genomics insights into ecotype formation of ammonia ...sihua.ivyunion.org/qt/genomics insights into...

Genomics insights into ecotype formation ofammonia-oxidizing archaea in the deep ocean

Yong Wang ,1* Jiao-Mei Huang,1,2 Guo-Jie Cui,1,2

Takuro Nunoura,3 Yoshihiro Takaki,3,4 Wen-Li Li,1,2

Jun Li,1 Zhao-Ming Gao,1 Ken Takai,4 Ai-Qun Zhang1

and Ramunas Stepanauskas51Institute of Deep-Sea Science and Engineering,Chinese Academy of Sciences, Sanya, Hainan, China.2University of the Chinese Academy of Sciences,Beijing, China.3Research and Development Center for MarineBiosciences, Japan Agency for Marine-Earth Scienceand Technology (JAMSTEC), 2-15 Natsushima-cho,Yokosuka, 237-0061, Japan.4Department of Subsurface Geobiological Analysis andResearch, Japan Agency for Marine-Earth Science andTechnology (JAMSTEC), 2-15 Natsushima-cho,Yokosuka, 237-0061, Japan.5Bigelow Laboratory for Ocean Sciences, 60 BigelowDrive, East Boothbay, ME, 04544, USA.

Summary

Various lineages of ammonia-oxidizing archaea(AOA) are present in deep waters, but the mecha-nisms that determine ecotype formation are obscure.We studied 18 high-quality genomes of the marinegroup I AOA lineages (alpha, gamma and delta) fromthe Mariana and Ogasawara trenches. The genomesof alpha AOA resembled each other, while those ofgamma and delta lineages were more divergent andhad even undergone insertion of some phage genes.The instability of the gamma and delta AOA genomescould be partially due to the loss of DNA polymeraseB (polB) and methyladenine DNA glycosylase (tag)genes responsible for the repair of point mutations.The alpha AOA genomes harbour genes encoding athrombospondin-like outer membrane structure thatprobably serves as a barrier to gene flow. Moreover,the gamma and alpha AOA lineages rely on vitaminB12-independent MetE and B12-dependent MetH,respectively, for methionine synthesis. The delta AOA

genome contains genes involved in uptake of sugarand peptide perhaps for heterotrophic lifestyle. Ourstudy provides insights into co-occurrence of clado-genesis and anagenesis in the formation of AOA eco-types that perform differently in nitrogen and carboncycling in dark oceans.

Introduction

Nutritional deprivation in deep-sea environments maylimit the productivity of an ecosystem. In addition to recal-citrant materials recycled by deep-sea bacteria, organiccarbon produced by CO2 fixation can be supplied byammonia-oxidizing and nitrite-oxidizing prokaryotes(Pachiadaki et al., 2017). Ammonia-oxidizing archaea(AOA) are ubiquitous in oxic environments on Earth rang-ing from shallow waters to deep-sea zones and soils(Karner et al., 2001; Leininger et al., 2006; Park et al.,2008). They outcompete ammonia-oxidizing bacteria(AOB) due to their higher affinity for ammonia (Martens-Habbena et al., 2009), which probably permits fitnessand wider distribution of AOA in deep oceans comparedwith AOB (Wang et al., 2017). As a major autotrophicinhabitant of deep oceans, AOA represented by MarineGroup I (MGI) Thaumarchaeota are critical for maintain-ing both the carbon and nitrogen cycles in the water col-umn and in sediments of the full-ocean depth (Kato et al.,1997; Stahl and de la Torre, 2012; Nunoura et al., 2018).

Initially assigned to MGI Crenarchaeota, MGI AOAwere shown to include at least three lineages: alpha, betaand gamma (Massana et al., 2000). Now affiliated with anew phylum of Thaumarchaeota, at least five lineages(with epsilon and delta introduced as new lineages) ofwater column AOA have been shown to be distributed atdifferent depths in different environments (Takai et al.,2004; Nunoura et al., 2015). The distribution depth of thebeta lineage is less than 500 m, that of the delta lineageis primarily between 500 m and 2000 m, and that of thealpha lineage is largely restricted to >6000 m; the gammalineage is ubiquitous in the full-ocean depths (Nunouraet al., 2015). The environmental factors and genomic fea-tures that determine the depth stratification pattern of theselineages are presently unknown. Complete genomes ofMGI AOA were obtained for Nitrosopumilus maritimus and

Received 29 August, 2018; revised 18 December, 2018; accepted26 December, 2018. *For correspondence. E-mail [email protected];Tel. (86) 898-88381062; Fax (86) 898-88222506.

© 2018 Society for Applied Microbiology and John Wiley & Sons Ltd.

Environmental Microbiology (2019) 00(00), 00–00 doi:10.1111/1462-2920.14518

https://orcid.org/0000-0003-3595-3867

mailto:[email protected]

‘Candidatus Nitrosopelagicus brevis’ (Walker et al., 2010;Santoro et al., 2015), and this paved the way to under-standing the metabolism of the model marine MGI AOAspecies and its lifestyle of autotrophic growth. Using atransmembrane hydroxylamine: ubiquinone reductasemodule (HURM) complex, AMO and plastocyanin,N. maritimus is able to generate NADH and oxidize ammo-nia to nitrite. During this process, ammonia serves as anelectron donor to generate reducing force for CO2 fixationby the 3-hydroxypropionate/4-hydroxybutyrate (HP/HB)cycle (Walker et al., 2010; Santoro et al., 2015). Ammoniaconcentration/flux seems to be the limiting factor for thesurvival of AOA in deep oceans and may also contribute tothe diversification of their lineages (Sintes et al., 2013).Nevertheless, alternative sources of ammonia, includingurea and glycine, have been suggested by recent geno-mics studies (León-Zayas et al., 2015; Kerou et al., 2016).In the mesopelagic zone of the Mariana Trench, a consid-erable percentage of the microbial community is composedof the MGI AOA (Nunoura et al., 2015), but current dataprovide only a glimpse of the genetic disparities present inthe AOA (León-Zayas et al., 2015; Kerou et al., 2016).Without high-quality genomes, the underlying mechanismsthat resulted in the divergence of the MGI AOA lineagescannot be completely understood. The MGI AOA lineagesmight be ecologically distinct populations or ecotypes, butthe differences in their metabolic potential have not yetbeen studied. According to the theory of microbial specia-tion, new ecotypes are formed by cladogenesis (splitting ofa population into ecologically distinct populations) and byanagenesis (adaptive improvements of a single population)(Koeppel et al., 2013). Whether the five AOA lineagesevolved into various ecotypes through cladogenesis and/oranagenesis is still a question. Some clues may be providedby comparative genomic analysis of AOA lineages.In this study, we obtained full-ocean water samples

from the Mariana and Ogasawara trenches and usedthem to investigate the vertical distribution and genomicsof AOA lineages. Metagenome assembled genomes(MAGs) and single amplified genomes (SAGs) for threeAOA lineages (alpha, gamma and delta) were used in apan-genomic study, the results of which provided hintsregarding the mechanisms that have led to the emer-gence of ecotypes in the lineages.

Results

Ubiquitous ammonia-oxidizing archaea in theMariana trench

Water samples from the full-ocean depth were collectedfrom two trenches in the western Pacific Ocean duringfour cruises (Supporting Information Fig. S1). A total of75 water samples were obtained from the Mariana

Trench, and four were obtained from the OgasawaraTrench (Supporting Information Table S1). Ammonia con-centration, ion concentrations and other environmentalfactors in the waters overlying the Challenger Deep andthe Ogasawara Trench have been measured previously(Nunoura et al., 2015; Kawagucci et al., 2018). Tagsequencing of the 16S rRNA gene for the Mariana sam-ples showed that Thaumarchaeota, represented by MGIAOA, was one of the dominant phyla in the communities,as reported previously (Nunoura et al., 2015). Classifica-tion of operational taxonomic units (OTUs) at 97%similarity identified the four most abundant MGI AOAOTUs in the Mariana waters: OTU272811, OTU31185,OTU294048 and OTU53278. At depths of 500–6500 m,these four OTUs accounted for up to 17% of the retrieved16S rRNA genes (Supporting Information Fig. S2). Thetotal percentage of the AOA decreased in the hadal zone>6500 m, where Marinimicrobia (SAR406) dominated themicrobial community based on the sequencing of 16SrRNA gene amplicons. From 500 m depth to the bottom ofthe Challenger Deep, the relative abundance of AOAOTU272811 correlated negatively with that of OTU31185.AOA OTU294048 and OTU53278 seem to be an abyssallineage. Regarding diversity within the AOA OTUs, furthergrouping of the tag sequencing reads at 1% dissimilarityrevealed that the highest diversity exists within OTU53278(Supporting Information Table S2). For the subtypes inOTU272811, they were more diversified in upper layersbetween 500 m and 2000 m (Supporting InformationTable S2).

To determine the stratified distribution of the AOA line-ages, we obtained 14 metagenomes for waters from theMariana Trench (Supporting Information Table S3) andthen binned 14 MAGs for the MGI AOA. Three metagen-ome assemblies allowed genome binning of two lineages(Table 1 and Supporting Information Fig. S3). A totalof 18 SAGs for MGI AOA were identified among theSAGs from the Ogasawara Trench (Supporting Informa-tion Table S4). The complete isolate genomes ofN. maritimus and ‘Ca. N. brevis’ contain 85 conservedsingle-copy genes (CSCGs) that are universally main-tained in archaeal genomes (Supporting InformationTable S5). Eight of the AOA MAGs also possess thesame number of CSCGs (Table 1). Two AOA MAGs har-boured potentially contaminant contigs and were lessthan 80% complete. The AOA SAGs contain at most77 unique CSCGs, resulting in a completeness of 91.8%.

Major lineages of AOA in the western Pacific Ocean

The 16S rRNA genes were extracted from the MAGs,SAGs and the representative amplicon reads of thefour OTUs and used to infer phylogenetic relationshipswith reference sequences from the five known MGI AOA

© 2018 Society for Applied Microbiology and John Wiley & Sons Ltd., Environmental Microbiology

2 Y. Wang et al.

lineages. The AOA in the Mariana and Ogasawarawaters were phylogenetically associated with the alpha,gamma and delta lineages (Fig. 1). The most abundantAOA at the depth of 500–2000 m, OTU272811, includedthe reads of both gamma and delta lineages, indicatingthat the short 16S rRNA amplicon reads could not beused to distinguish the gamma and delta lineages. Inmost of our metagenomes, both alpha and gamma AOAwere detected (as exemplified in Supporting InformationFig. S3). Note that the contigs of the gamma AOAgenome had a much wider range of %GC content andtetranucleotide frequency (TNF) than those of the alphagenome as shown in the binning process of the T3 L15metagenome (Supporting Information Fig. S3). The alphalineage was distributed in the abyssal and hadal layersand showed high pairwise similarity of the 16S rRNAsequences in samples from the two trenches. Thegamma lineage was identified at all depths and was morediversified, as suggested by the diversity of the gammaAOA OTUs (Supporting Information Table S2). All theSAGs affiliated with the gamma lineage were obtainedfrom waters at a depth of <6000 m, overcoming the prob-lem of misassembly of genomic fragments from differentcells. Using the 16S rRNA genes, only one of our MAGswas grouped with delta MGI AOA; this MAG was abun-dant in the D11 sample from the Mariana 1000-m depth.In the phylogenetic tree, ‘Ca. N. brevis’ was also placedin the beta group composed of species inhabiting<500-m waters. With the exception of the epsilon lineage,

all of the MGI AOA lineages inhabiting the water columnhave complete genomes, SAGs or MAGs at present(Walker et al., 2010; Santoro et al., 2015).

Phylogenetic relationships and pangenomics of the MGIAOA lineages

Average nucleotide identity (ANI) values for the genomesof the alpha AOA, gamma AOA, delta AOA, N. maritimus(alpha lineage) and ‘Ca. N. brevis’ (beta lineage) werecalculated. Within the alpha MGI AOA lineage, pairwiseANIs averaged 99.2% with a small standard deviation of0.41. The SAG oa1 from the Ogasawara Trench alsohighly resembled the alpha AOA MAGs from the MarianaTrench (Fig. 2A). The pairwise ANI values for gammaMGI AOA were scattered over a wide range (85%–98%),largely accounting for the divergence of the gamma MGIAOA obtained from the two trenches. The ANIs betweengamma and delta MGI AOA genomes were narrowly dis-tributed, ranging from 80% to 83% and were at least 10%higher than those between the gamma and alpha line-ages. This indicates that there is a closer relationshipbetween the gamma and delta lineages than between thegamma and alpha lineages. The pairwise ANIs betweenthe lineages coincided with the similarity between homol-ogous regions of the genomes (Supporting InformationFig. S4). The alpha AOA MAG ma7 and the N. maritimusgenome share regions of high similarity (>80%), while thegamma AOA MAG mg3 contains some lineage-specific

Table 1. Summary of AOA MAGs and SAGs.

Genome/sampleDepth(m)

AOAtype Name

Genomesize (bp)

No. ofcontigs

No.of CDSs

TotalCSCGs

UniqueCSCGs

Compl.(%)

Contam.(%)

cG13 505 γ og1 949 320 27 1113 77 76 90.6 0aD11 1000 δ md1 1 279 113 215 1577 78 75 80.5 6.8cF20 2015 γ og2 1 075 201 19 1264 77 77 91.8 1.2aD3 3000 γ mg1 1 373 839 319 1898 74 60 74.9 22.4cN8 5010 γ og3 884 242 86 1099 55 53 67.1 2.4aT1 L11 5080 α ma1 1 273 336 21 1557 87 85 99.1 0.97aD1 5400 α ma2 1 196 229 15 1477 86 85 98.1 0aD1 5400 γ mg2 883 129 154 1224 81 72 78.9 8.09aD17 5900 γ mg3 1 050 488 37 1235 87 85 99.8 2.91aT1C4 6000 α ma3 1 134 945 46 1386 86 82 96.1 5.83aT1 L9 6890 α ma4 1 176 403 13 1412 87 85 96.1 0.97aT3 L1 7100 α ma5 1 267 893 16 1585 86 85 98.1 0aT3 L15 8150 α ma6 1 522 689 44 1967 86 85 100 0.97aT3 L15 8150 γ mg4 1 424 723 155 1740 86 83 92 4.21cC4 9697 α oa1 819 148 47 988 46 46 54.1 0aT3 L14 10 900 α ma7 1 202 873 17 1492 86 85 100 0aT3 L19 10 890 α ma8 1 256 128 6 1532 86 85 100 0aT3 L19 10 890 γ mg5 1 009 376 138 1313 86 77 90.3 9.06bN. mari - α - 1 645 259 1 1910 86 85 100 0.97bCa.N.br <500 β - 1 232 128 1 1434 85 85 99.5 0

a. Metagenome assembled genomes (MAGs) of the AOA from the Mariana Trench (see Supporting Information Table S3 for information onsampling sites).b. Cultivated AOA.c. Single amplified genomes (SAGs) for the AOA isolated from the Ogasawara Trench (only >800 Kbp were retained). N. mari: N. maritimus

(CP000866); Ca.N.br: ‘Ca. Nitrosopelagicus brevis’ (CP007026); CSCGs: conserved single-copy genes; compl.: completeness; contam.:contamination.


Genomics of MGI AOA lineages 3

http://ca.n.br

http://ca.n.br

regions that are not present in other lineages. Thegenetic distance between the MGI AOA lineages was fur-ther displayed using a maximum likelihood (ML) tree builtwith 24 conserved proteins (Fig. 2B). Two clades formedin the tree, one for alpha and another composed of theremaining lineages. The short distance between the newalpha MGI AOA dwelling in the two trenches indicatestheir homogenous genetic background. This presents astrong contrast to the relatively larger distances betweenthe gamma AOA. In this tree, N. maritimus was sepa-rated from these deep-sea alpha AOA, reflecting geno-mic heterogeneity within the alpha lineage. The deltaAOA and ‘Ca. N. brevis’ remained in the clade with thegamma lineage but were distantly associated with thegamma lineage (Fig. 2B).MAGs and SAGs from the alpha and gamma lineages

were aligned to display conserved gene clusters anddivergent genomic contents. The size of the conservedgene cluster was much larger in the alpha AOA pangen-ome than in the gamma (Fig. 3), supporting the genomichomogeneity of the alpha lineage. Moreover, the ratio ofgenes with unknown function was significantly higher inthe gamma lineage than in the alpha (chi-square test;P = 0.012). The MAGs ma6 and ma8 in the alpha lineageharbour a large number of genomic regions that are

specific in the two samples from >10 000 m depth. Thisis ascribed to the fact that the highest number of single-ton gene clusters (352 in ma6 and 380 in ma8) wasobserved in the two alpha MAGs. Most of the genes thatevolved at the bottom of the Challenger Deep are associ-ated with unknown functions. The SAG og1 also containssome specific gene clusters that were perhaps obtainedindependently in the Ogasawara Trench (Fig. 3). In thepangenome of the gamma lineage, a large fraction of thegene clusters is shared by two but not all of the MAGs orSAGs; this is attributable to the genomic heterogeneity ofthis lineage, as indicated in Fig. 3B. The MAGs mg3 andmg4 include some singleton gene clusters in the gammapangenome. In addition, most of the singleton genes ofthe MAG mg4 remain hypothetical.

Prediction of the metabolic profiles of AOA lineages

Metabolic pathways were depicted for the AOA lineagesaccording to the gene annotation results of the MAGsand SAGs (Fig. 4). All the MGI AOA genomes from thisstudy contain the genes necessary for the essentialmetabolism of AOA, including genes encoding the pro-teins essential for ammonia oxidation, the HP/HB cycle,respiration complexes and sulfate assimilation (Fig. 4).

Fig. 1. Maximum-likelihood phyloge-netic tree of 16S rRNA genes.The bootstrap values of the ML treeare indicated by the size of the greydots on the branches. The speciesassociated with blue diamonds had acomplete genome; those with stars ofdifferent colours correspond to agenome bin for the three lineages inthis study; purple dots refer to thosefrom SAGs. The accession numbersof the sequences from this study arelisted in Supporting InformationTable S6.


4 Y. Wang et al.

The functional genes amoABCX for ammonia oxidationwere detected in the AOA MAGs and in some of theSAGs. The deduced AmoA proteins were distributed intothree phylogenetic clades (Supporting InformationFig. S5). Those of the gamma and delta lineages weregrouped with the low-ammonia cluster (LAC) Ba and Bbtypes, respectively, while that of the alpha lineage waslocated in a new branch of type E within the high-ammonia cluster (HAC) (Sintes et al., 2013). AOAdepend on ammonia permease to import ammonia that ispresent in very low concentrations; as expected, theammonia transporter gene (amt) was identified in most ofthe AOA MAGs and SAGs. All the MGI AOA lineagesfrom the Mariana and Ogasawara trenches are likely ableto obtain ammonia by catalysing the cleavage of carba-moyl phosphate, urea, glycine and probably cyanide, aswe identified the genes encoding urease, aminomethyl-transferase (GcvT) and nitrilase (Fig. 4). GcvT catalyses

the cleavage of glycine to CO2 and NH3 with the involve-ment of lipoylprotein (Supporting Information Fig. S6)(Okamura-Ikeda et al., 1987), which is rarely reported inArchaea. The genes that mediate the lipoylprotein cycleand glycine cleavage were not specialized to any of theAOA lineages; instead, they were present in some MAGsof the alpha AOA (oa1, ma1, ma2, ma4, ma6 and ma7),the delta AOA (md1) and the gamma AOA (mg1). Serinemay be derived from the cleavage of peptides by pepti-dase S8, the gene for which was frequently detected ingenomes of the alpha AOA. Choline, betaine and glycineimported via cross-membrane transporters may undergofurther degradation via the lipoylprotein cycle in some ofthe gamma and delta MGI AOA (Supporting InformationFig. S6). However, none of the MGI AOA lineages is ableto catalyse the full pathway of choline/betaine degrada-tion as their genomes do not contain some of the neces-sary genes.

Fig. 2. Average nucleotide identityand phylogenetic tree of concatenatedconserved proteins.The MAGs and SAGs used in theaverage nucleotide identity (ANI) cal-culation (A) are shown in Table 1.N. mari refers to Nitrosopumilus mari-timus; the beta AOA are representedby ‘Ca. N. brevis’. Blue dots indicateMAGs from the Mariana, and red dotsindicate SAGs from the OgasawaraTrench. A total of 24 conserved pro-teins were extracted and concatenatedfor construction of an ML tree (B). Themarked species shown in the treewere collected from the OgasawaraTrench; the other species wereobtained from metagenomes of theMariana Trench.



By comparing the KEGG genes in the MAGs andSAGs, we examined lineage-specific genes for the deep-sea AOA lineages (Fig. 5). Remarkably, the gamma anddelta AOA genomes lack the genes encoding the twosubunits of DNA polymerase B (PolB), which is critical forthe repair of deaminated adenine and cytosine bases(Jozwiakowski et al., 2014; Abellon-Ruiz et al., 2016).There was no tag gene encoding a 3-methyladenineDNA glycosylase in these gamma and delta genomes.DNA point mutations caused by 3-methyladenine can be

repaired by the enzymes encoded by tag and MPG(Wyatt et al., 1999). The identification of the MPG genein all of the AOA genomes might to some extent compen-sate for the loss of the tag gene in the gamma and deltalineages (Fig. 5). The ogg1 gene, which functions in theexcision of 8-oxoguanine (a type of point mutation involv-ing guanine), was identified in all of the AOA lineages.Homologous recombination and double-strand breaksmight be amended by the RadA (Seitz et al., 1998) andMre/Her proteins (Zhang et al., 2008) respectively. Therelevant genes were present in all the AOA SAGs andMAGs. We were unable to find the Pop4 gene (K03538)in the gamma and delta genomes. As a ribonuclease,POP4 can facilitate pre-tRNA cleavage by forming acomplex with RPR2 protein that probably plays a majorrole in pre-tRNA processing (K03540) (Esakova andKrasilnikov, 2010). The genomes of the alpha AOA con-tain both Pop4 and Rpr2 genes. The gamma and deltalineages probably depend solely on RPR2 for pre-tRNAcleavage since most of the MAGs possess the geneencoding RPR2.

The genes that function in vitamin B12 synthesis wereidentified in almost all the AOA MAGs. The proteinencoded by the methionine synthesis gene metH requiresB12 as a coenzyme, whereas methionine synthesis by theprotein encoded by metE is B12-independent (Pejchal andLudwig, 2005). Surprisingly, the gamma AOA appear tobe independent of B12 for methionine synthase, as themetE gene was identified in all their MAGs. In contrast, allof the alpha AOA in the study harboured at least onemetH gene (Fig. 5).

Four of the alpha AOA MAGs lack ilvA and ilvD genesfor the synthesis of branched-chain amino acids (Fig. 5).In contrast, all the alpha AOA genomes in the study pos-sess the pepA gene, which encodes a protein that isinvolved in the cleavage of leucine from periplasmic pep-tides. The released leucines are likely imported directlyinto the cytoplasm with the assistance of a permeaseencoded by the livM gene (K01998 in Fig. 5).

Two types of ATP-binding cassette transporter genesthat encode proteins responsible for the binding of phos-phate compounds, phnD and pstS, were identified. phnD(K02044), which encodes a protein involved in phospho-nate binding, was significantly more abundant in thegamma AOA than in the alpha AOA genomes (two-tailedt-test; P = 0.004). However, pstS (K02040) for phosphatebinding is present in all the alpha AOA genomes. Itappears that these deep-sea AOA lineages are diversi-fied in their utilization of phosphate sources.

The delta AOA MAG in particular was found to carrygenes that are probably related to bioluminescence (Leeand Meighen, 2000). These include luxC (gene ID:2569_3; acyl-CoA reductase), luxE (gene ID: 2569_2;long-chain fatty acid-CoA ligase/acyl-protein synthetase)

Fig. 3. Pangenomics analysis of the alpha and gamma AOA.The predicted genes among MAGs and SAGs from the alpha(A) and gamma (B) AOA, respectively, were compared (see Table 1for names). The outermost circle displays the COGs in the genomes(genes with known functions are shown in red; unknown are ingreen). The brown bar denotes the cluster of COGs for the singleconserved genes (SCGs) that were shared by all the genomes. TheANVIO also provided the number of singleton gene clusters thatwere present in only one of the genomes, the number of gene clus-ters (gene number), the number of genes per Kbp genomic region,and an evaluation of genome completion. The numbers in thebrackets following the items are the maximums among the MAGsand SAGs.


6 Y. Wang et al.

and fadD (gene ID: 2569_1; acyl-CoA synthetase).These genes were also present in an SAG of a deep-seaMGI AOA isolate (NCBI Bioproject accession numberPRJNA248555). The sequence identity of the homo-logues ranged from 51% to 59%, suggesting persistenceof the lux genes in some of deep-sea MGI AOA species.

The delta AOA MAG contains four sets of gtsABCgenes that encode proteins responsible for the uptake ofglucose (Supporting Information Table S7), indicating amixotrophic lifestyle for the delta MGI AOA. The gtsgenes were probably obtained through horizontal genetransfer (HGT) from a Candidatus Pelagibacter that pos-sessed the homologues with the highest sequence iden-tity (80%). The presence of bacterial genes in the deltaAOA genome appears to be a result of horizontal genetransfer between the deep-sea bacteria and archaea(López-García et al., 2015). Our estimate was that about11.7% of all the CDSs of the delta AOA MAG wereobtained by HGT (Supporting Information Table S8).Considering that 56% of the potentially transferred CDSswere located at the ends of the contigs, some of themwere probably a result of misassembly. In contrast, onlyabout 1% and 0.25% of the CDSs were estimated to betransferred through HGT in the MAGs and SAGs of thegamma and alpha AOA respectively.

Different adaptation strategies as indicated by AOAcomparative genomics

The MGI AOA lineages in this study adapted to theirenvironments through different strategies with respect totheir genomic features. The alpha AOA MAGs possess

the cheY gene, which encodes a protein that serves as aregulator of the chemotaxis response to environmentalchanges (Parkinson, 2003), but the gamma and deltagenomes lack this gene (Fig. 5). In the alpha AOAgenomes, the neighbouring gene of cheY encodes thetype II secretion system protein E, whereas methyl-accepting chemotaxis mcp gene and flagella-encodinggenes are absent. The MCP protein initiates the signaltransfer to CheY, which triggers the subsequent move-ment (Parkinson, 2003). Therefore, the cheY, mcp andflagella-encoding genes are typically neighbours, as seenin the shallow-water AOA genomes. Hence, the cheYgene in the alpha lineage probably encodes a signalresponse regulator that functions in signal transduction tostimulate the secretion system rather than motility. Thebeta, gamma and delta AOA lineages lack the cheY gene(Fig. 5). Likewise, an ompA gene (K02368) was identifiedin all of the alpha MGI AOA genomes but not in the otherlineages. Several other genes related to peptidoglycanmetabolism (K00012, K02472, K16881 and K02563) mayalso distinguish the alpha AOA from the other lineages(Fig. 5). Almost all of the deep-sea MGI AOA in this studycan probably generate di-myo-inositol-phosphate (DMIP)as an osmolyte to deal with hydrostatic pressure (Carioet al., 2015), since the functional gene ipct involved inDMIP synthesis was uniquely found in these deep-seaMGI AOA genomes. The alpha AOA genomes in thisstudy probably do not encode the universal stress proteinUspA, which is critical for survival in the stationary phase(Freestone et al., 1997). Expression of UspA is activatedby nutrient starvation and by osmotic shock (Nyströmand Neidhardt, 1992). In contrast, approximately a dozen

Fig. 4. Schematic model of AOA metabolism.The pathways or enzymes marked by numbers are specific to a lineage represented by the alpha, gamma or delta AOA. HURM: hydroxylamine: ubi-quinone reductase module. The functional genes of the pathways and their closest annotated species are listed in Supporting Information Table S7.



uspA genes were identified in the gamma, delta andshallow-water alpha AOA genomes.Numerous phage genes appear to have been integrated

into the AOA genomes of the gamma and delta lineages.Phage proteins such as site-specific phage integrase,sigma factor, capsid assembly protein and phage DNApolymerase (89% identity with the homologue of phageSPO1) were present among the predicted proteins of thegamma and delta AOA. Most of the phage proteins aremost similar to those of deep-sea phages isolated fromthe Mediterranean Sea (López-Pérez et al., 2017). Thephage genes were not present at the ends of contigs andtherefore seem not to be a result of misassembly.

Discussion

In this study, we obtained MAGs and SAGs from twotrenches in the western Pacific. Based on the genomicfeatures and phylogenetic trees of the alpha and gammaAOA lineages, we argue that many ecotypes exist in these

lineages. It is likely that different strategies have contrib-uted to the development of the ecotypes. Pangenomicanalysis revealed extreme genomic stability of the alphaAOA that is probably a result of adaptation to the hadalenvironment. The alpha AOA lineage in this study wasfound to be an abyssal and hadal ecotype, but the factorsthat have determined its niche specification are unclear.The alpha lineage uniquely contains genes related toouter-membrane protein A glycoprotein (OmpA) and throm-bospondin, but its role in the archaea remains unknown.The thrombospondin protein contains five Ca2+-bindingdomains (Supporting Information Fig. S7A) that may reg-ulate the cellular structure for adhesion and evade mac-rophage attach (Dai et al., 2017). Unlike the genesencoding known thrombospondins, the homologues ofthe alpha AOA genes contain several short insertions(Supporting Information Fig. S7B). Previous studies ofthe gram-negative bacterium Pseudomonas aeruginosarevealed the importance of thrombospondin in virulence-related functions such as adhesion and protection

Fig. 5. Lineage-specific KEGG genes.Na: Nitrosopumilus adriaticus; Np: Nitrosopumilus piranensis; Ns: Nitrosopumilus salaria; CNb: ‘Ca. Nitrosopelagicus brevis’. Na, Np and Ns arerepresentative genomes for the alpha MGI AOA lineage from shallow waters. They are reference (ref.) genomes for the comparison of KEGGgene profiles. The deep-sea alpha, gamma and delta MAGs and SAGs were obtained in this study (see Table 1 for MAG and SAG names).


8 Y. Wang et al.

against macrophages (Dai et al., 2017). Whether or notthe alpha AOA gain immunity against phage invasionand genomic stability due to a thrombospondin structureis an interesting question. A larger number of unknowngenes emerged in the hadal alpha AOA genomes. Thesegenes are probably crucial for the survival of the alphaAOA and experienced high selection pressure by theextreme environment. The high stability of the alpha AOAgenomes, together with the presence of some niche-specific novel genes, suggests anagenesis of the alphaAOA ecotypes. By following the deep-sea water current,the alpha AOA could spread in global deep waters. In thepresent study, the MGI AOA lineages from two trenchesof the western Pacific resembled each other in phyloge-netic status and predicted metabolism. Recently, alphaAOA SAGs were obtained from the Puerto Rico Trench(Leon-Zayas et al., 2015); these SAGs showed >98%ANI to the MAGs from this study. The gamma AOA dis-play high genomic plasticity that may endow them with fit-ness at almost full-ocean depths. For the delta AOA, itsgenome harbours some genes for organic carbon uptakeand assimilation. The acquisition of these genes possiblythorough horizontal gene transfer was likely driven by theunstable ammonia concentration in the mesopelagiclayer brought about by various types of sinking organicmatter. It is difficult to estimate the rate of horizontal genetransfer events in a genome. The recently acquiredgenes show abnormal %GC content, TNF and codonusage pattern (Lawrence and Ochman, 2002; Techtmannet al., 2012). In our binning results, the gamma and deltaAOA contigs are more diversified in %GC and TNF thanthe alpha contigs, indicating that the gamma and deltaAOA genomes are more dynamic due to acquisition oflarger amounts of foreign DNA. However, the delta AOAgenome could perhaps reflect a broader ecologicalpotential, a topic for future study. The instability of thegamma and delta AOA genomes provides strong evi-dence for the involvement of cladogenesis in the forma-tion of ecotypes showing high genetic diversity. Given thefull-depth distribution of the gamma lineage, we proposethat different ecotypes have formed in the gamma lineageacross the depth due to allopatric speciation.

The genomic comparison conducted in this studyrevealed the presence of lineage-specific genes andpathways that reflect the genomic divergence of the line-ages under different stresses and to some extent explaintheir vertical distribution in the deep waters. Lack of PolB,which is essential for DNA template stability, is believedto provide the delta and gamma AOA with more geneticvariants due to the resulting accumulation of point muta-tions caused by the deamination of single-stranded DNA;such mutations are a prerequisite for the emergence ofvarious ecotypes at different depths and in differentmicroenvironments. We cannot exclude the possibility

that unknown genes that encode proteins with functionssimilar to those of PolB are present in our MAGs for thegamma and delta AOA. Another DNA repair system,MutSL, which is specialized for the repair of point muta-tions (Busch and DiRuggiero, 2010), was also notdetected in gamma AOA or in the other lineages in thisstudy. Together with the absence of the tag gene forrepair of 3-methyl-adenine mutations, the genomes of thegamma and delta AOA lineages might have experiencedextensive genomic changes. Furthermore, phage inser-tions, which may have brought new genes into theirgenomes, are present. The loss of the Pop4-encodinggene in the gamma and delta AOA is likely to haveresulted in low translation efficiency. Moreover, the lackof extracellular structures (such as thrombospondin in thealpha AOA) in the gamma and delta AOA might facilitategene flow among members of the microbial community.The cellular structures and the molecular machinery pre-sent in the gamma and delta AOA lineages make the for-mation of ecotypes through cladogenesis an easyprocess. The occurrence of genomic variants has alsopermitted a wide distribution of the gamma ecotypes inthe full-ocean depths. In contrast, the alpha AOA lineagewas restricted in the hadal zone and maintained homoge-nous genomes. The low rates of gene flow and mutationof the alpha AOA genomes likely hindered anagenesisand sympatric speciation of the alpha AOA in thetrenches. In this study, the gamma and delta AOA wereshown to employ the MetE, which acts much more slowlythan the MetH, for methionine synthesis. The lack of aB12-dependent metH gene could impose a burden on thetranslational machinery in the gamma and delta AOA(Danchin and Braham, 2017). If true, this may be an indi-cator of an overall low metabolic level of the gamma anddelta AOA. UspA, which was shown in this study to beencoded by the genomes of the gamma and delta AOA,may regulate these ecotypes into dormancy under star-vation conditions. Overall, the metabolic incompletenessof some MGI AOA ecotypes in this study can beexplained by the ‘black queen hypothesis’ (Morris et al.,2012). The loss of some genes, such as those involvedin glycine degradation, might benefit AOA ecotypes withreduced genomic content. The products generated by theleaky functions would be supplied by helpers in thecommunity.

The MGI AOA lineages conduct a critical nitrificationprocess in deep-sea waters. The low ammonia concen-tration in the abyssal layer is a limiting factor for the che-molithoautotrophic life mode of MGI AOA; ammonia isoften present in deep waters at levels below the limit ofdetection (Nunoura et al., 2015). Hence, ammonia per-mease was highly expressed by the AOA, perhaps toendow them with a high affinity for ammonia in the water,as shown in this study. A previous study suggests that



enrichment of organic matter by suspended sedimentsincreases the ammonia flux of the hadal zone and subse-quent ammonia oxidation by the alpha AOA (Nunouraet al., 2015). In this study, we did not detect genesthat support the prevalence of members of the alpha line-age with a heterotrophic lifestyle in the hadal depths.Recently, AOA were argued to be chemolithomixotrophic(Mussmann et al., 2011; Hatzenpichler, 2012). However,recent work has shown that α-keto organic acids such aspyruvate are simply used to detoxify H2O2 (Kim et al.,2016). Our results indicate that some AOA lineages haveevolved to utilize versatile carbon sources and haveprobably developed heterotrophic lifestyles. Furthermore,amino acids such as serine and glycine might be used asammonia, nitrogen and carbon sources by these deep-sea AOA. In previous studies, detrital material was sug-gested to be degraded by deep-sea microorganisms(Lloyd et al., 2013). With the flourishing of the MGI AOAlineages in the deep oceans, biogenic nitrite flux thereinwas sustained by AOA. Such a nitrification process isperhaps a further driving force for inorganic carbon fixa-tion by nitrite oxidizers (Pachiadaki et al., 2017). In return,nitrite oxidizers would probably increase the supply ofammonia in the water by their own degradation of urea,cyanate and amino acids (Pachiadaki et al., 2017).Overall, based on the results obtained in this study, we

propose that there is an association between stratificationpattern, ecotype formation scheme and genomic differ-ences of the AOA lineages in the deep-sea layers. Ourstudy also suggests that these AOA lineages contributeto the carbon and nitrogen cycles in different ways. In situfiltration of the collected water samples in this studymade it possible to conduct a metatranscriptomic studyof different AOA lineages, which may permit a full evalua-tion of their contribution in production of organic carbonfor the deep-sea ecosystem.

Experimental procedures

Sample collection

Water samples were collected from the OgasawaraTrench by R/V Kairei in December 2011 and from thesouthern Mariana Trench by R/V DY37II, Tansuo01 andTansuo03 between June 2016 and March 2017(Supporting Information Fig. S1 and Table S1). Duringthe R/V Kairei KR11-11 cruise, ROV ABISMO was usedby the Japan Agency for Marine-Earth Science and Tech-nology (JAMSTEC) to collect 5-L samples of water inindividual Niskin bottles from depths ranging from 505 mto 9697 m. For water collection from the Mariana Trench,Niskin bottles on a conductivity, temperature and depth(CTD) unit and landers were used to collect 70-L watersamples from depths between 500 m and 6500 m. The

waters were filtered through 0.22 μm polycarbonate mem-branes on board (diameter 45 mm; Whatman, Clifton, NJ,USA). For samples obtained below 6500 m, an in situautomatic apparatus was employed to filter microbes in80 l-300 l of bottom waters (1 m above the seafloor)through a 0.22 μm membrane (diameter 142 mm; What-man, Clifton, NJ, USA) (Supporting Information Table S1and Fig. S8A). The filtration was driven by an electromag-netic pump controlled by the landers. Immediately after thefiltration, ~300 ml RNALater (Ambion, Carlsbad, CA, USA)was injected automatically by the pump into the filtrationchamber to preserve the RNA in the microbial cells(Supporting Information Fig. S8B). The RNALater injectionwas also controlled by the hadal lander (Supporting Infor-mation Fig. S8). For the single-cell genomics studies,water samples from the Ogasawara Trench were storedas described previously (Nunoura et al., 2013). The sam-ples were maintained at −80 �C on board until furtherprocessing.

16S rRNA amplicons and metagenomic sequencing

Genomic DNA was extracted from the filtration membranesusing a Powersoil DNA isolation kit (Qiagen, Hildon, Ger-many). 16S rRNA genes were amplified using the universalprimers 341F (50-CCTAYGGGRBGCASCAG-30) and 802R(50-TACNVGGGTATCTAATCC-30) (Claesson et al., 2010;Klindworth et al., 2013), which target the V3-V4 region of16S rRNA genes. The PCR conditions for 16S rRNAamplification were as follows: initial denaturation at 98 �Cfor 10 s; 30 cycles of denaturation at 98 �C for 10 s,annealing at 50 �C for 15 s and extension at 72 �C for30 s; and a final extension at 72 �C for 5 min. Two PCRreplicates were performed for each sample. After librarypreparation using a TruSeq Nano DNA LT kit (Illumina,San Diego, CA, USA), the DNA samples and 16S rRNAgene amplicons were sequenced on an Illumina MiSeqplatform via 2 × 300 bp paired-end sequencing (Illumina).The amplicons were less than 500 bp in length and thuscould be fully sequenced using the MiSeq platform.

Separation of AOA OTUs for different samples

The raw reads for 16S rRNA gene amplicons were sepa-rated based on the 6-nt barcodes assigned to the differentsamples. Low-quality reads were removed by NGSQCToolkit (v2.3.3) (Patel and Jain, 2012). The quality-filteredpaired-end reads were assembled using PEAR (v0.9.5)(Zhang et al., 2014). The sequenced amplicons were thenanalysed via the QIIME pipeline with the SILVA rRNAdatabase version 128 as a reference (Caporaso et al.,2010). The reads were grouped into OTUs at 97% similar-ity. The OTUs used in the classification of MGI wereextracted for further phylogenetic examination with AOA


10 Y. Wang et al.

references. The percentages of the AOA OTUs in differentreplicates and at different depths were averaged for differ-ent depth ranges, and the standard deviation of the aver-aged values was calculated.

SAG library construction and sequence analyses

Single-cell sorting and whole genome multiple displace-ment amplification (MDA) for the Ogasawara Trenchsamples were conducted in the Bigelow Laboratory Sin-gle Cell Genomics Center (SCGC) as described previ-ously (Swan et al., 2011). One SAG library in a 384-wellplate was constructed for each sample. PCR screeningusing primer sets for archaeal and bacterial SSU rRNAgenes (Swan et al., 2011) was conducted in the BigelowLab and at JAMSTEC. To prepare a template for Sangersequencing in JAMSTEC, 10 cycles of PCR amplifica-tion were conducted to add M13 sequencing primers(forward, 5´-GTTTTCCCAGTCACGAC-30 and reverse,5´-CAGGAAACAGCTATGAC-30) to SSU rRNA primersand metabolic genes. The amplified SSU rRNA andmetabolic genes were then directly sequenced usingM13 primers on a 3730xl genetic analyser (ThermoFisher Scientific, Waltham, MA, USA). In the BigelowLab, the sequencing of selected SAGs was performedusing the Illumina MiSeq platform as previously described(Swan et al., 2011).

De novo genome assembly with SPAdes (v3.6) (Nurket al., 2013) and assembly QC were performed at SCGCas previously described (Stepanauskas et al., 2017). Thispipeline was evaluated for assembly errors using fivemicrobial benchmark cultures with diverse genomic com-plexity and %GC content of DNA; the results indicated60% average genome recovery, an absence of nontargetor undefined bases, and low frequencies of misassem-blies, indels, and mismatches (Stepanauskas et al.,2017). Functional annotation of the obtained SAGs wasperformed using a combination of Prokka (Seemann,2014) and UniProt (Kiefer et al., 2009). For taxonomicassignments, 16S rRNA gene regions were identifiedusing local alignments provided by Blastn (Altschul et al.,1997) against the SILVA reference database (Pruesseet al., 2007). Identified regions were annotated usingBlastn hits for each region against CREST’s (Lanzénet al., 2012) curated SILVA reference database (Pruesseet al., 2007). Taxonomic assignment was based on appli-cation of a reimplementation of CREST’s LCA algorithmto the top 2% (sorted by bit score) of hits per region.

Binning and analyses of MAGs

The quality-filtered metagenomics data were assembledusing SPAdes with default parameters for individualwater samples (Nurk et al., 2013). Coverage of the

contigs was calculated by searching the reads that couldbe mapped to the contigs using Bowtie 2 with default set-tings (Langmead and Salzberg, 2012). Genome binningof AOA lineages was performed by separation of contigswith consistent coverage level and %GC content followedby an examination of TNF by CA analysis (Albertsenet al., 2013). CDSs in the contigs were predicted byProdigal (v2.6.2) (Hyatt et al., 2012). The 162 Pfammodels (Rinke et al., 2013) were searched in 29 genomesfrom 13 archaeal phyla to obtain CSCGs for assessmentof completeness of the AOA genomes. A total of87 archaeal CSCGs were collected from the Pfammodels (Supporting Information Table S2). TheseCSCGs could be identified in at least 27 genomes.According to our survey, known complete AOA genomeshave 86 or 87 CSCGs. The CSCGs in the AOA bins werecounted to provide a preliminary evaluation of the com-pleteness of the MAGs. A further evaluation of the AOAgenomes was conducted using CheckM (v1.0.5) (Parkset al., 2015). The CDSs were annotated against the NCBInr and KEGG databases using Blastx with an e-valuecutoff of 1E-5. In the BLASTP search results, the CDSspossibly obtained through HGT were identified usinginformation of the homologues with the highest BLASTPscore. A name list was created for the MGI AOA. If thespecies of the homologue was out of the list and the simi-larity was higher than >50%, a case of HGT was pre-dicted for the CDS.

Pangenomics analysis

ANI between the genomes was calculated in www.ezbiocloud.net. A pangenomic comparison was con-ducted using Anvio-4 according to the tutorial (Erenet al., 2015). The gene annotation of the pangenome wasperformed using BLASTP against the COG databasereleased in December 2014. Proteins with a minimal simi-larity of 35% to known proteins in the KEGG databasewere collected to display genomic disparity. For the con-served hypothetical genes, functions were inferred from3-dimensional structural patterns predicted by SWISS-MODEL (Waterhouse et al., 2018).

Construction of phylogenetic trees

The 16S rRNA genes of AOA were extracted from meta-genomes, SAGs and MAGs using rRNA_HMM (Huanget al., 2009). They were used to build an ML phylogenetictree with representative reads of major AOA OTUs andreference sequences. The rRNA sequences were firstaligned using MAFFT L-INS-i (v7.294b) (Alva et al., 2016)and then adjusted with trimAl (v1.4) (Capellagutiérrezet al., 2009). The trees were inferred from 1492 alignedpositions by the ML algorithm using raxmlGUI v1.5



http://www.ezbiocloud.net

http://www.ezbiocloud.net

(Silvestro and Michalak, 2012) (GTRGAMMA model). Toobtain bootstrap values, 1000 replicates were performed.An ML phylogenetic tree was also constructed using24 protein sequences. The archaeal conserved proteinswere aligned using MAFFT and concatenated, whichresulted in 3825 alignment positions. PROTGAMMA+BLOSUM62 was selected as the substitution model forthe construction of a phylogenetic tree with bootstrapvalues calculated based on 1000 replicates.

Availability of data

The reads of 16S amplicons were deposited in the NCBISRA database under the accession numbers SRR6466501,SRR6466502, SRR6466503 and SRR6466504. The MAGsfor AOA lineages were deposited in the BIGD database (BigData Center Members, 2017) under the accession numbersSAMC021084-SAMC021097. The BIGD accession num-bers of the four SAGs listed in Table 1 are SAMC026833-SAMC026836.

Acknowledgements

We are thankful to the team members aboard the R/VDY37II, Tansuo01, Tansuo03 and Kairei KR11-11 for theirinvaluable efforts during the sampling cruises. We are alsograteful to J. Chen and D.S. Cai for the operation of landersand for in situ sampling work. This study was supported bythe National Science Foundation of China (No. 41476104and No. 31460001), the Strategic Priority Research ProgramB of the Chinese Academy of Sciences (No. XDB06010201and XDB06040101) and the National Key Research andDevelopment Program of China (2016YFC0302500) toY.W. T.N. was supported in part by a Grant-in-Aid for Scien-tific Research (B) (30070015) from the Japan Society for thePromotion of Science (JSPS). We thank the staff of the Bige-low Laboratory Single-Cell Genomics Center for the genera-tion of single-cell genomics data. The single-cell genomicswork was supported by NSF grants OCE-1232982, OCE-1335810 and EF-0826924 to R.S.

References

Abellon-Ruiz, J., Ishino, S., Ishino, Y., and Connolly, B.A.(2016) Archaeal DNA polymerase-B as a DNA templateguardian: links between polymerases and base/alternativeexcision repair enzymes in handling the deaminatedbases uracil and hypoxanthine. Archaea 2016: 1510938.

Albertsen, M., Hugenholtz, P., Skarshewski, A., Nielsen, K.L., Tyson, G.W., and Nielsen, P.H. (2013) Genomesequences of rare, uncultured bacteria obtained by differ-ential coverage binning of multiple metagenomes. NatBiotechnol 31: 533–538.

Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J.H.,Zhang, Z., Miller, W., and Lipman, D.J. (1997) GappedBLAST and PSI-BLAST: a new generation of protein data-base search programs. Nucleic Acids Res 25: 3389–3402.

Alva, V., Nam, S.Z., Söding, J., and Lupas, A.N. (2016) TheMPI bioinformatics toolkit as an integrative platform foradvanced protein sequence and structure analysis.Nucleic Acids Res 44: W410–W415.

Big Data Center Members. (2017) The BIG data center: fromdeposition to integration to translation. Nucleic Acids Res45: D18–D24.

Busch, C.R., and DiRuggiero, J. (2010) MutS and MutL aredispensable for maintenance of the genomic mutation ratein the halophilic archaeon Halobacterium salinarum NRC-1. PLoS One 5: e9045.

Capellagutiérrez, S., Sillamartínez, J.M., and Gabaldón, T.(2009) trimAl: a tool for automated alignment trimming inlarge-scale phylogenetic analyses. Bioinformatics 25:1972–1973.

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K.,Bushman, F.D., Costello, E.K., et al. (2010) QIIME allowsanalysis of high-throughput community sequencing data.Nat Methods 7: 335–336.

Cario, A., Mizgier, A., Thiel, A., Jebbar, M., and Oger, P.M.(2015) Restoration of the di-myo-inositol-phosphate path-way in the piezo-hyperthermophilic archaeon Thermococ-cus barophilus. Biochimie 118: 286–293.

Claesson, M.J., Wang, Q., O’Sullivan, O., Greenediniz, R.,Cole, J.R., Ross, R.P., and O’Toole, P.W. (2010) Compar-ison of two next-generation sequencing technologies forresolving highly complex microbiota composition usingtandem variable 16S rRNA gene regions. Nucleic AcidsRes 38: e200.

Dai, S., Sun, C., Tan, K., Ye, S., and Zhang, R. (2017) Struc-ture of thrombospondin type 3 repeats in bacterial outermembrane protein a reveals its intra-repeat disulfidebond-dependent calcium-binding capability. Cell Calcium66: 78–89.

Danchin, A., and Braham, S. (2017) Coenzyme B12 synthe-sis as a baseline to study metabolite contribution of animalmicrobiota. J Microbial Biotechnol 10: 688–701.

Eren, A.M., Esen, Ö. C., Quince, C., Vineis, J.H., Morrison,H.G., Sogin, M.L., and Delmont, T.O. (2015) Anvi’o: anadvanced analysis and visualization platform for ‘omicsdata. PeerJ 3: e1319.

Esakova, O., and Krasilnikov, A.S. (2010) Of proteins andRNA: the RNase P/MRP family. RNA 16: 1725–1747.

Freestone, P., Nyström, T., Trinei, M., and Norris, V. (1997) Theuniversal stress protein, UspA, of Escherichia coli is phos-phorylated in response to stasis. JMol Biol 274: 318–324.

Hatzenpichler, R. (2012) Diversity, physiology, and niche dif-ferentiation of ammonia-oxidizing archaea. Appl EnvironMicrobiol 78: 7501–7510.

Huang, Y., Gilna, P., and Li, W. (2009) Identification of ribo-somal RNA genes in metagenomic fragments. Bioinfor-matics 25: 1338–1340.

Hyatt, D., Locascio, P.F., Hauser, L.J., and Uberbacher, E.C. (2012) Gene and translation initiation site prediction inmetagenomic sequences. Bioinformatics 28: 2223–2230.

Jozwiakowski, S.K., Keith, B.J., Gilroy, L., Doherty, A.J., andConnolly, B.A. (2014) An archaeal family-B DNA polymer-ase variant able to replicate past DNA damage: occur-rence of replicative and translesion synthesis polymeraseswithin the B family. Nucleic Acids Res 42: 9949–9963.


12 Y. Wang et al.

Karner, M.B., DeLong, E.F., and Karl, D.M. (2001) Archaealdominance in the mesopelagic zone of the Pacific Ocean.Nature 409: 507–510.

Kato, C., Li, L., Tamaoka, J., and Horikoshi, K. (1997) Molec-ular analyses of the sediment of the 11000-m deep Mari-ana trench. Extremophiles 1: 117–123.

Kawagucci, S., Makabe, A., Kodama, T., Matsui, Y.,Yoshikawa, C., Ono, E., et al. (2018) Hadal water biogeo-chemistry over the Izu-Ogasawara trench observed with afull-depth CTD-CMS. Ocean Sci 14: 575–588.

Kerou, M., Offre, P., Valledor, L., Abby, S.S., Melcher, M.,Nagler, M., et al. (2016) Proteomics and comparativegenomics of Nitrososphaera viennensis reveal the coregenome and adaptations of archaeal ammonia oxidizers.Proc Natl Acad Sci USA 113: E7937–E7946.

Kiefer, F., Arnold, K., Künzli, M., Bordoli, L., and Schwede, T.(2009) The SWISS-MODEL repository and associatedresources. Nucleic Acids Res 37: D387–D392.

Kim, J.-G., Park, S.-J., Sinninghe Damsté, J.S., Schouten,S., Rijpstra, W.I. C., Jung, M.-Y., et al. (2016) Hydrogenperoxide detoxification is a key mechanism for growth ofammonia-oxidizing archaea. Proc Natl Acad Sci USA113: 7888–7893.

Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast,C., Horn, M., and Glöckner, F.O. (2013) Evaluation of gen-eral 16S ribosomal RNA gene PCR primers for classicaland next-generation sequencing-based diversity studies.Nucleic Acids Res 41: e1.

Koeppel, A.F., Wertheim, J.O., Barone, L., Gentile, N.,Krizanc, D., and Cohan, F.M. (2013) Speedy speciation ina bacterial microcosm: new species can arise as fre-quently as adaptations within a species. ISME J 7:1080–1091.

Langmead, B., and Salzberg, S. (2012) Fast gapped-readalignment with bowtie 2. Nat Methods 9: 357–359.

Lanzén, A., Jørgensen, S.L., Huson, D.H., Gorfer, M.,Grindhaug, S.H., Jonassen, I., et al. (2012) CREST - clas-sification resources for environmental sequence tags.PLoS One 7: e49334.

Lawrence, J.G., and Ochman, H. (2002) Reconciling themany faces of lateral gene transfer. Trends Microbiol10: 1–4.

Lee, C.Y., and Meighen, E.A. (2000) Expression and DNAsequence of the gene coding for the lux-specific fatty acyl-CoA reductase from Photobacterium phosphoreum. JMicrobiol 38: 80–87.

Leininger, S., Urich, T., Schloter, M., Schwark, L., Qi, J.,Nicol, G.W., et al. (2006) Archaea predominate amongammonia-oxidizing prokaryotes in soils. Nature 442:806–809.

León-Zayas, R., Novotny, M., Podell, S., Shepard, C.M.,Berkenpas, E., Nikolenko, S., et al. (2015) Single cellswithin the Puerto Rico trench suggest hadal adaptation ofmicrobial lineages. Appl Environ Microbiol 81: 8265–8276.

Lloyd, K.G., Schreiber, L., Petersen, D.G., Kjeldsen, K.U.,Lever, M.A., Steen, A.D., et al. (2013) Predominantarchaea in marine sediments degrade detrital proteins.Nature 496: 215–218.

López-García, P., Zivanovic, Y., Deschamps, P., andMoreira, D. (2015) Bacterial gene import and mesophilicadaptation in archaea. Nat Rev Microbiol 13: 447–456.

López-Pérez, M., Haro-Moreno, J.M., Gonzalez-Serrano, R.,Parras-Moltó, M., and Rodriguez-Valera, F. (2017)Genome diversity of marine phages recovered from Medi-terranean metagenomes: size matters. PLoS Genet 13:e1007018.

Martens-Habbena, W., Berube, P.M., Urakawa, H., de laTorre, J.R., and Stahl, D.A. (2009) Ammonia oxidationkinetics determine niche separation of nitrifying Archaeaand bacteria. Nature 461: 976–979.

Massana, R., DeLong, E.F., and Pedros-Alio, C. (2000) Afew cosmopolitan phylotypes dominate planktonicarchaeal assemblages in widely different oceanic prov-inces. Appl Environ Microbiol 66: 1777–1787.

Morris, J.J., Lenski, R.E., and Zinser, E.R. (2012) The blackqueen hypothesis: evolution of dependencies throughadaptive gene loss. MBio 3: e00036–e00012.

Mussmann, M., Brito, I., Pitcher, A., Sinninghe Damste, J.S.,Hatzenpichler, R., Richter, A., et al. (2011) Thaumarch-aeotes abundant in refinery nitrifying sludges expressamoA but are not obligate autotrophic ammonia oxidizers.Proc Natl Acad Sci USA 108: 16771–16776.

Nunoura, T., Nishizawa, M., Kikuchi, T., Tsubouchi, T., Hirai,M., Koide, O., et al. (2013) Molecular biological and isoto-pic biogeochemical prognoses of the nitrification-drivendynamic microbial nitrogen cycle in hadopelagic sedi-ments. Environ Microbiol 15: 3087–3107.

Nunoura, T., Takaki, Y., Hirai, M., Shimamura, S., Makabe,A., Koide, O., et al. (2015) Hadal biosphere: insight intothe microbial ecosystem in the deepest ocean on earth.Proc Natl Acad Sci USA 112: E1230–E1236.

Nunoura, T., Nishizawa, M., Hirai, M., Shimamura, S.,Harnvoravongchai, P., Koide, O., et al. (2018) Microbialdiversity in sediments from the bottom of the challengerdeep, the Mariana trench. Microbes Environ 33: 186–194.

Nurk, S., Bankevich, A., Antipov, D., Gurevich, A.A.,Korobeynikov, A., Lapidus, A., et al. (2013) Assemblingsingle-cell genomes and mini-metagenomes from chimericMDA products. J Comput Biol 20: 714–737.

Nyström, T., and Neidhardt, F.C. (1992) Cloning, mapping andnucleotide sequencing of a gene encoding a universal stressprotein in Eschericha coli. Mol Microbiol 6: 3187–3198.

Okamura-Ikeda, K., Fujiwara, K., and Motokawa, Y. (1987)Mechanism of the glycine cleavage reaction. Properties ofthe reverse reaction catalyzed by T-protein. J Biol Chem262: 6746–6749.

Pachiadaki, M.G., Sintes, E., Bergauer, K., Brown, J.M.,Record, N.R., Swan, B.K., et al. (2017) Major role ofnitrite-oxidizing bacteria in dark ocean carbon fixation. Sci-ence 358: 1046–1051.

Park, S.J., Park, B.J., and Rhee, S.K. (2008) Comparativeanalysis of archaeal 16S rRNA and amoA genes to esti-mate the abundance and diversity of ammonia-oxidizingarchaea in marine sediments. Extremophiles 12: 605–615.

Parkinson, J.S. (2003) Bacterial chemotaxis: a new player inresponse regulator dephosphorylation. J Bacteriol 185:1492–1494.

Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P.,and Tyson, G.W. (2015) CheckM: assessing the quality ofmicrobial genomes recovered from isolates, single cells,and metagenomes. Genome Res 25: 1043–1055.



Patel, R.K., and Jain, M. (2012) NGS QC toolkit: a toolkit forquality control of next generation sequencing data. PLoSOne 7: e30619.

Pejchal, R., and Ludwig, M.L. (2005) Cobalamin-independentmethionine synthase (MetE): a face-to-face double barrelthat evolved by gene duplication. PLoS Biol 3: e31.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B.M., Ludwig, W.,Peplies, J., and Glockner, F.O. (2007) SILVA: a compre-hensive online resource for quality checked and alignedribosomal RNA sequence data compatible with ARB.Nucleic Acids Res 35: 7188–7196.

Rinke, C., Schwientek, P., Sczyrba, A., Ivanova, N.N.,Anderson, I.J., Cheng, J.F., et al. (2013) Insights into thephylogeny and coding potential of microbial dark matter.Nature 499: 431–437.

Santoro, A.E., Dupont, C.L., Richter, R.A., Craig, M.T.,Carini, P., McIlvin, M.R., et al. (2015) Genomic and prote-omic characterization of "Candidatus Nitrosopelagicusbrevis": an ammonia-oxidizing archaeon from the openocean. Proc Natl Acad Sci USA 112: 1173–1178.

Seemann, T. (2014) Prokka: rapid prokaryotic genome anno-tation. Bioinformatics 30: 2068–2069.

Seitz, E.M., Brockman, J.P., Sandler, S.J., Clark, A.J., andKowalczykowski, S.C. (1998) RadA protein is an archaealRecA protein homolog that catalyzes DNA strandexchange. Genes Develop 12: 1248–1253.

Silvestro, D., and Michalak, I. (2012) raxmlGUI: a graphicalfront-end for RAxML. Organisms Div Evol 12: 335–337.

Sintes, E., Bergauer, K., De Corte, D., Yokokawa, T., andHerndl, G.J. (2013) Archaeal amoA gene diversity pointsto distinct biogeography of ammonia-oxidizing Crenarch-aeota in the ocean. Environ Microbiol 15: 1647–1658.

Stahl, D.A., and de la Torre, J.R. (2012) Physiology anddiversity of ammonia-oxidizing archaea. Annu Rev Micro-biol 66: 83–101.

Stepanauskas, R., Fergusson, E.A., Brown, J., Poulton, N.J.,Tupper, B., Labonté, J.M., et al. (2017) Improved genomerecovery and integrated cell-size analyses of individual,uncultured microbial cells and viral particles. Nat Comm8: 84.

Swan, B.K., Martinez-Garcia, M., Preston, C.M., Sczyrba, A.,Woyke, T., Lamy, D., et al. (2011) Potential for chemo-lithoautotrophy among ubiquitous bacteria lineages in thedark ocean. Science 333: 1296–1300.

Takai, K., Oida, H., Suzuki, Y., Hirayama, H., Nakagawa, S.,Nunoura, T., et al. (2004) Spatial distribution of marinecrenarchaeota group I in the vicinity of deep-sea hydro-thermal systems. Appl Environ Microbiol 70: 2404–2413.

Techtmann, S., Colman, A., Lebedinsky, A., Sokolova, T.,and Robb, F. (2012) Evidence for horizontal gene transferof anaerobic carbon monoxide dehydrogenases. FrontMicrobiol 3: 00132.

Walker, C.B., de la Torre, J.R., Klotz, M.G., Urakawa, H.,Pinel, N., Arp, D.J., et al. (2010) Nitrosopumilus maritimus

genome reveals unique mechanisms for nitrification andautotrophy in globally distributed marine crenarchaea.Proc Natl Acad Sci USA 107: 8818–8123.

Wang, J., Kan, J., Zhang, X., Xia, Z., Zhang, X., Qian, G., etal. (2017) Archaea dominate the ammonia-oxidizing com-munity in deep-sea sediments of the eastern IndianOcean—from the equator to the bay of Bengal. FrontMicrobiol 8: 415.

Waterhouse, A., Bertoni, M., Bienert, S., Studer, G.,Tauriello, G., Gumienny, R., et al. (2018) SWISS-MODEL:homology modelling of protein structures and complexes.Nucleic Acids Res 46: W296–W303.

Wyatt, M.D., Allan, J.M., Lau, A.Y., Ellenberger, T.E., andSamson, L.D. (1999) 3-methyladenine DNA glycosylases:structure, function, and biological importance. Bioessays21: 668–676.

Zhang, J., Kobert, K., Flouri, T., and Stamatakis, A. (2014)PEAR: a fast and accurate Illumina paired-end readmerge R. Bioinformatics 30: 614–620.

Zhang, S., Wei, T., Hou, G., Zhang, C., Liang, P., Ni, J., etal. (2008) Archaeal DNA helicase HerA interacts withMre11 homologue and unwinds blunt-ended double-stranded DNA and recombination intermediates. DNARepair 7: 380–391.

Supporting Information

Additional Supporting Information may be found in the onlineversion of this article at the publisher’s web-site:

Fig. S1. Sampling sites in the western Pacific trenches.Fig. S2. Distribution of AOA lineages along water column.Fig. S3. Genome binning as exemplified by T3L15metagenome.Fig. S4. Distribution of homologous genomic regionsamong representative genomes.Fig. S5. Phylogenetic tree of AmoA proteins.Fig. S6. Betaine and glycine degradation.Fig. S7. Calcium binding domains of thrombospondin-likeprotein identified in alpha AOA genomes.Fig. S8. Schematic of in situ microbial filtration and fixa-tion apparatus.Table S1. Sampling sites.Table S2. Diversity of major AOA OTUs.Table S3. Summary of metagenomes for binning ofMGI AOA.Table S4. Summary of single amplified genomes (SAGs)from Ogasawara Trench.Table S5. CSCGs for Archaea.Table S6. Accession numbers of 16S rRNA genes inFigure 1.


14 Y. Wang et al.

genomics insights into ecotype formation of ammonia ...sihua.ivyunion.org/qt/genomics insights into...

Documents