me31B regulates stem cell homeostasis by preventing excess dedifferentiation in the Drosophila male germline

Tissue-specific adult stem cells play a critical role in sustaining tissue homeostasis by continuously providing differentiated cells throughout the life of organisms (He et al., 2009; Nystul and Spradling, 2006). The loss of stem cells or their functions underlie tissue degeneration under physiological and pathological conditions. The stem cell pool is primarily maintained by self-renewal. In addition, dedifferentiation, a process whereby differentiated and/or differentiating cells revert back to a stem cell identity, also helps to maintain the stem cell population beyond the lifetime of individual stem cells (de Sousa and de Sauvage, 2019; Merrell and Stanger, 2016). However, the misregulation of dedifferentiation is thought to underlie tumorigenesis (Landsberg et al., 2012; Schwitalla et al., 2013). Therefore, dedifferentiation must be tightly controlled to ensure stem cell maintenance, while preventing transformation. However, the molecular mechanisms that regulate dedifferentiation are not well understood.

The Drosophila testis serves as an excellent model system to study dedifferentiation. Notably, this model offers unambiguous identification of stem cells [germline stem cells (GSCs)] and their differentiating progeny (Fuller and Spradling, 2007; Yamashita, 2018). GSCs are attached to postmitotic somatic hub cells, which function as a major component of the stem cell niche (Fig. 1A). The hub cells secrete two major signaling ligands that promote GSC self-renewal: a cytokine-like ligand Upd that activates the JAK-STAT (Janus kinase-signal transducer and activator of transcription) pathway, and a BMP (bone morphogenic protein) ligand Dpp that activates the downstream Tkv receptor (Kawase et al., 2004; Kiger et al., 2001; Schulz et al., 2004; Shivdasani and Ingham, 2003; Tulina and Matunis, 2001). Upon GSC divisions, daughter cells that are displaced away from the hub initiate differentiation as gonialblasts (GBs), which then continue with proliferative mitotic divisions (or transit-amplifying divisions) as spermatogonia (SGs) before entering a meiotic program as spermatocytes (SCs). SG divisions are characterized by incomplete cytokinesis, connecting all sister cells as a cluster (i.e. cyst). A membranous organelle called the fusome runs through the stabilized contractile ring, called a ring canal, connecting SGs within a cyst (Fig. 1A; Yamashita, 2018).

Although GSCs are maintained relatively stably through consistent asymmetric divisions, which generate one GSC and one GB (Yamashita et al., 2003), GSCs can occasionally be lost (Wallenfang et al., 2006). Upon GSC loss, SGs can respond to niche vacancy, and dedifferentiate to replenish the GSC pool. During dedifferentiation of SGs, the fusome that connects SGs fragments into a more spherical structure, referred to as ‘spectrosome’, as typically observed in GSCs (Fig. 1A) (Brawley and Matunis, 2004). Fragmenting fusomes in >2 cell SGs are observed only during dedifferentiation, not during differentiation, and these features can be used to unambiguously identify dedifferentiating SGs without lineage tracing (Brawley and Matunis, 2004; Sheng et al., 2009; Sheng and Matunis, 2011). Dedifferentiation was first shown in an experiment that artificially removed all GSCs via transient overexpression of Bam, a master regulator of differentiation (Brawley and Matunis, 2004; Sheng et al., 2009; Sheng and Matunis, 2011). Although temporally controlled overexpression of Bam induced all GSCs to differentiate, withdrawal of Bam allowed SGs to repopulate the stem cell niche and produce GSCs. Subsequently, it was shown that SG dedifferentiation occurs naturally and increases during aging in unperturbed tissues (Cheng et al., 2008), suggesting that dedifferentiation is likely a mechanism that helps to maintain the GSC population throughout the lifetime of organisms, particularly with age. More recent work showed that dedifferentiation is important for sustaining the GSC population under conditions that repeatedly induce GSC replenishment and challenge tissue homeostasis, such as cycles of starvation and refeeding (Herrera and Bach, 2018). SG dedifferentiation under these conditions required Jun N-terminal kinase (JNK) signaling (Herrera and Bach, 2018). However, whether mechanisms exist to prevent excess dedifferentiation remain poorly understood.

Maternal expression at 31B (me31B) encodes an RNA helicase of the DEAD-box family that regulates translation (Kugler et al., 2009; Kugler and Lasko, 2009; Nakamura et al., 2001). In particular, me31B silences the translation of oocyte-localizing mRNAs, such as oskar, in nurse cells prior to their transport to the oocyte (McDermott et al., 2012; Nakamura et al., 2001). me31B has also been shown to repress translation of nanos (nos) (Gotze et al., 2017; Jeske et al., 2011), a translational regulator that is critical for germ cell specification and maintenance of GSCs (Li et al., 2009; Wang and Lin, 2004). Here, we show that me31B is a critical negative regulator of dedifferentiation in the Drosophila testis. In the absence of me31B, SGs frequently dedifferentiated even in the absence of known triggers, such as the induced removal of GSCs. We further show that me31B suppresses SG dedifferentiation by repressing nos. Our study reveals that dedifferentiation is actively repressed under normal conditions, likely to protect the native GSC population, and identifies me31B as a previously unknown negative regulator of dedifferentiation.

To study the role of me31B in the testis, we used two independent RNAi constructs [UAS-me31B^TRiP.GL00695 and UAS-me31B^{TRiP.HMS00539}, available from Bloomington Drosophila Stock Center (see Materials and Methods)]. Using these constructs and the nos-gal4 driver, we knocked down me31B in germ cells (Fig. S1, nos-gal4>UAS-me31B^TRiP.GL00695 and nos-gal4>UAS-me31B^{TRiP.HMS00539}, hereafter nos>me31B^TRiP.GL00695 and nos>me31B^{TRiP.HMS00539}, respectively, or simply nos>me31B^RNAi, as essentially the same results were obtained with both RNAi constructs). We found that Me31B-GFP was expressed in both germline and somatic cells in the testis, and the GFP signal was substantially reduced in the germline upon expression of the RNAi construct using nos-gal4, confirming the efficiency of these RNAi constructs (Fig. S1). Although Me31B has been reported to be a component of nuage (germ granules) (DeHaan et al., 2017; Liu et al., 2011; Thomson et al., 2008), we observed diffuse cytoplasmic localization of Me31B-GFP in germ cells in the adult testis and Me31B-GFP did not colocalize with the nuage marker Vasa in control flies. Moreover, me31B knockdown did not affect nuage morphology (Fig. S1).

As expected, GSCs in control testes surrounded the hub and were either single cells or connected to their immediate daughter cells (GBs) prior to completion of cytokinesis (Fig. 1B). Intriguingly, nos>me31B^RNAi testes often contained dedifferentiating SG cysts that were attached to the hub cells (Fig. 1B). Their identity as dedifferentiating SG cysts is based on the fact that they contained ≥3 germ cells that were connected to each other (Fig. 1C-D′; see Materials and Methods for details for identifying dedifferentiating cysts). The fusomes in these SG cysts at the hub in nos>me31B^RNAi testes were fragmented (Fig. 1C-D′), a well-established hallmark of dedifferentiating SGs (Brawley and Matunis, 2004; Sheng et al., 2009; Sheng and Matunis, 2011), rather than continuous as in differentiating SGs (Fig. 1E,E′). We observed dedifferentiating SG cysts, identified by their fragmented fusomes and attachment to the hub, in ∼80% of nos>me31B^RNAi testes but not in any control testes (Fig. 1F). The number of SGs within dedifferentiating SG cysts was not always 2ⁿ: often they contained three SGs, indicating that some SGs might have already dedifferentiated into single GSCs or died during dedifferentiation.

We considered two possibilities that could explain this phenotype. First, me31B may be required in SGs to directly prevent their dedifferentiation. Second, me31B may be required to maintain GSCs in the niche, which would indirectly prevent SG dedifferentiation. To determine whether me31B acts directly in SGs, we used the bam-gal4 driver to deplete me31B only in the four-cell SGs and later stages (Chen and McKearin, 2003b). We found that ∼50% of bam>me31B^RNAi testes contained dedifferentiating SG cysts (Fig. 1D,F). These results demonstrate that me31B is required in SGs in a cell-autonomous manner to prevent their dedifferentiation; however, we note that the frequency of dedifferentiation is higher when RNAi constructs were driven by nos-gal4 than by bam-gal4, suggesting that me31B may have additional functions in early germ cells to indirectly prevent dedifferentiation (see below).

GSC identity in the Drosophila testis is specified by JAK-STAT and BMP signaling (Kawase et al., 2004; Kiger et al., 2001; Schulz et al., 2004; Shivdasani and Ingham, 2003; Tulina and Matunis, 2001). We examined whether the activation of these pathways was altered upon knockdown of me31B.

In wild-type testes, activation of BMP signaling triggers phosphorylation of Mad (pMad) in GSCs and in GBs that are still connected to GSCs (Kawase et al., 2004) (Fig. 2A-A″). We found that knockdown of me31B, either by nos-gal4 or bam-gal4, resulted in a high pMad signal in germ cells outside GSCs and GBs (Fig. 2B-C″). Moreover, in me31B^RNAi testes, we observed high pMad signal in all the germ cells within a dedifferentiating SG cyst attached to the hub (Fig. 2B-C″) and even in SGs that were not yet attached to the hub (Fig. 2B-B″). We observed pMad⁺ germ cells outside the niche in only 7.7% of control testis (n=39 testes), but in over 50% of me31B^RNAi testes (91.7% in nos>me31B^{TRiP.HMS00539}, n=48; 66.7% in nos>me31B^TRiP.GL00695, n=18; 58.8% in bam>me31B^{TRiP.HMS00539}, n=34; 54.8% in bam>me31B^TRiP.GL00695, n=31). These results indicate that the activation of BMP signaling precedes the re-acquisition of GSC identity during dedifferentiation in me31B^RNAi testes, and may mediate dedifferentiation. Indeed, we found that overexpression of constitutively active Tkv (Tkv*) (Nellen et al., 1996), the receptor of BMP ligands, either by nos-gal4 or bam-gal4, was sufficient to induce dedifferentiation (Fig. 2D). Taken together, we propose that me31B may prevent dedifferentiation of SGs by directly or indirectly downregulating BMP signaling.

In contrast to the deregulation of BMP signaling upon knockdown of me31B, we found that GSCs in bam>me31B^RNAi testes had similar STAT expression as control testes (Fig. S2A-B″), suggesting that dedifferentiation induced in bam>me31B^RNAi testes is not due to altered STAT signaling. However, STAT expression was reduced in GSCs of the nos>me31B^RNAi testes compared to control (Fig. S2C-D″), suggesting that me31B may have an additional role in GSCs to maintain STAT activation (see Discussion).

Previous work showed that Me31B silences nos mRNA translation during embryonic development of Drosophila (Gotze et al., 2017; Jeske et al., 2011). In the adult germline, Nos instructs germ cell identity and GSC maintenance via translational repression of critical targets, such as Bam (Li et al., 2009; Wang and Lin, 2004), and a regulatory feedback exists between nos, Mad and bam to control germ cell differentiation (Harris et al., 2011).

To investigate whether Me31B might regulate nos mRNA translation during spermatogenesis, we examined Nos protein levels upon knockdown of me31B. In control testes, we detected Nos protein in early-stage germ cells (GSC to four-cell stage SGs) (Fig. 3A). In contrast, upon knockdown of me31B either by nos-gal4 or bam-gal4, we observed Nos protein even in eight-cell SGs (Fig. 3B-D), consistent with Me31B downregulating nos mRNA translation in the Drosophila testis. Nos and Bam, a master regulator of differentiation (McKearin and Ohlstein, 1995; McKearin and Spradling, 1990), are expressed in a reciprocal manner and act antagonistically in stem cell maintenance and differentiation in the Drosophila germline (Chen and McKearin, 2005; Li et al., 2009). Indeed, me31B knockdown in the testes led to delayed Bam expression and a dramatic increase in the frequency of four-cell SGs that lacked Bam protein (Fig. S3).

To determine whether Me31B regulates nos mRNA levels, we conducted single molecule RNA in situ hybridization to quantify nos mRNA levels (see Materials and Methods). We did not detect any difference in nos mRNA levels when comparing control versus nos>me31B^RNAi testes, either in GSCs or SGs (Fig. 3E-J), suggesting that Me31B does not regulate nos mRNA levels. Taken together, these results suggest that Me31B regulates nos mRNA translation but not mRNA levels, consistent with other contexts in which Me31B acts as a regulator of translation (Nakamura et al., 2001; Peter et al., 2019; Wang et al., 2017).

To determine whether Me31B might regulate nos mRNA translation via direct binding, we performed RNA immunoprecipitation (RIP)-qPCR with testes expressing Me31B-GFP or GFP as a control [see Materials and Methods: note that we also ectopically expressed Dpp to cause SG overproliferation (Kawase et al., 2004; Schulz et al., 2004; Shivdasani and Ingham, 2003) to increase the starting material]. We found that nos mRNA co-immunoprecipitated with Me31B-GFP (Fig. 3K). Interestingly, bam mRNA also co-immunoprecipitated with Me31B-GFP (Fig. 3K). These results indicate that nos mRNA is likely a direct target of Me31B in the testis, and identify bam mRNA as a potential additional target. Overall, we conclude that me31B prevents dedifferentiation of SGs by reducing Nos protein levels and potentially increasing Bam protein levels.

Based on the results described above, we hypothesized that Me31B prevents dedifferentiation in late SGs by silencing nos mRNA translation. This hypothesis predicts that nos downregulation would rescue the elevated dedifferentiation caused by knockdown of me31B. Indeed, we found that simultaneous knockdown of nos and me31B greatly reduced dedifferentiation to the level of the control (Fig. 4A; Fig. S4). These data suggest that nos is the major functional target of me31B in preventing dedifferentiation. To verify that the reduced dedifferentiation in the double knockdown lines is not due to the presence of two UAS-driven transgenes and thus dilution of the gal4 driver, we tested a control genotype expressing me31B^RNAi and a GFP transgene under the control of UAS. This genotype maintained the high frequency of dedifferentiation (Fig. 4A; Fig. S4). These results support the notion that nos is necessary for the dedifferentiation induced by depletion of me31B.

Moreover, we found that upregulation of nos was sufficient to induce dedifferentiation. We employed a nos transgene in which the 3′ untranslated region (UTR) is replaced by the tubulin 3′ UTR (UAS-nos-tub3′UTR), which disrupts the regulation of nos by translational repressors, such as Me31B (Gavis and Lehmann, 1994). When the UAS-nos-tub3′UTR transgene was expressed with the nos-gal4 driver, we found that ∼40% of testes contained dedifferentiating SGs, as opposed to ∼3% in control (Fig. 4B-D′). Moreover, when the UAS-nos-tub3′UTR transgene was driven by bam-gal4, we observed an even higher frequency of dedifferentiation (∼70%) (Fig. 4B,E,E′). These results suggest that upregulation of nos is sufficient to induce dedifferentiation.

Interestingly, when me31B knockdown was combined with nos-tub3′UTR expression under the control of the nos-gal4 driver, it led to a near complete block of differentiation (nos>nos-tub3′UTR, me31B^{TRiP.HMS00539}; Fig. 5). The differentiation block was so severe that our criteria of dedifferentiation used above (i.e. connected cells at the hub with fragmented fusomes) was not applicable, although we frequently observed cysts with fragmenting fusomes, indicative of dedifferentiation. Twenty-nine percent of testes (n=45 testes) contained SGs but never progressed to SC differentiation (which can be recognized by growth in cell size) (Fig. 5B). In addition, 91% of testes (n=45 testes) contained SG cysts with ≥32 cells, further suggesting the failure in differentiation into SC stage (Fig. 5C,C′). It cannot be determined whether these SGs continue to proliferate (e.g. to 64 SG, 128 SG, etc.), as such cysts may also break apart by dedifferentiation. The fact that nos overexpression enhances me31B-knockdown phenotype implies that additional targets of me31B cooperate with misregulated nos to enhance the phenotype. Alternatively, further upregulation of endogenous nos due to me31B depletion and the nos-tub3′UTR transgene may enhance the effect.

Regulation of nos mRNA translation has been well documented and intensively studied, particularly in the context of germ cell specification (Gavis and Lehmann, 1992, 1994; Kugler and Lasko, 2009). The regulation of mRNA translation is critically important during oocyte development: the mRNAs that specify germ cell fate in the embryos, including nos and osk mRNA, are transcribed in nurse cells, transported into developing oocytes and stored in mature oocytes to be translated later (Lehmann, 2016). Accordingly, mRNA synthesis (transcription) is spatially and temporally separated from protein production (translation), making it critically important to control the timing of translation by both translational repression and activation.

Whether nos transcription is spatiotemporally distinct from Nos protein production during the development of male germ cells in the testis is not known. To address this question, we generated a nos promoter reporter by driving a destabilized GFP (d2EGFP) fused to the hsp70 3′UTR from the nos promoter (Fig. 6A). Because neither the mRNA nor protein products are stable in this reporter, the GFP signal closely recapitulates the activity of the nos promoter. Interestingly, we found that the nos promoter is active only in GSCs and GBs that are still connected to GSCs (Fig. 6B,B′), suggesting that nos is transcribed only in these early germ cells. These data suggest that Nos protein that is observed in two- to four-cell stage SGs is primarily produced by translation of nos mRNA inherited from GSCs and GBs (Fig. 6C). In addition, stable Nos protein generated in GSCs and GBs may contribute to its persistence through the four-cell SG stage.

These results reveal dynamic regulation of nos expression through multiple layers (Fig. 6C): (1) GSCs and GBs actively transcribe nos mRNA, which is translated to produce Nos protein; (2) two- and four-cell SGs no longer transcribe nos but inherit nos mRNA, which is translated to produce Nos protein; and (3) ≥eight-cell SGs do not transcribe nos mRNA, and translation of inherited nos mRNA is inhibited by Me31B, leading to overall downregulation of Nos protein. Loss of Me31B leads to increased translation of nos mRNA, thus increased levels of Nos protein, promoting dedifferentiation at later stages.

Stem cell maintenance is critically important for long-term tissue homeostasis. Despite their ability to self-renew, stem cells are not immortal and their life span is often shorter than that of the organism. Dedifferentiation can replenish stem cell pools via conversion of more differentiated cells back into stem cell identity. However, uncontrolled dedifferentiation can lead to tumorigenesis (Landsberg et al., 2012; Schwitalla et al., 2013), thus proper control of dedifferentiation must be essential. Despite its importance, the mechanisms that regulate dedifferentiation are poorly understood.

This study identified me31B as a previously unknown and key negative regulator of dedifferentiation through its ability to regulate nos mRNA. Both nos and bam mRNAs co-immunoprecipitated with Me31B-GFP (Fig. 3G). Me31B may reinforce the known antagonistic relationship between nos and bam in the germline (Chen and McKearin, 2005; Li et al., 2009) by independently regulating these transcripts (Fig. 6C). In addition to extending Nos protein expression to eight-cell SGs and delaying Bam protein expression during germline development, depletion of me31B resulted in upregulation of BMP signaling, leading to an increased frequency of dedifferentiating SG cysts (Fig. 2). It remains unknown whether me31B directly regulates any components of BMP signaling. However, given the antagonistic relationship between nos and bam, and that BMP signaling represses bam expression (Chen and McKearin, 2003a,b, 2005; Harris et al., 2011; Li et al., 2012, 2009; Song et al., 2004; Wang and Lin, 2004), it is possible that BMP upregulation can be explained as a downstream effect of misregulated nos and/or bam.

In contrast to the deregulation of BMP signaling upon knockdown of me31B, STAT does not appear to be a relevant target of me31B in inducing dedifferentiation (Fig. S2). bam>me31B^RNAi testes did not detectably alter STAT signaling. Importantly, when a cyst of dedifferentiating bam>me31B^RNAi SGs was attached to the hub cells, only the germ cells that were in direct contact with the hub had high STAT levels (Fig. S2B, arrow). These results indicate that germ cells in ≥four-cell SG cysts can reestablish STAT signaling upon homing into the niche during dedifferentiation triggered by depletion of me31B. Although downregulation of JAK-STAT signaling is reported to prevent SG dedifferentiation (Sheng et al., 2009), our data suggest that the dedifferentiation induced by depletion of me31B does not directly involve the activation of the JAK-STAT pathway. We speculate that JAK-STAT signaling might help maintain GSCs that were generated by dedifferentiation, instead of inducing dedifferentiation per se. Interestingly, however, STAT expression was reduced in GSCs of the nos>me31B^RNAi testes compared to controls (Fig. S2C-D″), suggesting that me31B has an additional role in GSCs to maintain STAT activation. Reduced STAT in nos>me31B^RNAi testes, which may deplete native GSCs, might explain the higher frequency of dedifferentiation with nos-gal4-driven me31B^RNAi compared to bam-gal4-driven me31B^RNAi (Fig. 1F).

It remains elusive what controls me31B to promote differentiation and/or prevent dedifferentiation. Is me31B downregulated by conditions that trigger dedifferentiation? We did not observe any changes in Me31B-GFP protein level or localization when dedifferentiation was artificially induced by transient expression of Bam (not shown). In future studies, it will be of interest to investigate whether and how Me31B senses niche vacancy (missing GSCs) to trigger dedifferentiation of SGs.

The right balance of differentiation and dedifferentiation must be achieved to ensure maintenance of the stem cell pool, while minimizing the risk of tumorigenesis. The results presented in this study suggest that SGs are in a state of transition from stem cell identity to full commitment to differentiation (SC). Whereas GSCs produce Nos protein via nos mRNA transcription and its translation, two- and four-cell SGs produce Nos protein only via translation of inherited nos mRNA. We propose that two- and four-cell SGs represent a critical cell population/developmental time window that is not yet fully committed to differentiation but maintains the potential to dedifferentiate, as they still have Nos protein like GSCs, but no longer transcribe nos unlike GSCs (Fig. 6C). These SGs may hit a perfect balance of Nos protein that maintains their potential to dedifferentiate into GSCs as necessary, but prevents tumorigenesis by shutting down nos transcription. Indeed, two- and four-cell SGs are known to be most potent for dedifferentiation (Sheng and Matunis, 2011): although this was speculated to be mostly due to their physical proximity to the hub cells, it is also possible that their ‘Nos production state’ (actively producing Nos protein from inherited mRNA) is more suited for dedifferentiation than later SGs. We propose that stepwise transitions from the stem cell state to the differentiated state are key for maintaining the stem cell pool while preventing tumorigenesis. In summary, the present study provides a new insight into how gradual commitment to differentiation is ensured by transcriptional and translational control of a key regulator of cell fate.

Unless otherwise stated, all flies were raised on standard Bloomington medium at 25°C, and young flies (1- to 3-day-old adults) were used for all experiments. See Table S1 for the list of stocks used in this study.

Immunofluorescence staining was performed as described previously (Cheng et al., 2008). Briefly, tissues were dissected in PBS, transferred to 4% formaldehyde in PBS and fixed for 30 min. Tissues were then washed in PBS-T (PBS containing 0.1% Triton X-100) for at least 30 min (three 10 min washes), followed by incubation with primary antibody in 3% bovine serum albumin (BSA) in PBS-T at 4°C overnight. Samples were washed for 60 min (three 20 min washes) in PBS-T, incubated with secondary antibody in 3% BSA in PBS-T at 4°C overnight, washed as above, and mounted in Vectashield with DAPI (Vector Labs). The antibodies used are described in Table S2. Images were taken using a Leica TCS SP8 confocal microscope with 63× oil-immersion objectives (NA=1.4). Images were processed using Adobe Photoshop and ImageJ software.

Dedifferentiating SG cysts were identified as the cysts containing ≥3 SGs that are connected to each other by the fragmented fusome (Fig. 1C,D). In contrast, normally differentiating SGs contain continuous fusome that connects all cells within the cyst (Fig. 1E). Thus, the morphology of the fusome distinguishes differentiating SGs versus dedifferentiating SGs. The connectivity of cells within the dedifferentiating SGs was determined by the presence of fusome fragments between two cells within a cyst: for example, the connection between cell 1 and cell 2 can be confirmed by the presence of the fusome fragment between these two cells. Cell 2 may be then connected to cell 3 with another fragment of the fusome, establishing the connectivity of cell 1, 2 and 3, and so on. In rare cases, when two cells clearly shared the cytoplasm by continuous Vasa staining, such cell pairs may be determined as connected without the presence of fusome in between. When ≥3 cells were determined to be connected to each other with this method, and found at the hub cells, such SG cysts were scored as ‘dedifferentiating’. Significance was determined using a Fischer's Exact Test in comparison to a control.

To detect nos mRNA, single molecule fluorescent in situ hybridization (smFISH) was conducted by following a previously described protocol (Fingerhut et al., 2019). All solutions used for smFISH were RNase free. Testes from 2-3-day-old flies were dissected in 1× PBS and fixed in 4% formaldehyde in 1× PBS for 30 min. Then testes were washed briefly in PBS before being rinsed with wash buffer [2× saline-sodium citrate (SSC) and 10% formamide] and then hybridized overnight at 37°C in hybridization buffer [2× SSC, 10% dextran sulfate (Sigma-Aldrich, D8906), 1 mg ml⁻¹E. coli tRNA (Sigma-Aldrich, R8759), 2 mM vanadyl ribonucleoside complex (New England Biolabs, S142) and 0.5% BSA (Ambion, AM2618), 10% formamide]. Following hybridization, samples were washed three times in wash buffer for 20 min each at 37°C and mounted in Vectashield with DAPI (Vector Labs). Images were acquired using an upright Leica TCS SP8 confocal microscope with a 63× oil immersion objective lens (NA=1.4) and processed using Adobe Photoshop and ImageJ software. Fluorescently labeled probes were added to the hybridization buffer to a final concentration of 50 nM (for satellite DNA transcript targeted probes). Probe set against nos exons was designed using the Stellaris RNA FISH Probe Designer (Biosearch Technologies) available online at www.biosearchtech.com/stellarisdesigner. The Stellaris RNA FISH (Biosearch Technologies) probes were labeled with Quasar 670. Probe set was added to the hybridization buffer at a 50 nM final concentration. For smFISH probe sequences, see Table S3.

We thank the Bloomington Drosophila Stock Center, the Developmental Studies Hybridoma Bank and Drs Dennis McKearin and Liz Gavis for reagents. We thank Yamashita lab members and Dr Angela Anderson (Life Science Editors) for discussion and/or comments on this manuscript.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258757