Drosophila Clu ribonucleoprotein particle dynamics rely on the availability of functional Clu and translating ribosomes Open Access

Ribonucleoproteins (RNPs) often associate with cytoplasmic particles or bodies to form RNP particles, which play crucial roles in the post-transcriptional regulation of mRNAs (Gehring et al., 2017). These RNP particles, including stress granules and processing bodies (P-bodies), are highly conserved across species and function to sequester translation machinery and mRNAs, thereby regulating mRNA stability or active translation (Anderson and Kedersha, 2006; Bauer et al., 2023; Kato and Nakamura, 2012; Keene, 2007). Under normal conditions, stress granules and P-bodies are present in limited quantities, but their numbers increase significantly in response to cellular stress (Lin et al., 2008; Schisa, 2019).

Drosophila Clueless (Clu) and its vertebrate analog CLUH are highly conserved multidomain ribonucleoproteins abundantly found in the cytoplasm (Fig. 1A–C; Cox and Spradling, 2009; Gao et al., 2014; Sen and Cox, 2016). Clu forms robust particles in vivo, especially in female germ cells, which exhibit high metabolic activity (Fig. 1A,B; Cox and Spradling, 2009; Sheard et al., 2020). The loss of Clu/Cluh in Drosophila and mice leads to profound mitochondrial dysfunction, with flies living only a few days and mice dying on postnatal day 1 (Cox and Spradling, 2009; Schatton et al., 2017). clu mutants also have reduced mitochondrial proteins (Sen and Cox, 2022). Studies of CLUH have shown that a significant portion of CLUH-bound transcripts encode nucleus-encoded mitochondrial proteins, suggesting that CLUH is involved in the regulation of mRNAs that are crucial for mitochondrial function (Gao et al., 2014). Drosophila Clu also binds transcripts encoding mitochondrial proteins (Sen and Cox, 2022). The mechanism through which Clu/CLUH regulates these associated mRNAs is not yet fully understood (Hémono et al., 2022a; Schatton et al., 2017; Vardi-Oknin and Arava, 2019). Moreover, the association of Clu with ribosomal proteins and translation factors suggests that it plays a role in active translation, potentially involving mitochondria-associated ribosomes for co-translational or site-specific import, as well as non-mitochondria-associated cytoplasmic ribosomes (Hémono et al., 2022a; Vardi-Oknin and Arava, 2019; Bennett et al., 2022; Pla-Martin et al., 2020; Sen and Cox, 2022).

In Drosophila, Clu forms mitochondria-associated particles of various sizes that are highly dynamic and require intact microtubules for movement (Sheard et al., 2020). Although Clu self-associates, it remains unclear whether Clu forms multimers or simply aggregates within these particles (Sen and Cox, 2016). These particles, which we have named ‘bliss particles’, do not colocalize with common subcellular organelle markers such as the autophagosome marker Atg8 or the P-body protein Trailer Hitch (Sheard et al., 2020). Unlike stress granules and P-bodies, Clu particles are exquisitely sensitive to stress in vivo and only form under optimal conditions in well-fed flies (Sheard et al., 2020). Particle disassembly can be induced by starvation, oxidative stress and mitochondrial stress caused by mutations in Superoxide Dismutase 2 (Sod2), PTEN-induced putative kinase 1 (Pink1) and parkin (Fig. 1A′,B′; Cox and Spradling, 2009; Sheard et al., 2020; Sen et al., 2015). Remarkably, removing stress conditions, such as by refeeding the flies or by adding insulin ex vivo, restores Clu particles (Sheard et al., 2020). Despite the dynamic nature of these particles, Clu levels remain constant, indicating that Clu particle disassembly is not due to protein degradation (Sheard et al., 2020). Furthermore, insulin signaling is necessary and sufficient for particle assembly, suggesting that the metabolic state of the cell significantly influences the presence of Clu particles (Sheard et al., 2020).

Mitochondria, as central hubs of metabolism, house critical pathways including heme biosynthesis, fatty acid β-oxidation and steroidogenesis, in addition to generating ATP (Bennett et al., 2022; Rahman, 2020; Scarpulla, 2008). Although these organelles have their own mitochondrial DNA, the majority of proteins required for these pathways are supplied by nuclear DNA-encoded mRNAs that are translated on cytoplasmic and mitochondrial-associated ribosomes (Bennett et al., 2022; Devaux et al., 2010; Dimmer et al., 2002). Given binding of Clu to mRNAs and its association with ribosomes, as well as the juxtaposition of Clu particles to mitochondria, this study aims to determine how disrupting translation affects Clu particle dynamics in Drosophila female germ cells. We employed ectopic transgenic expression in vivo to demonstrate that three conserved domains of Clu are necessary for its assembly into particles and that an excess of full-length, but not deletion constructs of Clu disassemble particles, suggesting particle stability or assembly is altered with non-physiological Clu levels. Manipulating translation activity using the translation inhibitors puromycin (PUR) and cycloheximide (CHX) revealed that PUR treatment quickly led to disassembly of Clu particles whereas CHX treatment did not. In addition, CHX treatment stabilized existing Clu particles in the presence of nutritional and oxidative stress. Together, these observations support a model whereby Drosophila Clu particle assembly requires translating ribosomes and that particles could function as sites of active translation of mRNAs encoding mitochondrial proteins to sense changes to cellular metabolism and subsequently regulate mitochondrial function. In addition, our observed particle dynamics are unique from those observed for other RNP particles, underscoring the distinctive response of Clu particles to translation inhibition in Drosophila germ cells.

To identify the protein domains and ribosome activity requirements of Clu particle dynamics, we examined particle assembly and disassembly in female germ cells using fixed and live imaging (Fig. 1A–B′). Clu particles are always present in the nurse cells of follicles from well-fed flies (Fig. 1A,B; Cox and Spradling, 2009; Sheard et al., 2020). However, they look larger or smaller, or more or less numerous, depending on imaging conditions, including the length of dissection, whether the tissue is imaged while fixed or live, and the type of microscope (laser confocal versus spinning disc). Also, importantly, Clu particles completely disassemble in response to stress (Fig. 1A′,B′); thus we used a binary decision for assembly or disassembly – either Clu particles were present or absent – to determine how various conditions affect dynamics. Drosophila females have a pair of ovaries composed of 16–20 ovarioles (Fig. 1D,E; Spradling, 1993; Sarikaya et al., 2012). Ovarioles are strings of developing follicles that contain all follicle stages (2–14) when dissected from well-fed females. For consistency, we predominantly analyzed data from stage 6 and 7 follicles, which are large enough to be readily imaged and contain many Clu particles but are previtellogenic and have not yet undergone many complex developmental events (Fig. 1E).

Clu is a large, multi-domain protein (Fig. 1C); however, the function of each domain and how they controls particle dynamics is poorly understood. The melanogaster-specific (ms) domain is unique to Drosophila and is not required to rescue the clu null mutant (Sen and Cox, 2016). The β-grasp fold (βGF) domain is so-called based on the predicted structure, and the domain of unknown function (DUF) is predicted based on sequence homology (Sen et al., 2015). The Clu domain is highly conserved, but we do not yet understand its role. We have previously demonstrated that the tetratricopeptide repeat (TPR) domain is crucial for Clu to bind mRNA (Sen and Cox, 2016). Finally, the middle (M) domain sequence is unstructured, which is a feature commonly found in proteins associated with membraneless organelles (Shin and Brangwynne, 2017; Alberti et al., 2019; Li et al., 2012).

Previously, we used co-immunoprecipitation to show that full-length Drosophila Clu can self-associate (Sen and Cox, 2016). To test which domains are required for Clu particle assembly in vivo, we used the GAL4/UAS system (Brand and Perrimon, 1993). We created transgenic lines that ectopically express Clu-tagged with mScarlet at the C-terminus under the control of the conditional UASp promoter (Fig. 2A). Each construct was expressed at the appropriate molecular mass (Fig. S1A,B). clu null mutants are weak and infertile with very small ovaries, and die after ∼7 days (Cox and Spradling, 2009). Expressing full-length (FL)-Clu::mScarlet ubiquitously, rescued the clu null mutant, producing healthy, fertile females (Fig. S1C). We visualized endogenous Clu using GFP inserted at the clu locus (clu^CA06604, Clu::GFP; Fig. 1C) (Buszczak et al., 2007). To simultaneously visualize ectopically expressed mScarlet-labeled constructs and endogenous Clu, each construct was combined with germline-specific nanos GAL4 (nosGAL4) in a clu^CA06604 background (Van Doren et al., 1998). The nosGAL4 line we used is clonally expressed in the nurse cells (Rorth, 1998). Using live imaging, we found that full-length (FL)-Clu::mScarlet reliably formed Clu particles (Fig. 2B′,F,G), which colocalized with endogenous Clu::GFP particles, indicating that both Clu species exist within the same particle (Fig. 2B″; Fig. S2A,A′, Movie 1). To determine whether the DUF, Clu, and TPR domains are required for particle assembly, we ectopically overexpressed mScarlet-labeled Clu transgenes with each domain deleted (ΔDUF::mScarlet, ΔClu::mScarlet and ΔTPR::mScarlet) (Fig. 2A,C–E″, Movies 2–4). Using live imaging, we were unable to see particle assembly for any of these deletion constructs (Fig. 2C′,D′,E′,F). Furthermore, none of these deletions co-labeled with endogenous Clu::GFP particles (Fig. 2C″,D″,E″; Fig. S2B–D′). This suggests that these three domains are necessary to assemble Clu particles and to associate with endogenous particles that had already assembled.

When we tested expression levels of ectopic FL-Clu using different GAL4 drivers, we noticed a dose-dependent effect on particle assembly. daughterless (da) GAL4 is a ubiquitous GAL4 driver that when combined with FL-Clu::mScarlet rescued the clu null mutant (Fig. S1C) (Wodarz et al., 1995). daGAL4 induced high uniform expression of Clu in germ cells in contrast to lower expression levels of nosGAL4, consistent with da expression pattern (Fig. 3A–A″; Figs S1A, S3A–D″) (Cummings and Cronmiller, 1994). When ectopic FL-Clu::mScarlet was expressed using daGAL4 (Fig. 3A′,E), neither ectopic Clu nor endogenous Clu assembled into particles in germ cells (Fig. 3A versus Fig. S3D). Mitochondrial distribution remained normal, indicating that the germ cells were not experiencing a stress to cause mislocalization (Fig. S3F′ versus E′ control). We also observed that endogenous Clu assembled into particles in the somatic follicle cells, where daGal4-induced ectopic Clu expression was lower compared to the germ cells (Fig. 3A,A″, insets; Fig. S3B,B″). To test whether this was dependent on functional FL-Clu, we ectopically overexpressed a Scarlet-labeled ΔDUF or ΔTPR domain (ΔDUF::mScarlet, ΔTPR::mScarlet, Fig. 2A) using daGAL4. Consistent with our observations using nosGAL4, neither was able to mediate assembly of mScarlet particles (Fig. 3B′,C′), but unlike ectopic FL-Clu, ΔDUF and ΔTPR overexpression did not lead to disassembly of endogenous Clu particles (Fig. 3B,C). Finally, to ensure that expressing high concentrations of any protein does not cause cell stress that inhibited particle assembly, we sought an unrelated cytoplasmic protein. We used mCherry-labeled Capping Protein β(mCherry::CPB). CPB is a well-characterized and abundant actin-binding protein (Ogienko et al., 2013), and mCherry and mScarlet are highly similar monomeric red fluorescent proteins derived from Discosoma sp. (Bindels et al., 2017; McCullock et al., 2020; Shaner et al., 2004). mCherry::CPB also did not disrupt endogenous Clu particles (Fig. 3D–D″). This suggests that the mechanism assembling Clu particles depends on regulating the concentration of functional Clu but does not respond to non-functional Clu (ΔDUF and ΔTPR).

Stress granules and P-bodies have distinct assembly–disassembly responses downstream of various stressors and different translation inhibitors, depending on the mechanisms of action for the drugs (Buddika et al., 2020; Eulalio et al., 2007; Patel et al., 2016). This is thought to be due to manipulation of available levels of associated messenger RNPs (mRNPs) and polysome presence and activity (Hofmann et al., 2021; Hubstenberger et al., 2017; Kato and Nakamura, 2012; Ivanov et al., 2019). Puromycin (PUR) is a commonly used and unique translation inhibitor. PUR forms a stable peptide bond with nascent polypeptides, resulting in premature translation termination which leads to ribosome complex disassembly, polysome loss and increased free mRNAs (Fig. 4A) (Aviner, 2020). Given that Clu is a ribonucleoprotein crucial for mitochondrial function, binds mRNAs encoding mitochondrial proteins and associates with the ribosome, we wanted to determine how puromycin treatment affected Clu particle dynamics. To do this, we treated ovarioles dissected from well-fed clu^CA06604 females with PUR and found that the plentiful Clu particles were quickly and completely disassembled within 10 min (Fig. 4B–C′,G; Movie 5). Ex vivo imaging has many advantages for investigating cellular dynamics. To determine the effect of PUR treatment in vivo, we switched well-fed females to PUR mixed in wet yeast paste for 24 h, then used fixed imaging to analyze Clu particles. Feeding females PUR recapitulated our ex vivo result, resulting in disassembly of Clu particles (Fig. 4D–F″,H). Given the molecular action of PUR, this suggests that decreased translating ribosomes, disassembled ribosomes and/or increased concentrations of mRNPs cause Clu particles to disassemble.

CHX is another well-known and frequently used translation inhibitor that binds to the exit site of the ribosome, which stalls the ribosome and blocks translation elongation, leading to increased concentrations of polysomes (Fig. 5A) (Duncan and Mata, 2017). Under normal conditions, CHX treatment causes P-bodies to disassemble (Eulalio et al., 2007; Lin et al., 2008; Moutaoufik et al., 2014; Patel et al., 2016; Kedersha et al., 2000). To ensure our method of CHX treatment was effective, we dissected ovarioles from well-fed trailer hitch^CA06517 (tral) females that expressed GFP inserted at the endogenous tral locus, thus labeling P-bodies (Kato and Nakamura, 2012; Eulalio et al., 2007; Buszczak et al., 2007). We confirmed that Tral-labeled P-bodies decreased in size and number with CHX addition compared to mock control, as previously shown (Fig. S4; Eulalio et al., 2007; Patel et al., 2016). To determine the effect of CHX on Clu particles ex vivo, we tested whether CHX treatment caused disassembly of particles and found that this was not the case (Fig. 5B–D).

To determine the effect of CHX on Clu particles in vivo, we switched well-fed females to yeast paste supplemented with CHX for 24 h. Feeding CHX reduces protein synthesis (Hwang and Cox, 2024). To determine the effect of dietary CHX on Clu particle dynamics in vivo, we fixed and immunolabeled ovarioles from CHX-fed females (Fig. 5F–I″). As expected, untreated well-fed females displayed robust Clu particles (Fig. 5F,F″). Like our ex vivo experiment, CHX-fed females also exhibited robust Clu particles, indicating that CHX feeding does not disassemble particles (Fig. 5G,G″,H,H″,I,I″,J). We previously demonstrated that mitochondria mislocalize in nurse cells when the females are exposed to various stressors (Cox and Spradling, 2009; Sen and Cox, 2016; Sen et al., 2015; Sheard et al., 2020). CHX-feeding not only maintained Clu particles, but also maintained mitochondrial morphology and distribution, supporting that feeding CHX did not cause stress signals sufficiently acute to cause mitochondrial mislocalization in nurse cells (Fig. 5F–I″).

Given that CHX did not disassemble Clu particles ex vivo and in vivo, we wanted to test the effect of CHX on particle assembly. To do this, we dissected ovaries from starved clu^CA06604 females, which have completely disassembled Clu particles (Fig. 5K–M′; Movies 8,9) (Sheard et al., 2020). Normally, adding insulin to the medium causes Clu particles to quickly assemble (Fig. 5L,L′,N; Movie 8; Sheard et al., 2020). However, with preincubation of CHX, insulin-induced Clu particle assembly was blocked (Fig. 5M,M′,N; Movie 9). Taken together, these data suggest that the stalled ribosomes occurring with CHX treatment do not lead to the disassembly of Clu particles that have already formed. In addition, insulin signaling is insufficient to drive particle assembly in the absence of pre-existing particles after CHX treatment, when ribosomes are stalled.

For effective Drosophila ex vivo imaging, insulin must be added to the medium to fully support the tissue (Morris and Spradling, 2011). If it is omitted, egg chamber development is not normal, and the samples start to degenerate (Prasad and Montell, 2007; Prasad et al., 2007). Given that Drosophila Insulin-like peptides secreted by the brain are required for normal egg chamber development, incubating follicles without insulin does not supply effective nutritional signaling and is stressful to the tissue (LaFever and Drummond-Barbosa, 2005; Richard et al., 2005). Given that CHX treatment did not abolish Clu particles, in contrast to PUR, we tested whether CHX treatment could confer a protective effect to maintain Clu particles in the presence of nutritional stress (Fig. 6A–C). Ovarioles dissected from well-fed clu^CA06604 females and incubated in insulin-free medium (Complete Schneider's; CS) showed disassembly of Clu particles within 30 min, supporting the crucial role of nutrition in maintaining particles (Fig. 6B,B′). Surprisingly, adding CHX after dissection was sufficient to maintain Clu particles for 30 min, even incubating in the absence of insulin (Fig. 6B″,C). To test whether CHX could also protect Clu particles from nutritional stress in vivo, we switched well-fed females to yeast paste supplemented with CHX for 24 h, then subjected the flies to a mild 5-h starvation on water only (Fig. 6D). The 5-h starvation led to completely disassembly of Clu particles (Fig. 6E,E″) (Sheard et al., 2020). However, providing CHX for 24 h before starvation was sufficient to maintain Clu particles and protect them from disassembly (Fig. 6F,F″,G,G″,I). This was not the case with the highest concentration of CHX (Fig. 6H,H″). Although 10 mM CHX feeding did not affect Clu particles under well-fed conditions (Fig. 5I,I″), adding nutritional stress led to disassembly of the particles (Fig. 6H,H″,I). With 3.5 mM and 7.0 mM feeding, not only were particles maintained, CHX feeding before starvation appeared to decrease cellular nutritional stress as indicated by normal mitochondrial localization (Fig. 6E′ versus F′,G′,I′) (Sheard et al., 2020). In contrast, 10 mM feeding caused mitochondrial clumping (Fig. 6H′,H″,I′). This observation supports that stalled ribosomes maintain and protect Clu particles and mitochondrial distribution even with decreased nutrition ex vivo and in vivo, but that this protection becomes ineffective with a high dose of CHX.

Hydrogen peroxide (H₂O₂) is highly toxic to cells, immediately causing a sharp increase in reactive oxygen species and oxidative damage (Vona et al., 2021). Previously, we have demonstrated that adding H₂O₂ to cultured ovarioles quickly disassembles Clu particles (Fig. 7B–B‴; Movie 10; Sheard et al., 2020). Given that CHX treatment of cultured ovarioles maintained Clu particles even in the absence of insulin (Fig. 6B″), we tested whether CHX treatment could maintain particles with high levels of oxidative stress (Fig. 7). We cultured ovaries from well-fed clu^CA06604 females, added CHX, and then exposed the ovarioles to H₂O₂ (Fig. 7A). Surprisingly, adding CHX maintained Clu particles in the presence of H₂O₂ (Fig. 7C–C‴,E; Movie 11). This was also observed for a higher CHX concentration (Fig. 7D–D‴,E; Movie 12). This observation suggests that CHX treatment and stalled ribosomes are sufficient to maintain and protect Clu particles from oxidative stress-induced particle disassembly, in addition to protection from nutritional stress.

Clu could act as a scaffold protein within particles as this is a common feature of cytoplasmic granule proteins (Buchan, 2014). We have previously demonstrated that Clu can self-associate, but we do not yet know whether Clu forms dimers or multimers or whether self-association is due to protein aggregation in Clu particles (Sen and Cox, 2016). Determining which domains are required for self-association using ectopic Clu expression has proved challenging in vivo given that endogenous Clu must be absent and clu null mutants are quite sick. Although FL-Clu rescues the clu null mutant, ΔDUF, ΔClu and ΔTPR do not (Fig. S1) (Sen and Cox, 2016). We have identified that the DUF, Clu and TPR domains are required for Clu to assemble in particles and for association with endogenous Clu particles, but we do not yet know whether they play a role in molecular self-association in Drosophila. The deletion constructs were expressed at the correct molecular masses but could be misfolded. In addition, ΔDUF and ΔClu proteins were expressed at lower levels using the ubiquitous daGAL4, which could explain their inability to assemble in particles, although ΔTPR was expressed at normal levels, suggesting that the missing protein domains rather than reduced protein levels are the reason. The TPR domain of CLUH has been reported to be necessary for CLUH self-association, although it appears that the self-association is not direct, suggesting CLUH forms multimers (Hémono et al., 2022a). In addition, we have previously shown that, in Drosophila, the TPR domain is essential for Clu–mRNA association (Sen and Cox, 2016), which was confirmed to also be the case for CLUH (Hémono et al., 2022a). Many cytoplasmic bodies and granules have associated proteins containing low-complexity or prion-like domains. The large ‘M’ domain in Clu is predicted to be intrinsically disordered, a property frequently found in RNA granule-associated proteins that is important for condensate dynamics (Lin et al., 2017). The M-domain of Clu could also be important for particle dynamics, but this has yet to be tested. Ectopically expressing FL-Clu at high levels caused Clu particles to disassemble through an unknown mechanism, which differs from what has been observation with CLUH granules (discussed below; Pla-Martin et al., 2020). Although this could be due to toxicity, that seems unlikely given that overexpression of ΔTPR and ΔDUF did not cause endogenous particle disassembly, nor did overexpression of the unrelated protein CPB. An alternative explanation could be that high concentrations of FL-Clu somehow disrupt the required stoichiometry for forming Clu particles due to the non-physiological levels. Clu levels remain the same with nutritional disruption to particles (Sheard et al., 2020). Too much functional Clu could disrupt potential liquid–liquid phase separation, which might regulate Clu particle assembly and disassembly through a gain-of-function effect caused by sequestering factors necessary for particle assembly (Alberti et al., 2019). There are few proteins known to regulate condensate disassembly; thus, a better understanding of this observation might shed light on condensate dynamics (Bard and Drummond, 2024; Brumbaugh-Reed et al., 2024).

As Clu associates with mRNA and ribosomal proteins, we tested the effect of translation inhibitors on assembly and disassembly with and without stress. CHX and PUR treatment are powerful tools for assessing how stalled translation affects the dynamics of particles and granules involved in the posttranscriptional regulation of mRNA. PUR decreases the amount of polysomes, terminating translation, releasing ribosomes and increasing the amount of mRNPs. In contrast, CHX prevents elongation. This results in an inhibition of polysome-to-mRNP conversion and thus increases stalled ribosomes and decreases levels of mRNPs. Stress granules are composed of mRNPs, stalled preinitiation complexes and other proteins involved in translation. Under normal conditions, low concentrations of PUR rarely leads to assembly of stress granules, but longer incubation with higher concentrations does (Kedersha et al., 2000; Bounedjah et al., 2014; Martinez et al., 2016; Ihn et al., 2024 preprint). In the presence of stress, PUR increases the number of stress granules, whereas CHX disassembles them. This dynamic is caused by an increased number of mRNPs available with PUR and a decreased number available with CHX. P-bodies generally contain proteins that are associated with mRNA decay or silencing. P-body maintenance depends on the presence of translationally repressed mRNA. In the presence of CHX, mRNA trapped in polysomes causes P-body loss. However, upon PUR treatment, the number and size of P-bodies increase due to the increase in non-translatable mRNPs (Eulalio et al., 2007; Patel et al., 2016). When treating follicles with PUR and CHX, we found that the availability of translating ribosomes regulates assembly and disassembly of Clu particles. We demonstrated this is the case with CHX feeding, which lasts for 24 h (Hwang and Cox, 2024). However, our observations for Clu particle assembly and disassembly ex vivo take place on the order of 7–30 min. Administering PUR ex vivo and in vivo led to disassembly of Clu particles, whereas CHX had no effect ex vivo and in vivo. This suggests that in the presence of PUR, Clu particles disassemble in response to increased mRNPs and decreased numbers of translating ribosomes. In contrast, our observations using CHX treatment suggest that stalling translation has no effect on Clu particles when they are already present. However, when particles are absent, CHX treatment blocks insulin-induced assembly ex vivo suggesting that increased numbers of stalled ribosomes alone cannot drive Clu particle assembly. The increased stalled ribosomes might be physically prevented from separating into Clu particles despite the unchanged Clu levels. This also suggests that insulin signaling, which initiates a strong signaling cascade, cannot overcome the CHX-mediated block in assembly.

The effect of PUR and CHX treatment on Clu particle assembly and disassembly could be due to signaling pathways induced by treatment. CHX directly stimulates TOR signaling, a potent master regulator of metabolism, by increasing the pool of amino acids due to decreased protein synthesis, but PUR treatment indirectly stimulates TOR in a context-dependent manner (Finlay et al., 2006; Iiboshi et al., 1999). Given that we have previously shown that insulin signaling is necessary and sufficient for Clu particle assembly, it is possible that particle dynamics occurs through the TOR pathway. Presently, we cannot rule this out. However, if CHX treatment upregulated TOR using our methods, we would expect that CHX treatment would drive Clu particle assembly in the absence of particles, but we did not see this (Fig. 5M′). We do not know whether our PUR and CHX treatment methods decrease global protein levels equally, although they appear to do so in cultured cell experiments (Sidhu and Omiecinski, 1998; Croons et al., 2008). The objective in this study was to observe the effect of two translational inhibitors that use different mechanisms of action on Clu particle dynamics, similar to studies of P-bodies and stress granules. The formation of stress granules occurs through the integrated stress response (ISR) through eIF2α and by TOR signaling. However, stress granule dynamics also depends on the physical characteristics of participating proteins, RNAs and mRNPs that allow condensation to occur (reviewed in Baymiller and Moon, 2023).

Nutritional and oxidative stress disassemble Clu particles. Both stressors cause decreased translation rates through ISR signaling. During nutritional stress ex vivo and in vivo, CHX treatment maintained Clu particle assembly, suggesting that, once particles are formed, stalled ribosomes present in Clu particles stabilize the particles. This was also true for oxidative damage ex vivo caused by H₂O₂. Together, these data support a model whereby Drosophila Clu particles harbor actively translating mRNAs, and particle assembly relies on the presence of translating ribosomes. We have yet to successfully localize ribosomes to the particles. However, data from the Arabidopsis Clu homolog FRIENDLY (FMT) has shown that FMT particles associate with cytoplasmic ribosomes at the mitochondrial surface (Hémono et al., 2022b). Evidence for Clu particles as sites of active translation is supported by observations of CLUH granules (Pla-Martin et al., 2020).

Clu forms large, prominent and highly dynamic particles in germ cell cytoplasm (Cox and Spradling, 2009; Sheard et al., 2020). These particles are also found in Drosophila somatic tissues (Sen et al., 2013; Sheard et al., 2020; Wang et al., 2015). We have previously demonstrated that Clu particles are highly sensitive to stress. Live imaging has shown that Clu particles quickly disassemble in the presence of H₂O₂, and for germ cells lacking particles due to starvation, adding insulin to the medium causes particle assembly in minutes (Sheard et al., 2020). Nutritional stress in vivo from starvation causes particle disassembly, with no decrease in protein, and particles reassemble upon feeding (Sheard et al., 2020). Various studies in vertebrate systems have demonstrated CLUH localization in the cell. In COS7 cells, CLUH exhibits a granular pattern, particularly after Triton-X 100 extraction (Gao et al., 2014). In primary hepatocytes and HeLa cells, CLUH has been reported to form granules that colocalize with some, but not all, components of stress granules, and these granules increase with stress (Pla-Martin et al., 2020). In addition, in contrast to the data shown here, CLUH overexpression induced the formation of peripheral CLUH granules in ∼40% of transfected HeLa cells (Pla-Martin et al., 2020). However, additional reports have shown CLUH is broadly diffuse in the cytoplasm in HCT116 cells, and CLUH does not localize with the stress granule component G3BP1 (Hémono et al., 2022a). In agreement with our observation, CLUH granules did not disassemble in response to CHX treatment, although these granules were the peripheral granules assembled by overexpression of CLUH (Pla-Martin et al., 2020). Finally, two groups have shown CLUH colocalizes with one or two structures composed of SPAG5 (also known as astrin), which is a mitotic spindle protein during mitosis that localizes to microtubule plus-ends in the cytoplasm during interphase (Hémono et al., 2022a; Dunsch et al., 2011; Schatton et al., 2022). It is not clear at present why Clu localization and dynamics are so different from those of vertebrate CLUH. One possibility could be that vertebrates might simply have different CLUH dynamics from Drosophila Clu due to differences in cell types and species. Another possibility could be that Clu particles might act differently in vivo compared to CLUH in cell culture experiments due to differences in cell physiology.

Starvation causes germ cell mitochondria to cluster (Sheard et al., 2020). This occurs not only with nutritional stress but appears to be a hallmark of additional cellular stressors, such as oxidative damage or stress caused by mutation in genes involved in mitochondrial function. When Clu particles are absent, mitochondria are clumped, but when the particles are present, mitochondria disperse throughout the germ cell cytoplasm in the normal pattern. This indicates that normal mitochondrial localization highly correlates with physiological levels of Clu and the presence of Clu particles (Cox and Spradling, 2009; Sheard et al., 2020; Sen et al., 2015). We found similar mitochondrial dynamics with CHX treatment. Females fed a rich CHX diet maintained assembled Clu particles and mitochondrial distribution (Fig. 5F–I″). Clu particles that remained assembled after CHX treatment followed by starvation also had normal mitochondrial localization (Fig. 6F–G″). This could be attributed to CHX increasing the level of amino acids and thus stimulating TOR activity, a downstream component of the insulin signaling pathway (Beugnet et al., 2003), although, CHX is insufficient to drive particle assembly suggesting this is not the case (Fig. 5M′). Another possibility is that, because Clu particles associate with mitochondria (Cox and Spradling, 2009; Sen et al., 2015), Clu particles assembled by CHX treatment itself could affect mitochondrial physical distribution even under stress.

In this study, we identified three protein domains of Clu that are necessary to assemble Clu particles. We also found that overexpression of functional FL-Clu causes particle disassembly. In addition, we found that Clu particles require the presence of translating ribosomes to remain assembled and that stabilizing mRNA-associated ribosomes protects particles from disassembly. Given that Clu associates with ribosomes and is a ribonucleoprotein, this supports a model whereby Clu particles are active sites of translation under non-stressed conditions but disassemble when cellular stress increases. As Clu is closely tied to mitochondrial function and binds mRNAs encoding mitochondrial proteins, particle dynamics likely affects mitochondrial function. Disassembly of particles could lead to decreased translation of the associated mRNAs, fewer mitochondrial proteins actively translated in the cytoplasm and a shift in metabolism in response to stress. There is evidence supporting this idea through studies on CLUH (Pla-Martin et al., 2020). For a better understanding of Drosophila Clu particles, several challenges must be overcome. There are many proteins known to associate with P-bodies and stress granules (Anderson and Kedersha, 2006; Ivanov et al., 2019). Although we and others have identified Clu/CLUH-associated proteins using co-immunoprecipitation and mass spectrometry, we have yet to localize any of them to Clu particles. In addition, we have yet to determine whether Clu particles are active sites of translation. Particles are easily seen in the nurse cells, yet using fluorescent in situ hybridization is challenging. Nonetheless, given the crucial role for Clu in mitochondrial function, fully understanding how these unique and novel RNP particles function will ultimately deepen our knowledge of how mitochondria respond to stress.

Fly stocks were maintained on standard cornmeal fly medium [Bloomington Drosophila Stock Center (BDSC), Bloomington, IN, USA, https://bdsc.indiana.edu/information/recipes/bloomfood.html]. Animals were grown at room temperature. The following stocks were used for experiments: w¹¹¹⁸; clu^d08713/CAG; daGAL4/TM6c, w*; clu^CA06604/CyO; nanosGAL4/TM3 Sb, w¹¹¹⁸; UASp-cluΔDUF::mScarlet/TM3 Sb, w¹¹¹⁸; UASp-cluΔClu::mScarlet/TM3 Sb, w¹¹¹⁸; UASp-cluΔTPR::mScarlet/TM3 Sb, w¹¹¹⁸; UASp-FLclu::mScarlet/TM3 Sb (this study). w¹¹¹⁸, w*; Kr^If-1/CyO; P{w[+mW.hs]=GAL4-da.G32}UH1 (BSC #55850), w*; M{w[+mC]=UASp-mCherry.cpb}ZH-86Fb/TM3 Sb¹ (BSC #58728), and w¹¹¹⁸; P{w[+mC]=GAL4::VP16-nanos.UTR}CG6325[MVD1] (BSC #4937) (Van Doren et al., 1998) were obtained from the Bloomington Drosophila Stock Center (BDSC), Bloomington, IN, USA. w¹¹¹⁸; clu^CA06604 (Cox and Spradling, 2009) and w*; P{PTT-GA} tral^CA06517 are described in (Buszczak et al., 2007). A newly eclosed fly is considered day 0.

For the C-terminal fusion of mScarlet for live-imaging, we created a Gateway destination vector pPgateWmScarlet-i for subcloning. Briefly, the mScarlet-i coding sequence was amplified from pCytERM_mScarlet-i_N1 (Addgene #85068) using the following primers, 5′-TAGGCCACTAGTGTGAGCAAGGGCGAGGCAGT-3′and 5′-TGCTTAGGATCCTTACTTGTACAGCTCGTCCA-3′. The amplicons were subcloned into a pQUASp (Addgene #46162), placing the UASp promoter upstream of the mScarlet-i coding sequence. Ampicillin resistance was used to select positive clones, which were verified by restriction digest and sequencing. The resulting pUASp-mScarlet-i construct was converted into a Gateway destination vector using the Gateway™ Vector Conversion System (Invitrogen, cat. no. 11828029). Chloramphenicol resistance was used to select positive clones, which were verified by sequencing. The pQUASp vector was Addgene #46162 (RRID: Addgene_46162, deposited by Christopher Potter), and pCytERM_mScarlet-i_N1 vector was Addgene #85068 (RRID: Addgene_85068, deposited by Dorus Gadella; Bindels et al., 2017). Gateway entry vectors with full-length Clu (Clu_pENTR) or domain-deleted Clu (ΔDUF_pENTR, ΔClu_pENTR, and ΔTPR_pENTR) were previously described (Sen and Cox, 2016). Each entry vector was cloned into pPgateWmScarlet-i using LR Clonase mix (Invitrogen, cat. no. 11791020) following the manufacturer's directions. The resulting expression vectors were selected by ampicillin resistance and verified by sequencing. For transgenic flies, the vectors were commercially injected (BestGene Inc., Chino Hills, CA, USA).

Live imaging with ovarioles was performed as previously described (Sheard et al., 2020) with some modifications. Ovaries were dissected in a live-imaging medium composed of Complete Schneider's (CS) medium and 200 µg/ml of insulin (insulin from bovine pancreas, Sigma-Aldrich, Burlington, MA, USA, cat. no. I6634). The CS medium was Schneider's Drosophila medium (Thermo Fisher Scientific, Hampton, NH, USA, cat. no. BW04351Q) supplemented with 15% heat-inactivated fetal bovine serum (CPS Serum, Parkville, MO, USA, cat. no. FBS-500HI) and penicillin (100 U/ml)-streptomycin (100 µg/ml) (Thermo Fisher Scientific, cat. no. BW17602E). After separating ovarioles from an ovary and eliminating the connected muscle sheath and nerve fibers, the tissues were transferred into a 35 mm MatTek glass bottom dish (MatTek Corporation, Ashland, MA, USA, cat. no. P35G-0-20-C) with a live-imaging medium. The follicles were mainly chosen between stages 5 to 7, previtellogenic stages, for better observation with nurse cells. The focal plane was selected to have at least three to four nurse cells with a clear visibility of nuclear and cytoplasmic areas, with ∼25% depth from the top surface of a follicle. Follicle stages were determined by follicle size as described previously (Spradling, 1993). Live images were obtained using a Nikon A1 plus Piezo Z Drive Confocal microscope with a 60× objective lens (Nikon Corporation, Tokyo, Japan) or a Nikon Eclipse Ti2 spinning disk microscope with a 100× objective lens (Nikon Corporation, Tokyo, Japan). Fiji ImageJ was utilized to analyze confocal images (Schindelin et al., 2012).

Fiji Image J (Schindelin et al., 2012) was utilized to detect and count the Clu particles co-labeled with endogenous Clu::GFP and ectopic Clu::mScarlet. Briefly, a line was applied to a single-frame image, and a multi-channel plot histogram was generated with the tools, Plot Profile and Visualization_toolset. After getting the image by Difference of Gaussian for each fluorescence channel, the tool DiAna was used to analyze Clu particles co-labeled with endogenous and ectopic Clu proteins (Gilles et al., 2017).

The working solution for each chemical was prepared just before performing an experiment as follows: 10 mg/ml puromycin (PUR, Gibco™, Waltham, MA, USA, cat. no. A1113803) was diluted to 20 µM in a medium composed of CS medium and 100 µg/ml of insulin (CS/Ins); cycloheximide powder (CHX, Sigma-Aldrich, Burlington, MA, USA, cat. no. C7698) was dissolved in CS for 14 mM stock solution, which was further diluted to 3.5 mM and 7 mM CHX-containing CS or CS/Ins media; and 30% H₂O₂ (Sigma-Aldrich, Burlington, MA, USA, cat. no. H1009) was diluted to 4% H₂O₂ in CS/Ins or CHX-containing CS/Ins. Ovaries were dissected as described above with a corresponding medium, CS or CS/Ins, depending on the purpose of each experiment. To test the PUR effect on Clu particle dynamics, dissected ovaries with CS/Ins medium were transferred into a 35 mm MatTek glass bottom dish with 50 µl of CS/Ins medium, and live imaging was performed after adding 50 µl of CS/Ins containing 20 µM puromycin to the ovaries to make a final concentration of 10 µM puromycin. To test the CHX effect on Clu particle dispersion, dissected ovaries with CS/Ins medium were incubated for 20 min with 3.5 mM CHX-containing CS/Ins (CS/Ins/CHX) following two washes with the same medium, and then a live-imaging was performed. To test the CHX effect on Clu particle formation, ovaries were dissected from starved animals in CS, incubated for 20 min with 3.5 mM CHX-containing CS (CS/CHX) following two washes with the same medium, re-washed twice with CS/CHX containing 100 µg/ml insulin (CS/CHX/Ins), and then switched to CS/CHX/Ins for immediate live imaging. To test the CHX effect on particle dispersion without insulin, ovaries were dissected from w¹¹¹⁸; clu^CA06604 in CS, washed twice with CS containing 3.5 mM CHX (CS/CHX), incubated for 30 min with the same medium and then live images were obtained. To produce oxidative stress, dissected ovaries were incubated with 50 µl of CS/Ins or CS/CHX/Ins for 20 min following two washes with a corresponding medium. Live imaging was performed after adding 50 µl of each corresponding medium containing 4 mM H₂O₂ to make a final concentration of 2 mM H₂O₂. Selection of follicle stages and a focal plane was performed as described above. Images were obtained using a Nikon A1plus Piezo Z Drive confocal microscope with a 60× objective lens or a Nikon Eclipse Ti2 spinning disk microscope with a 100× objective lens.

PUR and CHX were dissolved in water to prepare a 2 mM stock solution and 10 mM stock solution, respectively. The stock solutions were aliquoted and stored at −20°C until use. The desired concentration of PUR or CHX was prepared by serial dilution of the stock solution. 0.3 g of active dry yeast powder (Red Star^® Yeast) was mixed with 450 µl of inhibitor solution to create a yeast paste, which was provided daily as needed.

Ten female day 0 adult flies and ten male day 0 adult flies were collected in a standard food vial with wet yeast paste, and on day 3, female adults were separated from males. The food vial with fresh wet yeast paste was switched every day to make flies fatten until day 4. On day 4, all female flies in each vial were transferred to a standard food vial containing freshly made yeast paste with PUR or CHX. At 24 h after feeding yeast paste containing each translation inhibitor, fly ovaries were dissected in Grace's insect medium (Invitrogen, cat. no. 1595030). To determine the effect of starvation after CHX feeding, after feeding CHX for 24 h, the flies were transferred to an empty vial with a wet piece of Kimwipe with water and maintained for 5 h. The flies were then dissected and immunostained.

Ovaries were dissected with Grace's insect medium and fixed for 20 min (4% paraformaldehyde and 20 mM formic acid in Grace's insect medium). After washing with antibody wash solution (AWS; 0.1% Triton X-100 and 1% BSA in phosphate-buffered saline) three times for 20 min, the tissue was stained with primary antibody overnight at 4°C. After washing with AWS three times for 20 min, the tissues were stained with secondary antibodies overnight at 4°C, then washed with AWS twice for 20 min and stained with 5 ng/ml DAPI solution for 10 min. After removing the DAPI solution, the tissues were mounted in Vectashield Antifade Mounting Medium (Vector Laboratories, Newark, CA, USA, cat. no. H-1000). Images were obtained using a Zeiss LSM 980 confocal laser scanning microscope (Carl Zeiss Microscopy LLC, White Plains, NY, USA). The follicles were mainly chosen between stages 5 to 7, the previtellogenic stages, for better observation in nurse cells. The focal plane was selected to ensure at least three to four nuclei were clearly visible in a nurse cells, with ∼25% depth from the top surface of a follicle but also to avoid a dim fluorescence signal due to a deeper depth of imaging. The numbers of dissected and observed follicles and replicates are described in a graph within each figure. The following antibodies were used: guinea pig anti-Clu N-terminus (1:2000; Cox and Spradling, 2009), rat anti-mScarlet-i sdAb-FluoTag-X2 (1:1000, NanoTag Biotechnologies, Goettingen, Germany, cat. no. N1302-At488-L, RRID:AB_3075967), chicken anti-mCherry (1:1000, Novus Biologicals, Centennial, CO, USA, cat. no. NBP2-25158 RRID:AB_2636881), mouse anti-ATP5A1 (1:1000, Abcam, Cambridge, UK, cat. no. 14748, RRID:AB_301447), or Thermo Fisher Scientific, Waltham, MA, USA, cat. no. 439800, RRID:AB_2533548), goat anti-guinea pig IgG conjugated to Alexa Fluor 488 (1:1000, Molecular Probes, Waltham, MA, USA, cat. no. A11073, RRID:AB_2534117), goat anti-guinea pig IgG conjugated to Alexa Fluor 633 (1:1000, Thermo Fisher Scientific, Waltham, MA, USA, cat. no. A21105, RRID:AB_2535757), goat anti-chicken IgY conjugated to Alexa Fluor 568 (1:500, Thermo Fisher Scientific, Waltham, MA, USA, cat. no. A11041, RRID:AB_2534098), goat anti-mouse IgG2b conjugated to Alexa Fluor 488 (1: 500, Thermo Fisher Scientific, Waltham, MA, USA, cat. no. A21141, RRID:AB_2535778), goat anti-mouse IgG2b conjugated to Alexa Fluor 568 (1: 500, Thermo Fisher Scientific, Waltham, MA, USA, cat. no. A21144 RRID:AB_2535780).

Flies overexpressing Clu in the clu null background were generated by crossing w¹¹¹⁸; clu^d08713/CAG; daGAL4/TM6c virgins with w¹¹¹⁸; clu^d08713/CAG; UASp-FLclu::mScarlet/TM3 Sb males. From this cross, ten virgin females of w¹¹¹⁸; clu^d08713/clu^d08713; daGAL4/UASp-FLclu::mScarlet were crossed with ten w¹¹¹⁸ males in a standard food vial, then removed after 3 days. The vials were kept at room temperature for an additional 5 days and then imaged. Sibling females, w¹¹¹⁸; clu^d08713/CAG; daGAL4/TM3b and w¹¹¹⁸; clu^d08713/CAG; daGAL4/UASp-FLclu::mScarlet, were used for control.

Ten ovary pairs were dissected from day 5–7 well-fed adult females and homogenized with a whole-cell lysis buffer composed of 50 mM Tris-HCl pH 8.0, 150 mM sodium chloride, 1 mM EDTA, 1% NP-40, 0.1% sodium dodecyl sulfate and Complete-mini EDTA-free protease inhibitor (Roche Applied Science, Indianapolis, IN, USA, cat. no. 11836170001) by 30 times of strokes with a pestle. Insoluble material was removed by centrifugation (12,000 g for 20 mins at 4°C), and the subsequent supernatant was collected as a fraction of whole cellular proteins. After determining protein concentration using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA, cat. no. 23227), 40 µg of each sample was separated by SurePAGE™, Bis-Tris gel (GensScript, Piscataway, NJ, cat. no. M00653) and transferred to a nitrocellulosemembrane. The nitrocellulose membrane was probed with guinea pig anti-CluN antibody (1:10000, Cox and Spradling, 2009), chicken anti-mCherry antibody (1:1000, Novus Biologicals, Centennial, CO, USA, cat. no. NBP2-25158, RRID:AB_2636881), and mouse anti-tubulin antibody (1:5000, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA, cat. no. AA4.3). Western blots were performed in triplicate and images were analyzed using Fiji ImageJ (Schindelin et al., 2012). To analyze the expression levels of ectopic proteins, mScarlet detected by anti-mCherry antibody was normalized with tubulin. The data was analyzed using GraphPad Prism, as described below. See Fig. S5 for uncropped images of blots presented in this study.

To test the CHX effect on P-body disassembly, ovaries were dissected from w*; P{PTT-GA} tral^CA06517 flies in CS medium, and then incubated for 30 min with CS containing 3.5 mM CHX (CS/CHX) following two washes with the same medium. Live images were obtained using a Nikon Eclipse Ti2 spinning disk microscope with a 100× objective lens (Nikon Corporation, Tokyo, Japan). The follicles were mainly chosen in stages 7–8 that show greater numbers of P-bodies compared to earlier stages. The focal plane was selected to have at least three to four nurse cells with a clear visibility of nuclear and cytoplasmic area, with ∼25% depth from the top surface of a follicle. Changes in P-bodies were determined as a subjective measurement by a researcher who was aware of the experimental conditions.

The graphs and statistical analysis by the unpaired two-tailed t-test and one-way ANOVA test with multiple comparisons followed by Tukey's post hoc test were generated using GraphPad Prism (GraphPad Software, Boston, Massachusetts USA, www.graphpad.com).

We would like to thank Matthew Gillen for critically reading the manuscript. We would like to thank the USUHS Biomedical Instrumentation Core and Dr Dennis McDaniel for imaging support. Antibodies obtained from The Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology were used in this study. Stocks obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) were used in this study. Cartoons in Figs 4 and 5 are courtesy of BioRender.