Embryonic gonadal development is conserved between Ae. Aegypti and D. melanogaster
Under the standard breeding condition (see the “Methods” section), embryogenesis took about 60 to 72 h in Ae. aegypti (NEA-EHI strain) (Fig. 1A). In order to label PGCs during germline development, we generated antibodies against Ae. aegypti Vasa (AaegVasa or Vasa herein, AAEL004978), a conserved germ cell marker across metazoan [22]. Between 0 and 2 h after egg laying (AEL), embryos contained a few interiorly localized nuclei. During this stage, Vasa was detected as a crescent located at the posterior end of the embryos (Fig. 1B, B’), similar to that observed in D. melanogaster [23, 24]. Pole cell formation was observed at 4 h AEL, with pole buds detected (Fig. 1G and Additional file 1: Fig. S1A-S1A”). Following that, the PGC number increased from 4 h (3.3 ± 0.7, n = 50) to 8 h AEL (12.6 ± 0.4, n = 26) (Fig. 1C, C’, G). PGCs were subsequently carried by surrounding tissues during germband extension. By 12 h AEL, a closely packed group of PGCs (12.1 ± 0.3, n = 47) was observed at the tip of the extending germband (Fig. 1D, D’, G). After reaching the midgut pocket within the invaginated embryo, PGCs underwent a trans-midgut migration into the interior of the embryo. PGCs then split into two groups, migrated, and eventually coalesced with gonadal somatic cells to form two embryonic gonads (Additional file 1: Fig. S1B-B”). Some PGCs failed to migrate properly during this process (Fig. 1F and Additional file 1: Fig. S1C-S1D’), and similar observations were previously reported during D. melanogaster PGC migration [25, 26]. By 24 h AEL, each gonad located on the dorsal side of the retracting germband contained an average of 4.8 ± 0.1 PGCs/gonad (n = 50) (Fig. 1E, E’, G). By the completion of gonad formation, about 24% of PGCs failed to reach embryonic gonads. PGC numbers increased slightly and reached 5.3 ± 0.2 PGCs/gonad (n = 58) by 48 h AEL (Fig. 1G). The proliferation of PGCs in each gonad was observed from 48 h AEL (5.3 ± 0.2/gonad, n = 58) and reached an average of 9.0 ± 0.3 PGCs/gonad (n = 54) by 60 h AEL, shortly before the completion of embryogenesis (Fig. 1F, G). Collectively, embryonic germline development in Ae. aegypti shares many features with that of D. melanogaster, including PGC formation, migration, gonad formation, and limited PGC proliferation [6].
PGC number continues to increase during larval development
From L1 to L3 stages, each larval stage took about 20 to 24 h, while L4, which is arbitrarily divided into early L4 (eL4) and wandering L4 (wL4), took about 40 to 48 h (Fig. 1A).
Shortly after hatching, the number of larval PGCs continued to increase, similar to that observed in D. melanogaster [1]. PGCs increased rapidly with a doubling time of around 20 h, resulting in an average of 17.4 ± 0.8 (n = 23), 20.3 ± 1.5 (n = 20), 40.8 ± 2.0 (n = 22), 90.8 ± 5.6 (n = 17), and 187.4 ± 8.2 (n = 5) PGCs at L1, L2, L3, eL4, and wL4, respectively (Fig. 2F and Additional file 2: Table S1).
During L1, we were not able to distinguish the ovary from the testis by morphological features. Thus, the PGC number was reflected by the average number of all L1 gonads examined (Fig. 2A, A’, F), while from L2 onwards, the ovaries showed distinct morphologies from the testes. First, the ovaries were oval-shaped while the testes were rounder in shape (Fig. 2B, B’ and Additional file 3: Fig. S2). Second, the ovaries but not the testes contained multiple layers of somatic cells at both ends, a structure similar to that of L3 ovary shown in Fig. 2B, B’, C, C’ and Additional file 3: Fig. S2 (marked by yellow dashed lines). From L2 to eL4, the PGC number further increased within a tight cluster, which is referred to as “PGC mass” (Fig. 2B, B’, C, C’, D, D’, F). Meanwhile, gonadal somatic cells increased, and the ovary grew along the A/P axis with an elongated shape (Fig. 2B–D). During the wL4 stage, a rapid increase of somatic cells was observed, and the ovary was further elongated. Meanwhile, the large “PGC mass” was reorganized to form a long stretch with a width of about 3 PGCs before breaking down into small groups containing 2–6 PGCs. Some somatic cells were subsequently observed between these small PGC groups and surrounded them to form separate units, referred to as “pre-ovarioles” (Fig. 2E, E’, marked by a yellow rectangle, also refer to the “Ovariole is formed via a distinct mode during metamorphosis in Ae. aegypti” section).
PGCs undergo germline cyst-like proliferation during larval development
In D. melanogaster, PGCs are separated by somatic ICs, and each PGC has its own cell cycle program independently of other PGCs. The D. melanogaster ICs play an important role in controlling the proliferative activity of the enclosed PGCs [7, 27]. Notably, Ae. aegypti PGCs seem to form only one tightly packed PGC mass without observed IC-like cells during early larval stages (Fig. 2A, A’, B, B’, C, C’, D, D’). In order to investigate the structure of the PGC mass, we searched for additional germ cell markers and focused on the components of fusome, a germ cell-specific intracellular organelle observed in several insect orders [28]. Fusome is an endoplasmic reticulum (ER) extension enriched in Actin and membrane cytoskeletal proteins, including α-Spectrin [29]. In D. melanogaster, the fusome appears spherical in PGCs and GSCs, referred to as spectrosome, but transforms into an elongated and branched structure in germline cysts as a result of incomplete cytokenesis [30]. Thus, the morphology (spherical vs. branched) of the fusome can serve as an indicator of distinct developmental states (germline stem cell/progenitor vs. differentiating germline cyst) of germ cells. After screening available fusome markers from D. melanogaster, we found that phalloidin and anti-Drosophila α-Spectrin antibody label fusome of Ae. aegypti germ cells (Additional file 4: Fig. S3A-B) [28]. Their signals were largely overlapped with a subtle difference, thus providing a better visualization of fusome when both were applied. We thus used a double labeling of these two markers in the following experiments.
In Ae. aegypti, the fusomes appeared spherical in shape and were weakly detected during the embryonic stage (Fig. 3A–C). Unexpectedly, during the PGC proliferation period at the early larval stages, branched fusomes were observed (Fig. 3D–G), instead of spectrosomes which are normally detected in proliferating PGCs of D. melanogaster [31]. A careful examination showed that these fusomes run through the ring canals to connect PGCs within a cluster (Additional file 4: Fig. S3A). Through 3D reconstruction, we found that L1 and L2 ovaries contain 4–5 fusomes, each connecting a group of PGCs which we name as “PGC cyst” (Additional file 5: Movie S1 and Additional file 6: Movie S2). Of note, the number of fusomes is close to 4.8 PGCs in the newly formed embryonic gonad (Fig. 1G), suggesting that each PGC cyst is likely derived from a single PGC through incomplete cytokinesis. From the L3 stage onwards, some of the less branched fusomes were also observed, which might derive from the breakdown of highly branched fusomes (Fig. 3F, G). These data show that during the early larval stages, Ae. aegypti PGCs form a large PGC mass containing several interconnected PGC cysts, different from that observed in D. melanogaster PGCs, which divide as individual cells.
Another hallmark of interconnected germline cysts is the shared cytoplasm and synchronized cell cycle progression [32]. We next investigated whether these fusome-connected PGC cysts undergo a synchronized proliferation. We used anti-phospho-histone H3 (pSer10) or PH3 antibody, a mitotic marker, to directly investigate the mitotic synchrony of the PGC cysts. At 24 h AEL, no PH3-positive PGC was detected (Fig. 3A, H, I). Some single/two-cell PH3-positive PGCs were detected at 48 h and 60 h, consistent with the window of PGC proliferation (Figs. 1G, 3B, C, H). However, from the L1 stage onwards, most PH3-positive PGCs formed clusters (Fig. 3D–G). Co-labeling with fusome markers revealed that these mitotic PGCs belong to the same PGC cyst in a similar mitotic phase, supporting the notion of the mitotic synchrony (Fig. 3D–G and Additional file 6: Video S2). Of note, the size of PH3-positive PGC cysts increased concomitantly with germline development and frequently corresponded to a power of 2, a key feature of germline cyst division (Fig. 3H). For instance, L1 ovaries typically contained 2-cell or 4-cell clusters, and L2 ovaries contained 4-cell or 8-cell clusters, while 16-cell cysts were observed only in L4 ovaries. However, the portion of mitotic PGC cysts corresponding to a power of 2 reduced gradually from L1 to L4 (Fig. 3I), suggesting a PGC cyst breakdown during proliferation. This is consistent with previous studies on germline development in mice [33, 34].
Mitotic synchrony was further supported by EdU pulse-chase labeling experiment. By exposing L3 larvae to EdU-containing solution for a short period, EdU is incorporated into the chromatin undergoing DNA synthesis. It allows to label PGCs that are in the S phase but not those in the mitotic phase. Indeed, while the majority of PGCs were labeled by EdU, some PGC cysts were EdU-negative, indicating the mitotic synchrony during the pulse period (Additional file 4: Fig. S3C). Furthermore, shared cell fate within a PGC cyst was further supported by synchronized cell death triggered by acute starvation (Additional file 4: Fig. S3D).
Collectively, these data reveal that Ae. aegypti PGC mass consists of several interconnected PGC cysts, which undergo multiple rounds of synchronous divisions with incomplete cytokinesis during the larval stage. In contrast to D. melanogaster in which PGCs proliferate by dividing independently of each other [3], Ae. aegypti expands its PGC pool via a unique cyst-like proliferation manner.
PGC cyst-like proliferation rapidly responds to nutritional status
We next investigated the physiological significance of this PGC cyst-like proliferation during Ae. aegypti larval development. In the field, one of the most common environmental challenges is food availability, when mosquito larvae can survive with a growth arrest for days without food and resume to grow shortly after refeeding [35]. Considering that a cyst-like mitotic synchrony might be an efficient way to regulate PGC proliferation, we designed a starvation/refeeding regime to examine the dynamics of larval PGC cyst proliferation under an irregular food supply condition. Under normal breeding conditions with a constant food supply, the percentage of larval ovaries containing PH3-positive proliferating PGC cysts was consistent throughout the L3 stage (22.0%, n = 41, Fig. 4A, F). However, no L3 ovary (0%, n = 23) examined contained PH3-positive PGCs after 3-day starvation (Fig. 4B, F), indicating proliferation arrest of PGCs. Few ovaries (2.3%, n = 88) contained PH3-positive PGC cysts at 8 h after refeeding (Fig. 4C, F). Notably, the ovaries resumed normal division rates 16 h (21.4%, n = 42) and 24 h (20.0%, n = 70) after refeeding (Fig. 4D–F), indicating a swift response to the availability of food. Likewise, L2 ovaries behaved similarly in the starvation/refeeding experiments (Additional file 7: Fig. S4A-F). These data suggest that a cyst-like proliferation strategy may provide flexibility to the regulation of PGC proliferation in response to food availability.
In both vertebrates and invertebrates, TOR signaling is the key regulator of nutritional status, cellular growth, and metabolism in response to environmental inputs [36,37,38]. Hence, we investigated whether the prompt response of PGC cyst to nutrients is mediated by TOR signaling using rapamycin, an inhibitor of TOR protein kinase. Notably, only 9.3% of L3 ovaries with rapamycin treatment (n = 75) showed PH3-positive PGC cysts, compared to 26.9% in mocked-treated ovaries (n = 52) (p = 0.0139; Fig. 4G–I). Similarly, a significant drop of proliferating ovaries from 16.5% in mock-treated (n = 85) to 4.8% in rapamycin-treated (n = 84) L2 larvae was observed (p = 0.0226; Additional file 7: Fig. S4G-I). Together, these results suggest the involvement of TOR signaling in regulating PGC cyst proliferation in response to nutritional status.
Next, we examined the effects of larval nutritional status on adult fecundity. To this end, we developed a long-term starvation regime, with a minimal amount of food supply during larval growth. Indeed, the ovaries in long-term starved larvae were significantly smaller than those in control larvae and contained fewer PGCs (Fig. 4J, K). In line with this, adult ovaries from the larval starvation group contained fewer ovarioles (19.2 ± 0.8, n = 16) than those from controls (67.7 ± 1.8, n = 11, p < 0.0001; Fig. 4L–N). Consistently, a significant reduction of an average egg number produced by the starvation group (25.6 ± 1.2, n = 93) was observed, in comparison with the control group (102.6 ± 5.3, n = 40, p < 0.0001; Fig. 4O). Of note, the hatch rate of eggs from the starvation group was comparable to that of the control group (p = 0.3144; Fig. 4P). Collectively, these results show that nutritional status during the larval stage has a strong impact on ovarian development, as well as on the fecundity of adult mosquitoes.
In summary, Ae. aegypti evolves in a germline cyst-like proliferation manner during the larval stage to regulate PGC dynamics, presumably via the TOR pathway, in response to nutritional status, which is also important for its fecundity.
Ovariole is formed via a distinct mode during metamorphosis in Ae. aegypti
In D. melanogaster, ovariole formation starts with the formation of TF at the apical side of the ovariole, and the number of TF stacks predetermines the ovariole number [39]. Although the overall structure of Ae. aegypti ovariole is similar to that of D. melanogaster, to our surprise, no TF stack equivalents were observed in adult ovarioles (Fig. 5N–Q). Hence, we investigated how Ae. aegypti ovarioles are formed in detail.
During the mid-L4 stage, a PGC mass was asymmetrically surrounded by somatic cells, with one side covered by a single layer of somatic cells and the other side enclosed with multiple layers (Fig. 5A, A’). During the early wL4 stage, the ovary grew, and PGC mass elongated to form a long stretch along the A/P axis (Fig. 5B, B’). Subsequently, somatic cells of the multiple layers underwent a reorganization to form multiple stacks, the TF stack equivalents (marked by yellow dashed lines), with one end attaching to the PGC stretch (Fig. 5B, B’). During the late wL4 stage, the PGC mass broke down into smaller clusters which contain 2–6 PGCs; meanwhile, extensively branched fusomes were also fragmented into spectrosomes (Fig. 5C, C’). Some somatic cells migrated to enclose those 2–6 PGC clusters to form pre-ovarioles (Fig. 5C, C’). One narrow cavity, the developing oviduct, was evident between the PGC stretch and the single layer of somatic cells (Fig. 5B, C and Additional file 8: Fig. S5A). During the pupal stage, the PGC number increased, and the pre-ovarioles contained more PGCs (Fig. 5D, D’). Notably, those pre-ovarioles also underwent extensive migration and rotation, a process not observed during D. melanogaster ovariole formation. During wL4, the pre-ovarioles are formed with TF stacks pointing to the lateral side of the ovary, orthogonal to the A/P axis of the ovary (Fig. 5C, C’). Concomitant with ovarian growth, pre-ovarioles migrated and resided on the surface of the ovary with an oviduct located in the central position of the ovary (Fig. 5D, E). Meanwhile, pre-ovarioles underwent about a 90° rotation from the lateral to the A/P orientation (Fig. 5D, D’, E, E’, F, F’).
Although TF stacks were observed during ovariole formation (Fig. 5C, G), no TF stacks were observed in adult ovarioles (Fig. 5N–Q). During wL4, TF stacks formed on one side of the elongated PGC mass before fusome breakdown and the formation of pre-ovarioles (Fig. 5G). During larval/pupal transition, PGCs of pre-ovarioles increased without differentiation (Fig. 5G–I). Meanwhile, somatic cells on the basal side (opposite end of TF) intercalated to form an ovariole stalk (equivalent to the basal stalks in D. melanogaster), which connects pre-ovarioles to developing oviduct (Fig. 5H–I). During the late pupal stage, PGCs continued to increase, and the ovariole elongated to form two segments: the germarium and the developing primary follicle, separated by somatic cells in between (Fig. 5J–M). The germarium contained spectrosome-containing germ cells, while the developing primary follicle harbored an 8-cell cyst (Fig. 5M). Of note, the TF stack underwent a morphological change and became thin during the late pupal stage (Fig. 5K–L) and was eventually degenerated 30 h after pupa formation (Fig. 5M). Consequently, adult ovarioles did not have TF stacks attached to the germaria (Fig. 5N–Q), indicating a complete loss of TF during development. After emergence, the ovarioles underwent a maturation process. First, the primary follicle separated from the germarium, grew, and developed to the resting stage (Fig. 5N–P). Second, a new 8-cell cyst emerged from the germarium (Fig. 5Q). Third, the ovariole stalk appeared to undergo a morphological change and became relatively rounded in shape (Fig. 5O–P).
Collectively, Ae. aegypti exhibited distinct features during ovariole formation, including PGC mass elongation and breakdown, pre-ovariole migration and rotation, and TF stack degeneration during metamorphosis, in comparison with D. melanogaster.
Ovariole formation is triggered by ecdysone signaling
In Ae. aegypti, the ovariole formation begins from the wL4 stage and continues throughout the pupal stage, concurrent with metamorphosis, a critical developmental transition from aquatic larvae to terrestrial adults. It has been well known that the steroid hormone ecdysone is the master regulator of this transition during insect development [40, 41]. In addition, ecdysone signaling involves in the establishment of stem cell niche during the ovariole formation in D. melanogaster [42]. We thus investigated whether the formations of TF stacks and pre-ovarioles are triggered by the ecdysone hormone. To induce an early onset of ecdysone pulse mimicking larvae/pupae transition, we subjected larvae at different developmental stages to a pulse treatment of biologically active ecdysteroid 20-hydroxyecdysone (20E). Similar to those untreated ovaries, mocked-treated eL4 ovaries contained an intact oval PGC mass (100.0%, n = 20; Fig. 6A, A’, D). However, 20E-treated eL4 ovaries exhibited precocious PGC mass elongation (weak phenotype, 53.7%, n = 54; Fig. 6B, B’, D), TF formation and PGC mass breakdown (strong phenotype, 42.6%, n = 54; Fig. 6C, C’, D), and hallmarks of precocious pre-ovariole formation, which is significantly different from the mock control (p < 0.0001). In contrast, when L2 or L3 larvae were subjected to the same treatment, no noticeable change in ovarian development was observed (Additional file 9: Fig. S6A-D), suggesting inhibitory mechanisms in place to prevent 20E-mediated PGC mass breakdown during early developmental stages or prerequisites of additional developmental processes for 20E function during L4 stage. These data showed that the early onset of ecdysone pulse during eL4 is sufficient to induce a precocious PGC mass breakdown, suggesting that ovariole formation is triggered by ecdysone signaling in Ae. aegypti.
Ovarian development is conserved in mosquitoes
So far, our data revealed a novel mode of insect ovarian development in Ae. aegypti, which exhibits two main unique features, a PGC cyst-like proliferation during larval development and a distinct process of ovariole formation during metamorphosis (Fig. 7G). To test if this strategy is conserved in other mosquito species, we examined the ovarian development in Culex quinquefasciatus and Anopheles sinensis. Of interest, in both species, the ovary contained a “PGC mass” and underwent cyst-like proliferation with a mitotic synchrony during the larval stage (Fig. 7A, D). Furthermore, the PGC mass elongation, TF formation, and PGC mass breakdown were observed during ovariole formation (Fig. 7B, E). Lastly, pre-ovarioles also underwent a massive migration and rotation to form an ovarian structure during the pupal stage (Fig. 7C, F). Hence, this mode of ovarian development appears to be conserved in various mosquito species.
