Metabolic potential of the imperfect denitrifier Candidatus Desulfobacillus denitrificans in an anammox bioreactor

Abstract The imperfect denitrifier, Candidatus (Ca.) Desulfobacillus denitrificans, which lacks nitric oxide (NO) reductase, frequently appears in anammox bioreactors depending on the operating conditions. We used genomic and metatranscriptomic analyses to evaluate the metabolic potential of Ca. D. denitrificans and deduce its functional relationships to anammox bacteria (i.e., Ca. Brocadia pituitae). Although Ca. D. denitrificans is hypothesized to supply NO to Ca. B. pituitae as a byproduct of imperfect denitrification, this microbe also possesses hydroxylamine oxidoreductase, which catalyzes the oxidation of hydroxylamine to NO and potentially the reverse reaction. Ca. D. denitrificans can use a range of electron donors for denitrification, including aromatic compounds, glucose, sulfur compounds, and hydrogen, but metatranscriptomic analysis suggested that the major electron donors are aromatic compounds, which inhibit anammox activity. The interrelationship between Ca. D. denitirificans and Ca. B. pituitae via the metabolism of aromatic compounds may govern the population balance of both species. Ca. D. denitrificans also has the potential to fix CO2 via an irregular Calvin cycle and couple denitrification to the oxidation of hydrogen and sulfur compounds under chemolithoautotrophic conditions. This metabolic versatility, which suggests a mixotrophic lifestyle, would facilitate the growth of Ca. D. denitrificans in the anammox bioreactor.

frequently been detected in other anammox bioreactors (Bae et al., 2010;Lawson et al., 2017) and its genome showed 99.0%-average nucleotide identity (ANI) to the draft genome of Rhodocyclaceae bacterium UTPRO2 from the metagenome of an anammox bioreactor (Lawson et al., 2017).
Recently, it has been reported that denitrification synergized with anammox could accelerate the anaerobic degradation of benzene, and Rhodocyclaceae bacteria might play a role in benzene degradation (Peng et al., 2017). Although one possible contribution of anammox bacteria could be to remove the nitrite that accumulates as a result of denitrification, benzene and metabolic intermediates such as toluene, phenol and benzoate were found to inhibit anammox activity (Peng et al., 2018). In addition to toxic aromatic compounds, anammox activity is also inhibited by non-toxic organic matter, salts, heavy metals, phosphate, and sulfide, which are commonly present in practical applications such as wastewater treatment (Jin et al., 2012). On the other hand, because 16S rRNA genes with more than 99% identity to that of Ca. N. proteolyticus have also been detected in many other anammox bioreactors (Liu et al., 2009;Park et al., 2017), this aerobic species was inevitably suggested to be an anammox bacterial community (ABC) member responsible for nitrite oxidation via consumption of O 2 in the anammox bioreactor (Okubo et al., 2021). Since Ca. N. proteolyticus possesses multiple secretory lytic enzymes and type II secretion systems, it was suggested that proteolysis of biomass from autolyzed cells and also the lysis of active cells sensitive to lytic enzymes may provide nutrients for itself as well as other heterotrophic members of the ABC (Okubo et al., 2021). Therefore, these predominant bacteria may be important cooperators that help to maintain a balanced population of ABC members and stable anammox activity in the bioreactor. Indeed, cooperative relationships were suggested in cross-feedings of nutrients such as amino acids, carbohydrates, and vitamins, and also in cell aggregation by supplying exopolysaccharides (Lawson et al., 2017;Zhao et al., 2019). However, the lifestyles of major cooperators in the ABC are still unclear, as complete genome sequences of non-isolatable cooperators have not yet been obtained although NO production by incomplete denitrification is not unusual (Schuster & Conrad, 1992). In this study, we performed a detailed genomic analysis of Ca. D. denitrificans and examined the expression profile of its genes to determine why Ca. D. denitrificans is selected as a predominant species in the ABC and how it interacts with anammox bacteria, that is, Ca. B. pituitae.
Among them, the genome sequence of Ca. D. denitrificans was used for detailed analysis in this study. To determine the phylogenetic position of Ca. D. denitrificans, a total of forty complete or draft genome sequences of bacteria, classified mainly into the orders Rhodocyclales, Burkholderiales, and Nitrosomonadales, were obtained from the DDBJ/EMBL/GenBank database.

| Evaluation of the metabolic and physiological potential
The pattern of the metabolic and physiological potential of Ca. D.
denitrificans was investigated using Genomaple TM (formerly MAPLE) ver. 3.2 (Arai et al., 2018;Takami et al., 2016). Genomaple TM is available through a web interface (https://maple.jamst ec.go.jp/maple/ maple -2.3.1/) and as a stand-alone package from Docker Hub (https://hub.docker.com/r/genom aple/genom aple). Genes were mapped to 795 functional modules defined by the KEGG (pathways, 305; complexes, 294; functional sets, 157; and signatures, 40), and the module completion ratio (MCR) was calculated according to a previously described Boolean algebra-like equation (Takami et al., 2012). To evaluate the MCR, Q-values suggesting the working probability of the modules were also calculated by Genomaple TM . Qvalues near zero indicate a high working probability of the module (Takami et al., 2016).

| Analysis of RNA-seq data
To identify the actively working metabolic pathways in the anammox reactor, metatranscriptomic reads obtained in a previous study (Okubo et al., 2021) were mapped to the genome sequences with a cutoff identity of 95% using the Magic-BLAST program (Boratyn et al., 2019). The numbers of mapped reads were counted by SAMtools (Li et al., 2009) and HTSeq (Anders et al., 2015). The RPKM (reads per kilobase of exon per million mapped sequence reads) ratio, calculated by dividing the RPKM of each gene by the mean RPKM of all ribosomal proteins, was used to determine relative gene expression levels. Physiological and biochemical features of hydrogenases were estimated with the HydDB program (Søndergaard et al., 2016).

| Denitrification pathway
Ca. D. denitrificans possesses all genes necessary for denitrification except for NO reductase (i.e., norBC) as shown in Figure 1.
The expression level of nosZ was much higher than that of nirS.
Accordingly, it is thought that nitrous oxide (N 2 O) is used as a major electron acceptor in Ca. D. denitrificans; however, unexpectedly Ca.
D. denitrificans lacks norBC, which encodes the enzyme responsible for the reduction of NO to N 2 O. Thus, we explored the possibility that N 2 O is produced by alternative enzymes. It is known that nitric oxide reductase is structurally similar to cytochrome oxidase (Zumft, 1997) and indeed, the cytochrome cbb 3 -type oxidase of Pseudomonas stutzeri is known to have nitric oxide reductase activity (Forte et al., 2001). Although Ca. D. denitrificans possesses two genes encoding a cytochrome cbb 3 -type oxidase (  (Gardner et al., 2003), but Ca. D. denitrificans has no genes encoding this enzyme (norVW). These results suggest that Ca. D. denitrificans does not utilize self-produced N 2 O, but presumably it can use N 2 O supplied by other community members because norBC genes from minor community members were detected in the anammox bioreactor (Okubo et al., 2021) although alternative ways of producing N 2 O have been reported in other heterotrophic nitrifying bacteria (Zhang et al., 2015), aerobic ammonium oxidizing bacteria (AOB) (Caranto et al., 2016). On the other hand, since the nosZ gene is highly expressed, Ca. D. denitrificans is thought to reduce N 2 O emissions from the anammox bioreactor. It has been reported that only 0.0037% of the total nitrogen load in the anammox reactor was emitted as N 2 O even though N 2 O was detected within anammox granules (Rathnayake et al., 2018).
N 2 O has a greenhouse effect more than 300 times that of carbon dioxide on a 100-y timescale and it also depletes the ozone layer (Solomon et al., 2007).
The haoA (DSYM_27790) gene product (i.e., hydroxylamine oxidoreductase) from Ca. D. denitrificans clustered with those of aerobic AOB in the genera Nitrosospira and Nitrosomonas ( Figure A1). The haoB (DSYM_27800) gene, encoding another whose functional role remains to be elucidated, is adjacent to haoA and the genes encoding cytochrome c 554 (DSYM_27810) and cytochrome c (DSYM_27820), proteins that would relay the electrons to the quinone pool. Among these neighboring genes, only haoA and haoB were expressed ( Figure 1). Although the Hao protein catalyzes the oxidation of hydroxylamine to NO (Caranto & Lancaster, 2017) in aerobic AOB, it has also been reported to catalyze the reduction of NO to hydroxylamine and ammonium, which are substrates for anammox bacteria (Kostera et al., 2010). Accordingly, Ca. D. denitrificans is thought to provide not only NO, but also hydroxylamine and ammonium to anammox bacteria. However, because the reaction pathway catalyzed by the Hao protein is still unclear, further study on the role of this protein in Ca. D. denitrificans is required to fully understand the nitrogen flow in the anammox community.
F I G U R E 1 Nitrogen metabolism pathway. Values in parentheses represent RPKM ratios. The dotted arrow shows the missing process. By-products though imperfect denitrification such as NO, NH 2 OH, and NH 4 + are thought to be supplied to anammox bacteria  Table 1). Benzoyl-CoA is a key in the anaerobic degradation of many aromatic compounds (Fuchs et al., 2011). Expression of the genes encoding the conversion of 3-hydroxypimeloyl-CoA to acetyl-CoA (step 5-10 in Figure 2) was also observed except for step 6. The acetyl-CoA generated through this pathway would be used in the TCA cycle and glyoxylate cycle for not only ATP and NADH production but also carbon assimilation (Table 1) Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 7 Step 8 Step 9 Step 10 the reduction of NAD + , providing reducing equivalents in the form of NADH (Maia et al., 2015). The genes encoding this enzyme were organized in the order γ-, β-, α-and δ-subunits (DSYM_25570-600), but the gene encoding the accessory protein (fdsC) was missing. In addition, only the α-subunit gene was expressed with an RPKM ratio of 0.1.

| Inorganic compound metabolism
Because inorganic compounds can act as electron donors for denitrification (Capua et al., 2017), we search for genes that would suggest potential inorganic electron donors that could drive denitrification. Ca. D. denitrificans possesses two sox gene clusters, soxXAYZ (DSYM_07930-60) and soxYZAXB (DSYM_23290-330) ( Table 2), but soxCD was not found. The reaction catalyzed by the Sox system in the absence of SoxCD proteins is considered to convert thiosulfate to elemental sulfur (S0) or polysulfide, which produces 2 mol of electrons per mol of thiosulfate (Friedrich et al., 2005).  TA B L E 1 Module completion ratios of carbon metabolism pathways enzymes are phylogenetically distinct (Müller et al., 2015), a phylogenetic analysis was performed to determine its type. Because the dsrAB genes from Ca. D. denitrificans clustered with oxidative type enzymes ( Figure A3), Ca. D. denitrificans is predicted to oxidize sulfide into sulfate via elemental sulfur (S0), sulfite, and adenosine 5′-phosphosulfate (Kappler & Dahl, 2001;Russ et al., 2014). The oxidation of sulfur compounds can be coupled with denitrification (Chung et al., 2014;Russ et al., 2014) and these reactions can also be catalyzed by enzymes encoded by cysDN and sat, fccAB, and aprAB, respectively (Table 2).
Thus, Ca. D. denitrificans may also contribute to the removal of sulfide, which is known to inhibit anammox activity (Russ et al., 2014). lacks the gene encoding sedoheptulose-bisphosphatase, which catalyzes the 9th reaction step in the 11-step reductive pentose phosphate cycle (Calvin cycle), the completion ratio of this module was 90.9% (Table 1). This missing enzyme is considered to be unique to the Calvin cycle (Atomi, 2002). On the other hand, it was recently shown that transaldolase (EC 2.2.1.2) can substitute for sedoheptulosebisphosphatase and sedoheptulose-1,7-bisphosphate aldolase. Ca.
Therefore, Ca. D. denitrificans presumably has the potential to carry out CO 2 fixation via this irregular Calvin cycle using transaldolase instead of sedoheptulose-bisphosphatase. The fixed carbon seems to be used for the biosynthesis of amino acids ( Figure A4), nucleotides, and sugars (Table 1). Expression of the genes for nine of the ten steps of the irregular Calvin cycle was observed, including genes encoding ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) (DSYM_24900) and phosphoribulokinase (DSYM_24930) ( Figure 5).
Considering the limited number of metatranscriptomic reads that   Thus, it appears that these species assigned as Sterolibacteriaceae species in the Rhodocyclales cluster are misidentified and should be reclassified as members of the order Rhodocyclales. Although Ca. D.
denitrificans is seemingly close to species in the Sterolibacteriaceae cluster, Ca. D. denitrificans is phylogenetically distant from Sterolibacteriaceae species due to the low bootstrap value of 50% ( Figure A5). Therefore, Ca. D. denitrificans is presumed to be a member of a new family within the order Rhodocyclales.

| CON CLUS ION
Through a series of analyses, it was found that Ca. D. denitrificans has versatile potential to exploit various compounds such as aromatic compounds, glucose, sulfur compounds, and hydrogen as electron donors for denitrification, but the most favorable compounds were aromatics, which inhibit anammox. In addition, Ca. D.
denitrificans also possessed hydroxylamine oxidoreductase, which catalyzes the oxidation of hydroxylamine to NO and potentially the reverse reaction, and the potential for CO 2 fixation via an irregular Calvin cycle, implying mixotrophic potential. Thus, we revealed the metabolic versatility that may facilitate the colonization of Ca. D. denitrificans in the anammox bioreactor. Our findings will not only boost our understanding of the functional relationships between incomplete denitrifiers and anammox bacteria, but also a potential isolation strategy for non-isolatable anammox community members.

ACK N OWLED G M ENTS
We thank Professors Y. Suwa of Chuo University and M. Kuroiwa of Tokyo University of Agriculture and Technology for providing us useful information on the conditions of anammox bioreactor.

This work was supported by KAKENHI Grants-in-Aid for Scientific
Research to H.T. (17H00793 and 15KT0039).

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
RNA sequence (RNA-seq) data for the anammox bacterial commu-