Development of germline progenitors in larval queen honeybee ovaries

Apis mellifera (Honeybees) are eusocial insects, characterised by cooperative brood care, overlapping generations and a division of reproductive labour (Wilson and Hölldobler, 2005). Individuals undergo enforced altruism (Ratnieks and Halenterä, 2009), becoming aggressive to workers with more developed ovaries (Visscher and Dukas, 1995) and killing pathogen-infected (Starks et al., 2000). The two female castes of honeybees, workers and queens, are exemplars of the polyphenism underlying the division of reproductive labour. Queen bee anatomy differs from workers, especially in the case of the abdomen, which is larger in queens, and contains two large active ovaries (Snodgrass, 1956). The queen's ovaries are divided into 116-219 ovarioles each (Jackson et al., 2011), while workers have less than 10 per ovary (Snodgrass, 1956). This polyphenism in ovary size is a result of differential feeding during larval developmental stages (Rachinsky et al., 1990). After eclosion, honeybee queens remain in the hive for a few days (Plate et al., 2019) to a week (Jackson and Robinson, 2018), before undertaking nuptial flights (Woyke, 1964). After mating, queens usually take a few days to begin laying (Kocher et al., 2008). The queen produces thousands of embryos that become either workers or drones depending on fertilization (Woyke, 1963). She is the only individual in a colony that can produce fertilized eggs, and thus female workers. Therefore, the queen's ability to lay is crucial for long-term hive success (Gill and Hammond, 2011).

Honeybees have polytrophic meroistic ovaries (Gutzeit et al., 1993; Tanaka and Hartfelder, 2004). This ovary structure is defined by the presence of nurse cells, which provide nutrients and materials to the developing oocyte, and accompany it along the ovariole (Büning, 1993). In honeybees, the ovariole is divided into three major sections: the terminal filament, germarium and vitellarium (Cullen et al., 2023; Dearden, 2006). The terminal filament consists of flattened somatic cells, that appear to provide the progenitors of the somatic follicle cells of the ovary (Hartfelder and Steinbrück, 1997; Martins et al., 2011; Tanaka and Hartfelder, 2004). The germarium contains the first germ-line progenitors, which are clusters of eight germline cells joined by a polyfusome (Cullen et al., 2023). These 8-cell clusters can divide synchronously at a rate that maintains the numbers of these clusters between laying seasons (Cullen et al., 2023).

The progenitor 8-cell clusters in the germarium will divide multiple times (Cullen et al., 2023). Upon entering the posterior half of the germarium, polyfusomes cease and ring canals, another cytoplasmic bridge, appear (Gutzeit et al., 1993). Eventually, germline clusters become oocytes and accompanying nurse cells (Cullen et al., 2023; St Johnston, 2008). This, and the further maturation of the oocyte, takes place in the vitellarium (Chapman, 1998). The oocyte must undergo meiosis, and accompanying nurse cells will divide and endoreplicate, producing polyploid nurse cells linked to each oocyte via the ring canals. The oocytes will then grow, and become provisioned with yolk until they are ready to be laid (Aamidor et al., 2022). The 8-cell cluster is a key point in reproduction as these are the primary store of germline cells in the ovary and the source of oocytes.

Despite the importance of honeybees to the economy and the environment (Khalifa et al., 2021; Tanaka and Hartfelder, 2004), honeybee populations are declining (Brosi et al., 2017; Yang et al., 2023). They experience a range of threats, such as parasites (Warner et al. 2024), viruses (Abbo et al., 2017; Kang et al., 2016), insecticide exposure (Manzoor and Pervez, 2022; Stuligross and Williams, 2021), and the reduction of natural ecosystems and diverse flora for agriculture and urbanization (Goulson et al., 2015). Due to these threats, and others, the costs of beekeeping are increasing. A better understanding of how queen honeybees consistently produce large numbers of offspring, and how to optimise conditions to support that reproduction, will help reduce the effects these threats have on the global populations.

To understand how to best support honeybee reproduction we need to understand the dynamics of germline and oocyte production. Previous studies have shown that germline cells are specified late in honeybee embryogenesis (Dearden, 2006) and that the bulk of ovary growth occurs during larval and pupal stages (Chapman, 1998; Hartfelder and Steinbrück, 1997).

Here we set out to trace ovary development in queen honeybees. Understanding when and where ovary structure appears, how the ovarioles form, and, perhaps most importantly, where/when the 8-cell clusters form, is crucial information if we are to better support reproduction in queen bees and understand the evolution of the unusual structures of the honeybee ovary.

The precursors to honeybee ovaries form in the embryo as two lines of cells in the most dorsal regions of stage 9 embryos, between abdominal segments 3 and 6 (Dearden, 2006). These consolidate into two arches of cells underlying the dorsal surface of just-hatched (L1) larval honeybees (Dearden, 2006). Considerable growth and differentiation occur during larval and adult stages.

Stage 1 and 2 ovaries are difficult to visualize and dissect out of the larval body and do not stain well if kept as whole mounts. L2 larval ovaries lie on either side of the midline between abdominal segments 3 and 6. The ovaries are small and transparent. The cells of the ovary are linked with trachea and buried in the fat layer of the L2 larva (Fig. 1A).

L3 queen larval ovaries have contracted somewhat and are now found between abdominal segments 4 and 5 on either side of the dorsal line (data not shown). When stained with DAPI and phalloidin, the cells are organized into ‘fingers’ of tissue that resemble small future ovarioles, although they are not separated from each other (Fig. 1B).

From this stage onward, the ‘fingers’ of tissue elongate, and, by early L5, separate into individual ovarioles. The ovarioles lengthen through late larval stage 5, with connective tissue forming to connect the posterior of all the ovarioles (Fig. 1E-G). This is likely tissue that will go on to be, or connect to, calices (Kozii et al., 2022). Calices provide the transfer of eggs between the ovaries and the lateral oviduct (Kozii et al., 2022). This region connects both lateral oviducts from either ovariole with the spermatheca (median oviduct) (Kozii et al., 2022).

Growth continues in the pupal stages, with terminal filaments and germaria present. On hatching, the ovary does not appear to contain nurse cells or differentiated oocytes.

To understand the regions of the developing honeybee ovaries, we used markers of specific cell types to identify each region as it develops. In adults, ovary cells that express castor (cas) are somatic (Cullen et al., 2023) and so we used this gene as a potential marker of somatic cells in the ovary. Germline cells express nanos (nos) and vasa (vas) (Dearden, 2006) from late embryonic stages through to adults (Cullen et al., 2023), allowing us to use these as markers of germline fate. RNA expression from the nos gene was used in these experiments to mark germline cells

The odd skipped (odd) gene encodes a zinc finger transcription factor that in Drosophila melanogaster acts as a pair-rule gene (Coulter et al., 1990). Odd is a member of a family of similar transcription factors encoded by the brother of odd with entrails limited (bowl) and sister of odd (sob) genes (Hart et al., 1996). To determine the most likely orthologue of odd in honeybees, we identified homologues of all these genes using blastp (Altschul et al., 1990), and then carried out maximum likelihood phylogenetics. Honeybee XP_001120949 clusters with odd proteins from other holometabolous insects against sob and bowl proteins (Fig. S1). We thus designate the gene encoding this protein, honeybee odd skipped (odd).

In Drosophila, RNA from odd is expressed in a variety of cell types, but not germline or ovary cells (Coulter et al., 1990; Ward and Coulter, 2000). In experiments to examine the expression of genes involved in segmentation in honeybees (data not shown), we identified that honeybee odd is expressed during segmentation, but is also expressed in the same presumptive ovary region as nos and vas RNA in late embryos (Fig. 2). This region, located in the most dorsal regions of the germband, between segments 3 and 6, is where primordial germline cells can first be detected via expression of nos and vas RNA (Dearden, 2006).

To attempt to determine what cells odd might be marking in the honeybee embryonic ovary, we investigated where odd is expressed in the adult queen ovary. RNA from odd is expressed in the cells of the terminal filament (Fig. 2). It is detected in the cells around the boundary of the terminal filament and the germarium but absent in the follicle cells surrounding the germarium (Fig. S2). These cells are thought to be the descendants of terminal filament cells (Dobens and Raftery, 2000). This encouraged us to use odd as a marker for presumptive terminal filament cells during ovary development.

At larval stage 3, odd, cas and nos RNA expression can be used to identify the cell-type structure of the developing ovary (Fig. 3). The ovary is a curved structure with the outside of the curve oriented to the midline of the larva. Many of the cells of the ovary appear to express castor RNA, especially the trachea that infiltrate the tissue. Expression of odd RNA appears in clusters of cells on the outside of the curved ovary. RNA from nos appears in groups of cells juxtaposed to odd expressing cells. Close analysis of these expression patterns (Fig. 3F, G and H) show odd RNA expression at the end of the small fingers of tissue in the ovary at this stage. RNA from nos is expressed in a band of cells abutting the odd domain further down the structure. Cells do not co-express odd and nos RNA. This relationship is very similar to the final pattern of gene expression of odd and nos, with odd RNA being expressed in the terminal filament, and its expression ceasing at the boundary of the germarium, where groups of nos RNA-expressing cells first appear in the ovary. These expression patterns imply that the finger structures are presumptive ovarioles, and these are initially separated into terminal filament and germline cells.

Larval stage 4 ovaries show a very similar arrangement as those at L3 (Fig. 4). Though the ‘fingers’ are slightly longer. By early larval stage 5, however, the ‘fingers’ have separated into ovarioles, with the terminal tips of the ovarioles expressing odd and cas RNA, nos RNA being expressed by clusters of cells further down each ovariole, and cas RNA expressing cells surrounding these. No cells co-express cas and nos RNA.

As the larval ovary extends during L5, the arrangement of cells seen in early L5 is maintained but extended. Odd and cas RNA-expressing cells make up the developing terminal filament. Moving along each ovariole, in the region in which clusters of nos RNA expressing cells appear, odd RNA-expressing cells disappear, with cells surrounding the nos RNA expressing cells expressing only cas RNA. These structures and gene expression patterns remain throughout larval and pupal development as the ovary grows. By analogy to the adult form, the somatic and germline cells of the ovary are defined in the embryo, they become organized by L3, and then divide and grow to produce the structures of the terminal filament and germarium in the adult ovary. There is no evidence of a vitellarium in larval or pupal ovaries.

These in situ hybridization data indicate that there are germline cells in groups at L3. However, it does not tell us whether these germline cells are linked via cytoplasmic bridges, and divide synchronously, as in adult germline cell clusters.

Having described the regionalisation of the ovary into terminal filaments and germaria, we next investigated when and where the germline cells form into the 8-cell clusters that comprise the germline in adult queens (Cullen et al., 2023). These clusters have two key characteristics. The first is that they are joined by a polyfusome, a cytoplasmic bridge that links the 8 cells together (Cullen et al., 2023). The second characteristic is that they divide synchronously (Cullen et al., 2023). To determine when 8-cell clusters begin to form in the ovary, we stained for polyfusomes with phalloidin (Cullen et al., 2023), and for dividing cells with antibodies directed to phosphorylated histone H3 (Hans and Dimitrov, 2001).

Fusomes and polyfusomes are cytoplasmic intercellular bridges between cells forming cystocytes and germline cell clusters (Ventelä, 2006). In Drosophila melanogaster, fusome precursors are called spectrosomes (Deng and Lin, 2001). The cystoblast spectrosome grows into a fusome, a distinct region of the cytoplasm, which bridges two germline cells (King, 1979), as germline cysts form (de Cuevas and Spradling, 1998). Polyfusomes, in Drosophila, are assembled from the further joining of fusomes creating branched structures that can link together the cytoplasm of several cells (King, 1979; Pritsch and Büning, 1989; Trauner and Büning, 2007). The appearance of polyfusomes during development implies the formation of germline cysts (Cullen et al., 2023).

Phalloidin staining should mark spectrosomes, fusomes, polyfusomes and ring canals. In the earliest larval stage ovaries we can image with confidence, larval stage 2, we cannot find accumulations of phalloidin staining consistent with any of these structures (Fig. 5A). By larval stage 3, however, phalloidin can be seen staining concentrations of material that appear to link neighbouring cells (Fig. 5B), reminiscent of Drosophila fusomes. These potential honeybee larval stage 3 fusomes are 5-7.6 μm wide comparable to Drosophila fusomes, (6-10 μm wide during cystoblast to 2-cell cyst stages) (de Cuevas and Spradling, 1998). The similarity in size, staining and shape implies the presence of fusomes at larval stage 3, and that some of the germline cells have formed 2 or 4-cell cysts.

Polyfusomes, similar in size and shape to those in adult ovaries, are present from larval stage 4 (Fig. 5C,D,E). Ring canals, which are present in the posterior germarium of adult queen honeybee ovaries, are not present in the ovary until late pupal stages (Fig. 5F).

Phalloidin staining implies that fusomes, and thus the development of cysts, appear from larval stage 3. If these are the developing germline cell clusters, we should be able to detect synchronously dividing cells at the same stages.

Germline cell clusters in adult honeybee ovaries are joined by polyfusomes and divide synchronously (Cullen et al., 2023). Synchronously dividing cells, especially if associated with polyfusomes, in the developing ovary may be signs of clusters forming. As noted (Fig. 1), the ovaries grow significantly during larval and pupal stages. This growth is associated with large amounts of cell divisions in all tissues of the developing ovary (Fig. 6) making identifying synchronously dividing cells challenging.

Synchronously dividing cells can be detected during larval ovary development. At larval stages 3 (Fig. 6D and Movie 1) and 4 (Fig. 6E and Movie 2), there are clusters of simultaneously dividing cells in the region of nos positive cells in the ovary. At larval stages 3 and 4, these are clusters of 4 dividing cells. At larval stage 5 (Fig. 6F and Movie 3) these are clusters of eight dividing cells, having undergone duplications. The eight dividing cells, while in the same regions of polyfusomes, have very faint, or difficult-to-see polyfusomes. When the germline cell clusters in adult ovaries divide, the phalloidin staining of the polyfusomes joining them is reduced, making them hard to discern (Cullen et al., 2023). It is unclear why.

The appearance of fusomes, and then polyfusomes, as well as clusters of synchronously dividing cells, implies that 8-cell germline clusters in the adult ovary begin to form in larval stages 2 and 3, with the advent of cells joined by fusomes, which then, in larval stage 4, produce clusters of eight cells joined together by polyfusomes. This clustered organisation of the germline in larval development remains the core structure of the germline in adult queen honeybees.

As noted previously, the pupal ovary does not contain the vitellarium region of the ovary. In adults, the vitellarium contains meiosis and cysts of germ-line nurse cells and oocytes. Oocytes are provisioned and enlarge in the vitellarium, finally producing mature oocytes for laying. To determine when the vitellarium begins to form, we followed the expression of cas, nos and odd through late pupal development and into newly emerged queens as a way to understand the further differentiation of the ovary (Fig. 7).

The patterns of gene expression noted during larval development continue into late pupal and newly emerged queen ovaries. RNA expression from odd marks the terminal filaments, with all terminal filament cells expressing odd RNA, until the anterior of the germarium, where clusters of nos expressing germline cell clusters reside. Expression of cas is restricted to somatic cells of both the terminal filament and the germarium. Expression of nos RNA is limited to the germarium.

Even in newly emerged queen ovaries, there is little development of the vitellarium, supporting the idea that the ovary undergoes significant changes in structure and gene expression after mating (Kocher et al., 2008). Expression of cas RNA in newly emerged queens marks a small triangular structure at the end of each germarium (Fig. 7C), perhaps a sign of the beginnings of differentiation of the vitellarium. Cas marks somatic cells (Cullen et al., 2023), such as follicle cells. These ‘triangles’ could signify the production of follicle cells that will envelope developing oocytes (Storto and King, 1989) as they move into the vitellarium.

While the vitellarium does not appear to form until after mating (Kocher et al., 2008), we wanted to know if the division of germline cell clusters is occurring in newly emerged queens. By staining newly emerged queen ovaries with an antibody against PHH3 we can see 8-cell clusters undergoing duplication in the germarium (Fig. S2).

Honeybee queens have been reported as laying ∼500-3000 (Allen, 1955; Avni et al., 2014) eggs per day. This egg-laying occurs in bursts, with permissive environmental conditions, and space in the hive required to allow these high rates of reproduction (Fine et al., 2018; Shehata et al., 1981). While the rate of laying is subject to environmental factors, e.g. nutrition and climate (Avni et al., 2014), structures that might affect egg production, such as ovariole number, can differ between queens (Jackson et al., 2011; Tarpy et al., 2000). It is also possible that the number, and quality, of germ-cell clusters can be manipulated during development, alongside ovariole number.

Understanding the development of the larval ovaries in bees required suitable markers. While nos and cas have been used previously, we serendipitously discovered that odd is expressed from early embryonic stages, in the cells of the developing terminal filament. This expression does not occur in Drosophila, and we have, as yet no function for odd in the terminal filament. As a marker of terminal filament cells and development, odd expression demonstrates that, in bees at least, the terminal filament is an early developing component of the ovary; perhaps providing the somatic precursors to each ovariole. The initial ovary is thus a collection of terminal filament cells as somatic precursors to the ovary, and germline cells, all of which form in the embryo. Terminal filament cell number determines ovariole number in Drosophila melanogaster (Sarikaya et al., 2012), and co-varies with ovariole number in other Drosophila species (Green and Extavour, 2012). Given the unusually large number of ovarioles in Apis mellifera, differing from most other Hymenoptera (Hartfelder et al., 2018), odd may provide a key marker in understanding the genetic mechanism underlying the evolutionary increase in ovariole number. Comparing the numbers of odd RNA-expressing cells in late embryogenesis between honeybees and Hymenoptera with smaller numbers of ovarioles and determining how they are specified may help us understand how, and why, queen honeybees produce so many ovarioles.

Tracing the regionalization of the honeybee ovary during larval development indicates that the ovariole structure is determined by larval stage 3, with terminal filament and germline cells residing in different domains by this early stage (Fig. 8). In honeybee testes, the testioles are visible as finger-like structures by the first larval stage (Lago et al., 2020), so the testioles are likely defined in late embryonic stages. Due to the similarities in morphology and structure between honeybee ovarioles and testioles (Hartfelder et al., 2018), ovarioles are likely determined in the embryonic stages.

Later larval and pupal development involves the expansion and elongation of these domains to produce an ovary in newly emerged queens that is made up of an extensive terminal filament, and a germarium, but with no sign of the vitellarium. It is unclear if meiosis has begun in these ovaries as we have no markers to date that identify cells in meiosis in honeybees.

The key feature of the honeybee queen germline is that germline precursors are organized into 8 cell clusters, joined by a polyfusome (Cullen et al., 2023). These structures reside in the germarium, where they divide at a rate that can replace these germline cell precursors over the years of a queen's life (Cullen et al., 2023). We see evidence of the formation of these clusters beginning in larval stage 2, where we first see fusomes linking two cells in the germline regions of the ovarioles together, and stage 3 where we find synchronously dividing 4-cell clusters and polyfusomes. By early stage 5, there are numerous polyfusomes in the germarium and clusters of 8 cells dividing synchronously (Fig. 8). The germline cell clusters crucial to fertility in the adult queen honeybee thus form during early larval development and are in place in the germarium before the emergence of virgin queens.

The honeybee queen ovary is immature on emergence, as the vitellarium, where oocytes are specified and resourced, has not yet formed. Germline cell clusters are dividing at this time, but it is unclear if these clusters are being stored ready for reproduction to begin, or apoptosed. At this stage, after a few weeks in the hive (Koeniger, 1986), the virgin queens go on a mating flight, mating with multiple males on the wing, and then return to the hive. Egg laying begins 2-3 days (Koeniger, 1986) after mating. Changes in gene expression and morphology of the ovary occur after mating (Kocher et al., 2008), which produces an active vitellarium, and active egg laying.

Queen bees are constrained somewhat by their social biology. They need to be able to lay bursts of eggs when conditions are appropriate, reaching remarkable levels of egg laying (Harbo, 1986), and thus need large and active ovaries. Despite this, virgin queen bees also need to go on a metabolically challenging nuptial flight early in life. It seems likely that the immature ovaries in newly emerged queen bees reflect a trade-off in resource need for flight and reproduction. A similar trade-off occurs at the offset of swarming, as the queen decreases her egg-laying rate (Allen, 1956), thus reducing weight (Seeley and Fell, 1981), to allow for flying with the colony during swarming. By suppressing, or delaying, the development and provisioning of eggs before mating, queens have the resources for an expensive nuptial flight. Trade-offs between reproduction and flight have been identified in many insect species (for examples see Guerra, 2011). After repressing reproduction for a nuptial flight, queens need to rapidly activate their ovaries and begin laying large numbers of eggs to continue populating the hive with workers to replace the nurse workers as they age out of that role to become foragers (Seeley, 1982). The number of ovarioles in bees is unusual, especially for Hymenoptera (Church et al., 2021; Khila and Abouheif, 2010; Kugler et al., 1976; Martins and Serrão, 2004), a testament to the need to rapidly lay large numbers of eggs in ideal situations. Eight-cell clusters provide a ‘timing’ advantage to reproduction, as rather than having to develop and lay eggs from a single germ stem cell, honeybees are further along in the duplication timeline. Perhaps the large number of ovarioles in honeybee queens, and the 8-cell cluster form of the germline in the ovary (Cullen et al., 2023), are adaptations to allow the rapid activation and high rates of reproduction in young honeybee queens.

Honeybees were sourced from Otto's Bees, a bee-keeping supplier based in Dunedin, Otago, New Zealand. Otto's Bees keeps two stocks of bees, an ‘Italian’ strain, derived from a closed breeding programme maintained by Betta Bees Research Ltd, and a ‘Carniolan’ strain. Replicates of experiments were performed using unrelated hives.

Grafting of embryos into queen cells and queen cell raising was carried out as per Cameron et al. (2013).

Embryos and larvae were staged according to Cridge et al. (2017). We investigated larval stages 2, 3 and 4. The larvae are in stage 5 for 4 days (Cridge et al., 2017) so we investigated ‘early’ and ‘late’ periods. We also split the pupal stage, also 5 days, into early and late stages (Table S1).

The staging was performed by timing individuals from days post-laying, and based on morphology, as described below.

Embryos were collected, fixed and dissected using a protocol modified from Osborne and Dearden (2005), as described below. Embryos were collected from worker cells with a paintbrush into 1:1 heptane:4% formaldehyde in 1× PBS. Embryos were shaken overnight at moderate speed on a nutating mixer. After shaking, the lower phase was removed and replaced with 100% methanol at −20°C, this was then shaken vigorously for 3 min by hand. The solution was then removed, and embryos were washed three times in fresh −20°C 100% methanol. Embryos were then stored until dissections in 100% methanol at −20°C.

Embryos and early larva were transferred to a thin-walled glass test tube; 100% methanol was replaced with 2 ml PTw (1× PBS+0.1% Tween 20). Embryos were allowed to settle and then sonicated in a sonic cleaning bath for 5-10 s. Embryos were transferred to a microcentrifuge tube with 200 μg/ml of Pronase (Sigma) in PTw. After 5 min the embryos and larva were washed twice with 500 μl of PTw. The embryos were transferred to a plastic Petri dish with PTw and the outer membranes were dissected away using sharp forceps. Dissected embryos were collected into a microcentrifuge tube with PTw. Embryos were then fixed again in 10% formaldehyde in PBS for 35 min. Embryos were then rinsed six times in 500 μl of PTw.

Larva and pupa were collected from queen cells. Larval samples were kept in the hive until necessary to continue brood care. Pupae were removed from the hive and kept in an incubator with at least 50% ROH at 34°C. Upon removal from queen cells, larvae and pupae were immediately dissected. Larval stages 2 and 3 were dissected by cutting a slit along the centre of the dorsal midline and loosening the internal organs with tweezers. The entire larva was then stained as below. For Larval stages 4 and 5, ovaries were dissected by cutting a slit along the centre of the dorsal midline and dissecting them from the larval body. For pupal stages, the head and thorax were removed from the abdomen. Then segments three and four were pulled apart and ovaries were removed from the abdomen.

Newly emerged queens were always dissected at less than 12 h old. Newly emerged queens were placed at 4°C until movement ceased, and then the heads were removed from the thorax before immediate dissection. Ovaries were dissected into 1× PBS under a dissection microscope and kept in 1× PBS on ice until fixation (less than 2 h). Once ready for fixation, ovaries were fixed in 1:1 heptane:4% formaldehyde in 1× PBS for 12 (larval stages)-15 (pupal stages and newly emerged queens) minutes, then rinsed in 1× PTx (PBS+0.1% Triton X-100) six times. All ovaries were used immediately in experiments.

Honeybee embryos were stained via a modification of that used for Nasonia embryos in Taylor and Dearden (2022). Embryos were pre-hybridised in 200 μl of probe hybridisation buffer (2.4 M urea, 5×sodium chloride sodium citrate (SSC), 9 mM citric acid (pH 6.0), 0.1% Tween 20, 50 μg/ml heparin, 1×Denhardt's solution, 10% dextran sulfate) for at least 30 min at 37°C and probe solution was prepared by adding 6μL of each 1 μM probe (Molecular Instruments) to 200 μl probe hybridisation buffer at 37°C. The pre-hybridisation solution was removed, and the probe solution was added. These samples were then incubated for 2 days at 37°C. Unbound probes were removed by 4×15-min wash steps with 200 μl of probe wash buffer [2.4 M Urea, 5×SSC, 9 mM citric acid (pH 6.0), 0.1% Tween, 50 μg/ml heparin] at 37°C and 3×5-min washes with 200 μl 5X SSCT (5X SSC, 0.1% Tween 20) at room temperature. Embryos were pre-amplified in 200 μl of amplification buffer (5X SSC, 0.1% Tween 20, 10% dextran sulfate) for at least 30 min at room temperature. Six μL of h1 and h2 hairpins (Molecular Instruments) were prepared separately, snap-cooled by heating to 95°C for 90 s then left to cool in the dark for 30 min. The hairpin solution was then prepared by adding all the hairpins to 200 μl of amplification buffer at room temperature. The pre-amplification solution was then removed and replaced with the hairpin solution. Embryos were incubated for 2 days in the dark at room temperature. Unbound hairpins were then removed by 2×5-min washes with 200 μl of SSCT, 2×30-min washes with 200 μl of SSCT and 1×5-min wash with 200 μl of SSCT. SSCT was then removed and replaced with 500μL 70% glycerol. Embryos were then bridge-mounted for microscopy.

Hybridization chain reactions of larva and pupal ovaries were performed as described in Cullen et al. (2023), with probe concentrations of 2 μl of 1 μM probes (nanos, castor, oddskipped). Hairpins were changed between replicates (−488,-546,-647), to ensure no signal bias in specific wavelengths of light.

Immunohistochemistry, phalloidin and DAPI staining were performed as described in Cullen et al. (2023), with primary antibody for dividing cells [1:200 α-mouse-ph3 (Anti-Histone H3; phospho S10) antibody (abcam 14955)].

Confocal microscopy was performed on an Olympus FV3000 confocal microscope. Ovaries were serially optically sectioned with the number of slices/depth optimized using FV3000 software. Images were produced using Icy (de Chaumont et al., 2012) or FIJI (Schindelin et al., 2012) and images were trimmed for ease of visualisation. All raw data images and media are available at Zenodo (DOI 10.5281/zenodo.10972330).

Proteins with similarities to odd skipped-like proteins were identified using blastp (Altschul et al., 1990), and aligned using Clustal (Thompson et al., 1994). Maximum likelihood phylogenetic analysis was carried out using RAxML (Stamatakis, 2014).

The authors would like to that Dr Otto Hyink for the provision of bees and grafting services and Amanda Austin for honeybee embryo staining. We thank Petra Dearden for critical review of the manuscript.