SbO enrichment incubation -to monitor the shift in the bacterial community after amending Sb(III) and NO 3 − ; (iii) SIP incubation -to identify nitrate-dependent SbOB by DNA-SIP.
Nitrate-dependent SbO activity incubation: Soil cultures were prepared by mixing 2 g of paddy soil and 40 ml of mineral salts medium [53]  Sb species (i.e., Sb(III) and Sb(V))) were adapted from the methods for measurement of As species using LC-AFS (AFS-920, Haiguang, Beijing) equipped with a hallow cathode lamp for Sb (Shuguangming, Beijing) [54]. The remaining soil sample pellets were subsampled for analyses of Sb species adsorbed in the soil (stored at 4 °C) and for molecular microbial community analysis (stored at -80 °C). To extract Sb species adsorbed in the soil phase, 0.5 g of the remaining soil sample pellets were mixed with 50 mL 1 M H 3 PO 4 and resuspended by 10 s ultrasonication, shaken for 2 h at 200 rpm, and followed by 10 s ultrasonication (modi ed from a method for As species extraction from soil [55]). The resuspended mixture was immediately centrifuged and the resulting supernatant was subject to LC-AFS for Sb species measurement as mentioned above.
Nitrate-dependent SbOB enrichment incubation: To investigate the shift of microbial communities after amending Sb(III) and NO 3 − , cultures inoculated with approximately 2 g paddy soil were prepared, purged and sealed as described above. Two treatments were set up with the following amendments: (i) 1 mM Sb(III) and 3 mM NO 3 − ; or (ii) 3 mM NO 3 − only. These cultures were respiked with 8 mM NaHCO 3 , 3 mM NO 3 − and 1 mM Sb(III) every 12 days when Sb(III) was completed oxidized and NO 3 − was depleted. All the cultures were incubated for a total of 60 days as mentioned above. Triplicate cultures of these two treatments were destructively sampled after incubation for 0, 12, 30 and 60 days. Each culture sample was immediately centrifuged and the soil pellet was stored at -80 °C for molecular analysis.

SIP gradient fractionation
Genomic DNAs from the 13 C-and 12 C-NaHCO 3 SIP incubations (day 0, 30 and 60) were separated into "heavy" (i.e., 13 C-DNA) and "light" (i.e., 12 C-DNA) fractions by isopycnic density gradient centrifugation as previous [34]. Brie y, approximately 2 µg of genomic DNA was added into CsCl solution (buoyant density (BD) = 1.714 g mL − 1 ) in an OptiSeal polyallomer tube (Beckman Coulter, Palo Alto, USA). The mixture was then ultracentrifuged at 409,000 g for 24 h using a VTi 90 vertical rotor in an Optima XPN-100 Ultracentrifuge (Beckman Coulter). The resulting CsCl gradients were then fractionated into 24 equal volumes (~ 200 µL) with a fraction recovery system (Beckman Coulter, USA) [57,58]. The BD value of each fraction was immediately determined by measuring the refractive index using a digital refractometer (Palette, ATAGO, Japan). DNA in each fraction were precipitated with 6 µL glycogen (ZOMANBIO, China) dissolved in 30% cold ethanol and then nally eluted in 30 µL of TE buffer (pH 8.0) [57,58]. From the eluted DNA, qPCR was used as described above to determine the copy numbers of aioA gene in each of the 24 fractions collected.
In addition, the bins phylogenetic tree was generated based on conserved protein sequences with PhyloPhlAn [72].

Activity of nitrate-dependent SbO
Oxidation of Sb(III) to Sb(V) was only observed in the treatment amended with both Sb(III) and NO 3 − but not in the treatment amended with Sb(III) or NO 3 − only ( Fig. 1a and b). Approximately 0.90 ± 0.08 mM Sb(III) was fully oxidized to Sb(V) (0.87 ± 0.04 mM) after an incubation period of 12 days, with 2.26 ± 0.06 mM NO 3 − reduced. In contrast, 0.11 ± 0.00 mM Sb(V) (probable carryover from the paddy soil) was detected at day 0 in the treatments amended with Sb(III) only, but no discernable SbO was detected over the course of incubation ( Fig. 1a and b). Nitrate reduction (1.41 ± 0.03 mM) was also observed in the treatment amended with NO 3 − only, but was signi cantly less than that in the treatment amended with both Sb(III) and NO 3 − (P < 0.05) (Fig. 1c). These observations suggest that anaerobic SbO in these cultures was nitrate-dependent. In addition, neither SbO nor NO 3 − reduction was detected in the sterile controls inoculated with autoclaved soil (Fig. 1), indicating that anaerobic SbO was mainly driven by microorganisms in the paddy soil.
Increase in the abundance of the aioA gene over the course of nitrate-dependent SbO Given that arsenite oxidase AioA has been reported to catalyze SbO in previous studies [20,35], the aioA gene was quanti ed in all treatments from nitrate-dependent SbO activity incubations at day 0, 3, 6 and 12. The copies of the aioA gene signi cantly increased by 1.4-fold throughout the 12-day incubation in the treatments amended with Sb(III) and NO 3 − (P < 0.05), while no such change in aioA genes was observed in the treatments amended with Sb only (Fig. 2a). A signi cant decrease in the abundance of the aioA gene was observed in the treatments amended with NO 3 − only (P < 0.05) (Fig. 2a), suggesting that aioA gene-containing microorganisms might be outcompeted by others when no stress of Sb contamination occurred. In addition, a signi cantly positive correlation between the copies of the aioA gene and concentrations of Sb(V) was found in the treatment amended with Sb(III) and NO 3 − (R = 0.88, P < 0.05) (Fig. 2b), while no such correlations were found in treatments amended with Sb or NO 3 − only (data not shown). No PCR products were obtained by amplifying genes encoding antimonite oxidase (anoA) in any of these three treatments (data not shown).

Shift of bacterial communities over the course of nitratedependent SbO
The bacterial communities of two treatments (treatment amended with Sb(III) and NO 3 − and treatment amended with NO 3 − only) were further characterized. Accordingly, an obvious shift in the bacterial communities was detected in the treatment amended with Sb(III) and NO 3 − over the course of the incubation (day 0, 12, 30 and 60) (Fig. 3). Speci cally, the relative abundance of Azoarcus signi cantly increased from undetectable at day 0 to 78 ± 2% at day 60 in treatments amended with Sb(III) and NO 3 − (P < 0.05) (Fig. 3). The proportion of bacteria associated with Azospira reached their peak (42 ± 6%) at day 30 then decreased to 5 ± 2% at day 60 in treatments amended with Sb(III) and NO 3 − (P < 0.05) (Fig. 3).
In contrast, bacterial communities were relatively stable in treatments amended with NO 3 − only, with Herbaspirillum and Gemmatimonas gradually increasing from undetectable at day 0 to 10 ± 2% at day 60 and from 4 ± 0% at day 0 to 9 ± 1% at day 60, respectively (Fig. 3). Further, PERMANOVA demonstrated that the bacterial communities were signi cantly different between the treatment amended with Sb(III) and NO 3 − and the treatment amended with NO 3 − only (P = 0.001), suggesting that Sb(III) plus NO 3 − shapes the bacterial communities and possibly enriched those involved in nitrate-dependent SbO. Therefore, DNA-SIP combined with amplicon sequencing was subsequently performed to identify the microorganisms responsible for nitrate-dependent SbO.
SbOB identi ed by DNA-SIP and amplicon sequencing of 16S rRNA gene The maximum abundance of aioA gene was detected in the light fraction (BD = 1.708 g ml − 1 ) of 12 CSbN during the SIP incubation (Fig. 4). Compared to 12 CSbN, the highest abundance of the aioA gene gradually shifted to the heavier fractions (BD = 1.712 at day 30, BD = 1.727 g ml − 1 at day 60) in the 13 CSbN treatment only (Fig. 4), implying that nitrate-dependent SbOB incorporated 13 C over the course of the incubation. In contrast, no obvious shifts in abundance of the aioA gene to the heavy fractions were detected in 12 CSbN, 12 CN or 13 CN treatments (Fig. 4).

Discussion
Rice has been suggested as a major route for Sb exposure, especially in mining areas [14,36], and is reported to be more e cient in taking up Sb(III) than Sb(V) [14]. SbO will generate the less mobile Sb(V) and thus reduce the uptake of Sb by the rice. Therefore, SbO may be bene cial to attenuate the consequences Sb contamination in rice paddies. The anoxic conditions and the high levels of nitrate in rice paddies may facilitate nitrate-dependent SbO, which, however, has never been reported in rice paddies. Therefore, the current study tackles this important but less understood environmental issue to investigate the potential of nitrate-dependent SbO in the Sb-contaminated rice paddies.
Our current understanding of anaerobic Sb(III) oxidizers is mainly based on three isolates [13,18,27].
Culture-independent tools such as DNA-SIP may be available to expand the list of nitrate-dependent SbOB. However, clear Sb-dependent growth is di cult to observe because SbOB require high concentrations (millimolar range) of Sb(III) to prompt signi cant increases in biomass [26]. The slow growing nature of SbOB may increase the required duration of 13 C incorporation and thus incur crossfeeding, which complicates the interpretation of SIP data. The current study aims to examine the proof of concept of using DNA-SIP combined with 16S rRNA gene amplicon sequencing to reveal nitratedependent SbOB. Given that the long incubation time for DNA-SIP may cause cross-feeding among microorganisms, shotgun metagenomic sequencing was further performed on the heavy DNA fractions of 13 CSbN to examine whether some key genes responsible for nitrate-dependent SbO (i.e., Sb(III) oxidation, nitrate reduction, and carbon xation) were present in the putative nitrate-dependent SbOB to con rm their role in nitrate-dependent SbO.
Nitrate-dependent SbO potential of the Sb-contaminated rice paddy soil Anaerobic Sb(III) oxidation to Sb(V) was clearly demonstrated in the anoxic rice-paddy cultures amended with Sb(III) and NO 3 − , but not in the cultures amended with NO 3 − or Sb(III) only (Fig. 1), suggesting that the addition of nitrate may facilitate SbO. Nitrate reduction with concomitant SbO is further supported by the conversion of nitrate to nitrite ( Fig. 1c and d). Sterile controls showed no formation of Sb(V) supporting that nitrate-dependent SbO is a biotic process. Overall, this observation con rms that bacteria can mediate nitrate-dependent SbO in rice paddy soils.
aioA may be the key gene for nitrate-dependent SbO Given similar chemical properties shared by Sb and As elements, it has been proposed that microbes may drive Sb transformation by using similar metabolic pathways with As. Previous studies suggest that arsenite oxidase (encoded by the aioA gene) may be responsible for SbO. For example, the aioA gene has been detected in some known nitrate-dependent SbOB including Hydrogenophaga taeniospiralis strain IDSBO-1 and Sinorhizobium sp. GW3 [13,27]. In addition, the transcription level of the aioA gene in Sb(III)oxidizing Sinorhizobium sp. GW3 signi cantly increased upon Sb(III) addition under anaerobic conditions and a mutation in the aioA gene reduced the anaerobic SbO rate by over 70% [13]. Consistently, several observations in this study also support that the aioA gene is involved in anaerobic SbO: (i) the copy numbers of the aioA gene increased only in the cultures with nitrate-dependent SbO (Fig. 2a). The abundance of the aioA gene showed signi cant positive correlations with the concentration of Sb(V) produced over the course of nitrate-dependent SbO in the treatment amended with both Sb(III) and NO 3 − (R = 0.88, P < 0.05) (Fig. 2b), whereas such correlation was not detected in two other treatments where nitrate-dependent SbO was not observed (data not shown); (ii) following a 60-day incubation period, the highest relative abundance of the aioA gene was observed to gradually shift to the heavier DNA fractions only in the 13 CSbN treatment where nitrate-dependent SbO occurred, while no obvious shifts to the heavier fractions were found in other treatments (i.e., 13 CSbN, 13 CN and 12 CN). These observations collectively support that the aioA gene is responsible for nitrate-dependent SbO. A Sb(III) oxidase, encoded by anoA gene, belonging to the short-chain dehydrogenase/reductase family was recently identi ed and proposed to be responsible for aerobic SbO [20]. In this study, the anoA gene, however, was neither successfully ampli ed from any of the treatments nor observed in the metagenome, implying that anoA may not be responsible for nitrate-dependent SbO in this rice paddy soil.
Putative nitrate-dependent SbOB identi ed by DNA-SIP A number of genera, such as Azoarcus, Azospira and Chelativorans, were proposed as putative nitratedependent SbOB in the current study. The relative abundance of Azoarcus spp. increased from undetectable in the original rice paddy soil inoculum to 78% at day 60 in the cultures amended with Sb(III) and NO 3 − (Fig. 3). Since Azoarcus was not enriched in the cultures amended with NO 3 − only, it suggests that SbO likely supported its growth. Furthermore, as seen in the DNA-SIP result (Fig. 5), Azoarcus dominated (close to 50%) in the heavy DNA fractions of the 13 CSbN treatment, but was not found in the 13 CN treatments, thus demonstrating that Azoarcus incorporated 13 C-NaHCO 3 only during nitratedependent SbO. Azoarcus spp. are well known for their capability to mediate nitrate-dependent As(III) oxidation via AioA in paddy soils and other environments [34,38]. Consistently, aioA genes were observed in the bin associated with Azoarcus (bin9), supporting their role also in SbO. In addition, genes for denitri cation and carbon xation were observed in the Azoarcus-associated bin9, suggesting its capability for denitri cation and autotrophy (Fig. 6). Collectively, these results support that Azoarcusassociated bacteria are responsible for the autotrophic oxidation of Sb(III) linked to nitrate reduction in the paddy soil. Bacteria associated with Azospira dominated (42 ± 6%) the bacterial communities in the treatment amended with Sb(III) and NO 3 − at day 30 ( Fig. 3) and was observed to be signi cantly enriched in the heavy DNA fractions of 13 CSbN treatment compared to 13 CN (Fig. 5). In addition, a bin associated with Azospira containing the aioA gene was detected in the 13 C-heavy-fraction metagenome (Fig. 6). These observations suggest that Azospira may be a putative nitrate-dependent SbOB. The detection of genes for denitri cation and carbon xation in the Azospira-associated bin also supported its capability for nitrate-dependent SbO. Nitrogen cycling by the genus Azospira has been previously described, including nitrogen xation and denitri cation [39,40]. Although Azospira spp. has not previously been shown to oxidize Sb(III) under either aerobic or denitrifying conditions, autotrophic Azospira sp. strain ECC1-pb2 isolated from sludge and sediment samples was capable of As(III) oxidation linked to chlorate reduction [41]. Our current study identi ed Azospira spp. as putative nitrate-dependent SbOB. Chelativorans-a liated bacteria were identi ed as putative nitrate-dependent SbOB in this study because of two reasons: (i) they were signi cantly enriched in the heavy fractions of 13 CSbN than their counterparts in 13 CN (Fig. 5); (ii) aioAB genes and denitrifying genes were observed in the Chelativoransassociated bin (Fig. 6), supporting their potential ability for nitrate-dependent SbO. Members of Chelativorans have been extensively identi ed as heterotrophic denitri ers and have been enriched in uranium-contaminated soil [42][43][44]. However, the detection of genes for carbon xation suggested that they hole the potential to oxidize Sb(III) autotrophically.
Metagenomic-binning of the 13 C-heavy-fraction metagenome provides an additional method to examine the physiological traits of the nitrate-dependent SbOB community. Many bins, such as those associated with Thauera, Ramlibacter and Anaeromyxobacter, contained an aioA gene. Although Thauera, Ramlibacter and Anaeromyxobacter have been detected in As-contaminated sites previously [45][46][47], their role in either As(III) or Sb(III) oxidation has not been reported. The presence of aioA and the genes responsible for denitri cation and carbon xation in the bins related with these genera (Fig. 6), suggests that they have the potential for nitrate-dependent SbO.
Relatively higher abundance of Gemmatimonas was observed in the heavy DNA fractions of both 13 CSbN and 13 CN than those in corresponding light fractions. Neither aioA nor aioB, however, was detected while genes involved in denitri cation and carbon xation were observed in the Gemmatimonas-associated bins (bin8). These observations suggested that Gemmatimonas may be more likely autotrophic denitri er without the capability to oxidize Sb(III). In addition, bacteria associated with Halomonas, Geobacter and Pelagibacterium were signi cantly enriched in the heavy fractions in the 13 CSbN treatment than that of 13 CN (Fig. 5). Although Halomonas was identi ed as As(III) oxidizers [48], Geobacter spp. are notable for their capability for metal reduction [49] and Pelagibacterium has never been associated with As or Sb transformation. Unfortunately, bins associated with these three genera were not detected by the metagenomic-binning, thus we cannot determine whether they are potentially nitrate-dependent SbOB.
Further investigation, such as isolation of members of these genera, are necessary to reveal their role in nitrate-dependent SbO.
The current study provided a proof of concept of using DNA-SIP to identify nitrate-dependent SbOB. The long incubation time (60 day) are necessary to observe obvious shift of 13 C-incorporating microbial communities. Because long incubation time may incur cross-feeding [50], shotgun metagenomics followed by DNA-SIP is suggested to provide the physiological traits of the putative nitrate-dependent SbOB and identify the scavenging denitri es or other microorganisms incorporating 13 C from crossfeeding.

Conclusions
Rice is a staple food in China and is a major route (over 30% of the daily intake) for Sb exposure in some Sb mining areas [8], especially when rice is grown close to Sb mines. Since Sb(III) is more e ciently taken up by rice than Sb(V) [8,14], oxidation of Sb(III) in the rice paddies can have health and environmental bene ts by reducing the uptake of Sb by rice. The current study showed that: (i) nitrate-dependent SbO can take place and carried out by the innate microbiota in rice paddy soils; (ii) the aioA gene may be the key gene responsible for nitrate-dependent SbO; (iii) a number of novel putative nitrate-dependent SbOB including bacteria associated with the genera Azoarcus, Azospira, and Chelativorans were identi ed by DNA-SIP.

Declarations
Ethics approval and consent to participate Availability of data and materials The nucleotide sequences generated by amplicon and shotgun metagenome sequencing in this study have been deposited in GenBank database (Bioproject: PRJNA640466).

Figure 1
Transformation of Sb(III) (a) to Sb(V) (b) and NO3-(c) to NO2-(d) in paddy soils amended with Sb(III) and/or NO3-. Sterile controls were performed with soils autoclaved before incubation. Data are shown in mean ± SE (n =3).

Figure 2
Abundance of the aioA gene (mean ± SE (n =3)) in treatments amended with Sb(III) and/or NO3-(a) and the correlation between copy number of the aioA gene and the concentration of Sb(V) in treatment amended with both Sb(III) and NO3-(b) under anoxic conditions.    Relative abundances of genera were shown as the bubble plot (a). Each bubble stands for representative fractions from one culture, and triplicate cultures were sequenced for each treatment. Linear discriminant analysis effect size (LEFSe) showed differentially abundant genera between the heavy fractions from 13CSbN and 13CN treatments (P < 0.05 and LDA score > 2.0) (b).

Figure 6
Counts of genes responsible for Sb(III) oxidation, denitri cation and carbon xation detected in the assembled bins according to shotgun metagenomic sequencing of heavy DNA fractions from the 13CSbN treatment.