A W chromosome-derived feminizing piRNA in pyralid moths demonstrates convergent evolution for primary sex determination signals in Lepidoptera

Abstract

Background

The primary signals of sex determination in insects are diverse and evolve erratically. This also appears to be the case with moths and butterflies. In the silkworm Bombyx mori, female development is triggered by a W chromosome-derived Feminizer (Fem) piRNA that degrades the mRNA of the Z-linked Masculinizer (Masc) gene, which controls male development. We investigated whether this mechanism is conserved in another group of Lepidoptera.

Results

We identified a putative feminizing piRNA and many partial copies of the EkMasc gene on the W chromosome of Ephestia kuehniella. The piRNA is generated by a repetitive W-linked sequence named E. kuehniella Moth-overruler-of-masculinization (EkMom). EkMom piRNA shows high similarity to a region of Z-linked EkMasc and is expressed at the onset of female development, but has no relationship to the B. mori Fem piRNA. We then mapped small RNA-seq data from embryos of the related Plodia interpunctella to the PiMasc gene and identified a single small RNA, a PiMom piRNA, able to target PiMasc and with high sequence identity to the EkMom piRNA. Both the PiMom and EkMom repeats are present in high copy number and form a single cluster on the W chromosome. In both species, the Mom piRNA is responsible for Masc mRNA cleavage, clearly demonstrating that the Mom piRNA triggers female development.

Conclusions

Our study provides multiple lines of evidence that Mom piRNA is the primary sex-determining signal in two pyralid moths and highlights a possible pathway for the origin of feminizing piRNAs in Lepidoptera. The similarity in female sex determination between the phylogenetically distant species suggests convergent evolution of feminizing piRNAs in Lepidoptera.

Background

In sexually reproducing organisms, sexual differentiation is one of the most important developmental processes that takes place during embryogenesis. Sexual differentiation is controlled by a hierarchy of genes that are differentially conserved. The genes at the bottom of the sex-determining cascade arose first and therefore tend to be conserved, whereas the genes at the top are recent additions and show great diversity [1,2,3]. The gene at the bottom of the sex determination cascade in insects is doublesex (dsx), which regulates sexual differentiation through sex-specific differential splicing [4, 5].

Primary signal genes for sex determination in insects, which are at the top of the sex determination cascade, are elusive due to their high turnover rate. Only in a few insect species has the primary signal been identified, and the identified genes do not appear to have homology or a common origin [6,7,8,9,10,11,12,13]. Even among closely related species, the primary signal gene does not appear to be conserved, with the exception of Maleness-on-the-Y (MoY), which was identified in Ceratitis capitata and is conserved over 111 million years of evolution in many fruit flies of the family Tephritidae [14]. This high turnover rate makes their identification difficult, so little is known about the functional mechanism of most of these sex-determining genes.

Moths and butterflies (Lepidoptera) have sex chromosome systems with female heterogamety, mostly with a WZ/ZZ (female/male) system, but numerical variants including Z0/ZZ also occur [15,16,17,18]. The sex-determining pathway in Lepidoptera has so far only been extensively studied in the silkworm Bombyx mori, which has a dominant feminizing W chromosome [19, 20]. However, it took a long time to identify the feminizing gene. This was partly due to the repetitive nature of the B. mori W chromosome, which contains predominantly transposable elements that complicate the assembly of the W chromosome sequence [21, 22]. Contrary to initial expectations, the Feminizer (Fem) gene turned out to be a non-protein-coding gene. The active region of Fem is a 29-nucleotide long Fem piRNA [23], which is the only non-protein-coding primary sex-determining trigger known in insects. Feminization relies on the post-transcriptional silencing of Masculinizer (Masc), a Z-linked gene that initiates male development in the absence of Fem piRNA [23]. In males, previous findings indicated that the Masc protein indirectly guides the splicing of Bmdsx into the male-specific isoform via the male-specific variant of IGF-II mRNA binding protein (BmImp^M) and the B. mori homologue of P-element somatic inhibitor (BmPsi), resulting in male development [22, 24]. Recently, however, the Masc protein was found to interact directly with the BmPSI protein, and the Masc-BmPSI protein complex binds to the Bmdsx pre-mRNA, thereby triggering male-specific splicing of Bmdsx [25]. The Masc protein not only controls male development, but also induces dosage compensation through the transcriptional downregulation of Z-linked genes in males [26].

The feminization process in B. mori is initiated by Masc piRNA, which is maternally provided to the offspring [23]. This Masc piRNA is bound to B. mori Argonaute3 (BmAgo3) and guides this protein to its target RNA, the Fem precursor RNA, which is subsequently cleaved [27]. The cleaved Fem RNA is bound by the Siwi protein and processed into an active piRNA-protein complex. The Fem piRNA guides this protein to its complementary target, i.e. the Masc mRNA, which in turn is cleaved and bound by BmAgo3 and continues the piRNA amplification cycle. This amplification process relies on a 10-nucleotide stretch of perfect sequence identity between the two piRNAs, Fem piRNA and Masc piRNA, referred to as the ping-pong signature ([23, 27, 28]. The Fem gene is present in high copy number and is located in a cluster on the W chromosome [22].

The function of the Masc gene appears to be conserved at least in the clade Ditrysia, which comprises about 98% of moths and butterflies [29], as its essential role in male development has been confirmed in several species from related, distant and basal families [30,31,32,33,34,35,36,37,38]. However, the primary signal for female development and the actual role of the W chromosome remain enigmatic, except in the butterfly Bicyclus anynana, where the Z-linked BaMasc gene itself is the primary switch, with BaMasc hemizygotes developing as females and heterozygotes as males [39]. Analysis of a Masc ortholog in Trilocha varians, a close relative of B. mori, revealed that the Fem piRNA-binding site in TvMasc is not conserved, suggesting that the primary signal gene in T. varians differs from that in B. mori [31]. The absence of a W-linked gene producing a piRNA targeting TvMasc was later confirmed by deep sequencing of T. varians embryonic piRNA libraries [40]. Furthermore, no female‐specific small RNA that maps to OfMasc mRNA has been identified in the more distant Ostrinia furnacalis [41]. In wild silkmoths, Samia cynthia ssp., the presence or absence of the W chromosome plays no role in sex determination, and sex is determined by the ratio between the number of Z chromosomes and the number of autosome sets, i.e. the Z:A ratio [42, 43]. The Z:A ratio is also expected to determine sex in species that lack or have lost the W chromosome [15, 17, 44, 45]. However, a sex‐determining mechanism similar to that in B. mori was recently identified in a basal Ditrysia species, Plutella xylostella, where a W‐linked locus (Pxyfem) consisting of “retrocopies” of the PxyMasc gene is a source of small silencing RNAs (ssRNAs) targeting the PxyMasc sequence [46]. In addition, a W-linked locus called Fet-W (female expressed transcripts of the W chromosome) was recently identified in the spongy moth Lymantria dispar japonica. The Fet-W locus is likely a precursor of the Fet-W piRNA targeting LdMasc mRNA [38].

We investigated sex determination in the Mediterranean flour moth, Ephestia kuehniella (Pyralidae), one of the early models of Lepidoptera genetics (see Anagasta kühniella in [47]) and later a model for sex chromosome research [15]. In E. kuehniella, we identified an ortholog of B. mori Masc, EkMasc, and showed that it is duplicated on the Z chromosome [35]. Both copies, EkMasc and EkMascB, have a similar structure and encode similar proteins that share 91.7% identity over the entire length. They follow the same expression pattern during early embryogenesis depending on the sex of the embryos. Their simultaneous knockdown by RNA interference (RNAi) resulted in female-specific splicing of Ekdsx in genetic males and a strongly female-biased sex ratio, demonstrating that one or both copies of the gene regulate male development [35]. However, little is known about how female development is initiated in E. kuehniella and what role the W chromosome plays in this process. Sequencing of the W chromosome of E. kuehniella revealed transposable elements, microsatellites and mitochondrial sequences, but no protein-coding genes [48]. Nevertheless, given the discovery of Fem piRNA in B. mori [23], the absence of protein-coding genes on the W chromosome does not rule out a function of this chromosome in the initiation of female development by piRNAs or other types of non-coding RNAs.

We used a bioinformatics approach to screen W-linked sequences of E. kuehniella for the presence of feminizing small RNAs. Here we report the identification and characterization of a feminizing gene in E. kuehniella. This gene is located on the W chromosome of E. kuehniella and produces a single piRNA capable of targeting EkMasc and EkMascB. In addition, we identified many partial copies of EkMascB on the W chromosome, which are also expressed and processed into small RNAs. We then investigated sex determination in a close relative of E. kuehniella, the Indian meal moth Plodia interpunctella. By mapping reads from small RNA transcriptomes to the P. interpunctella Masculinizer (PiMasc) sequence, we identified a single piRNA with high homology to PiMasc. In both species, we have experimentally demonstrated that these piRNAs are the primary sex determination triggers that control female development. A comparison of the feminizing piRNAs in E. kuehniella and P. interpunctella showed a high level of sequence identity. The W-linked gene producing these piRNAs has a similar structure to Fem in B. mori and Fet-W in L. dispar, but lacks sequence homology to Fem and Fet-W. Our study highlights a possible pathway for the origin of feminizing piRNAs not only in pyralid moths, but also in other evolutionarily distant Lepidoptera species.

Results

Identification of putative feminizing small RNAs in Ephestia kuehniella

To identify potential small RNAs that could target EkMasc and EkMascB, we used the complete mRNA sequences of EkMasc and EkMascB [35] for a BLASTn search against the available genomic reads of the W chromosome of E. kuehniella [48]. Hits (i.e. complementary sequences) obtained from the BLASTn search were mapped back to the original EkMasc and EkMascB query sequence and showed variable coverage along the EkMasc (Additional file 1: Fig. S1) [35] and EkMascB (Fig. 1A) genes. In particular, a 25-nucleotide fragment in exon IX appeared to be overrepresented compared to the rest of both genes. Based on the structure of Fem on the W chromosome of B. mori, i.e. a high copy number repeat containing the Fem piRNA sequence and organized in tandem [22], this overrepresented region in exon IX was identified as a sequence of interest. Reads containing this short sequence were extracted from the database and aligned to create a consensus sequence. Almost 94% of the reads (948/1009) could be used to create a single consensus sequence, termed E. kuehniella Moth-overruler-of-masculinization (EkMom), which has no homology to sequences in the male genome, apart from the short RNA sequence, EkMom piRNA, which is homologous to part of exon IX of the EkMasc and EkMascB genes. EkMom is therefore likely a precursor gene of EkMom piRNA. The absence of EkMom in the male genome was confirmed by PCR (Fig. 1B).

Fig. 1

Identification, structure and localization of the EkMom and EkMasc^W sequences in Ephestia kuehniella. A Identified sequences homologous to EkMascB on the W chromosome. The graph shows the read coverage of EkMascB on the W chromosome. Below is a representation of the full EkMasc transcript as reported by [35]; shown are the two degenerate zinc finger domains (pink), the bipartite nuclear localization signal (green), the masculinizing domain (blue) and the open reading frame (yellow). The predicted EkMom piRNA sequence is visible as a distinct peak marked with asterisk. Note the absence of reads homologous to exons I, XI and XII of EkMascB. B, C Presence of EkMom (B) and EkMasc^W (C) in gDNA of females (WZ) and absence in males (ZZ). Multiple bands are visible showing a tandem repeat structure for both genes. The DNA marker (m) with sizes in kb and the no template control (NTC) are indicated. D–G FISH mapping of the EkMom and EkMasc^W clusters on chromosomes of females, counterstained with DAPI (blue). D WZ bivalent at the pachytene stage after GISH, where the W chromosome is identified by a Cy3-labelled female gDNA probe (red). E The same bivalent after re-probing and TSA-FISH with probes for EkMom (red, arrow) and EkMasc^W (green, arrowhead). F Schematic representation of the WZ bivalent with hybridization signals in D and E. G Mitotic metaphase after TSA-FISH with hybridization signals of the EkMom (red) and EkMasc.^W (green) probes on the W chromosome, marked with asterisk (*). Bar = 10 µm. H Southern hybridization of digested female and male gDNA with the EkMom probe showing high sequence conservation of the EkMom cluster. Partially digested female DNA (F_p) shows an evenly distributed ladder pattern with a space size corresponding to the predicted EkMom monomer size (602 bp). Fully digested female DNA (F_c) shows a single strong band corresponding to the EkMom monomer and a few fairly faint additional bands. Fully digested male DNA (M_c) shows no signals, confirming the absence of EkMom in the male genome. Also shown is the DNA marker (m) with sizes in kb. Unprocessed images of panels B, C and H can be found in Additional file 2

The W chromosome read coverage spanning a large part of the EkMasc and EkMascB coding sequence indicated the presence of partial homologous sequences on the W chromosome of E. kuehniella (Additional file 1: Fig. S1; Fig. 1A). The absence of EkMasc and EkMascB intron sequences in the reads ruled out any genomic DNA (gDNA) contamination from the Z chromosome during the sequencing project, thus eliminating the possibility of false-positive results [48]. In addition, the presence of these EkMasc/EkMascB W-specific sequences (Fig. 1A) was confirmed by PCR using gDNA of both sexes (Fig. 1C). Sequencing of the PCR products confirmed the lack of introns in the W-linked EkMasc (EkMasc^W) sequences. Based on the sequence comparison of EkMasc, EkMascB and EkMasc^W, we found that EkMasc^W arose from a duplication of EkMascB after the divergence between EkMasc and EkMascB (Additional file 3: Fig. S2). Apart from the introns, EkMascB and EkMasc^W share significant homology from most of exon 2 to most of exon 10. In these regions, their sequences show 94–97% identity (Additional file 4: Fig. S3).

Structure of EkMom and EkMasc ^W on the W chromosome of Ephestia kuehniella

PCR targeting EkMom resulted in the amplification of multiple fragments in E. kuehniella (Fig. 1B). Sequencing of these amplicons revealed a tandem repeat structure of a 602-bp monomer with high sequence identity between the EkMom piRNA and the spacer sequence (98.2% between 8 copies; Additional file 5: Fig. S4). Similar to EkMom, amplification of EkMasc^W revealed multiple fragments indicating a tandem repeat structure (Fig. 1C). The estimated copy number of the EkMom repeat ranges from 123 ± 16 to 233 copies, depending on the method used, while the copy number of EkMasc^W ranges from 17 to 23 (see Additional file 6: Fig. S5 for details). To determine the chromosomal distribution of the EkMom and EkMasc^W repeats, we performed fluorescence in situ hybridization (FISH) experiments. In both mitotic and meiotic nuclei, a single strong hybridization signal was observed on the W chromosome for each of the gene-specific probes, confirming one large EkMom cluster and one EkMasc^W cluster on the W chromosome (Fig. 1D–G).

We performed a Southern hybridization assay to determine the level of conservation of the EkMom repeat in the cluster on the W chromosome. DNA digestion was performed using EcoT22I, a restriction enzyme predicted to cut only once in the EkMom repeat. Female gDNA was either partially or fully digested, while male gDNA was only fully digested. Fully digested female gDNA showed hybridization signals corresponding to the EkMom monomer, dimer, trimer and another fragment slightly shorter than the dimer (Fig. 1H). The intensity of the single monomer was by far the strongest signal, indicating high sequence conservation of both the restriction site in the monomers and the overall size of the monomers. In addition, the assay confirmed the absence of EkMom in the male genome. In the partially digested female gDNA, an evenly distributed hybridization signal with a spacing of approximately 600 bp was observed, which corresponds to the size of the EkMom monomer and confirms the tandem repeat structure of the cluster.

Expression of sex-determining genes and timing of sexual fate in Ephestia kuehniella

To find out when sexual fate is determined during development, we examined the expression of sex-determining genes during embryogenesis. The expression of EkMasc and EkMascB as well as E. kuehniella doublesex (Ekdsx) had been determined previously [35]. Briefly, no significant difference in the expression of EkMasc and EkMascB was detected between the sexes 12 h post oviposition (hpo). Then, the expression of both genes increased in male embryos, with statistically significant differences observed between males and females between 14 and 22 hpo. The expression of both genes peaked in male embryos at 16 hpo, while simultaneously decreasing in female embryos. After 16 hpo, expression decreased in males and reached a level comparable to females at 24 hpo (see Fig. 3 in [35]). Consistent with the expression pattern of EkMasc and EkMascB genes, female-specific splicing of Ekdsx in male embryos switched to male-specific splicing at 16–18 hpo (see S8 Fig. in [35]). Here we focused on the expression of other genes known to play a role in sex determination in B. mori. Using 3’ RACE-PCR, we identified an ortholog of the male-specific splice variant of Imp in E. kuehniella (EkImp^M). We used the same time frame as in the previous study [35], i.e. 12–24 hpo, to detect the presence or absence of EkMom and EkImp^M in both sexes by reverse transcriptase PCR (RT-PCR). In the WZ individuals, EkMom expression was detected during the entire time window examined, while EkImp^M expression remained below the detection limit of RT-PCR (Fig. 2A). Contrary to expectations, EkMom was detected in ZZ embryos at 12–16 hpo, but not from 18 hpo onwards. The presence of EkImp^M was detected in male embryos from 14 hpo, just before the transition of Ekdsx splicing from female- to male-specific [35]. Thus, based on the expression of EkImp^M, male development appears to be initiated between 12 and 14 hpo, while the transition of Ekdsx from female to male splicing occurs after 14 hpo, i.e. between 16 and 20 hpo [35].

Fig. 2

Expression of sex-determining genes during early embryogenesis of Ephestia kuehniella. A Sex-specific expression patterns of EkMom and EkImp^M in WZ and ZZ embryos. EkMom is present in both sexes during early embryonic development, but disappears in males between 16 and 18 hpo. Expression of EkImp^M is male-specific and occurs in males around 14 hpo. Expression of Ekrp49 is shown as a positive control for amplification of cDNA samples. The DNA marker (m) with sizes in kb and the no template control (NTC) are indicated. Unprocessed images can be found in Additional file 7. B, C Expression of small RNAs capable of targeting EkMascB during early development of unsexed embryos. The red peaks show small RNAs derived from EkMasc^W, the pink peak shows EkMom piRNA and the grey peaks are small RNAs of unknown origin. B Mapping of small antisense RNAs from embryos at 7–9 hpo against EkMascB. Note that no EkMom piRNA was detected at 7–9 hpo. C Mapping of small antisense RNAs from embryos at 11–13 hpo against EkMascB. The same small RNAs were detected in similar copy number as in B, but in addition EkMom piRNA was detected at this time point. The exon diagram and the open reading frame (ORF; in yellow) of EkMascB are shown below the graphs

Expression of EkMom and EkMasc ^W piRNAs in Ephestia kuehniella

To investigate whether the presence of EkMom RNA in eggs also leads to the production of putative EkMom piRNA, we sequenced the small RNA fraction at two time points during early embryogenesis, i.e. before the onset of sex determination (7–9 hpo) and at the onset of sex determination (11–13 hpo). Mapping of small RNAs to EkMasc and EkMascB revealed the presence of multiple small RNAs antisense to the mRNA sequences at both 7–9 hpo and 11–13 hpo (Fig. 2B, C and Additional file 8: Fig. S6). At 7–9 hpo, we identified several EkMasc^W-derived small RNAs that can target both EkMasc and EkMascB mRNAs (Fig. 2B and Additional file 8: Fig. S6A). We identified one exception of an EkMasc^W-derived small RNA that can exclusively target EkMascB in the 5’ region of exon II. We have also identified several other small RNAs that have the potential to silence either EkMasc, EkMascB, or both genes, but the origin of these small RNAs is unknown. Several of these small RNAs of unknown origin have the potential to target the 3’-UTR of EkMasc, but not EkMascB (cf. Figure 2B, C and Additional file 8: Fig. S6). When plotting the sequence logos of the antisense and sense reads mapped to EkMascB, we found a clear bias towards 1U in the antisense sequences to EkMascB. We also found a bias for 10 A in the sense oriented sequences, although this bias was less extreme (Additional file 9: Fig. S7). The small RNA sequences potentially targeting EkMascB (and EkMasc, similar but not shown) therefore show signs of piRNA-like sequences.

At 7–9 hpo, the putative EkMom piRNA was not detected. Data from embryos at 11–13 hpo showed largely the same small RNAs at similar expression levels, but at this time point the EkMom piRNA was detected (Fig. 2C and Additional file 8: Fig. S6B). In addition, we identified the full length of the EkMom piRNA, i.e. 28 nucleotides, and found that EkMom piRNA shares approximately 89% (25/28) and 86% (24/28) sequence identity with EkMasc and EkMascB, respectively (Additional file 10: Fig. S8A). In B. mori, Masc piRNA is maternally delivered to the eggs and initiates the ping-pong amplification system of Fem piRNA [23]. Masc piRNA is recognized by its 5’ ping-pong signature, i.e. a 10-nucleotide sequence that is perfectly complementary to Fem piRNA. However, we were not able to identify EkMasc piRNA in the datasets from either time point.

Identification of PiMom piRNA in Plodia interpunctella

Based on the results in E. kuehniella, we tested whether the putative EkMom piRNA is conserved in closely related species. To this end, we screened the available sequencing datasets of species from the subfamily Phycitinae and identified small RNA transcriptomes from the egg stage of the Indian meal moth, Plodia interpunctella. We mapped the small RNA transcriptomes to the mRNA sequence of PiMasc and identified a single small RNA sequence capable of targeting the coding region of PiMasc, which corresponds to a region in exon XI (Fig. 3A). This 26–27 nucleotide long sequence differed from PiMasc by only two nucleotides and was named Plodia interpunctella Moth-overruler-of-masculinization (PiMom) piRNA. In addition, we found several small RNAs that might target exon XIV of PiMasc, but these sequences can also be found in other regions of the P. interpunctella genome and therefore might not exclusively target PiMasc. The number of reads of PiMom piRNA is approximately ten times higher than that of EkMom in E. kuehniella at 11–13 hpo, but information on the post-oviposition time of the egg samples used for sequencing the P. interpunctella small RNA transcriptomes is not known. In contrast to E. kuehniella, we were able to identify PiMasc piRNA in three small RNA transcriptomes of P. interpunctella (Additional file 10: Fig. S8B), albeit in low numbers, i.e. 18–20 reads per dataset, corresponding to ~ 1.4 reads per million.

Fig. 3

Identification, structure and localization of the PiMom sequence in Plodia interpunctella. A Presence of antisense small RNAs with high sequence identity to PiMasc in eggs. The putative PiMom piRNA (pink) is present in high copy number, while several other small RNAs are present in low numbers scattered in the last two exons. The figure shows the data from dataset SRX1539920. The other two datasets analysed showed very similar patterns. B Presence of PiMom in females (WZ) and absence in males (ZZ). Multiple bands in females show the tandem repeat structure of PiMom. The DNA marker (m) in kb and the no template control (NTC) are indicated. C–E TSA-FISH mapping of PiMom (red) and PiSAT1 (green) probes on chromosomes of females, counterstained with DAPI (blue). C Pachytene WZ bivalent showing the location of a single PiMom cluster (arrow) and a single PiSAT1 cluster (arrowhead) on the W chromosome. D A schematic representation of the same WZ bivalent (W is shown in pink, Z in blue). E Mitotic metaphase showing both PiMom (red) and PiSAT1 (green) hybridization signals on the W chromosome marked with asterisk (*). Note that the chromosome number is 2n = 62. Bar = 10 µm. F Southern hybridization of female and male gDNA with the PiMom probe shows a high level of sequence conservation of the PiMom cluster. Partially digested female gDNA (Fp) shows a largely uniform ladder pattern with a space size corresponding to the PiMom monomer size. Fully digested female gDNA (Fc) shows a single strong band corresponding to the PiMom monomer and several additional bands of varying size and signal intensity. Fully digested male gDNA (Mc) shows no signals, confirming the absence of PiMom in the male genome. Also shown is the marker (m) with sizes in kb. Unprocessed images of panels B and F can be found in Additional file 11

The obtained PiMom piRNA sequence was used to identify genomic reads from a female sample by BLASTn search, and a consensus sequence was generated from the reads. This PiMom sequence was then confirmed by PCR (Fig. 3B). Like EkMom in E. kuehniella, the PiMom sequence is female-specific and appears to be present as a tandem repeat (Fig. 3B). Sequencing of several longer fragments amplified by PCR confirmed the tandem repeat structure of PiMom with 95–98% sequence identity between different copies of the monomer and a monomer size of 434–442 bp (Additional file 12: Fig. S9).

Structure of PiMom on the W chromosome of Plodia interpunctella

As in E. kuehniella, we investigated the genomic organization of PiMom. Amplification of PiMom from female gDNA revealed several DNA fragments (Fig. 3B) consisting of multiple copies of PiMom arranged in tandem. In addition, the estimated copy number of the PiMom repeat ranges from 12 ± 1 to 20 copies depending on the method used (see Additional file 13: Fig. S10 for details). Based on these data, we assumed that PiMom is localized in at least one cluster and can be visually mapped by FISH on the W chromosome of P. interpunctella. To identify the W chromosome, we used a previously published W-specific marker, PiSAT1 [49], and performed TSA-FISH with two probes. Both PiSAT1 and PiMom were visible as single distinct clusters on the W chromosome in both meiotic and mitotic nuclei (Fig. 3C–E). The W chromosome of P. interpunctella contains a large block of heterochromatin in the middle of the chromosome [50] and PiSAT1 was previously detected in close proximity to this block [49]. We confirmed this position of PiSAT1 and found that PiMom is located at the opposite end next to the heterochromatin block (Fig. 3C, D). The W chromosome position of PiMom was also confirmed in mitotic metaphase nuclei (Fig. 3E). However, in the two larvae examined, a diploid chromosome number of 2n = 62 was consistently observed in mitotic nuclei, which differs from the 2n = 60 chromosomes reported previously [50]. Whether this finding reflects an actual difference in chromosome number between populations of different origin (strains from Italy and Turkey were used in [50]) is currently unknown. However, a recent chromosome-level genome assembly of a P. interpunctella male (ilPloInte3.2, BioProject PRJNA782555) originating from the Savannah laboratory colony (GA USA) revealed a haploid number of n = 31 chromosomes, which is consistent with this study.

We analysed the structure of the PiMom cluster on the W chromosome by Southern hybridization, which revealed similar patterns as in E. kuehniella. Female gDNA partially digested with Eco32I (EcoRV) showed an evenly distributed ladder pattern with bands separated by the approximate size of the PiMom monomer (Fig. 3F), suggesting that the size of the PiMom monomer is consistent between copies. Fully digested female gDNA primarily showed a signal corresponding to the size of a single monomer, again indicating a high level of sequence conservation between PiMom copies. However, several additional hybridization signals were present in the fully digested female gDNA, indicating greater sequence divergence between PiMom monomers than between EkMom monomers. Male gDNA showed no signals, confirming the absence of PiMom in the male genome.

Comparison of feminizing piRNAs

After identifying the putative Mom piRNAs in E. kuehniella and P. interpunctella, we compared the Mom piRNA sequences and Mom repeats in both species in more detail. Alignment of the two piRNA sequences revealed a high level of sequence identity (Fig. 4A). We additionally aligned the two Mom monomer sequences and performed a sliding window analysis with a window size of 10 nucleotides, i.e. the length of a ping-pong signature, and a step size of 1 nucleotide to determine the level of homology between EkMom and PiMom repeats. The analysis revealed a single region of at least 10 nucleotides of consecutive sequence identity between the Mom repeats, corresponding to the region containing the putative Mom piRNAs of the species (Fig. 4B). Apart from the Mom piRNA region, the overall similarity between EkMom and PiMom is low. A comparison of the EkMom piRNA with the previously identified Fem piRNA of B. mori showed no homology between the sequences, not even in the target region of their respective target Masc genes (Additional file 14: Fig. S11).

Fig. 4

Sequence comparison and maternal provision of EkMom and PiMom. A The sequence alignment of the EkMom and PiMom regions containing the piRNA sequences (shown in grey) shows high sequence identity between the piRNA sequences. B Sliding window analysis of the aligned EkMom and PiMom sequences shows a low level of homology between the sequences over their entire length, with the exception of the feminizing region. C, D Maternal provision of EkMom and EkMasc/EkMascB was detected in cDNA samples from 0–1 hpo eggs of Ephestia kuehniella (C), but only maternal provision of PiMasc was detected in 0–1 hpo eggs of Plodia interpunctella (D). RNA samples, i.e. identical samples to the cDNA samples, except that no reverse transcriptase was added during the cDNA conversion reaction and no template control (NTC) was added as a negative control. In addition, gDNA served as a positive control for both PCRs. The marker (m) in kb is also indicated. Unprocessed images of panels C and D can be found in Additional file 15

Maternal provision of EkMom and PiMom

In the silkworm, Masc piRNA is maternally delivered to the eggs, while Fem piRNA and Fem precursor are probably not [23]. The presence of EkMom RNA in male embryos despite the absence of EkMom in the male genome suggests that EkMom RNA is maternally provided to the eggs. To confirm the presence of maternally provided EkMom RNA in the eggs, we extracted RNA from the eggs before the expected onset of transcription and tested its presence by RT-PCR. We confirmed that both EkMom and EkMasc (and EkMascB) RNAs are maternally provided to the eggs (Fig. 4C). Similarly, we tested the presence of PiMom RNA in freshly laid eggs of P. interpunctella. Using cDNA prepared from RNA extracted from pooled eggs up to 1 hpo, we were able to show that maternal PiMasc RNA is present in the eggs, but PiMom RNA is not (Fig. 4D).

Functional analysis of EkMom piRNA and PiMom piRNA

We performed two types of experiments to demonstrate the feminizing function of Mom piRNAs. (i) We performed modified RNA Ligase-mediated 5′ rapid amplification of cDNA ends (modified 5’ RACE), as previously performed in B. mori [23], to identify cleavage sites in EkMasc/EkMascB mRNA and PiMasc mRNA in embryos of E. kuehniella and P. interpunctella, respectively. (ii) We injected double-stranded RNAs (dsRNAs) targeting EkMom into freshly laid eggs of E. kuehniella to suppress the expression of the precursor RNA, as was recently done in L. dispar [38].

In the modified 5’ RACE experiment with 48 hpo eggs of E. kuehniella, sex was successfully determined in 15 of 16 eggs, of which 9 were female and 6 were male embryos. Total RNA of each sex was pooled, Ligated to the GeneRacer RNA Oligo adapter and converted to cDNA, which served as template for the modified 5’ RACE PCR. Two amplified products of the expected size were detected in female embryos, whereas no such products were obtained in male embryos (Additional file 16: Fig. S12A). Based on the study in B. mori [23], we predicted that the EkMasc and EkMascB mRNAs would be cleaved in the complementary sequence to the EkMom piRNA, opposite nucleotides 10 and 11, measured from the 5′-end of the EkMom piRNA (Additional file 10: Fig. S8A). Indeed, we found that all 16 cloned and sequenced EkMasc/EkMascB fragments, of which 7 were derived from EkMasc mRNA and 9 from EkMascB mRNA from female embryos, could be accurately mapped to the predicted EkMom piRNA cleavage site (Fig. 5A, B). In 48 hpo eggs of P. interpunctella, sex was successfully determined in 12 of 16 eggs, of which 6 were female and 6 were male embryos. The modified 5’ RACE PCR amplified a single product of the expected size only in female embryos, while no product was obtained in male embryos (Additional file 16: Fig. S12B). The sequenced PiMasc fragments of all 16 clones derived from PiMasc mRNA exactly matched the predicted PiMom piRNA cleavage site (Additional file 10: Fig. S8B; Fig. 5C). Taken together, these results clearly show cleavage of EkMasc/EkMascB mRNAs and PiMasc mRNA in female embryos, strongly suggesting that these mRNAs are targets of EkMom piRNA and PiMom piRNA, respectively.

Fig. 5

Analysis of the function of EkMom piRNA and PiMom piRNA. A–C Identification of the cleavage site of EkMasc mRNA (A), EkMascB mRNA (B) and PiMasc mRNA (C). The RNA fragments derived from EkMasc mRNA, EkMascB mRNA and PiMasc mRNA were amplified by a modified 5’ RACE method, cloned and sequenced. The cleavage sites are shown in red. The RACE adapters (in black) indicate the cloned 5’-ends. Asterisks indicate nucleotide matches in both strands of the cloned fragments. The numbers above the nucleotides indicate the position in the coding sequences. D Effect of RNAi-mediated knockdown (KD) of EkMom RNA on splicing of the Ekdsx gene. Splicing pattern of Ekdsx detected by RT-PCR in male (ZZ) and female (WZ) embryos injected with DsRed-dsRNA (DsRed KD) or EkMom-dsRNA (EkMom KD). Blue arrows indicate the male-specific transcript (Ekdsx^M) and red square brackets show the female-specific transcripts (Ekdsx^F). Also shown are the marker (M) with sizes in bp, untreated control embryos (♂ and ♀) and no template control (NC). Unprocessed images can be found in Additional file 18

In the dsRNA microinjection experiment, we investigated the effect of EkMom knockdown on the sex-specific expression of the Ekdsx gene during the sex-determining phase of early embryogenesis. We injected EkMom-dsRNA targeting EkMom RNA (Additional file 17: Fig. S13A) into 2–4 hpo eggs of E. kuehniella. As a negative control, we also injected DsRed-dsRNA targeting the Discosoma red fluorescent protein gene (Additional file 17: Fig. S13B) into E. kuehniella eggs. After injection, sex-specific splicing of Ekdsx was examined by RT-PCR in 24 hpo embryos. Male embryos (ZZ) injected with either EkMom-dsRNA or DsRed-dsRNA showed only the male splice form of Ekdsx, as in the untreated male control (Fig. 5D). However, female embryos (WZ) injected with EkMom-dsRNA showed the strong male splice form of Ekdsx visible on the gel as single band of high intensity and a reduced female splice form visible as bands of low intensity, while female embryos injected with DsRed-dsRNA showed only female splice forms, like the untreated female control (Fig. 5D). These results clearly show that EkMom is required for female development in E. kuehniella.

Discussion

Feminizer (Fem), which was discovered in the silkworm B. mori, is the first master sex-determining gene that does not encode a protein but a small feminizing RNA. This Fem piRNA targets the masculinizing gene, Masculinizer (Masc), leading to its post-transcriptional downregulation and thus triggering female development in WZ individuals [23]. However, the primary sex-determining genes in insects are often species-specific, suggesting that their turnover is rapid [5]. Therefore, the lack of Fem piRNA-binding sequence conservation in TvMasc of the closely related T. varians [31, 40] was not a surprise. Moreover, the absence of a female‐specific small RNA that maps to OfMasc mRNA in O. furnacalis [41] from the distant family Crambidae suggests the evolution of distinct primary signals for sex determination in Lepidoptera. Nevertheless, the recent discovery of W chromosome-derived short silencing RNAs complementary to Masc mRNA in P. xylostella, a member of the basal family Plutellidae [46], in L. dispar from the family Erebidae [38] and in E. kuehniella and P. interpunctella, both from the family Pyralidae (this study), suggests a convergent evolution of a similar molecular mechanism of sex determination as in B. mori.

Identification of feminizing piRNAs in pyralid moths

Previous studies of the W chromosome of E. kuehniella have shown that there are no W-linked protein-coding genes [48], similar to other W chromosomes of lepidopterans [15]. Therefore, we used a bioinformatic approach to screen the W chromosome of E. kuehniella for potentially feminizing small RNAs capable of targeting EkMasc and/or EkMascB. The mode of operation of small RNAs depends on a high homology to their target RNA [28]. Given the high copy number of Fem piRNA found on the W chromosome of B. mori, we attempted to identify a similar type of feminizing small RNA in E. kuehniella by mapping W-specific reads against the EkMasc/EkMascB sequence. Using this approach, we identified a short sequence, EkMom piRNA, with high sequence identity to EkMasc and EkMascB, which is present in high copy number on the W chromosome. We were also able to confirm that EkMom is specific to the W chromosome and is not present in the male genome. The presence of EkMom piRNA coincides with the onset of sex determination, suggesting its possible role in this process. Thus, our approach of using the W chromosome reads to identify a candidate feminizing piRNA was successful in E. kuehniella. In addition to the EkMom piRNA, we identified another cluster of sequences with high sequence identity to EkMascB on the W chromosome of E. kuehniella that are also present as small RNAs during early development, suggesting that other small RNAs may also be involved in sex determination in this species. In P. interpunctella, we examined small RNA-seq data of embryos for the presence of putative feminizing small RNAs. To our surprise, we not only identified a candidate feminizing piRNA sequence, but also found that this PiMom piRNA is a homologue of EkMom piRNA.

The structure of Mom on the W chromosome of pyralid moths

The evolution of W chromosomes in Lepidoptera is rapid due to their hemizygous state, their gene-poor nature and the absence ofrecombination in females [48, 50,51,52]. So far, W-linked protein-coding genes have been identified in a few species (e.g. [53,54,55,56,57]), showing that the W chromosome of Lepidoptera is a hostile environment for genes. Genes on hemizygous chromosomes are at risk of losing their function due to the accumulation of potentially deleterious mutations that cannot be restored by recombination [58,59,60]. One process by which gene integrity can be maintained is intrachromosomal gene conversion. This process plays a key role in maintaining the high level of sequence homogeneity between tandemly repeated genes and is involved in the concerted evolution of multigene families in eukaryotes, including those located on the Y or W chromosomes [59, 61,62,63,64].

The Feminizer gene in B. mori consists of many copies of the Fem precursor, arranged in a tandem repeat structure and located in a single cluster on the W chromosome according to the results of FISH mapping [22]. However, a recent genome assembly of a B. mori female identified 129 Fem copies, arranged in tandem in one large and 10 smaller clusters on the W chromosome [40]. We also identified a large number of highly similar EkMom and PiMom copies arranged in tandem in a single cluster on the W chromosomes of E. kuehniella and P. interpunctella, respectively, using FISH. However, the occurrence of further smaller clusters cannot be completely ruled out due to the limited resolution of FISH. The sequences of EkMom/PiMom piRNA and Fem piRNA in B. mori show no homology, suggesting that they have evolved independently. However, the similarity in structure suggests that this arrangement arose through convergent evolution and could therefore be adaptive. The high level of sequence conservation between copies of EkMom and PiMom is likely the result of intrachromosomal gene conversion.

Presence of feminizing piRNAs during early development of Ephestia kuehniella

In B. mori, Fem piRNA is not provided maternally [23]. Furthermore, Fem piRNA was not detected in WZ eggs during the first 12 hpo, but was detected at low levels just before the differential expression of Masc between the sexes [23]. Similarly, we could not detect EkMom piRNA in E. kuehniella before the onset of female development. We detected the presence of EkMom piRNA around the onset of sex-specific differential expression of EkMasc and EkMascB, which supports a role for this piRNA in the initiation of female development in E. kuehniella. Surprisingly, we identified many other small RNAs that could theoretically target EkMasc and/or EkMascB. Most of these additional small RNAs originate from EkMasc^W, but some originate from other regions of the genome. Interestingly, these additional small RNAs are present both before and during the onset of female development. The presence of these additional small RNAs at both time points suggests that these small RNAs do not initiate feminization in E. kuehniella.

Initiation of the Mom piRNA pathway in pyralid moths

The Fem piRNA pathway in B. mori females is initiated by the presence of maternally provided Masc piRNA and the expression of Fem precursor RNA from the W chromosome during early embryogenesis [23, 27]. The Fem piRNA is then processed and amplified via a biogenesis pathway known as the ping-pong cycle [28], which involves two PIWI proteins: B. mori Argonaute3 (BmAgo3) binds to Masc piRNA and silkworm Piwi (Siwi) binds to Fem piRNA [27]. This is different in E. kuehniella, where the EkMom precursor RNA, but not the EkMasc piRNA, is maternally supplied to the eggs. In contrast to the female (WZ) embryos, however, no EkMom RNA was detected in the male (ZZ) embryos from 18 hpo onwards. This suggests that the maternally supplied transcripts are degraded in the embryos during the so-called maternal-to-zygotic transition, whereas EkMom RNA expression in WZ embryos is maintained by transcriptional activation of the zygotic genome [65, 66]. The maternal provision of EkMom RNA and the absence of EkMasc piRNA may indicate that EkMom piRNA is a primary piRNA, produced in the absence of EkMasc piRNA. In P. interpunctella, similar to B. mori, we found no maternal provision of PiMom precursor RNA. This may indicate that feminization in P. interpunctella is more similar to B. mori than to E. kuehniella, even though PiMom piRNA and EkMom piRNA share a common origin. However, both Mom piRNAs start with uridine (1U) at the 5’ end (Additional file 10: Fig. S8A, B), which is characteristic for primary piRNAs [28, 67]. In addition, the PiMasc piRNA identified in the available transcriptomes of small RNAs has adenine at position 10 (10A) (Additional file 10: Fig. S8B), which is the hallmark of secondary piRNAs [67]. This supports the hypothesis that the activation of the piRNA mechanism in the two pyralids differs from each other and may also differ from that in B. mori. However, EkMasc piRNA, which is required for piRNA biogenesis, was not present in the small RNA datasets obtained from early embryos of E. kuehniella in this study. Nevertheless, the two sampling time points were most likely too early to detect it, because the later sampling time point was more or less at the beginning of EkMom piRNA (the putative primary piRNA) production, and the secondary piRNA (i.e. EkMasc piRNA) would only be produced afterwards. It also cannot be ruled out that the primary biogenesis pathway is sufficient to suppress the EkMasc/EkMascB mRNA and that the EkMasc piRNA is completely absent in E. kuehniella. Overall, many aspects of the Mom piRNAs of E. kuehniella and P. interpunctella are unknown, and further studies are needed to obtain a comprehensive picture of their biogenesis, including the proteins involved in this process.

Feminizing function of Mom piRNA in the sex determination of pyralid moths

The results of our study strongly suggest that EkMom piRNA in E. kuehniella and PiMom piRNA in P. interpunctela play the role of the primary trigger of sex determination, just like Fem piRNA in B. mori [23]. In female embryos, the W-linked clusters of Mom genes encode a precursor Mom RNA that is processed into Mom piRNA. This EkMom piRNA or PiMom piRNA mediates the cleavage of EkMasc/EkMascB mRNA or PiMasc mRNA, respectively, which ultimately prevents the expression of the male-specific doublesex splice variant, leading to female development. Interestingly, both transcripts of the duplicated EkMasc and EkMascB genes in E. kuehniella are cleaved in female embryos, as clearly shown by the results of the modified 5’ RACE (see Fig. 5A, B). This indicates that both genes are targeted by EkMom piRNA, supporting the hypothesis that both genes play an essential role in masculinization [35]. In male embryos, in the absence of the W chromosome and thus the Mom genes, the EkMasc/EkMascB or PiMasc mRNAs are translated into proteins that mediate the expression of the male-specific doublesex splice variant, leading to male development. This process was mimicked by RNAi-mediated silencing of the precursor EkMom RNA, which shifted the female- to the male-specific splice variant of Ekdsx (Ekdsx^M) in female embryos of E. kuehniella (see Fig. 5D), further demonstrating the feminizing function of the EkMom piRNA.

Origin of feminizing piRNAs in Lepidoptera

The presence of EkMasc^W on the W chromosome of E. kuehniella suggests a possible pathway for the origin of EkMom. It is generally assumed that the piRNA pathway has evolved to protect against transposable elements in germline cells [68, 69]. The W chromosome of B. mori has been shown to be a source of piRNAs containing both functional and defective transposons that can be expressed in either sense or antisense direction or both, leading to the production of piRNAs [70]. In E. kuehniella, translocation of Masc to the W chromosome could be a source of antisense expression of EkMasc and EkMascB, one of the first steps required for piRNA synthesis. Indeed, we observed piRNAs derived from EkMasc^W in antisense orientation to EkMasc and EkMascB, possibly maternally supplied, as indicated by their presence before the onset of sex determination. A similar origin of feminizing piRNAs was recently discovered in P. xylostella [46]. In this species, a W‐linked locus called Pxyfem consists of “retrocopies” of the PxyMasc gene and generates a variety of ssRNAs, predominantly piRNAs, some of which are complementary to PxyMasc mRNA and thus capable of silencing this male-determining gene by RNA interference (RNAi). We hypothesize that EkMom has a similar origin to EkMasc^W, i.e. a translocation of Masc to the W chromosome, but has undergone extensive differentiation over time so that only one short region of homology to EkMasc and EkMascB remains conserved, namely the EkMom piRNA.

Recently, however, a completely different origin for the B. mori Fem piRNA has been proposed. Analysis of the high‐quality assembly of the B. mori W chromosome revealed a high similarity of a large part of the transcription unit of Fem to the RTE‐type LINE BovB transposon. In addition, the first half of the inner Fem piRNA‐producing part shows a satellite DNA-like sequence, while the second half shows similarity to the LTR of the gypsy transposon. This finding suggests that Fem is a chimeric sequence formed by the random fusion of multiple transposons whose boundaries formed the sequence complementary to Masc, which produces Fem piRNA [40].

The complete lack of homology between the Pxyfem locus of P. xylostella and the Fem locus of B. mori as well as the origin of Pxyfem from the highly divergent PxyMasc sequence in P. xylostella argue for an independent origin of the sequence-specific feminizer systems in these two phylogenetically very distant species [46]. Also, the lack of homology between the Fem piRNA of B. mori and the EkMom piRNA/PiMom piRNA in the two pyralid moths discovered in this study suggests an independent acquisition of feminizing piRNAs between the species, although the pathway for initiation of feminization, i.e. via piRNA-mediated RNAi, remains largely the same. This would imply that the piRNA pathway to initiate feminization has either evolved several times independently or that this system was acquired in a common ancestor, but the piRNA has been transformed or replaced over time. Two species belonging to the same superfamily as B. mori, i.e. Bombycoidea, but to a different family (Saturniidae) have lost the W chromosome, and therefore the initiation of feminization in these species cannot be controlled by a feminizing piRNA on the W chromosome [43, 71, 72]. However, due to the lack of further information, it is not known whether this is because the feminizing piRNA pathway has been lost in these species, or whether it was never widely integrated into the sex determination system in Lepidoptera. However, the presence of EkMasc^W suggests that it is possible for new feminizing piRNAs to evolve in parallel with an already established feminizing piRNA. Such a newly evolved piRNA could replace the established piRNA in the event of a Masc mutation in the target sequence for the feminizing piRNA, leading to insensitivity to downregulation of Masc in female embryos, as shown in the transgenic strains of B. mori expressing a Fem piRNA-resistant Masc gene [73].

Conclusions

We applied a sequence mapping strategy to identify putative feminizing small RNAs on the W chromosomes of two pyralid moth species, E. kuehniella and P. interpuntella. Both identified candidate sequences, EkMom piRNA and PiMom piRNA, are generated from a single locus on the W chromosomes and their presence results in suppression of EkMasc/EkMascB and PiMasc gene expression, respectively, in female embryos and subsequent female development. They show sequence homology, indicating their common origin. Overall, their characteristics and experimental evidence support their role as primary sex-determining signals in these species. In E. kuehniella, we also identified multiple duplications of the EkMascB gene on the W chromosome, EkMasc^W, which are expressed as small RNAs during early embryonic development. The presence of EkMasc^W may shed light on how feminizing piRNAs arise on the W chromosomes of Lepidoptera. The basic prerequisite is the translocation of a copy of the Masc gene to the W chromosome, for example via a transposon or by ectopic recombination, which could be a relatively frequent event. The translocated W-linked Masc could become a source of antisense expression of small RNAs homologous to the mRNA of the Z-linked Masc. This could form the basis for the evolution of a piRNA targeting Masc mRNA via RNAi. Although the genomic organization of the EkMom and PiMom loci is reminiscent of the structure of Fem in B. mori, sequence comparison of the two pyralid piRNAs with the Fem piRNA of B. mori suggests their independent origin, supporting the hypothesis of convergent evolution of feminizing piRNAs in Lepidoptera.

Methods

Insects

We used a laboratory strain (WT-C02 [35]) of the Mediterranean flour moth, Ephestia kuehniella, maintained on ground wheat with a small amount of dried yeast following the procedure described previously [74]. A laboratory culture of the Indian meal moth, Plodia interpunctella, was established in 2018 from larvae collected from nuts and other stored food products in České Budějovice, Czech Republic. The culture was maintained on a mixture of walnuts and raisins. Both species were reared under a 12-h light/12-h dark regime at 20–22 °C. These rearing conditions were maintained in all experiments.

Identification of candidate feminizing small RNAs in Ephestia kuehniella and Plodia interpunctella

Transcript silencing by piRNAs relies on homology with their target RNA sequences [28]. Therefore, we screened the W chromosome reads of E. kuehniella generated by sequencing W chromosome DNA [48] for short sequences homologous to EkMasc and EkMascB [35]. We performed a BLASTn search with the full-length EkMasc and EkMascB sequences (GenBank Acc. No. MW505939 and MW505941, respectively) using low stringency settings (word size 7; expected e-value of 10) against the W chromosome sequence database (GenBank Acc. No. SRR926303). The hits obtained were mapped back to the respective sequence using Geneious 9.1.6 (https://www.geneious.com [75]) with default settings. The correct alignment of the reads was checked manually and corrected if necessary, and graphs were subsequently plotted using R version 3.5.2 [76] with the ggplot2 package [77]. To provide context about the location of reads that were mapped to the genes, exon maps and shading were added using Inkscape 0.92 (https://inkscape.org/). Based on the high coverage and short length, one putative feminizing piRNA sequence (EkMom piRNA) was identified and subsequently used in a BLASTn search against the E. kuehniella W chromosome database. We extracted the complete reads for all hits, aligned them and inferred a consensus sequence of EkMom (GenBank Acc. No. PP715506) corresponding to a single monomer of this repeat.

Primers EkMom_F1 and EkMom_R1 (Additional file 19: Table S1) were designed using Primer3 v2.3.4 [78] based on the EkMom consensus sequence to verify the presence of EkMom in the female genome. Genomic DNA (gDNA) was extracted from female and male pupae of E. kuehniella using the NucleoSpin DNA Insect kit (Macherey–Nagel, Düren, Germany) with slight modifications to the manufacturer’s protocol as previously described [17]. The PCR mixture consisted of 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each forward and reverse primer, 0.025 units of Ex Taq polymerase and 10 ng of gDNA. Amplification was performed using a thermocycling program with an initial denaturation at 94 °C for 3 min, 35 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 2 min, and a final extension at 72 °C for 5 min. Products were purified using the Wizard SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA), cloned using the pGEM-T Easy Vector (Promega) and plasmids were isolated using the NucleoSpin Plasmid kit (Macherey–Nagel) and sequenced as described in our previous study [35].

In addition to the EkMom sequence, a large segment of EkMasc and EkMascB showed coverage on the W chromosome. In our previous study [35], a genetic sexing method using Masc-specific primers was described in which additional fragments were amplified in females, suggesting that the observed coverage on the W chromosome is indeed W chromosome-specific. We designed primers EkMascW_F1 and EkMascW_R1 (Additional file 19: Table S1) to amplify the largest possible fragment of this W-linked Masc (EkMasc^W) sequence, performed PCR, cloned the amplified fragments and sequenced four clones as described above (GenBank Acc. No. PP715507-PP715510).

To search for putative feminizing small RNAs in P. interpunctella, the complete PiMasc sequence (annotated as maker-scaffold92-augustus-gene-0.127-mRNA-1) obtained from LepBase (https://lepbase.org/ [79]) was used in a BLASTn search against three miRNA-seq databases derived from pooled eggs (GenBank Acc. Nos. SRX1539920–SRX1539922), using low stringency settings (see above) to detect homologous sequences. The sequences obtained were then mapped back to the PiMasc sequence, and the results were plotted using R. To identify the full PiMom sequence, we used a putative PiMom piRNA sequence in a BLASTn search against the unassembled genomic reads of a P. interpunctella female (GenBank Acc. No. ERX334127). The entire reads of the hits as well as the forward and reverse reads of the pair were aligned to create a consensus sequence of PiMom.

To test whether the assembled PiMom sequence was female-specific, primers PiMom_F1 and PiMom_R1 (Additional file 19: Table S1) were designed and used in PCR. Genomic DNA was extracted from female and male pupae, and PCR and sequencing were performed as described above. The annealing temperature of the PCR was adjusted to 58 °C. Two PiMom sequence types were identified (GenBank Acc. Nos. PP715511 and PP715512).

Evolutionary history of EkMasc ^W

To determine the origin of EkMasc^W in E. kuehniella, we constructed a phylogenetic tree of EkMasc, EkMascB and four EkMasc^W sequences. The six sequences and the PiMasc sequence were aligned using MAFFT [80, 81]. The alignment was used to construct a phylogenetic tree using MrBayes [82] with default settings, with PiMasc selected as the outgroup.

Quantification of EkMom, EkMasc ^W and PiMom copy numbers

The copy numbers of genes on the W chromosomes of E. kuehniella and P. interpunctella were determined by quantitative PCR (qPCR) as described in our previous study [35]. For the autosomal reference gene, the E. kuehniella ortholog of Acetylcholinesterase 2 (EkAce-2; GenBank Acc. No. MW505944), the primers Ek_Ace2_F × Ek_Ace2_R (Additional file 19: Table S1) [35] were used. The P. interpunctella ortholog of Ace-2 (PiAce-2) was identified by a tBLASTn search in the available P. interpunctella transcriptome in Lepbase [79] using the EkAce-2 sequence [35]. Primers used for the amplification of EkMom, EkMasc^W, PiMom and PiAce-2 can be found in Additional file 19: Table S1. To estimate the copy number of EkMasc^W, primers were selected that amplified all EkMasc^W copies as well as EkMasc and EkMascB.

In both species, gDNA was extracted from female and male pupae using the NucleoSpin DNA Insect kit as described above. The experiments were performed using three biological and three technical replicates. The 10 µL qPCR mix consisted of 1 × Xceed SG qPCR Mix Lo ROX (Institute of Applied Biotechnologies, Prague, Czech Republic), 400 nM of each forward and reverse primer and 10 ng of genomic DNA. For PiMom, primer concentrations were increased to 4 μM to increase qPCR efficiency. The thermocycling program included a combined annealing-extension step at 60 °C for 20 s for E. kuehniella genes and 15 s for P. interpunctella genes to avoid amplification of multiple monomers. Copy number estimates were calculated by comparing the gene of interest to the autosomal reference gene as previously described [35]. For EkMasc^W, the female-to-male ratio was then calculated, and the female copy number estimate of EkMasc^W was obtained by multiplying the female-to-male ratio by the diploid Masc copy number in males (i.e. 4, two copies on each Z chromosome) and subtracting the haploid Masc copy number in males (i.e. 2, representing the single Z chromosome in females).

To obtain a second copy number estimate for EkMom and EkMasc^W in E. kuehniella, we used available genome sequencing data to compare the coverage of the two genes with the chromosome/genome sequencing depth. A single monomer of the consensus sequence of both genes was used in a BLASTn search against the unassembled W chromosome reads of E. kuehniella [48]. The reads obtained were then mapped against the respective sequences, and copy numbers were estimated by dividing the average coverage of the sequence of interest by the total chromosome sequencing depth estimated in the previous study [48].

To obtain a second copy number estimate for PiMom, we used a similar approach as described for E. kuehniella by using genomic reads from a female specimen of P. interpunctella (GenBank Acc. No. ERX334127). For P. interpunctella, the average coverage of PiMom was divided by the sequencing depth of the sequencing project. Since no genome size estimate is available for P. interpunctella, the genome size of E. kuehniella was used [83], as this is the closest relative with known genome size.

Chromosome preparations and fluorescence in situ hybridization (FISH)

Spread chromosome preparations were made from ovaries (for meiotic chromosomes) or wing imaginal discs (for mitotic chromosomes) of fifth instar female larvae of E. kuehniella and P. interpunctella, as described previously [84, 85].

To map the EkMasc^W and EkMom clusters on the W chromosome of E. kuehniella, we used two FISH techniques. First, we performed genomic in situ hybridization (GISH) to visualize the W chromosome, and then we re-probed the slides using tyramide signal amplification FISH (TSA-FISH) with two probes specific for EkMasc^W and EkMom. To map the PiMom cluster on the W chromosome of P. interpunctella, we used TSA-FISH with two probes, one for PiMom and one for the W-specific repeat PiSAT1 [49] to identify the W chromosome.

For GISH, high molecular weight DNA was isolated from individual female and male pupae of E. kuehniella using cetyltrimethylammonium bromide (CTAB) as previously described [86]. Female gDNA was fluorescently labelled using an improved nick translation protocol [87] with previously described modifications [17]. Approximately 1 μg of female gDNA was labelled with Cy3-dUTP (Jena Bioscience, Jena, Germany), the labelling reaction was incubated at 15 °C for 2 h 15 min and inactivated by incubation at 70 °C for 10 min. Male gDNA was sheared to act as a species-specific competitor in GISH, and therefore 12 μg DNA was incubated at 90 °C for 20 min. The GISH procedure followed the protocol for CGH [88] with the previously described modifications [44] and some additional modifications. The 10 μL hybridization mixture contained 300 ng of female gDNA probe, 3 μg of male competitor DNA and 25 μg of sonicated salmon sperm DNA (Sigma-Aldrich, St. Louis, MO, USA) in 50% deionized formamide and 10% dextran sulfate in 2 × SSC. This mixture was denatured at 90 °C for 5 min and immediately chilled on ice for at least 5 min. Slide pretreatments were more extensive to avoid background signals in subsequent TSA-FISH and followed previously described procedures [89] with some modifications. In particular, the treatment with 50 μg/mL pepsin in 0.01 M HCl at 37 °C was extended to 10 min and a final pretreatment by incubating the slides for 30 min in 5 × Denhardt’s solution (0.1% Ficoll, 0.1% polyvinylpyrrolidone, 0.1% bovine serum albumin) at 37 °C was added. In contrast to the published work [89], the slides were not passed through an ethanol series, but were immediately denatured with 70% formamide in 2 × SSC for 3.5 min at 68 °C, then passed through an ice-cold ethanol series (70%, 80% and 100%; 30 s each) and air-dried. The hybridization mixtures were applied to these slides and hybridization was carried out at 37 °C for 3 days. The slides were then incubated in 1% Triton X-100 0.1 × SSC at 62 °C for 5 min and washed in 1% Kodak PhotoFlo (Sigma-Aldrich) for 1 min. Finally, slides were mounted in antifade based on DABCO (1,4-diazabicyclo(2.2.2)-octane; Sigma-Aldrich) containing 0.5 μg/mL DAPI (4’,6-diamidino-2-fenylindole; Sigma-Aldrich). GISH results were observed using a Zeiss Axioplan 2 microscope (Carl Zeiss, Jena, Germany) equipped with an XM10 monochrome CCD camera (Olympus Europa Holding, Hamburg, Germany). Images were captured separately for each fluorescent dye using cellSens Standard software version 1.9 (Olympus), pseudo-coloured and merged using Adobe Photoshop CS6 (Adobe Systems, San Jose, CA, USA). The coordinates on the slide were recorded to return to the same coordinates after the second round of FISH, i.e. after TSA-FISH.

For TSA-FISH in E. kuehniella, we prepared gene-specific probes for EkMom and EkMasc^W. Plasmids containing a single monomer of EkMom or EkMasc^W (this study) were used to amplify the probe template with gene-specific primers (EkMom_F1 × EkMom_R1; EkMascW_F1 × EkMascW_R1; Additional File 19: Table S1). The PCR mixtures, 50 μL each, consisted of 1 × OneTaq PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each forward and reverse primer, 0.125 units of OneTaq polymerase, and 5 ng of plasmid DNA. For EkMom, the thermocycling program consisted of an initial denaturation at 94 °C for 3 min, 35 cycles of denaturation at 94 °C for 30 s, annealing at 58 °C for 30 s and extension at 68 °C for 30 s, and a final extension at 68 °C for 5 min. For EkMasc^W, the same thermocycling program was used, but the extension time was increased to 1 min 30 s to ensure complete amplification. Products were purified using the Wizard SV Gel and PCR Clean-Up System and samples were eluted in 30 μL of nuclease-free water. The probes were PCR-labelled using the same gene-specific primers. However, due to the size of the EkMasc^W sequence (1322 bp), we split the probe amplification for this sequence into three separate PCRs of 339 bp (primers EkMascW_F1 × EkMascW_R2), 492 bp (primers EkMascW_F2 × EkMascW_R3) and 525 bp (primers EkMascW_F3 × EkMascW_R1), each with a partial overlap of primers. All labelling reactions consisted of 1 × Ex Taq PCR buffer, 40 μM dATP, dCTP and dGTP, 30 μM dTTP, 10 μM labelled dUTP, 0.4 μM gene-specific forward and reverse primer, 0.625 units of Ex Taq polymerase and 5 ng of purified PCR product template. EkMom was labelled with dinitrophenol-11-dUTP (DNP; PerkinElmer, Waltham, MA, USA), while EkMasc^W was labelled with fluorescein-dUTP (Jena Bioscience). Labelling was performed using the following thermocycling profile: 94 °C for 3 min initial denaturation, 35 cycles of denaturation at 94 °C for 30 s, annealing at 60 °C for 30 s and extension at 68 °C for 1 min 30 s, and a final extension at 68 °C for 2 min. The probes were purified using an Illustra Sephadex G-50 column (GE Healthcare Life Sciences, Buckinghamshire, UK). Probe concentrations were measured with an Invitrogen Qubit 3.0 Fluorometer using the dsDNA BR Assay Kit.

To prepare the probes for TSA-FISH in P. interpunctella, template DNA for labelling was prepared by PCR using plasmid DNA containing a single monomer of PiMom (this study) and almost four complete monomers of PiSAT1 [49]. For both PiMom and PiSAT1, the 50-μL PCR reaction consisted of 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each forward and reverse primer (PiMom_F1 and PiMom_R1 for PiMom; M13 universal primers for PiSAT1, purchased from Thermo Fisher Scientific, Waltham, MA, USA), 0.125 units of Ex Taq polymerase, and approximately 25 ng of plasmid DNA. The same thermocycling program was used as described for testing the female specificity of PiMom. Products were purified using the Wizard SV Gel and PCR Clean-Up System and eluted in 30 μL of nuclease-free water. The probes were labelled by PCR with specific primers (PiMom_F1 × PiMom_R1; PiSAT1_F × PiSAT1_R; Additional File 19: Table S1) [49] in a volume of 25 μL, using the same composition of labelling reactions as for the labelling of EkMom and EkMasc^W. PiMom was labelled with DNP-11-dUTP and PiSAT1 with fluorescein-dUTP. Labelling was performed using a thermocycling profile of 94 °C for 3 min of initial denaturation, 35 cycles of denaturation at 94 °C for 30 s, annealing at 60 °C for 30 s and extension at 72 °C for 1 min 30 s, and a final extension at 72 °C for 2 min. Amplification was verified by gel electrophoresis and the products were purified on an Illustra Sephadex G-50 column. Due to the presence of multiple monomers in the PiSAT1 template and the relatively short monomer size (approximately 170 bp), the labelling reaction resulted in a large smear of DNA fragments that were too long to hybridize properly to chromosomes. Therefore, DNase I was added to the PiSAT1 probe at a final concentration of 2e − 04 units/μL, and the mixture was incubated at 15 °C for 25 min, followed by deactivation of the enzyme at 70 °C for 10 min. The size of the probe was verified by gel electrophoresis, and the probe was purified again on an Illustra Sephadex G-50 column. The probe concentrations were determined as described above.

TSA-FISH was performed according to the published protocol [89] with some modifications. For E. kuehniella, the probes from the GISH experiment were removed from the slides and the slides were prepared for reprobing as previously described [52], with one additional step: after dehydration, the slides were incubated in 5 × Denhardt’s solution at 37 °C for 30 min. For P. interpunctella, the slides were pretreated as described for GISH. For both species, the probe mixture consisted of 20 ng of each probe (EkMom probe and three EkMasc^W probes for E. kuehniella; PiMom probe and PiSAT1 probe for P. interpunctella) in 50% deionized formamide, 10% dextran sulfate, 2 × SSC in a total volume of 50 μL. The probe mixture was applied to the slides, covered with a coverslip and denatured at 70 °C for 5 min. Hybridization was performed in a humid chamber at 37 °C for 12–16 h. After hybridization, the coverslip was removed and the slides were washed three times in 50% formamide in 2 × SSC at 46 °C for 5 min each. The slides were then incubated three times in 2 × SSC at 46 °C for 5 min, three times in 0.1 × SSC at 62 °C for 5 min and once in TNT buffer (0.1 M Tri-HCl pH 7.5, 0.15 M NaCl, 0.05% Tween-20) at room temperature (RT) for 5 min. Slides were blocked with 200 μL of TNB buffer (0.1 M Tris–HCl pH 7.5, 0.15 M NaCl, 0.5% Blocking Reagent; PerkinElmer) at RT for 30 min in a humid chamber. Excess TNB buffer was poured off and the slides were incubated with 200 μL of antifluorescein-HRP conjugate (PerkinElmer), diluted 1:1000 with TNB buffer, in a humid chamber at RT for 1 h. Slides were washed three times in TNT buffer at RT for 5 min and tyramide amplification was performed using the TSA Plus Fluorescein kit (PerkinElmer) according to the manufacturer’s instructions. One hundred microliters of tyramide working solution was applied to each slide, and the slides were covered with a coverslip and incubated in a humid chamber at RT for 8 min. Slides were washed three times in TNT buffer at RT for 5 min each and then incubated in 1% H₂O₂ in 1 × PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄) to quench the antibody-conjugated peroxidase. Slides were washed three times in TNT buffer at RT for 5 min each and blocked in 200 μL of TNB buffer in a humid chamber at RT for 30 min. Excess TNB buffer was poured off again, but this time the slides were incubated with 200 μL of anti-DNP-HRP conjugate (PerkinElmer), diluted 1:1000 with TNB buffer, in a humid chamber at RT for 1 h. Slides were washed three times in TNT buffer at RT for 5 min, and tyramide amplification was performed using the TSA Plus Cyanine 3 (Cy3) detection kit (PerkinElmer) according to the manufacturer’s instructions. Again, 100 μL of the tyramide working solution was applied to each slide, the slides were covered with a coverslip and incubated in a humid chamber at RT for 8 min. The slides were washed three times in TNT buffer at RT for 5 min each and incubated in 1% Kodak PhotoFlo in H₂O. Finally, the slides were mounted in VECTASHIELD antifade mounting medium with DAPI (Vector Laboratories, Burlingame, CA, USA). Slides were observed and images were captured as described for GISH (see above).

Structural analysis of EkMom and PiMom by Southern hybridization

The genomic structure of EkMom and PiMom repeats was analysed by Southern hybridization. EkMom and PiMom probes were prepared as described for TSA-FISH using the same primers and gDNA templates, but labelled with digoxigenin-11-dUTPs (Roche Diagnostics, Mannheim, Germany). For each species, gDNAs were isolated from individual female and male larvae (E. kuehniella) or pupae (P. interpunctella) using CTAB as previously described [86], and their concentrations were measured with an Invitrogen Qubit 3.0 Fluorometer using the dsDNA BR Assay Kit. Extracted gDNAs were either fully (one female and one male sample) or partially (approximately 25%; one female sample) digested with restriction enzymes that cut once per monomer but not in the sequence of the probe. Samples of E. kuehniella were digested with EcoT22I (TaKaRa, Otsu, Japan), and for partial digestion the enzyme was diluted 4 × with nuclease-free water before use. Samples of P. interpunctella were digested with Eco32I (EcoRV) (Fermentas, Vilnius, Lithuania), and for partial digestion the enzyme was diluted 16 × with nuclease-free water before use. Digestion reactions were performed according to the manufacturer’s instructions supplied with the restriction enzymes. All reactions were incubated at 37 °C for 1 h, and enzymes were inactivated by addition of Gel Loading Dye, Purple (6 ×) (New England Biolabs, Ipswich, MA, USA). Hybridizations were performed as previously described [90] with some modifications [49].

Expression analysis of sex-determining genes during early embryogenesis of Ephestia kuehniella

To determine the time of onset of sexual differentiation in E. kuehniella, we analysed the expression of genes involved in sex determination during early embryogenesis. To this end, we analysed the expression of EkMom and the male-specific splice form of the E. kuehniella ortholog of the insulin-like growth factor II mRNA-binding protein (EkImp^M) gene. As a control for successful amplification of cDNA, we used primers for the Ekrp49 gene (GenBank Acc. No. MW505943) designed in our previous study [35]. The B. mori IMP protein sequence (GenBank Acc. No. LOC101745047) was used in a tBLASTn search against the male genome of E. kuehniella (GenBank Acc. No. PRJNA683200) to obtain partial exon sequences, after which primers were designed. Total RNA was isolated from approximately 50 pooled eggs collected within 24 h post oviposition (hpo), a single male pupa and a single female pupa using the NucleoSpin RNA kit (Macherey–Nagel) according to the manufacturer’s instructions with some modifications. Briefly, samples were homogenized in the presence of lysis buffer, then β-mercaptoethanol was substituted for tris(2-carboxyethyl)phosphine (TCEP) during lysis and elution was performed with RNase-free H₂O in a volume of 40 μL. Concentrations were measured with a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and 1 μg of RNA was converted to cDNA with the ImProm-II Reverse Transcription System kit (Promega) according to the manufacturer’s instructions using the Oligo(dT)₁₅ primer supplied with the kit and a final concentration of 3 mM MgCl₂. To isolate Imp in E. kuehniella, PCR was performed with primers EkImp_F1 and EkImp_R1 (Additional file 19: Table S1), using the PCR mixture and thermocycling program as described for the isolation of EkMom. The samples were then processed and sequenced in the same steps as described above. To identify the male-specific splice variant of EkImp, a set of primers (EkImp_3’RACE_F1 and EkImp_3’RACE_F2; Additional file 19: Table S1) was designed based on the identified sequence. Rapid Amplification of cDNA Ends PCR (RACE-PCR) was performed for the 3’-region of the cDNA as previously described [35] using RNA from the pooled sample of eggs. The male-specific splice variant, EkImp^M, was identified (GenBank Acc. No. PP715505) and a reverse primer in the male-specific exon (Additional file 19: Table S1) was designed and used in RT-PCR to confirm male-specific expression of this splice variant.

For expression analysis during embryogenesis, DNA and RNA were isolated simultaneously from individual embryos using TRI reagent (Sigma-Aldrich) as previously described [35]. Embryonic DNA was used for genetic sexing of embryos according to the previously described protocol [35]. DNA contamination was removed from the total RNA fraction using the Invitrogen TURBO DNA-free kit (Thermo Fisher Scientific) according to the manufacturer’s instructions, and total RNA was converted to cDNA using the ImProm-II Reverse Transcription System kit according to the manufacturer’s instructions, using a mixture of Oligo(dT)₁₅ primer and random primer (1:1) supplied with the kit and a final concentration of 3 mM MgCl₂. In our previous study [35], the expression of EkMasc and EkMascB peaked in males around 16 hpo, so we used the same time series of 12–24 hpo with 2-h sampling intervals. The PCR mixtures for the Ekrp49, EkMom and EkImp^M genes contained 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each forward and reverse primer (qrp49_F2 × qrp49_R2 or EkMom_F1 × EkMom_R1 or EkImpM_F1 × EkImpM_R1; Additional file 19: Table S1) [35], 0.025 units of Ex Taq polymerase and 2 μL of cDNA in a total reaction volume of 10 μL. The thermocycling program consisted of an initial denaturation at 94 °C for 3 min, 40 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for 40 s, and a final extension at 72 °C for 5 min. Amplified products were separated on a 1.5% agarose gel in 1 × TAE buffer.

To study maternal provision of EkMom, RNA was isolated from pooled samples of mixed-sex embryos. We isolated RNA from two samples, each containing 30–40 pooled eggs collected less than 1 hpo, with TRI reagent according to the manufacturer’s protocol, using chloroform for phase separation and 20 μL of diethyl pyrocarbonate (DEPC)-treated water to dissolve the RNA pellets. RNA concentrations were measured using a NanoDrop 2000 spectrophotometer, and total RNA was treated with DNase using the TURBO DNA-free kit. For each sample, two cDNA conversion reactions were prepared with 500 ng of RNA each using the ImProm-II Reverse Transcription System kit. Reverse transcriptase was added to one of the two reactions per sample, while the other served as a negative control for downstream purposes. All cDNA reactions were prepared using a mixture of Oligo(dT)₁₅ primer and random primer (1:1) and a final concentration of 3 mM MgCl₂ as described above. PCR for EkMasc (targeting EkMasc and EkMascB simultaneously) with the primer pair Masc_F1 and Masc_R1 (Additional file 19: Table S1) [35] was used to control for cDNA integrity. The primers EkMom_F1 and EkMom_R1 were used to amplify EkMom. In addition, a sample with gDNA was added to each PCR as a positive control to verify successful amplification. The same PCR mixture and thermocycling profile were used as described in the previous paragraph, but amplification was only performed for 35 cycles.

Expression of EkMom piRNA during embryogenesis of Ephestia kuehniella

To determine whether the putative EkMom piRNA is expressed during the onset of sex determination, we sequenced the small RNA fraction of eggs during early development. We chose two time periods for RNA extraction, one before sex-specific differential expression of EkMasc and EkMascB [35], i.e. 7–9 hpo, and one at the onset of sex-specific differential expression, i.e. 11–13 hpo. Since the expression of EkMasc and EkMascB decreases in females between 12 and 14 hpo, while it increases in males during the same period, we assumed that the expression in females must be reduced by EkMom piRNA. Total RNA was isolated from the two samples, each containing 200–300 pooled mixed-sex embryos, using TRI reagent according to the manufacturer’s protocol for tissue. To increase the purity of the RNA sample, two rounds of phase separation with chloroform and two rounds of ethanol washes were performed. The RNA pellets were dissolved in 30 μL RNase-free water, sample concentration and purity were assessed using NanoDrop 2000 spectrophotometer and RNA integrity was assessed using an RNase-free 1% agarose gel in 1 × TBE buffer stained with ethidium bromide. Total RNA samples were sent to Novogene (HK) Co., Ltd. (Hong Kong, China) for small RNA library preparation (Library type: 18–40 bp insert sRNA library) and sequencing of single-end 50-bp reads on an Illumina Novaseq 6000. Adapter trimming and quality control of reads were also performed by this company. Approximately 21–21.5 million reads were obtained for each sample (GenBank Acc. No. PRJNA1107079). Reads with high homology to EkMasc and EkMascB were identified using a BLASTn search in NCBI Genome Workbench (word size 7; expected e-value of 10), and all antisense reads were subsequently mapped back to the respective sequences using Geneious. For each peak, a representative sequence was extracted and compared to the EkMasc, EkMascB, EkMasc^W and EkMom sequences to determine their origin. Data for the graphs were exported and plotted using R.

Sliding window analysis of EkMom and PiMom

To compare the sequence homology between EkMom and PiMom, the consensus sequences of EkMom and PiMom monomers were aligned with MAFFT and the sequence identity between the two sequences was exported. These data were used to calculate the average sequence identity over a 10-nucleotide window corresponding to the size of the ping-pong signature. The window was shifted by one nucleotide until the end of the alignment. Data were plotted using R with the ggplot2 package and additional information was added using Inkscape 0.92.

Detection of maternal provision of PiMom and PiMasc in Plodia interpunctella

To test whether PiMom piRNA was present before transcription began, we isolated total RNA from two samples of pooled eggs. The first sample contained 28 eggs collected within 4 hpo, and the second sample contained 14 eggs collected within 1 hpo. Extraction of RNA using TRI reagent and all other procedures, including treatment with the TURBO DNA-free kit and conversion to cDNA, were performed in the same manner as described above for the study of maternal provision of EkMom. RT-PCR was performed in a volume of 10 μL consisting of 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each PiMom_F1 and PiMom_R1 primer, 0.025 units of Ex Taq polymerase and 1 μL of cDNA. The same thermocycling program was used as described above for testing the female specificity of PiMom. To verify correct cDNA conversion, P. interpunctella Masculinizer (PiMasc) was amplified with the primers Masc_F1 × Masc_R1 (Additional file 19: Table S1) [35] using the same PCR mixture and thermocycling program as in the PCR for EkMasc (see above).

PCR-based genetic sexing of embryos in Ephestia kuehniella and Plodia interpunctella

For E. kuehniella, the gDNA extracted from the embryos was used for PCR-based genetic sexing according to the previously described protocol [35] with the primers Masc_F1 and Masc_R1. For P. interpunctella, we developed a method for genetic sexing using two primer sets: PiMom_F1 and PiMom_R1 to amplify a female-specific fragment of the W-linked PiMom gene, and qPiMasc_F1 and qPiMasc_R1 to amplify a fragment of the Z-linked PiMasc gene in both sexes. The PCR mixture contained 0.2 μM of each primer, 0.2 mM dNTPs, 1 × Ex Taq PCR buffer, 0.025 units/μL of Ex Taq DNA polymerase (TaKaRa) and 2 μL of gDNA in a total reaction volume of 20 μL. The thermocycling profile consisted of initial denaturation at 94 °C for 5 min; 35 cycles of denaturation at 94 °C for 30 s, annealing at 58 °C for 30 s and extension at 72 °C for 30 s; with a final extension at 72 °C for 2 min. The primers used for PCR to sex the embryos of both species can be found in Additional file 19: Table S1 [35].

Detection of Mom piRNA-directed cleavage sites in Masc mRNA by modified 5’ RACE

To identify cleavage sites in EkMasc/EkMascB mRNA and PiMasc mRNA in embryos of E. kuehniella and P. interunctella, respectively, we performed modified RNA Ligase-mediated 5′ rapid amplification of cDNA ends (modified 5’ RACE), essentially following the previously described procedure [23, 91].

In both species, total RNA and gDNA were isolated simultaneously from 16 individual 48 hpo eggs using TRI reagent (Sigma-Aldrich) as previously described [35]. The gDNA was used for PCR-based genetic sexing of embryos as described above. Based on the results of molecular sexing, total RNA of each sex was pooled and purified using RNA Clean & Concentrator-5 (Zymo Research, Irvine, CA, USA) according to the manufacturer’s protocol, including DNase I treatment to remove DNA contamination [43]. The purified total RNA of each sex was then ligated with T4 RNA ligase (Thermo Fisher Scientific) to the GeneRacer RNA Oligo adapter (5’-CGACUGGAGCACGAGGACACUGACAUGGACUGAAGGAGUAGAAA-3’) custom synthesized by Azenta Life Sciences (Genewiz, Leipzig, Germany). The RNA ligated to the GeneRacer adapter was purified using the RNA Clean & Concentrator-5 and then converted to cDNA using SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific) according to the manufacturer’s instructions. A mixture of Oligo(dT)₂₀ primer and gene-specific reverse primer (EkMasc_5RACE_R1 for E. kuehniella or PiMasc_prime_R1 for P. interpunctella) was used. After reverse transcription, the modified 5’ RACE PCR was performed with the GeneRacer5Primer and a gene-specific primer (EkMasc_5RACE_R2 for E. kuehniella or PiMasc_prime_R2 for P. interpunctella). The PCR mixture consisted of 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.6 μM GeneRacer5Primer and 0.2 μM gene-specific primer, 0.025 units/μL of Ex Taq polymerase and 1 μL of cDNA in a total reaction volume of 50 μL. We used touchdown PCR conditions as described in the GeneRacer Kit user manual (Invitrogen, Carlsbad, CA, USA), with a gene-specific annealing temperature of 65 °C for E. kuehniella and 62 °C for P. interpunctella. The amplified products were separated on a 1.5% agarose gel in 1 × TAE buffer. As multiple bands were seen on the gel for E. kuehniella, we used the amplified products for a nested PCR, following exactly the conditions recommended in the GeneRacer Kit manual. The mixture for this nested PCR consisted of 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.2 μM GeneRacer5NestedPrimer and 0.2 μM EkMasc_5RACE_R3, 0.025 units/μL of Ex Taq polymerase and 1 μL of the original PCR product in a total reaction volume of 50 μL. For both species, the female-specific bands of the expected size were excised from the gel and purified using the Wizard SV Gel and PCR Clean-Up System (Promega). The obtained DNA fragments were cloned into the pGEM-T Easy Vector (Promega), the plasmids were isolated using the NucleoSpin Plasmid kit (Macherey–Nagel) and the samples from 16 clones of each species were sequenced (SEQme, Dobříš, Czech Republic). The primers used for the modified 5’ RACE procedure in both species can be found in Additional file 19: Table S1.

Preparation of dsRNAs and suppression of EkMom RNA by dsRNA microinjection into embryos

To investigate the function of EkMom in female sex determination, EkMom-dsRNA (320 bp) was designed according to the consensus sequence (GenBank Acc. No. PP715506) to suppress EkMom RNA in embryos of E. kuehniella (Additional file 17: Fig. S13A). In addition, DsRed-dsRNA (280 bp) targeting the Discosoma red fluorescent protein (GenBank Acc. No. FJ226078) was designed for the negative control experiments (Additional file 17: Fig. S13 B). The synthesis of double-stranded RNAs (dsRNAs) was performed as previously described [92]. Briefly, EkMom and DsRed fragments were amplified from cloned sequences by PCR with primers containing T7 promoter sites (Additional file 17: Fig. S13A, B). The PCR mixture consisted of 1 × Ex Taq PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each primer, 0.025 units/μL of Ex Taq polymerase and 20 pg of plasmid DNA in a total reaction volume of 50 μL. The thermocycling program consisted of an initial denaturation at 94 °C for 5 min, 35 cycles of denaturation at 94 °C for 30 s, annealing at 63 °C for 30 s and extension at 72 °C for 30 s, and a final extension at 72 °C for 10 min. The resulting PCR products were purified using the Wizard SV Gel and PCR Clean-Up System (Promega) and used as a template for in vitro transcription using the AmpliScribe T7 High Yield Transcription Kit (LGC Bioresearch Technologies, Middleton, WI, USA) according to the manufacturer’s instructions. Subsequently, the transcribed complementary RNA strands were purified using the RNA Clean & Concentrator-5 kit (Zymo Research) according to the manufacturer’s protocol and annealed by incubation at 70 °C for 10 min followed by slow cooling to RT for 20 min, allowing the formation of dsRNAs. Their concentrations were measured using a NanoDrop 2000 spectrophotometer and the synthesized dsRNAs were stored at − 80 °C.

Microinjections of dsRNA into freshly laid eggs were performed as previously described for E. kuehniella [35] with slight modifications. Briefly, E. kuehniella eggs were collected within 1 hpo and injected within 2–4 hpo with EkMom-dsRNA or with the control DsRed-dsRNA. Each dsRNA was administered to eggs at a final concentration of 1.75 μg/μL in a buffer containing 100 mM potassium acetate, 30 mM HEPES and 2 mM magnesium acetate. Eggs were injected with a FemtoJet Microinjector (Eppendorf, Hamburg, Germany) with needles made of borosilicate glass capillaries with filament, with an outer diameter of 1 mm and an inner diameter of 0.5 mm (Sutter Instrument, Novato, CA, USA) using a Magnetic Glass Microelectrode Horizontal Needle Puller PN‐31 (Narishige, Tokyo, Japan). After injection, the eggs were transferred to Petri dishes with moistened filter paper to ensure high humidity and incubated under standard rearing conditions. Total RNA and gDNA were isolated simultaneously from individual 24 hpo eggs using TRI reagent (Sigma-Aldrich) as previously described [35]. The gDNA was used for PCR-based genetic sexing of the embryos as described above. Subsequently, RNA from at least three confirmed WZ and three confirmed ZZ samples was used to evaluate the effects of EkMom RNA suppression on sex determination by detecting the splicing patterns of the E. kuehniella doublesex (Ekdsx) gene using reverse transcription PCR (RT-PCR). The Ekdsx gene was identified and its splicing pattern characterized in our previous study [35].

To assess the splicing pattern of Ekdsx, RNA samples from injected embryos were converted to cDNA. The individual cDNA samples were then used as templates for RT-PCR with dsx_dR_F2 and dsx_dR_R2 primers previously designed for Ekdsx (see Additional file 19: Table S1) [35]. RT-PCR was performed in a final volume of 20 μL containing 0.2 μM of each primer, 0.2 mM of each dNTP, 1 × Ex Taq PCR buffer, 0.025 units of Ex Taq DNA polymerase (TaKaRa, Otsu, Japan) and 2 μL of cDNA. PCR amplifications were performed under the following conditions: 94 °C for 5 min, 35 cycles of 94 °C for 30 s, 60 °C for 30 s, 72 °C for 40 s, and a final extension step at 72 °C for 1 min.