The piRNAs in Drosophila germline cells are produced and amplified in the membraneless organelle, nuage, which is assembled by orderly recruitment of the corresponding components to ensure its proper function. Although its precise function has not been clarified, our findings demonstrate that Tej plays a crucial role in recruiting RNA helicases Vas and Spn-E to nuage through distinct domains, namely, Lotus and SRS. Our results provide new insights into the regulation of stepwise piRNA precursor processing by Tej, Spn-E, and Vas in the initial phase of piRNA biogenesis prior to the ping-pong amplification cycle. Tej recruits these helicases for the engagement of the precursors involved in further processing of nuage, thereby also controlling the dynamics of these nuage components (Fig. 6).

Our results confirmed that the Tej Lotus domain recruited Vas to nuage, which is consistent with the fact that it enables Vas to hydrolyze ATP for RNA release (Jeske et al., 2017). We newly identified that the SRS motif in Tej is responsible for Spn-E recruitment to nuage. Full deletion or single amino acid substitution of SRS significantly disrupted Spn-E recruitment to Tej granules in S2 cells, whereas further deletions of eight amino acids other than SRS, eSRS, were critical for recruiting Spn-E to nuage in the ovaries. This result raises a possibility that Tej, as well as other factors, may assist the recruitment of Spn-E to nuage in the ovaries. Another protein known as Tap, which is a fly counterpart of TDRD7 and harbors Lotus and eTudor domains, has previously been reported to participate in the piRNA pathway and interact with Vas (Jeske et al., 2017; Patil et al., 2014). However, since Tap lacks the SRS found in Tej, it is unlikely to be involved in the recruitment of Spn-E. The mouse homolog of Spn-E (TDRD9) is localized in both nuage and the nucleus in prespermatogonia (Shoji et al., 2009; Wenda et al., 2017), and might perform different functions that remain elusive. Our finding suggests a possibility that the intrinsically nuclear protein Spn-E was deliberately recruited to nuage via Tej to exert a unique function, such as piRNA precursor processing. In contrast, the eTudor domain mainly contributes to Tej aggregation (Fig. 2 B and Fig. S6 A), which is consistent with previous studies showing that the eTudor domain is engaged in granulation by binding to its ligand sDMA (Courchaine et al., 2021).

Despite the unusual nuage granules of Tej-ΔeTudor, it mildly suppressed transposon expression (Fig. 4, B and F). Notably, Tej-ΔeTudor displays interaction with Vas and Spn-E, albeit to a lesser extent, especially with Spn-E (Fig. S5 C). Our CL-IP results also supported these interactions as reported in S2 cells (Patil and Kai, 2010). Alternatively, Tej-ΔeTudor possibly may facilitate the association of other components with nuage activity for piRNA processing. Unlike the mutation of precursor transporter, nxf3 (ElMaghraby et al., 2019; Kneuss et al., 2019), and the ping-pong cycle assistant, krimp (Sato et al., 2015; Webster et al., 2015), tej, as well as spn-E and vas mutants, exhibited the accumulation of piRNA precursors in the perinuclear region and a collapse of the ping-pong amplification. These results suggest that they function upstream during ping-pong amplification. Stalling of piRNA precursors was also observed when the recruitment of Vas or Spn-E to nuage was abolished by the loss of the Lotus or eSRS domains, respectively. Precursor accumulation was concentrated in the malfunctioning nuage or perinuclear region, which would result in a failure in precursor processing and cause TE upregulation.

Genetic analysis of nuage organization revealed that Spn-E and Tej occupy a higher hierarchical position than Vas at an earlier stage (Fig. S1 D), which is inconsistent with a previous observation (Patil and Kai, 2010), possibly due to the fluctuation of nuage assembly and/or structure at a later stage in the mutants. In contrast, Tej and Spn-E are mutually dependent for the proper assembly of nuage granules because Spn-E is required for the proper localization of Tej within nuage (Fig. S1 D and Fig. 4 B). Moreover, Tej may form a relatively stable scaffold with Spn-E for nuage assembly, while a mobile fraction of Tej may contain Vas. These results suggest that Tej may facilitate the compartmentalization of Vas and Spn-E, as shown in CL-IP experiments (Fig. 1 D) and also reported in Bombyx germ cells (Nishida et al., 2015), while we cannot exclude the possibility of simultaneous binding among these proteins. Our further results with S2 revealed that the weak hydrophobic interaction between the proteins may contribute to the formation and regulation of membraneless structures on nuage. DEAD-box RNA helicase family members, including Vas homolog, reportedly form non-membranous, phase-separated organelles in both prokaryotes and eukaryotes (Hondele et al., 2019), and the large IDR at the N-terminal region facilitates their aggregation by LLPS (Nott et al., 2015). In addition, the loss of IDR in Tej significantly suppressed the mobility of Tej and Vas; nevertheless, the TE repression was only mildly attenuated (Fig. 5 F). Thus, Tej-ΔIDR may remain colocalized with Vas and Spn-E, facilitating the processing of piRNAs (Fig. 5 C). Alternatively, the reduction of Vas mobility by the loss of Tej IDR could be compensated by other components in nuage. Only the localization of Vas was remarkably changed upon 1,6-HD treatment in S2 cells, further supporting the finding that weak hydrophobic interaction controlled the dynamics of Vas, although we cannot exclude a possibility of the unexpected effects by the 1,6-HD treatment. We also cannot exclude the possibility that 1,6-HD treatment might have impaired kinase and/or phosphatase activity (Düster et al., 2021). Hence, localization might have been affected by the changes in their phosphorylation status. The behavior of these proteins is seemingly influenced by their respective binding modes and properties with Tej. The interaction of Vas with Tej is affected by 1,6-HD and IDR region of Tej through the hydrophobic association, whereas that of Spn-E with Tej is more rigid, possibly contributing to the formation of the scaffold of nuage. In conclusion, Tej utilizes the eTudor domain for granule formation, whereas the IDR of Tej appears to maintain the assemble of Tej granules, controlling the mobility of Vas in nuage.

Membraneless macromolecular nuage contains more than a dozen components, including Vas and Tej that harbor IDRs, which could contribute to the dynamics of nuage and impact the efficient production of piRNAs. Nuage also contains piRNA precursors and TE RNAs that are processed therein; their unique or specific propensities may affect nuage assembly and function. Further investigation of those proteins and RNA components will shed light on the regulatory mechanisms underlying the formation and dynamics of nuage to promote each sequential step of piRNA biogenesis.