Functional insight into Maelstrom in the germline piRNA pathway

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2628886/

Abstract

• MAELのconserved residues(Glu-His-His-Cys-His-Cys, EHHCHC) は脊椎動物、ホヤ、昆虫、線虫、原生生物で見つかっている

Maelstrom (MAEL) plays a crucial role in a recently-discovered piRNA pathway; however its specific function remains unknown. Here a novel MAEL-specific domain characterized by a set of conserved residues (Glu-His-His-Cys-His-Cys, EHHCHC) was identified in a broad range of species including vertebrates, sea squirts, insects, nematodes, and protists.

• 硬骨魚類では失われている <- 卵? <- 硬骨魚類から進化した哺乳類で復活した？

It exhibits ancient lineage-specific expansions in several species, however, appears to be lost in all examined teleost fish species.

• MAELドメインの機能は HMG, SR-25-like and HDAC_interactとの相互作用で明らかになりうる

Functional involvement of MAEL domains in DNA- and RNA-related processes was further revealed by its association with HMG, SR-25-like and HDAC_interact domains.

• 全てのMaelドメインが標準的なcanonical RNase H foldをとっていることから、RNase H foldを持つDnaQ-H 3'–5' ヌクレアーゼとの遠い類縁性が発見された。

• 原生生物のMAEL domainsはDEDHDドメインを保存していた。

A distant similarity to the DnaQ-H 3'–5' exonuclease family with the RNase H fold was discovered based on the evidence that all MAEL domains adopt the canonical RNase H fold; and several protist MAEL domains contain the conserved 3'–5' exonuclease active site residues (Asp-Glu-Asp-His-Asp, DEDHD).

• この進化的な繋がりは、構造的研究と合わせて考えると、MAELドメインがpiRNA 生合成に関わる潜在的なnuclease activityまたはRNA-binding abilityを持っているかもしれないという仮設に繋がった。

This evolutionary link together with structural examinations leads to a hypothesis that MAEL domains may have a potential nuclease activity or RNA-binding ability that may be implicated in piRNA biogenesis.

• 祖先であるDnaQ-Hと子孫であるMAEL domainsの間で独特な２残基の移り変わりが観察されたことは、active site switchというタンパク質の進化的なモデルを示唆する。

• active site switch: 「原生生物のMAELホモログは、3'–5' エキソヌクレアーぜとMAEL domainsの特徴を失っているので、進化の中間体である」 <- phase seperationは保持されてる？3'–5' エキソヌクレアーぜはphase seperationする物がある？

The observed transition of two sets of characteristic residues between the ancestral DnaQ-H and the descendent MAEL domains may suggest a new mode for protein function evolution called "active site switch", in which the protist MAEL homologues are the likely evolutionary intermediates due to harboring the specific characteristics of both 3'–5' exonuclease and MAEL domains.

Background

• 生殖細胞はnuage（germ plasm）という特有のorganelleにより異種間で特徴付けられてきた。

• piRNAやrasiRNAsがトランスぽゾン抑制してる

• VASA [11], MAEL [11], SPN-E [12,13], Oskar [14], Tudor domain proteins, Armitage [13], Krimper [11], Cutoff [16], Dead end [17] and Zucchini and Squash [18].のようなgerm plasmタンパク質がある.

• トランスぽゾン抑制タンパクのノックアウトによるspindle-class gene phenotypes: 卵母細胞の極性消滅、Oskar, Gurken and BiocoidにおけるmRNA局在の崩壊、核小体期への進行の失敗

Germline cells among different species are characterized by the presence of a morphologically unique organelle called the germ plasm (also referred to as nuage, polar granules or mitochondrial cloud) [1,2]. This organelle has been considered the determinant of germline development. Very recently a germ plasm-specific small RNA pathway has been identified, in which a new type of small RNAs called PIWI-interacting RNAs (piRNAs) or repeat-associated small interfering RNAs (rasiRNAs) play a role in ensuring the genomic stability of germline cells by silencing certain endogenous genetic elements such as retrotransposons and repetitive sequences [3-8]. Different from short interfering RNAs (siRNAs) and microRNAs which are usually 21–22 nt long, piRNAs or rasiRNAs have longer nucleotide composition (26–31 nt) and 2'O-methyl modification in 3' end. Many germ plasm proteins are functionally important in piRNAs synthesis and function, including PIWI proteins (PIWI, Aubergine and AGO3) [4,9,10], VASA [11], MAEL [11], SPN-E [12,13], Oskar [14], Tudor domain proteins [15], Armitage [13], Krimper [11], Cutoff [16], Dead end [17] and Zucchini and Squash [18]. Their loss-of-function mutations commonly cause a huge reduction in the amount of piRNAs or rasiRNAs and an increase in transcript level of transposable elements in the germline cells [11,19,20] as well as the spindle-class gene phenotypes: failure in establishing anterior/posterior polarity in early oocytes, disrupted asymmetric subcellular mRNA localization of Oskar, Gurken and Biocoid, ectopic expression of Oskar and Gurken, failure to proceed to the karyosome stage [8,11,13,21].

• PIWIはPAZ および PIWIドメインを含むので、single-stranded RNAの認識および配列特異的なtarget nucleotideのendonucleolytic分解をそれぞれ示す、

• VASA, SPN-E, ArmitageはDEAD RNA helicase domainsを含むので、piRNA productionおよびretrotransposon silencingのためのhelicase activitiesを提供する

• Zucchini and SquashはpiRNA maturationに関わるとされるヌクレアーゼ

• Krimper and Tudor proteinsはmultiprotein RNA-induced silencing complex (RISC) や分解時のtargeting substrate RNA recognitionを助けるとされる

• MAEL in piRNA pathway の役割は未知

• MAELは最初にgenetic loss-of-function Drosophila mutantで見つかった。

• 生殖細胞でSPN-E, VASA, Aubergine, Tudor or Krimperの位置がMAELの場所を決める

• MAELはDicer and Argonaute2の場所を決める

• MAEL は germ plasm と the nucleus [21]の間を移動する.

• MAEL と chromatin remodeling proteins SNF5/INI1 and SIN3Bは直接相互作用すると知られている <- 知らなかった

• MAELはgerm plasm や piRNA pathwayをchromatin remodeling（トランスポゾンサイレンシング）と結びつける唯一のタンパク質である <- 知らなかった

• homologous sequence miningやphylogenetic analysis、domain architecture、protein fold recognition, and structure modelingによりMAELの機能を確かめるのが目的。

The molecular functions of most germ plasm proteins in the piRNA pathway have been assigned based on domain examination, biochemical and genetic characterizations. For instance, PIWI proteins contain the PAZ and PIWI domains, which contribute to recognition of single-stranded RNA [22] and sequence-specific endonucleolytic cleavage of target nucleotide [23,24], respectively. VASA, SPN-E and Armitage share DEAD RNA helicase domains, which provide helicase activities for piRNA production or retrotransposon silencing [13,25]. Zucchini and Squash are putative nucleases, which are believed to be involved in piRNA maturation [18]. Other Dead end, Krimper and Tudor proteins, contain RNA binding domains RRM [26] or Tudor [27] which may facilitate the assembly of multiprotein RNA-induced silencing complex (RISC) and targeting substrate RNA recognition during cleavage. In contrast, although many studies including specific knockouts, protein interaction and cellular distribution experiments have been conducted, the definitive function of MAEL in piRNA pathway remains unknown. MAEL was initially identified in a genetic loss-of-function Drosophila mutant, whose germline cells exhibit incorrect posterior localizations of several transcripts (i.e., Gurken, Oskar and Bicoid) [12]. It is a germ plasm-specific protein with all spindle-class gene phenotypes [12,13,21,28] and directly involved in the piRNA pathway [11,29]. The correct location of either SPN-E, VASA, Aubergine, Tudor or Krimper in germ plasm determines the location of MAEL [11], which in turn delineates the location of Dicer and Argonaute2 [21]. MAEL can shuttle between germ plasm and the nucleus [21]. Direct interaction between MAEL and chromatin remodeling proteins SNF5/INI1 and SIN3B during heterochromatin formation has also been demonstrated [30]. Therefore, MAEL is the only known protein connecting germ plasm and piRNA pathway to chromatin remodeling, a process required for piRNA-initiated genome transposon silencing [31]. In the present study, we were motivated to understand the putative function of MAEL using combined bioinformatic strategies including extensive homologous sequence mining, phylogenetic analysis, domain architecture, protein fold recognition, and structure modeling.

Results

A conserved MAEL-specific domain and its unique lineage-specific evolutionary expansion and loss

• ドメインのアノテーションはマウスのMAELタンパク質はN末端にHMG domainを含むことを示している。HMGは非ヒストン部位や転写調節因子で働くDNA-binding moduleである。

• MAELのC末端についてはドメイン情報が得られなかった。

• NCBI NR databaseに対して、PSI-BLASTを用いてホモログを検索した <- ホモログ検索の方法

• さらに、NCBI nucleotide and Ensembl genome databasesを用いて、追加で８個のホモログを見つけた<- ホモログ検索の方法②

• GeneDB databaseにより３つの原生生物のホモログが発見された。

• multiple sequence alignmentを実行した。（結果: Figure ​1） <- Figureに記されてるのはMAELのみ、C末端も見たい。C末端でドメイン検索してみる。

• MAEL homologues間ではEHHCHCモチーフが特に保存されていた。

Domain annotation showed that mouse MAEL protein contains a HMG domain in its N-terminal segment, which is a DNA-binding module in many non-histone components and transcriptional regulators [32]. However, no domain information could be assigned for the C-terminal segments of MAEL proteins (240 amino acids long). We conducted homologous sequence searching for this region using PSI-BLAST against the NCBI NR database. Many unique homologues were identified in a broad range of species from veterbrates, echinoderms, insects, nematodes, to the protists (Entamoeba histolytica, Entamoeba dispar, and Trypanosoma brucei). We also examined NCBI nucleotide and Ensembl genome databases and identified eight other homologues in insects and urochordates (Ciona intestinalis and Ciona savignyi). Three more protist homologues were obtained through searching GeneDB database. A multiple sequence alignment was built for all the retrieved sequences (additional file 1) and a condensed one is shown in Figure ​1. Although the overall sequence identity is very low, the conservation is apparent across all these MAEL homologues. Six residues Glu-His-His-Cys-His-Cys (EHHCHC) are highly conserved, suggesting that they may contribute to MAEL-specific activity. Thus the C-terminal segment appears to define a novel MAEL-specific domain that we now refer to as the MAEL domain.

• 大部分の種でMaelドメインは1コピーしか存在しなかった

• 他の種類ではいくつかのホモログが存在した; ホヤや蚊（A. aegypti）では２コピー、アカイエカ（Culex pipiens）では３コピー、アメーバ（E. dispar や E. histolytica）では５コピー存在した。

• 系統樹で解析すると系列特異的な重複によってmultiple MAEL copiesが発生したと考えられる。

• 既に調査された全ての種類のゲノムデータベースにおいてMAELは発見されなかった。 (Danio rerio, Gasterosteus aculeatus, Oryzias latipes, Takifugu rubripes and Tetraodon nigroviridis)

• 魚と哺乳類が種分化してから魚だけでMAELの欠失が起こった。

• 魚類にだけあったゲノム重複でMAELが失われたのかもしれない。

For the majority of species, only one copy of MAEL domain exists. However, there are multiple MAEL homologues in several other species; for instance, two copies are found in sea squirts (C. intestinalis and C. savignyi) and mosquito (A. aegypti), three copies in Culex pipiens, and five copies each in amoeba E. dispar and E. histolytica. Phylogenetic tree construction suggests that multiple MAEL copies are generated from a series of ancient lineage-specific duplication events (Figure ​(Figure2A).2A). Strikingly, no fish MAEL homologues could be identified. Its absence in teleost fish was confirmed by carefully examining the published whole genome databases in Ensembl for five different species (Danio rerio, Gasterosteus aculeatus, Oryzias latipes, Takifugu rubripes and Tetraodon nigroviridis). It can be inferred that it is the ancestor of the fish lineage after the divergence of teleost and tetrapod lineages that underwent the loss of MAEL domain. The timing of the loss is probably related to the ancient fish-specific genome duplication [33].

Functional insight from domain architectures

• HMG、HDAC_interact、SR-25-like domainがMAELドメインと関連している。

• HMGはクロマチン関連タンパク質によく見られるDNA結合 モジュールで、ゲノム再編成、転写開始、DNA修復での核タンパク質複合体に機能的に関わっている。

• 多くの種でMAELとHMGドメインが共存していることはMAELドメインがDNAに関連したプロセスで機能することを示唆している。 <- MDsimulationで相互作用見る？

• MAELドメインとHDAC_interactドメインが共存することもMAELがDNAに関連したプロセスに関わっていることを示唆する。 <- MDsimulationで相互作用見る？

• HDAC_interactドメインはクロマチン再編成でヒストン脱アセチル化酵素（HDACs）として中核的に働くことが知られている。

• ある生物で２つの異なるfusion homologueを持つということは、それらのドメインは他の生物では協働して働く可能性が高い。（ロゼッタストーン仮説） <- 現在のゲノムでfusion homologueをサーチしてみる。

• 蚊のMAELはロゼッタストーンタンパク質であり、他の生物でMAELとHDAC_interactドメインを含むタンパク質が相互作用している仮説が立てられる。

• SR-25-like ドメインとMAELドメインの相互作用はRNA関連プロセスとMAELとの関連を示す。

• SR-25-likeドメインはRNA認識モチーフ(RRM)やSR-25ドメインと離れた場所で関連しRNAスプライシングに関わるPRP38と関連づけられている。（SCOOP programにより明らかになった） <- 現代版のゲノムで関連因子を探索するためにSCOOP programも実行してみる。

• ドメインの構造はMAELドメインがDNA結合、RNA結合、クロマチン再編成に関わっている事を示唆する。

Three other domains are associated with MAEL domains, including HMG (SMART: SM00398), HDAC_interact (SMART: SM00761), and SR-25-like domain (DUF1777, Pfam: PF08648) (Figure ​(Figure2B).2B). HMG is a common DNA-binding module in a variety of chromatin-associated proteins and functionally involved in the nucleoprotein complex assembly during genome recombination, initiation of transcription, and DNA repair [32]. The association between MAEL and HMG domains in most species suggests that the MAEL domain may somehow function in a DNA-related process. This functional assignment is also suggested by the association of MAEL domain with HDAC_interact domain in two homologues from mosquitoes (A. aegypti and A. gambiae). The HDAC_interact domain is known to bind to histone deacetylases (HDACs), core enzymes for removing acetyl group from lysine residue of histones during chromatin remodeling process [34]. It has been observed that pairs of interacting domains in one organism may have a fusion homologue composing of these two domains in another organism, known as the rosetta stone protein theory [35]. Mosquito MAELs may be rosetta stone proteins and it can be hypothesized that there are interactions between other MAEL and some HDAC_interact-containing proteins in other species. Indeed, it has been illustrated that mouse MAEL can interact with the SIN3B protein which contains an HDAC_interact domain [30]. The associated SR-25-like domain provides another link between the MAEL domain and RNA-related process. The SR-25-like domain is associated with RNA-binding modules, RNA recognition motif (RRM) [26] and PRP38 [36], It is also distantly related to SR-25 domain which may be involved in RNA splicing, as revealed by the SCOOP program [37]. Therefore, domain architecture suggests a potential involvement of MAEL domains in DNA binding, RNA binding and chromatin remodeling.

A distant similarity between MAEL domains and the DnaQ-H 3'–5' exonuclease family with the RNase H fold

• fold recognition strategyにより離れて（異なった配列をとるが同じ構造をとって）MAEL domainsと関連するものを特定しようとした。

• remote homologyの場合、保存的なタンパク質の折り畳み構造はシークエンスが異なっていても維持されるという理論(rationale)。

• meta server（https://pubmed.ncbi.nlm.nih.gov/12824313/）が利用され、 3D-JURYシステムによってモデリングされた構造が評価された。

• human, X. tropicalis, Ciona and DrosophilaのMAEL domainsがクエリとして用いられ、MetaBasic, ORFeus and BasicDistでヒットした構造が特定された。

• カットオフスコアである50を下回っていたが、得られた全ての構造がRNase H foldを持つDnaQ-H 3'–5' エキソヌクレアーぜ ファミリーに属していることがわかった。

• 先祖である赤痢アメーバのMAELドメインをクエリとして用いてみた。

• 11個の構造が58–69のスコアでヒットし、全てDnaQ-H 3'–5'ぬくれあーぜファミリーに属していた。

• DnaQ-HとMAELドメインの構造的類似性が判明したので関係性をPSI-BLASTでもう一度調べてみた。

• PSI-BLASTによってDnaQ-HエキソヌクレアーぜがE valueが0.05以下ではないが、検出されていることに気づいた。

We applied a fold recognition strategy to identify remotely related homologues of MAEL domains. The rationale is that in the case of remote homology, conserved protein structural folds can be kept despite limited sequence identity [38]. A meta server was utilized, which assembles various state-of-the-art fold recognition methods and further evaluates modeled structures based on a consensus score computed by a 3D-JURY system [39]. MAEL domains from human, X. tropicalis, Ciona and Drosophila were first used as queries and several structural hits were identified by MetaBasic, ORFeus and BasicDist with consensus scores from 21 to 46. Although these 3D-Jury scores are below the cutoff 50, which corresponds to correct assignment with statistical significance [40], domain and fold examinations showed that all retrieved structures belong to the DnaQ-H 3'–5' exonuclease family with the RNase H fold [41,42]. We extended our search using an ancestral E. histolytica MAEL domain (GI: 67477376, residues 315–532) as a query. Eleven structural hits were identified with high scores around 58–69, and they all belong to the DnaQ-H 3'–5' exonuclease family. Structural fold similarities between DnaQ-H and MAEL domains encouraged us to re-examine this relationship using PSI-BLAST. We noticed that several DnaQ-H exonucleases can be retrieved as insignificant candidates in our initial PSI-BLAST searching with a profile inclusion expectation (E) value of 0.005. However, when we set inclusion E value at 0.05, significant similarity between the first 100aa segment of MAEL domains and several prokaryotic DnaQ-H exonucleases was achieved in the fourth iteration.

• 次にMAELドメインとDnaQ-H 3'–5' エキソヌクレアーゼで構造に基づいたmultiple sequence alignmentsを行い、このホモログの関係を調べた、

• DnaQ-H ドメイン間での配列類似性は低いので、「CE-MC server」で調査された構造情報をもとにアラインメントが作られた。

• その後、そのアラインメントをfold recognition resultsと予測された二次構造の情報をもとにアラインメントされたMAELドメインの情報と合わせた。

• 最終的なアラインメントはMAELとDnaQ-Hドメインの間で保存された残基を明らかにした。

• MAELドメインでRNase H foldの３個のベータシート構造に対応する構造はalpha helixと予測されたが、これは間違いだと考えられる。(RNase H foldでも予測間違いが起こったため)

• MAELドメインの二次構造はDnaQ-H 3'–5'エキソヌクレアーぜに似ている。（どちらもβ1- β2- β3- α1- α2- β4- α3- β5- α4- α5- α6の構成を持つ）

• さらに重要なことに、DnaQ-Hに特徴的な配列（Asp-Glu-Asp-His-Asp, DEDHD）を共有している。

• これらの残基は多様なDnaQ-H 3'–5'エキソヌクレアーぜで共有されており、２つの二価金属イオンと相互作用して活性部位を作る。

• 配列類似性は低いが構造的類似性と原生生物MAELでの共通モチーフ（DEDHD）があることから進化t系な関連が示唆された。

We next examined this homologous relationship by building structure-based multiple sequence alignments for MAEL domains and DnaQ-H 3'–5' exonucleases. Since sequence identity among different DnaQ-H domains is very low, their alignment was first generated based on structural information as assessed by a CE-MC server [43] followed by manual adjustment based on published literature. Thereafter, we combined this alignment with the aligned MAEL domains based on fold recognition results and predicted secondary structures. The final alignment showed the conserved residues among/between two domains and compositions of secondary structures (Figure ​(Figure3A).3A). It is to be noted (Figure ​(Figure3A)3A) that equivalents of beta sheet (β) 3 of the RNase H fold in most MAEL domains are predicted to be alpha helix (α). We believe that this is a wrong prediction since in the canonical RNase H fold, β3 is an edge β strand, which can usually be misidentified as an α helix because of its solvent sequence property [44]. As shown in Figure ​Figure3A,3A, the secondary structures of MAEL domains resemble those of DnaQ-H 3'–5' exonucleases; both have a β1- β2- β3- α1- α2- β4- α3- β5- α4- α5- α6 composition. More importantly, several ancestral protist MAEL domains also share all the critical DnaQ-H characteristic residues (Asp-Glu-Asp-His-Asp, DEDHD). These residues are commonly utilized by diverse DnaQ-H 3'–5' exonucleases and interact with two divalent metal ions to form an active site [45-47]. Thus, in contrast to a very low sequence identity (<15%) between MAEL domains and DnaQ-H 3'–5' exonucleases, the similar structural fold and the notable existence of DEDHD residues in protist MAEL domains strongly support a distant evolutionary relationship.

Structural examinations on active sites by DEDHD and EHHCHC residues in MAEL domains

• comparative modeling.. homology modeling

• DEDHD... 原生生物のMAELでは構造的中核になっていてDnaQ-H domainsの活性部位と似ていた。

• 他の種ではDEDHDはみられなかったが、EHHCHCはみられた。

• モデリングされたMAELの構造において、EHHCHCの場所を調べた。

• 全てのMAELに特異的な残基は似通った位置にあった。

• EHHchCはstructural coreを形成しうる上、他の２残基もそれに向かい合っている。

• 最後のCHCを形成するα5 and α6で似たような変化が見られた。

• C178 and C189 のCysでジスルフィド結合が見られるかもしれない。

• DnaQ-HのDEDHD残基はDnaQ-H 活性部位を形成するが、EHHCHC残基を持つ他のMAELドメインは標準的なRNase H 足場に基づいた新たな活性部位を形成する可能性がある。

The tertiary structures of protist and chicken MAEL domains were further constructed by comparative modeling. Like DnaQ-H domains (Figure 3B, C), these MAEL domains adopt a similar RNase H structural fold which is characterized by a compact α/β fold with open anti-parallel β sheets in the middle and several α helices surrounded (Figure 3D, E).
Moreover, the characteristic DEDHD residues in protist MAEL domains are clustered into a structural core, which resembles active sites of DnaQ-H domains (Figure 3A, B, C). In contrast, most other MAEL domains lack the DnaQ-H specific residues DEDHD. However, they are characterized by another conserved stretch of residues, EHHCHC. During evolution such conservation of MAEL-specific residues may reflect functional contributions most likely to a distinct active site. The spatial locations of EHHCHC residues were then examined in these modeled MAEL structures to check their possibility of forming an active site. Unexpectedly, we found that all MAEL-specific residues have very close spatial locations and they are clustered together at one side of the middle anti-parallel β sheets (Figure ​(Figure3E).3E). Four residues (EHHchC) can shape a structural core and other two residues may also potentially face down to it with slight structural rearrangements. Similar change of structural conformations of α5 and α6 comprising last CHC residues has been observed in crystal structures of DnaQ-H domains (additional file 2). There may exist another possibility that a disulfide bond (-S-S-) is formed between two Cys residues (C178 and C189 in the chicken MAEL domain) because of their close proximity. This is also supported by disulfide bond predictions [48]. Formation of a disulfide bond may facilitate the last His to approach other EHH residues, thus forming an active site with EHHH residues. Therefore, structural examinations suggest that protist MAEL domains with DEDHD residues may form a DnaQ-H active site whereas other MAEL domains with EHHCHC residues may potentially form a new active site based on the canonical RNase H scaffold.

Discussions and conclusion

Functional insight into MAEL in germline piRNA pathway

• 提案されたMAELドメインとRNase H foldを持つDnaQ-H 3'–5' エキソヌクレアーぜの進化的な繋がりによりMAEL domainsの機能が推測できる

• DnaQ-H 3'–5' exonuclease family（DEDDh exonuclease family or Exonuc_X-T domain）はRNase H fold superfamilyのmemberであり、RNase H, mu transposase, crossover junction resolvase RuvC, and PIWI domain familiesを含む。これらは全てcanonical RNase H foldを含むが、活性部位は異なる。

• DnaQ-H familyは５つの保存された残基（DEDHDドメイン、二価の陽イオンと共に活性部位を作る）により特徴付けられる。

• DnaQ-H familyのメンバーは核酸の代謝に関わる。（例: replicative proofreading (1J53:A) [47], DNA repair or RNA degradation (exonuclease I and oligoribonuclease) [45,46], and RNA interference (ERI-1) [53]. ）

• EDDHD残基により形成される活性部位は3'–5' exonuclease activityを形成する。酸性の DEDDは２つの金属イオンと3'末端を収納する負の電荷のポケットを形成する。次に結合した金属イオンと保存されたHが結合鎖と直接相互作用し、核酸の3'–5' directionリン酸ジエステル結合を切断する。したがって、DnaQ-H に特異的なDEDHD残基と活性部位を持つ原生生物のMAELドメインは3'–5'エキソヌクレアーぜ活性を持つかもしれない。（関連する金属イオンと標的は未知だが）

The proposed evolutionary link of MAEL domains to DnaQ-H 3'–5' exonuclease with RNase H fold may provide functional clues for MAEL domains. The DnaQ-H 3'–5' exonuclease family, also known as DEDDh exonuclease family or Exonuc_X-T domain (Pfam ID: PF00929), is one member of RNase H fold superfamily (SCOP: 53098) which also includes RNase H, mu transposase, crossover junction resolvase RuvC, and PIWI domain families [24,49-52]. They all share a canonical RNase H fold but contain different active site residues. The DnaQ-H family is characterized by five conserved residues, DEDHD, which form an active site in coordination with divalent metal ions (Figure ​(Figure3A).3A). Its members contribute to diverse nucleic acid metabolism processes such as replicative proofreading (1J53:A) [47], DNA repair or RNA degradation (exonuclease I and oligoribonuclease) [45,46], and RNA interference (ERI-1) [53]. Although different nucleotide targets (DNA or RNA) or diverse metal ions (Zn2+, Mg2+, or Mn2+) are involved [45-47], their active sites formed by the EDDHD residues delineates a common 3'–5' exonuclease activity. That is, the acidic DEDD together with two metal ions shape a negative pocket, which provides space for accommodating the 3' termini of oligonucleotide (DNA or RNA) chains. Thereafter, the coordinated metal ions and another conserved H are in direct contact with the bound chain, which induces a break of the phosphodiester bond of nucleotide in the 3'–5' direction [46]. Therefore, protist MAEL domains, harboring DnaQ-H specific DEDHD residues and active sites, may also employ a 3'–5' exonuclease activity, although their associated metal ions and nucleotide targets are still unknown.

• 原生生物のMAEL domainと異なり最近のMAELドメインはDnaQ-H に特異的な残基を持っていないが、特徴的なEHHCHC残基を持つ。

• EHHCHC または EHHHのような構造的なコアを形成しうるとわかった。

• MAELにはDnaQ-Hがないのでこの部位がRNA-binding abilityを持つかもしれない

• RNase H foldは配列の類似度は低く異なる活性部位残基を含むが、全て金属イオンと協働したDNA/RNA 3' または 5' end-directed nuclease活性を持つ。

In contrast to the protist MAEL domains, most recent MAEL domains do not contain the DnaQ-H specific residues but are characterized by the EHHCHC residues. What is the functional contribution of these residues to MAEL domains? Structural observations showed that a structural core can be potentially formed by the MAEL-specific residues EHHCHC or EHHH. This may provide a structural basis for an active site. On the one hand, this active site may confer RNA-binding ability for MAEL domains because of the lack of DnaQ-H specific residues. In this way, MAEL may contribute to stabilizing or positioning the RNA substrate in piRNA pathway. On the other hand, MAEL-specific residues and its potential active site may define another nuclease activity. We noticed that although all related families with the RNase H fold have low sequence identities and contain different active site residues, they all have DNA/RNA 3' or 5' end-directed nuclease activities with metal ion coordination in their own active sites [50,51]. For example, RNase H is a non-specific endonuclease whose catalytic activity requires divalent ions (Mg2+ or Mn2+) and is responsible for the hydrolysis of the RNA in a DNA/RNA duplex [52,54]. In contrast, PIWI domains contribute to 5'-3' exonulcease catalytic activity for the Argonaute family proteins (Slicer) in all types of small RNA pathways (siRNA, miRNA, and piRNA). The activity is achieved by three PIWI active site residues, DDH, in coordination with one divalent ion and used to cleave single-stranded RNA substrate guided by complementary double-stranded small RNAs (piRNA or siRNA) [23,24,55-57]. It seems that the RNase H structural fold is an efficient scaffold from which diverse nuclease families have evolved distinct nuclease activities by developing their own active site residues with metal ion coordination. Therefore, being one member of RNase H superfamily, the MAEL domain may share this characteristic, thus the residues EHHCHC may form an active site with a new nuclease activity. It has been shown in diverse proteins that H, C and E residues often interact with Zn2+ [58]. Moreover, the residue composition of EHHH is commonly utilized by several Escherichia coli proteins including ColE7 endonuclease [59], Zinc transport protein ZnuA [60], and Aldolase (1DOS) for their active sites, which also interact with metal ions, especially Zn2+ [61],

Experimental evidence have suggested that MAEL may be involved in piRNA biogenesis since its loss-of-function mutant impairs the production of piRNAs or rasiRNAs and increases the transcript level of transposable elements [11]. Different from siRNA and miRNA pathways, piRNAs biogenesis employs a Dicer-independent mechanism [4,10]. A ping-pong model has been recently proposed for this process and it is hypothesized that AGO3 bound to the sense strand of piRNAs catalyzes cleavage of the antisense strand that generates 5' end of antisense piRNAs. The 3' end of the resulting antisense piRNAs is subjected to a 3' cleavage by an unknown endonuclease or exonuclease and a HEN1-processed 3' methylation. Thereafter, the produced antisense piRNAs associate with Aubergine or PIWI and direct cleavage of transposon sequences, which then generates the sense strand piRNAs after 5' cleavage, 3' cleavage and 3' methylation [5,7,8]. This cycling model is not complete since the exonuclease or endonuclease enzyme responsible for the 3' terminal maturation remains uncharacterized [5,7,8]. Thus, because of its evolutionary relationship to 3'–5' DnaQ-H exonuclease and the potential (3'–5' exo-) nuclease activities, MAEL may be the nuclease candidate implicated in the cleavage of the 3' termini. Recently, the nucleases Zucchini and Squash have been proposed as the 3' termini nuclease candidate based on the evidence that they are also located in germ plasm and have a similar mutation phenotype in a loss of transposon silencing [18]. However, MAEL is distinct from those above two nucleases due to its translocation between germ plasm and nucleus and the direct interaction with chromatin remodeling proteins [21,30]. We believe that multiple nucleases are involved in the diverse steps of piRNA pathway in a sequential manner, similar to PIWI family members targeting 5' cleavage of piRNAs [62]; and MAEL is involved in a genomic DNA-related piRNA step, which may include chromatin remodeling process and initial transcriptions of transposon. In this way, MAEL-associated HMG domain or other chromatin remodeling proteins facilitate the access of piRNA complex to the genomic regions where are enriched with transposon sequences. The transposon transcripts undergoing processing interact with the piRNA complex in which PIWI, one RNase H member, generates 5' end of transposon transcripts via a piRNA-directed homologous cleavage whereas MAEL, another RNase H member, contributes to a 3' terminal cleavage of transposon transcripts.