Resuscitation of inactive ammonia-oxidizing archaea and complete nitriers by extracellular electrons from a heterotrophic bacterium, Bacillus amyloliquefaciens

Background: Nitrication in the soil is dominated by ammonia-oxidizing archaea (AOA), ammonia-oxidizing bacteria (AOB) and the newly discovered complete nitriers, which are members of complex communities. Complicated relations have been found in microbial communities, but the relations between heterotrophic bacteria and autotrophic ammonia oxidizers (AOM) are unclear, and experimental models are lacking. Here, we constructed a microcosm based on DNA stable isotope probing that was established between a heterotrophic bacterium, Bacillus amyloliquefaciens (BA), and the soil autotrophic AOM. Results: The inoculation of BA changed the activities of three indigenous nitriers. The originally marginal AOA and complete nitriers were active in the nitrication process as well as the AOB, which led to mitigation of N 2 O and an increase in NO. The network analysis showed that the inoculated BA was indirectly and positively linked to AOA through a Microbacteriaceae that contains genes encoding proteins responsible for transporting electrons. Further, the microbial fuel cell system indirectly conrmed the potential regulation of extracellular electron transfer (EET) between these two species. Conclusions: With our ndings, a useful model to investigate the relations between heterotrophic bacteria and AOM was constructed, and evidence of EET in complex regulations was provided.

In our previous work, a strain of Bacillus amyloliquefaciens (BA) that signi cantly reduced N 2 O emissions by 35%-50% after inoculation was isolated from rice rhizosphere, and the results suggested that the mitigation was mostly due to the inhibition of nitri cation [12]. However, the detailed process of inhibition and relations of organisms behind the regulation of BA have not been studied, especially those between heterotrophic BA and three groups of autotrophic AOM.
AOA amoA genes are abundant and widespread in soils, frequently outnumbering AOB amoA genes [15], and comammox bacteria are also abundant in some soils [16,17]. The relative abundance and dominance of these three types of ammonia-oxidizing microorganisms (AOM) are affected by abiotic factors, including pH, temperature and moisture content [18], and by several nitri cation inhibitors. AOM may also be in uenced by interactions with other microorganisms, e.g., through cross-feeding and quorum sensing [19,20], and by direct or indirect electron transfer between species. The latter is rarely studied and can be investigated using microbial fuel cells (MFCs). Geobacter sulfurreducens and Shewanella oneidensis are the two most commonly used model strains in MFCs because of their good e ciency in transferring electrons; in MFCs, protons are the only thing allowed to pass through the membrane between the anode and cathode chambers and complete the circuit [21].
In this study, we constructed microcosms to investigate the in uence of the heterotrophic bacterium BA on nitri cation and three different types of nitri ers. We also constructed MFCs to study the possible extracellular electron transfer between BA and AOA. Together, these results suggest a complex regulation strategy in which heterotrophic bacteria affect autotrophic organisms, and extracellular electron transfer may be a possible link between the two types of microbes.

Methods
Soil microcosms and stable isotope probing (SIP) The in uence of BA on nitri cation was investigated in soil microcosms consisting of 10 g of sieved acidic soil (mesh size of 3.35 mm) from Southwest China used in our previous work [12]. Microcosms were adjusted to 60% maximum water-holding capacity in 120-ml serum bottles, sealed with rubber stoppers and aluminium caps and incubated at 30 °C in the dark for 4 weeks. The headspace was sampled and adjusted to 5% (vol/vol) CO 2 by injection through the rubber septum, and each bottle was ushed for 5 min weekly with synthetic air (20% O 2 , 80% N 2 ) between sampling and injection. For every week, 100 µg urea-N g − 1 dry weight soil was added by liquid to each bottle.
Three treatments were established, each in triplicate microcosms. In Treatment 12C-B, the headspace contained 12 C-CO 2, and active BA was inoculated. The inoculum was grown in liquid batch culture to stationary phase in M9 Minimal Medium ( 13 C-glucose was used in treatments of 13 C-CO 2 to avoid additional 12 C into the microcosms) at 30 °C and 180 rpm for 28 h. Cells were separated by centrifugation at 8000 rpm and washed three times with sterilized water to remove the remaining culture medium. The nal cell suspension was adjusted to 3 × 10 8 CFU ml − 1 , and 1 ml of this suspension was Page 4/21 added weekly by injection through the rubber septum into each microcosm for 3 consecutive weeks. Treatment 13C-B was identical to Treatment 12C-B, except that the headspace contained 13 C-CO 2 .
Treatment 13C-BS was identical to Treatment 13C-B but was inoculated with B. amyloliquefaciens sterilized by autoclaving as a control.
Soil chemical analysis Soil pH was determined in water at a ratio of 1:2.5 (w/v) using a pH meter (Accumet Excel XL 60, Fisher Scienti c, Singapore). The NO 3 − and NH 4 + concentrations were determined after extraction in 2 M KCl (soil:solution ratio of 1:10 (w/v)) and measured using a continuous ow analyser (SAN++, Skalar, Breda, Holland deionized, nuclease-free water [24].

Quantitative PCR
Functional marker genes (bacterial and archaeal amoA) and amoA of Nitrospira clade A were quanti ed using Premix Ex Taq (TaKaRa, Japan) and gene-speci c primers. The primers used for Nitrospira clade A were coma-244f/coma-659r. Details of plasmid standards, gene-speci c qPCR primers, reaction mixtures and thermal programmes are described in previous studies [17,25]. Each sample was quanti ed in duplicate using the CFX Connect™ Real-Time PCR Detection System (Bio-Rad laboratories). The algorithm that was used to calculate the e ciency (E) and threshold cycle (CT) was based on the kinetics of each individual reaction.
Thermal cycling consisted of an initial denaturation at 95 °C for 30 s, followed by 40 cycles of denaturation at 95 °C for 5 s and annealing and extension at 60 °C for 34 s.

High-throughput sequencing
The heavy fractions (considered as the active species including ammonia oxidizers and their mostly closely linked species incorporating 13 C into their genomes during the DNA-SIP incubation) of Treatment 13C-B were sent for 16S rRNA sequencing of both bacteria and archaea. To construct a network to investigate the niches of AOA and AOB, we used universal primers for bacteria and archaea. The V3-V4 region of the bacterial 16S rRNA gene was ampli ed using the primers 341F (5'-CCTAYGGGRBGCASCAG − 3') and 806R (5'-GGACTACNNGGGTATCTAAT − 3') [27]. The V4 region of the archaeal 16S rRNA gene was ampli ed using the primers 524F10extF (5'-TGYCAGCCGCCGCGGTAA-3') and Arch958RmodR (5'-YCCGGCGTTGAVTCCAATT-3') [28]. Fragments from each soil sample were sequenced using MiSEq. Following gene ampli cation, 3 µl of the PCR product was used for agarose gel electrophoresis (1%) to con rm the results of ampli cation. PCRs of 4 replicates for each sample preparation were combined and quanti ed with PicoGreen. From each sample, 200 ng of the PCR product was collected and pooled with other samples for one sequencing run. The pooled mixture was puri ed with a QIAquick Gel Extraction Kit (QIAGEN Sciences, Germantown, MD, USA) and re-quanti ed with PicoGreen. The puri ed mixture was then diluted and loaded as described in the MiSeq Reagent Kit Preparation Guide (Illumina, San Diego, CA, USA).

Construction of a microbial fuel cell (MFC) system
A MFC system was constructed to assess possible extracellular electron transfer between heterotrophic BA and autotrophic AOM. Two-bottle MFC reactors (total volume of each bottle was 100 mL) were constructed as described by McAnulty in 2017 [21]. The anode side was the soil used in this experiment, and the cathode side was subjected to 3 different treatments: BA, sterile BA as the negative control and Shewanella oneidensis MR-1 as the positive control. The MFC systems were incubated at 30 °C in the dark for 14 days.

Data analysis
After assigning each sequence to the appropriate sample according to its barcode and allowing up to two mismatches, a total of 30,000 reads from both ends were obtained as a partitioned run for each sample.
All the OTUs belonging to ammonia-oxidizers were picked up from the bacterial and archaeal OTU The nitrate concentration in Treatments 13C-B and 12C-B was signi cantly (P < 0.05) lower than that in 13C-BS after incubation for 28 days. The nitric oxide concentration in 13C-B and 12C-B was signi cantly (P < 0.05) higher than that in 13C-BS. The N 2 O concentration in 13C-B and 12C-B was signi cantly (P < 0.05) higher than that in 13C-BS. In contrast, the CO 2 concentration decreased after inoculation of microcosms with active BA. The CO 2 concentration in 13C-B and 12C-B was signi cantly (P < 0.05) higher than that in 13C-BS. These factors all lead to the potential inhibition of the nitri cation process by BA.
However, NH 4 + in Treatments 13C-B and 12C-B was also less than that in Treatment 13C-BS. The ammonium concentration in Treatments 13C-B and 12C-B was signi cantly (P < 0.01) lower than that in Treatment 13C-BS.

DNA-SIP
To identify which ammonia oxidizers incorporated 13 CO 2 during incubation, both archaeal and bacterial amoA genes were quanti ed in DNA from different fractions ( Fig. 2a and b). Sequencing of 16S rRNA genes in the heavy fractions indicated that neither the AOA nor AOB were the most abundant species in the 13 C-labelled 16S rRNA-based communities. Within Thaumarchaeota, 93.6% of the 13 C-labelled AOA community fell within the Soil Crenarchaeotic Group and 1.5% within South African Gold Mine Gp 1. Within the bacteria, 80% of the AOB and NOB sequences fell within the Nitrospira group, while few traditional AOB, e.g., Nitrosococcus, or NOB, e.g., Nitrobacter, were observed in labelled DNA ( Fig. 2c and d).
Activity of comammox after stimulation by BA Given the relatively high proportion of Nitrospira 16S rRNA sequences in 13 C-labelled DNA, speci c amoA primers for comammox were quanti ed in DNA from heavy fractions. This analysis indicated that comammox, especially Nitrospira clade A, was active in our BA-induced systems (Fig. 3). The relative abundance of Nitrospira clade A was greatest around a buoyant density of ~ 1.71-1.72 g ml − 1 in Treatments 12C-B and 13C-BS and at ~ 1.73-1.75 g ml − 1 in Treatment 13C-B. Moreover, we ampli ed clade B at rst but did not obtain any positive results.
Network of the microcosms In order to elucidate the niche of BA after inoculation into the microcosms, we constructed a network of microorganisms, including archaea and bacteria (Fig. 4). The network showed that both nitri ers and BA were not heavily connected to other organisms because of their small size in the whole community. However, BA and Thaumarchaeota were linked indirectly by Microbacteriaceae, and the edge indicated positive relations between both BA-Microbacteriaceae and Microbacteriaceae-Thaumarchaeota.
AOA in the MFC system To further investigate whether AOA provides positive feedback to electron transfer from the outside, we constructed a MFC system, and the abundance of AOA was quanti ed by Q-PCR. The abundance of AOA in the presence of BA was signi cantly higher than that in Treatment SBA (Fig. 5), but it was not signi cantly higher between Shewanella and SBA. The abundance of amoA in Treatment C was 2.62 × 10 5 g − 1 dry weight soil, which was much lower than that in the other three treatments with inoculation of both active and sterile microbes. The abundance of AOA in Treatment BA was 6.65 × 10 5 g − 1 dry weight soil, which was signi cantly higher than that in the presence of Shewanella and Sterile BA (6.08 × 10 5 (P < 0.05) and 5.71 × 10 5 (P < 0.05), respectively).

Discussion
This study suggests that the three autotrophic AOM, AOA, AOB and comammox bacteria Nitrospira clade A, were active in ammonia oxidation. The system involved a heterotrophic bacterium (HB), BA, which suggested a special relation between the HB and autotrophic AOM in the soil to be con rmed in the future. This provides compelling evidence that the regulation of the extraneous HB on the nitri cation process involved the enhancement of oxidation of NH 2 OH to NO through AOA and Nitrospira clade A and the inhibition of the production of N 2 O through AOB, leading to a signi cant increase in NO and mitigation of N 2 O in the nitri cation process. It also provides evidence that the potential regulating strategy is extracellular electron transfer between HB and AOM, especially AOA.
Previous studies have suggested a complex relation between heterotrophic bacteria and autotrophic AOM, including competition for NH 4 + and cooperation where organic carbon is low [29][30][31]. There are different ways microbes interact, e.g., cross-feeding, co-metabolism, cell-to-cell communication [20,29,32] and interspecies extracellular electron transfer (IEET) [33]. In our constructed microcosms, we found the co-activity of AOA, AOB and the newly discovered comammox in the ammonia-oxidizing process. After BA was inoculated into the microcosms, it successfully resuscitated the inactive ammonia-oxidizing archaea and complete nitri er Nitrospira in the soil, which was previously dominated by ammoniaoxidizing bacteria [15,34].
Nitri cation is a two-step process: [14] [36]. The second is nitri er-denitri cation, an enzymatic process in which NO 2 − is reduced to N 2 O via NO; reduction of NO to N 2 O is catalysed by two classes of cytochrome c nitric oxide reductases (NORs). Chemodenitri cation is associated with ammonia oxidation by ammonia-oxidizing bacteria (AOB), ammonia-oxidizing archaea (AOA) and comammox, while nitri er denitri cation has only been reported in AOB [1,3,4,8,37], and all genome-sequenced AOA and comammox lack NOR [1,5,38,39]. The inoculation of BA and the change in AOM status in situ obviously changed the rate of nitri cation. The consumption of CO 2 and production of N 2 O and NO 3 − were lower in Treatment 13C-B compared to Treatment 13C-BS, indicating that BA inhibited the nitri cation process [40][41][42][43]. However, NH 4 + remained in the soil, and the higher emissions of NO in Treatment 13C-B after inoculation did not support the inhibition of nitri cation. The concentrations of NH 4 + in Treatment 13C-B were less than those in Treatment 13C-BS, and the concentration of NO was slightly higher in 13C-B, which led to an enhancement of the ammonia-oxidizing process. The analysis of ammonia-oxidizing microorganisms in this system helped us to resolve this problem. Our soil used in this work was agricultural soils collected from a typical acidic soil area in China, in which AOB play a dominant role, as previous works have suggested [34,44]. The results of SIP in Treatment 13C-BS con rmed the dominance of AOB in our experimental soils. However, the results also suggested that both AOA and Nitrospira clade A become other active organisms that oxidize NH 3 to NH 2 OH in Treatment 13C-B. This indicated that the initially inactive and feeble AOA in agricultural soils was resuscitated by the inoculation of BA and began to be active in function with AOB in situ.
The labelled AOA in Treatment 13C-B mainly belong to Soil Crenarchaeotic Group, which indicates their dominant function in the ammonia-oxidizing process. After that, the AOB in Treatment 13C-B were sequenced, and we found that only a few AOB were classi ed, including Nitrosospira and Nitrosococcus. We wondered if any other species participated in this process, especially the newly isolated comammox.
Interestingly, the results of Q-PCR of comammox showed that Nitrospira clade A was labelled in the heavy fractions. Taken together, after the inoculation of BA into the acidic soils, the whole ammonia-oxidizing process was rede ned; instead of the sole activity of AOB as usual, both AOA and comammox were more active and made use of NH 3 .
The results further revealed the abnormal increase in the production of NO in Treatment 13C-B. As we have mentioned before, NO is an important intermediate in the nitri cation process both in biotic and abiotic pathways [1,4,8,45]. However, in almost all the AOA and comammox genomes sequenced to date, no canonical nitric oxide reductases (NORs) have been detected, despite the wide presence of a nitrite reductase gene (nirK) in AOA. Although cytochrome P450 and other enzymes possibly involved in the production of N 2 O and acting as NOR may be detected in the future, it can only take action in some AOA in the presence of excess nitrite [1,8,46]. After the inoculation of BA and the recovery of AOA and comammox in the soil, more NO was produced in the microcosms (Fig. 6a).
The results and relations between the inoculated BA and ammonia-oxidizing microorganisms indicate a complex competence and cooperation among these species. BA is a widely isolated heterotrophic bacterium that holds a good NH 4 + a nity in oligotrophic environments, and the inoculation of BA consumes NH 4 + in soil, which makes it more competitive for nitri cation by AOM [47]. This in turn provides opportunities for AOA and comammox, whose a nity for ammonium is better than AOB [1,38,39,47,48], and makes the co-function of AOM in soils possible [49]. In addition, the inoculation of BA into the soils changed the original niche in situ, and new relations between BA and the ammonia-oxidizers were constructed. The network showed that BA and the AOA Thaumarchaeota are at the edge of the network, and they are indirectly and positively connected by a Microbacteriaceae.
Both BA and Microbacteriaceae contain genes encoding multiheme cytochrome c, which is good for the transport of electrons and facilitates ammonia oxidation, especially AOA [50][51][52]. To date, no cytochrome c has been found in AOA compared to its wide expression in AOB and comammox [53][54][55]. Previous studies have con rmed the in uence of IEET on anaerobic methane oxidizing microbes [33], which is similar to ammonia-oxidizing microbes. Therefore, we constructed a MFC system to investigate the potential IEET between BA and AOA. Although the MFC systems were carried out for only 14 days, the abundance of AOA in Treatment BA was signi cantly higher than that in the control. This indicated that the electrons transferred from BA were as high as the typical Shewanella at the cathode side, and the electrons may play a positive role in the abundance of AOA in the soil. Of course, this is only preliminary evidence for the IEET between these two species, but the difference between the 3 treatments did provide evidence that AOA reply positively to the electron transferred from other organisms.
Altogether, this suggests that BA is regulated by all autotrophic AOM in acidic soils via the competence of NH 4 + (competence resource) and the facilitation of electron transport (energy supplying), a hypothesis that warrants further experimental elucidation (Fig. 6b).

Conclusions
This is a novel discovery of the resuscitation of AOA and newly found complete nitri ers under the originally dominant AOB in the ammonia-oxidizing process. The results were detected based on the complex interactions between microbes. In our work, after the inoculation of BA, the whole nitri cation in soil was changed, the ammonia-oxidizing process was inhibited; in addition, the emission of N 2 O was much lower, and the production of NO increased owing to the lack of NOR in AOA and comammox. We provide a possible explanation for the special relationship between BA and AOM; both the competence of ammonium with AOB and the facilitation of electron transport with AOA may have led to these results. However, far more research needs to be performed, not only to testify and con rm the relations and electron transfer between these microorganisms but also to determine if some other microbes are involved in this process. Additionally, the mechanisms by which BA regulates the ammonia-oxidizing process at metabolic levels should be elucidated.      Network of bacteria and archaea in the heavy layer DNA from the soil incubated with 13CO2 and BA after incubation for 28 days. The red nodes belong to Bacillus, and the black nodes belong to AOA.

Figure 5
Abundances of AOA in the MFC systems. The error bars indicate standard errors of triplicate Q-PCRs.