Introduction

In sexually reproducing animals, germ cells give rise to the gametes, which carry genetic information to the next generation. As such, germ cells are often regarded as being immortal [1]. Hence, it is vital that germ cells are tightly regulated, both during development and adulthood, to safeguard the survival of the entire species.

In Drosophila, germ cells are generated in the developing embryo through a process known as preformation, via the inheritance of a specialized maternally provided cytoplasm termed the germplasm [2]. In the syncytial embryo, nuclei that migrate to the posterior pole will encounter the germplasm and cellularize to form primordial germ cells (PGCs) [3]. These PGCs proliferate asynchronously from the adjacent somatic cells, and by stage 5 of embryogenesis, cease their divisions to form the final pool of 30–40 PGCs [4]. During gastrulation, PGCs are initially passively brought into the embryo where they will begin their active migration towards the somatic gonadal precursors (SGPs) to form the embryonic gonads at stage 13 (Fig. 1a) [5]. Of the final pool of PGCs formed at stage 5, only a fraction of these PGCs successfully make it to the gonads [3]. Some of these PGCs die during their active migration whereas some mismigrate and eventually die [6,7,8,9]. Subsequently, during larval development, a homeostatic mechanism is in place to correct for any drop in the number of PGCs in the gonads [10]. Whether such PGC homeostasis occurs during early embryogenesis before they reach the gonads remains unclear.

Fig. 1: Maternal sisR-2 promotes PGC death.
figure 1

a Diagram highlighting key processes of PGC development during Drosophila embryogenesis. b Diagram showing the cell types in a Drosophila germarium in the ovary. TF terminal filament, CC cap cells, GSC germline stem cells, EC escort cells, CB cystoblasts. c Confocal images showing the germaria of the indicated genotypes stained with alpha-Spectrin (green) and Vasa (red). Control: sisR-2 RNAi parental. Maternal sisR-2 RNAi: vasa-Gal4 driven RNAi. GSCs were marked by asterisks (*). d Chart showing the percentage of germaria with the indicated number of GSCs in different genotypes shown in c. ***p < 0.001. Student’s t test was performed, comparing the mean number of GSCs of the indicated genotypes. e Confocal images showing the PGCs in stage 5 embryos of the indicated genotypes stained with Vasa (red). Control: sisR-2 RNAi parental. Maternal sisR-2 RNAi: vasa-Gal4 driven RNAi. f Chart showing the number of PGCs in different genotypes shown in e. Student’s t test was performed, comparing the mean number of PGCs of the indicated genotypes. Error bars depict SD. g Chart showing the total number of PGCs in different genotypes in stage 5, 11 and 13 as shown in e, h and (S1B). ***p < 0.001. Student’s t test was performed, comparing the mean number of PGCs of the indicated genotypes. Error bars depict SD. h Confocal images showing the PGCs in stage 13 embryos of the indicated genotypes stained with Vasa (red). Yellow arrowheads: PGCs in gonads. White arrowhead: mismigrated PGCs. Control: sisR-2 RNAi parental. Maternal sisR-2 RNAi: vasa-Gal4 driven RNAi. i Chart showing the number of PGCs in the gonads of different genotypes shown in h. ***p < 0.001. Student’s t test was performed, comparing the mean number of PGCs of the indicated genotypes. Error bars depict SD. j Confocal images showing the PGCs in stage 11 embryos of the indicated genotypes stained with Vasa (red) and pH3 (yellow). Control: sisR-2 RNAi parental. Maternal sisR-2 RNAi: vasa-Gal4 driven RNAi. k Chart showing the number of mismigrated PGCs in the embryos of different genotypes shown in h. ***p < 0.001. Student’s t test was performed, comparing the mean number of mismigrated PGCs of the indicated genotypes. Error bars depict SD. l Confocal images showing the PGCs in stage 11 embryos of the indicated genotypes stained with Vasa (red) and TUNEL (green). Panels on the right are magnified images of the dashed yellow box. White arrowhead: TUNEL-positive PGCs. Control: y w. m Chart showing the percentage of embryos with TUNEL-positive PGCs of different genotypes shown in l. ***p < 0.001. Fisher’s exact test was performed, comparing the number of embryos with or without TUNEL-positive PGCs. Scale bar: 10 μm.

Stable intronic sequence RNAs (sisRNAs) are proposed to function as an additional layer of gene regulation [11]. sisRNAs have been reported to participate in regulatory feedback loops to either enhance or repress gene expression, as well as acting as protein decoys to influence splicing of RNAs [12,13,14,15]. We previously characterized sisR-2, an ovary-enriched sisRNA that is maternally deposited into the oocytes [16]. We showed that sisR-2 represses germline stem cell (GSC) number in the adult females via a lipid metabolism gene dFAR1 [17]. In addition, we found that sisR-2 is upregulated during starvation, which would lead to a loss of GSCs [17]. However, we discovered a negative feedback loop involving the miRNA bantam, which represses the activity of sisR-2 [17]. As a result, this homeostatic mechanism prevents the loss of GSCs during starvation in Drosophila [17]. In addition, we found that maternally deposited sisR-2 influenced the number of GSCs in the female progeny, however, the cellular and molecular mechanisms are still unknown.

Here, we study the function of sisR-2 in PGC development and uncover the presence of a homeostatic mechanism regulating PGCs during embryogenesis. We found that sisR-2 represses PGC number by promoting PGC death. Surprisingly, increasing the maternally deposited pool of sisR-2 resulted in an increase in PGC death, but no apparent drop in PGC number. This is due to ectopic division of PGCs via the de-repression of Cyclin B, which is governed by a genetic pathway involving sisR-2, bantam and brat. Our study provides evidence for a mechanism that achieves germline homeostasis by preserving PGC number during embryogenesis.

Methods

Fly strains

Flies were maintained in standard cornmeal medium at 25 °C unless otherwise stated. The following Gal4 drivers were used to drive UAS-transgene expression in the germline: MTD-Gal4 [18], nanos-Gal4-VP16 [19], vasa-Gal4 (gift from Y. Yamashita) and NGT40-Gal4-VP16 [20]. vasa-Gal4/CyO;sisR-2 RNAi-1 and UAS-sisR-2 was generated previously [17, 21]. UAS-bantam sponge, UAS-bantam, and UAS-bantam sensor were gifts from Cohen [22, 23]. brat[11], UAS-brat, and UAS-brat[GD] were gifts from Ashe [24]. UAS-hid (Bloomington #65408), hid[05014] (Bloomington #83349) and TOR RNAi (Bloomington #339510) were obtained from the Bloomington Stock Center. dsRed-intron-myc overexpression flies was generated as described previously [16]. Mutations were introduced into the mbt intron using the Q5 site-directed mutagenesis kit (New England Biolabs). Injection was carried out by BestGene Inc. Oligonucleotides sequences are available in Table S1. Starvation experiments were done as described previously [17]. For embryo collections, females and males of the indicated genotypes were crossed in cages with apple juice plates, supplemented with wet yeast paste. Embryos were collected hourly, and allowed to develop to the required embryonic stage at either 18 °C or 25 °C [25].

Immunostaining

Immunostaining of ovaries was performed as described previously [17]. Ovaries were fixed in a solution of 16% paraformaldehyde and Grace’s medium at a ratio of 2:1 for 20 min, rinsed, and washed with PBX solution (PBS containing 0.2% Triton X-100) three times for 10 min each, and pre-absorbed for 30 min in PBX containing 5% normal goat serum. Ovaries were incubated overnight with primary antibodies at room temperature, washed three times for 20 min each with PBX before a 4 h incubation with secondary antibodies at room temperature. Ovaries were again washed three times for 20 min each with PBX. Staged embryos were dechorionated in 50% bleach (1:1, water:bleach) for 3 min and fixed in a scintillation vial containing heptane and fresh 4% paraformaldehyde in PBS with shaking for 20 min. Dechorionated fixed embryos were devitellinised in 1:3, heptane:methanol with vigorous shaking for 1 min. Devitellinised embryos were rinsed in methanol five times. Primary antibodies used in this study are as follows: guinea pig anti-Vasa (1:1000) [26], mouse monoclonal anti-a-Spectrin (3A9, 1:1; Developmental Studies Hybridoma Bank), mouse monoclonal anti-pH3 (ab14955, 1:200; Abcam), rabbit anti-Hid (gift from HD Ryoo) and mouse monoclonal anti-CyclinB (F2F4, undiluted; Developmental Studies Hybridoma Bank). For accurate counting of PGCs in stage 5 embryos, immunostained embryos were sliced using a sharp needle at the posterior end. Sliced embryo sections were mounted with the side containing PGCs facing the coverslip [7]. TUNEL assay was done using the In Situ Cell Death Detection Kit (Roche). Images were taken using Leica SPEII microscope and processed using either Adobe Photoshop, ImageJ or the Leica LAS X software.

Immunostaining signal quantification

Immunostaining of sample embryos were carried out in parallel. Images were taken under identical confocal settings and analyzed using ImageJ. PGCs were identified using Vasa staining. Anti-Hid, anti-Brat or anti-Cyclin-B fluorescence intensity was measured at three separate areas in the same cell, and an average was taken. Anti-Vasa fluorescence intensity was similarly measured at 3 separate areas in the same cell and used for normalization.

RNA extraction

RNA extraction was done as described previously [17]. Tissues were homogenized in 1.5 ml Eppendorf tubes using a plastic pestle and RNA was extracted using the TRIzol extraction protocol (Ambion) or the Direct-zol RNA miniprep kit (Zymo Research). RNA was quantified with the NanoDrop 2000 spectrophotometer (Thermo Scientific).

RT-PCR

For standard RT-PCR, total RNA was reverse transcribed with random hexamers or oligo-dT for 1 h using AMV-RT (New England Biolabs), M-MLV RT (Promega) or Superscript III (Invitrogen). PCR was carried out using the resulting cDNA. For qPCR, SYBR Fast qPCR kit master mix (2X) universal (Kapa Biosystems, USA) was used with addition of ROX reference dye high and carried out on the Applied Biosystems 7900HT Fast Real-Time PCR system. Oligonucleotides sequences are in Table S1 and reported previously [17].

Western blotting

Western blotting was performed as previously described [17]. Protein lysates were run on an SDS gel and transferred to a PVDF membrane. Antibodies used were mouse anti-GFP (1:1000; Invitrogen) and mouse anti-alpha Tubulin (1:10,000; Millipore). Western blot detection was done digitally using the ChemiDoc Touch Imaging System (BioRad) and under non-saturating conditions.

Statistics

In all experiments, the tests that were used and the number of independent biological replicates and gonads/embryos were indicated in the figure legends or figures. P values and definitions of error bars were indicated in the legends. Sample sizes were not pre-determined prior to the experiments. T-tests were performed on samples that are normally distributed.

Results

Maternal sisR-2 promotes PGC death

We previously showed that reducing the levels of maternally deposited sisR-2 led to more GSCs in the resulting female progeny dissected at day 8 [17]. Interestingly, the increase in number of GSCs was also observed in female flies dissected immediately after eclosion (Fig. 1b–d). Since GSCs are derived from the PGCs formed in the early embryos, and sisR-2 is maternally deposited, we hypothesized that maternal sisR-2 regulates PGCs during embryogenesis. First, to examine if sisR-2 regulates the formation of PGCs, we counted the number of PGCs in stage 5 embryos with reduced maternally deposited sisR-2 (hereafter referred to as sisR-2 RNAi embryos) (Fig. S1A and S1B). We did not observe a change in the number of PGCs between controls and stage 5 sisR-2 RNAi embryos (Fig. 1e–g), indicating that sisR-2 does not regulate the formation of PGCs in the early embryos (Fig. 1a).

Next, to investigate if maternal sisR-2 regulates the number of PGCs that reach the embryonic gonads, we counted the number of PGCs in stage 13 embryos (Fig. 1a). Interestingly, we observed a significant increase in the number of PGCs in the embryonic gonads of stage 13 sisR-2 RNAi embryos as compared to the controls (19.8 ± 4.9 PGCs in controls vs. 27.0 ± 5.3 PGCs in sisR-2 RNAi, yellow arrowheads) (Fig. 1g–i). Consistently, we also observed an increase in PGC number in stage 13 embryos with the same maternal sisR-2 RNAi driven by another Gal4 driver (MTD-Gal4) (Fig. S1B–S1E). Moreover, this increase in PGC number was also observed in stage 11 embryos, when the PGCs are still actively migrating (24.7 ± 3.5 PGCs in controls vs. 30.3 ± 3.8 PGCs in sisR-2 RNAi) (Figs. 1g, S1F, and S1G). PGCs do not undergo mitosis during these two stages although they are competent to divide (Fig. 1a) [4, 27]. Thus, we checked if the increase in the number of PGCs was due to premature proliferation by staining with a mitotic marker anti-phospho-Histone H3 (pH3). As expected, no pH3-positive PGCs were detected in the stage 11 control embryos (n = 100) (Fig. 1j). We did not observe any pH3-positive PGCs in stage 11 sisR-2 RNAi embryos (n = 100), indicating that PGCs were not dividing (Fig. 1j). Moreover, we did not detect an increase in the mitotic cyclin Cyclin B in PGCs from sisR-2 RNAi embryos further confirming that the increase in PGCs in sisR-2 RNAi embryos cannot be attributed to proliferation (Fig. S1H and S1I).

In wunen mutants, the number of PGCs that reach the embryonic gonads decreases due to a defect in PGC migration [9]. Since there was an increase in PGCs reaching the embryonic gonads in sisR-2 RNAi embryos, we wondered if sisR-2 represses PGC migration. We counted the number of mismigrated PGCs in stage 13 embryos and did not detect a decrease in mismigrated PGCs in sisR-2 RNAi embryos as compared to controls (Fig. 1h, k, white arrowheads). Instead, we observed a slight but significant increase in mismigrated PGCs in sisR-2 RNAi embryos (2.1 ± 1.6 PGCs in controls vs. 4.2 ± 2.2 PGCs in sisR-2 RNAi (vasa-Gal4) and 6.5 ± 2.4 in sisR-2 RNAi (MTD-Gal4)) (Figs. 1h, k, S1D and S1J). Thus, this observation rules out the possibility that the increase in PGCs observed in the embryonic gonads of sisR-2 RNAi embryos is due to enhanced PGC migration.

Since a fraction of PGCs die during their active migration to the embryonic gonads, we can assume that the increase in PGCs observed in sisR-2 RNAi embryos must be a result of a decrease in PGC death (Fig. 1a) [6, 28]. We examined if there was a decrease in PGC death in sisR-2 RNAi embryos using TUNEL labeling. However, we were unable to detect TUNEL-positive PGCs even in the control embryos. This is consistent with published literature reporting that it is difficult to detect dying PGCs in wild type embryos [8, 29]. Hence, we wondered if we could instead observe dying PGCs by overexpressing sisR-2. Indeed, when we looked for the presence of dying PGCs using TUNEL labeling in stage 11 embryos with increased maternally deposited sisR-2 (hereafter referred to as sisR-2 overexpressing embryos) (Fig. S1K and S1L), we were able to detect TUNEL-positive PGCs (0% of control embryos, n = 29 vs. 31% of sisR-2 overexpressing embryos, n = 51) (Fig. 1l, m). Furthermore, this increase in TUNEL-positive cells was specific to PGCs as we did not observe an increase in TUNEL-positive somatic cells in the sisR-2 overexpressing embryos (Fig. S1M). Taken together, our experiments suggest that maternal sisR-2 promotes PGC death during embryogenesis.

sisR-2 and bantam functionally interact

In Drosophila, the pro-apoptotic gene head involution defective (hid) is zygotically expressed in PGCs and is implicated in PGC death [8, 30]. Interestingly, hid is a verified target for the bantam miRNA [22]. We previously showed that bantam negatively feedbacks and represses sisR-2 activity to prevent the loss of GSCs during starvation (Fig. 2a) [17]. We speculated that bantam represses sisR-2 activity via a potential 17-nucleotide stable base pairing with sisR-2, consisting of a non-canonical seed region containing two G:U wobble base pairs (Fig. 2b, c) [17]. Overexpression of sisR-2 in the ovaries under fed conditions did not result in a loss of GSCs, possibly due to the repression by bantam [17]. To further examine the functional interaction between sisR-2 and bantam, we expressed a mutant form of sisR-2, which disrupts the base pairing with bantam (Figs. 2b–d, S1L and S2). As expected, overexpression of this mutant form of sisR-2 in the ovaries resulted in a loss of GSCs (Fig. 2e, f). Hence, this genetic evidence further suggests that sisR-2 and bantam can functionally interact by complementary base-pairing. However, we cannot totally exclude the possibility that the sequence also regulates other targets besides bantam. In addition, overexpression of the mutant form of sisR-2 (bantam base-pairing disrupted) no longer resulted in an increase in embryos with TUNEL-positive PGCs as observed in the wild type sisR-2 overexpression (1.8% of sisR-2 mutant overexpressing embryos, n = 53) (Fig. 2g, h). This experiment suggests that the bantam targeting sequence is required to induce cell death in the PGCs.

Fig. 2: sisR-2 and bantam functionally interact.
figure 2

a Model showing the negative feedback loop between sisR-2 and bantam in regulating GSC maintenance during starvation. b Sequence of bantam indicating potential stable base-pairing with sisR-2. The region mutated to disrupt the seed base-pairing is indicated in blue. c Predicted sisR-2 secondary structure. Green: bantam base-pairing region. Blue: mutated nucleotides as indicated in b. Red: complementary mutations introduced to preserve sisR-2 secondary structure. d qPCR showing the relative levels of sisR-2 in ovaries of the indicated genotypes. e Confocal images showing the germaria of the indicated genotypes stained with alpha-Spectrin (green) and Vasa (red). GSCs were marked by asterisks (*). f Chart showing the percentage of germaria with the indicated number of GSCs in different genotypes shown in e. **p < 0.01. Fisher’s exact test was performed, comparing the percentage of germaria with 2≤ or >2 GSCs. Scale bar: 10 μm. g Confocal images showing the PGCs in stage 11 embryos of the indicated genotypes stained with Vasa (red) and TUNEL (green). Panel below is a magnified image of the dashed yellow box. h Chart showing the percentage of embryos with TUNEL-positive PGCs of different genotypes shown in g and (1 L). The percentage of embryos with TUNEL-positive PGCs in the control (y w) and Maternal MTD > UAS-sisR-2 (WT) first appeared in Fig. 1m. ***p < 0.001. Fisher’s exact test was performed, comparing the number of embryos with or without TUNEL-positive PGCs.

Maternal sisR-2 promotes PGC death via bantam and hid

The Drosophila oocyte is transcriptionally quiescent, containing a store of maternally deposited proteins and mature RNAs important for early embryonic development [31]. Like sisR-2, bantam is also maternally deposited into the oocytes [32]. In transcriptionally active GSCs, sisR-2 promotes the transcription of bantam, while bantam targets sisR-2 to repress its activity [17]. We reasoned that in the oocyte where regulation of RNA occurs only at the posttranscriptional level, the negative feedback loop between sisR-2 and bantam is no longer active (Fig. 3a). Since sisR-2 and bantam can functionally interact via base-pairing, we wondered if the outcome of this interaction could be switched to sisR-2 inhibiting bantam from regulating its target gene(s). To monitor the activity of bantam in a sisR-2 RNAi background, we used a bantam sensor transgene [22]. The bantam sensor transgene expresses GFP under a tubulin promoter, with two copies of a perfect bantam target site in the 3’UTR. Thus, bantam activity is reported indirectly through the levels of GFP. Interestingly, we detected a decrease in the levels of GFP in sisR-2 RNAi ovaries indicating an increase in bantam activity (Fig. 3b, c). Because stage 14 oocytes are large, and well-fed flies contain abundant stage 14 oocytes in the ovaries, whole ovaries generally accurately reflect the status of stage 14 oocytes. Thus, our experiment suggests that sisR-2 acts as a negative regulator of bantam activity in the oocytes and early embryos.

Fig. 3: Maternal sisR-2 promotes PGC death via bantam and hid.
figure 3

a Diagram showing the transcriptional status of germ cells during the different stages of Drosophila oogenesis. b Western blot showing the level of bantam-GFP sensor in ovaries of control and sisR-2 RNAi flies. c Graph showing the relative levels of GFP normalized to Tubulin, as shown in b. *p < 0.05. Student’s t test was performed. Error bars depict SD from five biological replicates. d Working model. e, g Confocal images showing the PGCs in stage 13 embryos of the indicated genotypes stained with Vasa (red). Yellow arrowheads: PGCs in gonads. Maternal sisR-2 RNAi: vasa-Gal4 driven RNAi. Maternal bantam sponge: vasa-Gal4 driven bantam sponge. f, h Charts showing the number of PGCs in the gonads of different genotypes shown in e and g. Controls in both charts are from the same sample. *p < 0.05, ***p < 0.001. Student’s t test was performed, comparing the mean number of PGCs of the indicated genotypes. Error bars depict SD. i Confocal images showing the PGCs in stage 11 embryos of the indicated genotypes stained with Hid (green) and Vasa (red). j Chart showing the relative Hid fluorescence intensity in the PGCs (normalized to soma) shown in i. **p < 0.01. Student’s t test was performed. Error bars depict SD from five embryos. Scale bar: 10 μm.

Since hid is a target of bantam, we first examined if expression of hid is sufficient to induce PGC death (Fig. 3d). Consistent with a previous report, overexpression of hid using the NGT40 driver led to PGC death (20.9 ± 3.2 PGCs in controls vs. 10.9 ± 3.6 PGCs in NGT40 > UAS-hid) (Figs. 3e, f, and S3A) [9]. Furthermore, overexpression of maternal bantam led to an increase in PGCs, phenocopying sisR-2 RNAi (20.9 ± 3.2 PGCs in controls vs. 24.8.3 ± 5.8 PGCs in MTD > UAS-bantam) (Figs. 3g, h, and S3B). Next, we hypothesized that sisR-2 promotes PGC death by inhibiting bantam (Fig. 3d). Remarkably, we found that reducing the activity of the maternally deposited bantam using a bantam sponge transgene could rescue the PGC phenotype in stage 13 sisR-2 RNAi embryos (32.4 ± 5.3 PGCs in sisR-2 RNAi vs. 20.4 ± 3.7 PGCs in sisR-2 RNAi; bantam sponge) (Figs. 3g, h, and S4A and S4B). Expression of bantam sponge alone had no effect on PGC number, indicating a specific genetic interaction between sisR-2 and bantam (Figs. 3g, h, and S3B). To investigate if hid indeed acts downstream of sisR-2 and bantam in the regulation of PGCs, we examined the expression of Hid protein in sisR-2 RNAi embryos. As expected, Hid protein levels were down-regulated in sisR-2 RNAi embryos (Fig. 3i, j). Next, we elevated the expression of hid in the PGCs of sisR-2 RNAi embryos. Increase in hid expression rescued the PGC phenotype in stage 13 sisR-2 RNAi embryos, confirming a genetic interaction between sisR-2 and hid (34.4 ± 4.8 PGCs in sisR-2 RNAi vs. 20.6 ± 6.1 PGCs in sisR-2 RNAi; NGT40 > UAS-hid) (Figs. 3e, f, S4C and S4D). Finally, if sisR-2 promotes hid expression by repressing bantam, reducing sisR-2 is expected to inhibit PGC death by hid overexpression. Indeed, sisR-2 RNAi suppressed the PGC death phenotype caused by hid overexpression (10.9 ± 3.6 PGCs in NGT40 > UAS-hid vs. 20.6 ± 6.1 PGCs in sisR-2 RNAi; NGT40 > UAS-hid) (Fig. 3e, f).

Maternally deposited wunen2, p53 and nanos have been reported to play a role in PGC survival [6,7,8]. We wondered if sisR-2 promotes PGC death by regulating the expression any of these three genes as well. Levels of wunen2 and p53 remained unchanged in sisR-2 RNAi ovaries (Fig. S4E and S4F). Levels of nanos also remained unchanged when expression of sisR-2 was elevated in the ovaries [17]. Thus, it is unlikely that maternal sisR-2 promotes PGC death by regulating any of these genes. Taken together, maternal sisR-2 promotes PGC death by inhibiting bantam activity, which leads to the derepression of the pro-apoptotic gene hid (Fig. 3d).

TOR pathway inhibits sisR-2 expression

Previously, we have shown that nutritional deprivation promotes the expression of sisR-2 in the Drosophila ovaries [17]. In many organisms, including Drosophila, the PI3K/AKT signaling pathway is central in mediating changes in nutrition to cellular function [33]. Hence, we hypothesized that this pathway may be involved in regulating sisR-2 during starvation. During nutrient deprivation, reduced PI3K/AKT signaling results in the inhibition of target of rapamycin (TOR) (Fig. 4a). Thus, we asked if expression of sisR-2 is elevated when TOR was inhibited using TOR RNAi. Interestingly, we observed an upregulation of sisR-2 and its host gene mushroom bodies tiny (mbt) in nos-Gal4 > TOR RNAi ovaries suggesting that the PI3K/AKT pathway promotes the expression of sisR-2 via its host gene mbt during starvation (Fig. 4a–c).

Fig. 4: sisR-2 overexpression induced ectopic proliferation of PGCs.
figure 4

a Proposed model. qPCR showing the relative levels of b sisR-2 and c mbt pre-mRNA in ovaries of the indicated genotypes. ***p < 0.001. Student’s t test was performed. Error bars depict SD from three biological replicates. d Confocal images showing the PGCs in stage 13 embryos of the indicated genotypes stained with Vasa (red). Yellow arrowheads: PGCs in gonads. Control: y w. e Chart showing the number of PGCs in the gonads of different genotypes shown in d. ***p < 0.001. Student’s t test was performed, comparing the mean number of PGCs of the indicated genotypes. Error bars depict SD. f Confocal images showing the PGCs in stage 12 embryos of the indicated genotypes stained with Vasa (red) and pH3 (green). Panels below are magnified images of the dashed yellow box. Control: y w. Scale bar: 10 μm.

sisR-2 overexpression induced ectopic division of PGCs

Since starvation promotes the expression of sisR-2 [17], we wondered if embryos laid by starved female flies exhibit an increase in PGC death. However, starved female flies lay very few eggs, making it technically challenging to collect many staged embryos for the counting of PGCs. Hence, we quantified the number of PGCs in stage 13 sisR-2 overexpressing embryos. Although embryos with increased maternal sisR-2 derived from fed mothers is not equivalent to embryos derived from starved mothers, it has been demonstrated that oocytes produced by mothers exposed to these two contrasting nutritional conditions are rather similar [34]. As such, embryos with increased maternal sisR-2 can likely mirror embryos with starvation induced upregulation of sisR-2. We counted the number of PGCs in sisR-2 overexpressing embryos and did not detect a decrease in the number of PGCs at stage 13 (Figs. 4d, e, S1K and S5A). Instead, we detected a slight increase in PGC number in sisR-2 overexpressing embryos (Fig. 4e). However, this result was puzzling since we did detect an increase in TUNEL-positive PGCs in stage 11 embryos overexpressing sisR-2, which indicated an increase in PGC death (Fig. 1l, m).

During larval development, a decrease in the number of PGCs is corrected by a homeostatic mechanism that involves PGC proliferation [10]. We speculated that although there was an increase in PGC death in the sisR-2 overexpressing embryos, the inability to observe a decrease in PGCs might be due to ectopic PGC division. Consistent with this idea, we were able to detect the presence of pH3-positive PGCs in stage 12 sisR-2 overexpressing embryos, indicating that some PGCs were indeed undergoing mitosis (3% in sisR-2 overexpression vs. 0% in control embryos, n = 100) (Fig. 4f). Moreover, this increase in pH3-positive cells was specific to PGCs as we did not observe an increase in pH3-positive somatic cells in the sisR-2 overexpressing embryos (Fig. S5B). Remarkably, pH3-positive PGCs were not observed in stage 12 embryos overexpressing the mutant form of sisR-2 suggesting that the effect of sisR-2 on PGC division also involves bantam (Fig. S5C).

bantam represses brat in PGCs

From stage 5 till stage 14 of embryogenesis, PGCs are arrested in the G2 phase of the cell cycle [4]. Their transition to mitosis is prevented during migration due to the repression of Cyclin B production by the Nanos/Pumilio translational repressor complex [4, 27]. Overexpression of Cyclin B is sufficient to induce ectopic proliferation of PGCs [4, 27]. In the early embryo, the Nanos/Pumilio complex together with the NHL domain protein Brain tumor (Brat), represses the translation of hunchback mRNA to establish posterior patterning of the embryo. The repression of Cyclin B production however does not require Brat [35].

Interestingly, brat has been shown to be a target of bantam in the Drosophila larval brain [36]. Hence, we hypothesized that in the sisR-2 overexpressing embryos, increase in repression of bantam by sisR-2 might lead to an increase in Brat protein expression in PGCs (Fig. 5a). To examine this, we checked if bantam regulates brat in the PGCs. Increasing the maternally deposited pool of bantam led to a decrease in Brat protein expression (Figs. 5b, c, and S6). Interestingly, this drop in Brat was observed specifically in the PGCs and not in the neighboring somatic cells (Fig. 5b, c). To confirm this observation, decreasing the activity of maternal bantam using bantam sponge resulted in an increase in Brat protein levels specifically in the PGCs but not in the somatic cells (Fig. 5b, c). Together, these experiments indicate that maternally deposited bantam represses brat in PGCs (Fig. 5a).

Fig. 5: bantam represses brat in PGCs.
figure 5

a Working model. b Confocal images showing the PGCs in stage 5 embryos of the indicated genotypes stained with Vasa (red) and Brat (green). Yellow arrowheads: PGCs at the posterior tip. Control: y w. c Chart quantifying the Brat fluorescence intensity in PGCs (black bars) and somatic cells (white bars) of the indicated genotypes shown in b. **p < 0.01, ***p < 0.001. Student’s t test was performed. Error bars depict SD from five embryos. Scale bar: 10 μm.

Ectopic Brat expression disrupts cyclin B repression

Since Brat interacts with the Nanos/Pumilio complex, we speculated that the increase in Brat in PGCs might interfere with the repression of cyclin B by Nanos/Pumilio, resulting in an increase in Cyclin B production and division of PGCs (Fig. 5a). We asked if ectopically expressing Brat in the PGCs can disrupt the repression of cyclin B by overexpressing Brat in the embryos using the germline driver MTD-Gal4. Increasing production of Cyclin B eventually results in its shuttling and accumulation into the nucleus, marking the beginning of the progression from the G2 to M phase of the cell cycle [37, 38]. We observed PGCs with accumulation of Cyclin B protein in the nuclei in PGCs from embryos with germline overexpression of Brat, which was not observed in the control embryos (Figs. 6a and S7A). Moreover, germline overexpression of a mutant form of Brat, BratG774D, which has reduced binding ability to Pumilio, did not exhibit nuclear accumulation of Cyclin B protein [35] (Figs. 6a and S7A). This experiment suggests that the binding of Brat to Pumilio is required for the de-repression of Cyclin B production. Thus, these experiments suggest that Brat disrupts the repression of cyclin B by the Nanos/Pumilio complex in the PGCs (Fig. 5a).

Fig. 6: Ectopic Brat expression disrupts cyclin B repression.
figure 6

a, b Confocal images showing the PGCs in stage 12 embryos of the indicated genotypes stained with Vasa (red) and Cyclin B (green). Panels on the right are magnified images of the dashed yellow box. Yellow arrowheads: PGCs with nuclear translocation of Cyclin B. Control: y w. c Chart quantifying the Cyclin B fluorescence intensity in PGCs of the indicated genotypes shown in b. ***p < 0.001, *p < 0.05. Student’s t test was performed. Error bars depict SD from five to seven embryos. Scale bar: 10 μm.

Finally, we asked if Cyclin B protein levels are elevated in sisR-2 overexpressing embryos. We detected a significant increase in Cyclin B protein in the PGCs of stage 12 sisR-2 overexpressing embryos as compared to controls (Fig. 6b and c). Furthermore, we also observed PGCs with accumulation of Cyclin B in the nuclei (Fig. S7B). This experiment suggested that the ectopic division of PGCs observed in sisR-2 overexpressing embryos was due to the de-repression of Cyclin B production. To confirm the role of Brat in the de-repression of Cyclin B production, we reduced one copy of brat in the sisR-2 overexpressing embryos. As expected, we no longer detected an increase in Cyclin B protein levels, as well as the accumulation of Cyclin B in the PGC nuclei (Figs. 6b, c, S7B and S7C). In addition, we were no longer able to detect pH3-positive PGCs in these embryos (Fig. 4f). Finally, the number of PGCs is significantly decreased (17.6 ± 4.6 PGCs in maternal MTD > sisR-2 vs. 17.6 ± 2.7 PGCs in maternal MTD > sisR-2; brat[11]/+) (Fig. 4e). Taken together, our experiments suggest that overexpression of maternal sisR-2 promotes PGC division via the regulation of Cyclin B through Brat (Fig. 5a).

PGC death and proliferation are independent events

To achieve tissue homeostasis, it is possible that cell death and proliferation are linked. Induction of cell death may trigger proliferation and vice versa. Survival of embryonic PGCs are controlled by cell intrinsic factors [7]. We show that overexpression of hid led to a decrease in PGC number (Fig. S3A–C). It was also previously shown that overexpression of Cyclin B led to an increase in PGCs [4]. Thus, induction of cell death or proliferation in PGCs does not trigger a homeostatic mechanism.

We then asked if cell death and proliferation are linked in sisR-2 overexpressing embryos. If both processes are linked, a decrease in proliferation/death will not lead to a change in PGC number as it is expected to be compensated by a corresponding decrease in death/proliferation. We observed that by reducing brat, and consequently proliferation, in sisR-2 overexpressing embryos, the number of PGCs decreased significantly (Fig. 4e), suggesting that proliferation and death are independent events. We next reduced hid in sisR-2 overexpressing embryos and observed that it did not rescue the proliferation phenotypes as indicated by the continued presence of pH3-positive PGCs and nuclear accumulation of Cyclin B (Fig. 7a, b, yellow arrowheads, Fig. S8). Furthermore, it led to a significant increase in PGCs (Fig. 7c, d). Taken together, our experiments suggest that in the embryos, PGC death and proliferation are independent events.

Fig. 7: sisR-2 coordinates PGC homeostasis via bantam.
figure 7

a A confocal image showing the PGCs in stage 12 embryos of maternal MTD > sisR-2; hid/+ stained with Vasa (magenta) and pH3 (green). Yellow arrowhead points to pH3-positive PGC. Scale bar: 10 μm. b Confocal images showing the PGCs in stage 12 embryos of maternal MTD > sisR-2; hid/+ stained with Vasa (magenta) and Cyclin B (red). Yellow arrowheads: PGCs with nuclear translocation of Cyclin B. c Confocal images showing the PGCs in stage 13 embryos of the indicated genotypes stained with Vasa (red). Yellow arrowheads: PGCs in gonads. d Chart showing the number of PGCs in the gonads of different genotypes shown in c. *p < 0.05. Student’s t test was performed, comparing the mean number of PGCs of the indicated genotypes. Error bars depict SD. Scale bar: 10 μm. e Proposed model.

Discussion

In summary, our study reveals a homeostatic mechanism that preserves PGC number during embryogenesis in Drosophila. sisR-2 regulates both PGC death and division by modulating the activity of the miRNA bantam (Fig. 7e). We have shown that increasing maternally deposited sisR-2 results in a drop in bantam activity, leading to the derepression of hid and an increase in PGC death. Furthermore, a drop in bantam activity also leads to the derepression of brat, resulting in an increase in Cyclin B production and PGC divisions. How and which population of PGCs make the decision to die or divide remains unclear. Since PGCs formed furthest from the posterior tip (peripheral PGCs) inherit the lowest amounts of germplasm factors such as nanos [7], and both hid and cyclin B are under the translational control of Nanos [8, 27], it is tempting to speculate that peripheral PGCs are likely the most sensitive to changes in sisR-2 levels (Fig. 7e).

In the Drosophila larvae, germline homeostasis is achieved via a non-cell autonomous manner mediated by EGF receptor (EGFR) signaling between the PGCs and the somatic intermingled cells that they are in contact with in the gonads [10]. The mechanism in which germline homeostasis is accomplished during embryogenesis however appears to be different from that during larval development as the embryonic PGCs are not in contact with the SGPs before they reach the gonads. During migration, PGCs are primed or competent to divide during the short window of time. Embryonic PGCs are arrested in the G2 phase of the cell cycle [4] and express the mitosis-promoting cyclin, cyclin B [27], however it is translationally repressed during normal conditions. During starvation, although an increase in maternally deposited sisR-2 promotes PGC death, migrating PGCs can respond quickly by increasing Cyclin B translation and proliferate by entering mitosis.

In the wild, it is reasonable to assume that eggs laid by starved mothers are likely to be in an environment that is deprived of nutrients. Therefore, larvae that hatch from these eggs will be exposed to starvation conditions as well and hence exhibit reduced PI3k/Akt signaling [39]. It has been shown that there is crosstalk between the PI3k/Akt signaling pathway and the EGFR signaling pathway whereby expression of several EGFR pathway components are regulated by the PI3K/AKT pathway [40]. Since larval germline homeostasis is achieved through EGFR signaling, we speculate that this pathway may be impaired in larvae laid by starved mothers. Furthermore, germline knockdown of several PI3K/AKT pathway components resulted in a decrease in PGCs in the larval ovaries [41]. Hence, the embryonic germline homeostatic mechanism identified in this study may be crucial to preserve PGC number during periods of starvation.

Many protective mechanisms to preserve germ cells during nutritional deprivation in adulthood have been previously reported [17, 42, 43]. To the best of our knowledge, this is the first reported example of such a phenomenon occurring in the PGCs during embryonic development. The protective mechanism we uncovered in our study was carried out in flies raised in the laboratory, where starvation may seem like a rather harsh condition to be exposed to in otherwise well-fed animals. However, outside of the well-regulated laboratory, food availability is often limited in the wild [44,45,46]. Hence, in nature, periodic bouts of starvation might be the norm for most animals. Thus, it is important for animals to have evolved homeostatic mechanisms to preserve their germ cell pool during both development and adulthood to safeguard the survival of their entire species. Therefore, it is very likely that similar homeostatic mechanisms preserving PGCs during embryogenesis are present in other animals as well.