The sex determination pathways in insects are diverse [2]. The diversity arises mostly from primary signals, whereas transducers and executors are more conserved [29,30]. In D. melanogaster, M. domestica and C. capitata the primary sex determiner controls sex development via Tra-transduced signaling pathway, resulting in the alternative splicing of dsx [5,16,17]. The transducer gene usually encodes a splicing factor, and manipulation by the primary signals therefore exists or functions as a sex determiner only in one sex. The products of the executor gene dsx act downstream of these switches. There are female- and male-specific isoforms of DSX, which control sexual development in most insect species [31]. However, in the silkworm, the ortholog of tra does not exist, and no dsx cis-regulatory element binding sites are found, which indicates the sex determination pathways are markedly different from those seen in flies [20,32].

In the lepidopteran model species B. mori, the primary signal is the PIWI-interacting RNA (piRNA) Fem, which is derived from the W chromosome [24]. Fem silences a gene unique to Lepidoptera, Masc, which is essential for both male determination and repression of the vital process of dosage compensation of Z-linked genes during embryogenesis (Figure 1) [21,26,33,34,35]. In recent years, sex-determined factors encoded by PSI, Znf-2, Siwi and Gtsf1 have been identified along with preliminary established genetic cascade of the sex determination in the silkworm [25,27,28,33,36].

Research on sex determination factors in the silkworm began in 1933 when Hasimoto hypothesized that the F-factor originates from the W chromosome [39]. Subsequent genetic studies showed that one copy of the W chromosome is sufficient to determine femaleness [40]. However, in part due to the various transposable elements, their remnants, and simple repeats on W chromosome, the identity of the F-factor was a mystery [41,42]. Only recently was the F-factor identified when researchers used new generation sequencing technologies, bioinformatics and focused on non-coding RNAs (such as microRNAs, long non-coding RNAs and piRNAs). The F-factor is a single piRNA derived from a piRNA precursor named Fem, which is transcribed from the W chromosome. Fem functions as the primary determinant of female sex in the silkworm [24,43,44,45,46,47].

piRNAs are a class of small RNAs of 24–31 nucleotides in size. They are produced from transposons and from discrete genomic loci called piRNA clusters. The piRNAs guide PIWI proteins to target transcripts [48,49]. In flies, piRNAs and PIWI proteins mainly function in germ cells during gametogenesis to suppress transposable elements, which are selfish genomic elements that are able to jump around the genome [48,49]. It is fascinating that the Fem-derived piRNA participates in sex determination, which is a somatic cell fate event in the silkworm. The piRNA pathway is not well characterized in vivo, and as a result how Fem functions at the molecular level is still a puzzle [50]. piRNAs have no enzyme activity, and instead they assemble into piRNA-induced silencing complexes (piRISCs) with the PIWI proteins such as Siwi and Argonaute RISC Catalytic Component 3 (Ago3) identified in the silkworm [43]. After loading onto PIWI proteins, piRNAs are produced by a unique biogenesis pathway called the “ping-pong” cycle. The ping-pong cycle is a posttranscriptional gene-silencing mechanism in which RNAs degraded by piRNA-guided transcript silencing provide substrates for additional piRNA production [51]. We have discovered that SIWI is crucial for feminizing the silkworm, whereas Ago3 mutants display no phenotype involved in sex determination [25]. Our studies suggest that SIWI is dominant during Masc mRNA silencing via Fem-piRNA, whereas Ago3 have minor effects on Fem piRNA processing.

The demonstration that piRNAs have a function in sex determination in the silkworm prompted us try to understand the intricate biogenesis of PIWI-interacting RNAs. The sex determination cascades, as well as piRNA pathways, are distinct in D. melanogaster and B. mori [50]. The piRNA biogenesis pathway consists of a list of components which are specialized for their processing [52]. However, several key elements of the piRNA pathway in the fly, such as Yb, Rhino, Deadlock and Cutoff, which are crucial for piRNA transcription initiation and piRNA processing, are absent in the silkworm [52]. We have reported that a conserved component of the piRNA pathway called Gtsf1 is involved in female sex determination in the silkworm [28]. In Drosophila, Gtsf1 is required for female fertility and interacts with PIWI via its C-terminal end; and it is also essential for piRISCs-induced transposon silencing but not for piRNA biogenesis [53,54]. In the silkworm Gtsf1 is not only necessary for transposon silencing but also for piRNA biogenesis [28]. In addition, our co-immunoprecipitation experiments suggested that Gtsf1 interacts with SIWI. Such an interaction was also observed in Drosophila [28,53,54]. Interestingly, not every component from the piRNA pathway participates in feminization. For instance, Maelstrom (Mael) is essential for spermatogenesis and oocyte development in Drosophila as it is involved in piRNA-mediated silencing of transposable elements [55]. However, depletion of Mael in the silkworm leads to spermatogenesis defects but does not affect sexual development [56]. The poly(A)-specific ribonuclease-like domain-containing (Pnldc1) is necessary for piRNA maturation in silkworms, and mutation of Pnldc1 leads to abnormalities in nuclei of cells in eupyrene sperm bundles but not cells of other organs [57,58]. We have generated transgenic lines using CRISPR-Cas9 technology with mutations in Zucchini and Papi, which encode enzymes required for 3’-end processing of piRNAs, in Tdrd12 and Tudor-SN, which are involved in ping-pong cycling, and in Yu, Armi and Mino, which encode chaperonins that function during piRNAs biogenesis, but none of these lines exhibit any obvious phenotypes (unpublished data). However, all these genes are crucial for piRNA processing in Drosophila, and their mutation will cause sterility, whereas our results indicate that the piRNA processing pathway is also different between silkworms and flies, and the processing of the primary signal Fem-piRNA is independent of those elements [52]. Furthermore, we do not understand how Fem transcription is activated, and it is possible that piRNAs other than Fem or non-coding RNAs from W are involved in sex determination. If other non-coding RNAs from W are involved in the female sex determination of B. mori, their role is expected to be minor if compared with Fem. Indeed, Fem repression is sufficient to cause masculinization of dsx splicing in embryos.

Masc is the target gene of the primary signal mediated by Fem [24]. Evolutionary analysis revealed that Masc is conserved among the species in Lepidoptera but is not found in other insects, indicating that the sex determination pathway in Lepidoptera is likely distinct [18]. In the silkworm, Masc is located on the Z chromosome [24]. It encodes a CCCH-tandem zinc-finger protein and controls both masculinization and dosage compensation [59]. Recent studies have demonstrated that Masc is also required for masculinization in multiple lepidopterans including Trilocha varians, Ostrinia furnacalis and Agrotis ipsilon [21,34,35]. Besides, it is interesting that in O. furnacalis, Wolbachia-induced male-specific lethality is also caused by a failure of dosage compensation via suppression of Masc [60]. However, it remains to be determined whether Masc functions as masculinizer among all the species in the Lepidoptera order. A novel splice variant of Masc (Masc-S), which lacks the intact sequence of the cleavage site for Fem-piRNA, encodes a C-terminal truncated protein that has been identified in both sexes. The variant of Masc, Masc-S, participates in female genital development in the silkworm [61]. Moreover, there is still a gap between Masc and dsx, and further investigations that identify the direct targets of Masc will be able to clarify its role.

In Drosophila females, female-specific dsx is promoted by Tra and Tra-2 as trans-acting splicing activators [5,29] (Figure 2E). In Bombyx, lacking tra ortholog, the sex-specific splicing regulation of the dsx gene is different (Figure 2F). A splicing factor might have replaced the function as Tra in transformer-less species during evolution [9,62,63]. PSI is a KH-domain RNA-binding splicing factor that regulates tissue-specific alternative splicing of P-element transposon transcripts to restrict transposition activity to germ-line tissues and that influences development and mating behavior in flies [64,65,66,67,68,69]. Though PSI is not involved in the sex determination pathway in Drosophila, it is a key auxiliary factor in silkworm male sex determination. PSI binds to exon 4 of dsx pre-mRNA and facilitates its male-specific splicing [36]. Genetic evidence, by using a transgenic CRISPR/Cas9 system, has shown that the mutation of PSI will cause partial male-to-female defects, whereas it has been demonstrated that it is indispensable for masculinization in the silkworm [33]. It is still unclear that PSI affects only males, however, as it is present in both sexes. Besides, according to published reference genomes, PSI is present in multiple lepidopterans, but whether it is involved in sex determination still remains to be identified [70]. Recently, it was reported that a Znf-2 mutant has a phenotype similar to that of the PSI loss-of-function mutant [27]. Znf-2 encodes a CCCH-type zinc finger protein and contributes to male-specific splicing of dsx in silkworm cell lines [71]. The Znf-2 mutant males have feminized external genitalia, and the female isoform of dsx is detected in males, indicating that Znf-2 is essential for male sexual differentiation [27]. In addition, larval development is delayed and body size diminished in PSI and Znf-2 mutant males, demonstrating that these factors have functions in development in the silkworm.

There is strong evidence supporting that Masc, PSI and Znf-2 are indispensable for masculinization, but where and how exactly these factors function in the sex determination cascade is not known. These questions may be answered by dissection of the functional relationships among Masc, PSI and Znf-2, and by identification of additional players in the sex determination pathway. Recent studies have identified components as BxRBP1, BxRBP2 and BxRBP3 (Bombyx mori dsx RNA-binding proteins) are dsx pre-mRNA binding protein by using the strategy of yeast three-hybrid screening. It is interesting that the alternatively transcribed BxRBP3 isoforms are regulated by Masc and physically interact with PSI, but whether BxRBP3 is the transducer remains to be verified via genetic methods [72] (Figure 2D,F). However, based on the observation that females can be completely reversed to males when the signal transducer tra is knocked-down or knocked-out in Diptera, we speculate that the transducer in the silkworm either has not yet been identified or may act very differently in the sex determination cascade [19,73,74,75,76].

DSX protein sex-specific isoforms are conserved in insects and act as the downstream regulator in the sex determination pathway [31]. DSX controls sexual dimorphic development in insects, and mutation of sex-specific isoforms will cause male or female reproductive defects, respectively; hence, it has been targeted in development of sterile insect biotechnology [77,78,79,80]. As in Diptera, also in Lepidoptera, the dsx pre-mRNA is alternatively spliced to encode transcription factors that contain a common N-terminal domain and a sex-specific C-terminal domain [81]. The N-terminal domain possesses a DNA-binding motif that mediates the binding of DSX to target genes, whereas the C-terminal domain is crucial for DSX regulation of sexual development [37,82]. Disruptions of functions of certain isoforms of DSX lead to either sex-specific sexual-dimorphic defects or intersexual phenotypes in multiple Lepidoptera insects including B. mori, A. ipsilon, Plutella xylostella and H. cunea [38,83,84,85].

In Drosophila, the transcription cofactor Intersex (Ix) functions together with the female-specific product of dsx, DSXF, to implement female sexual differentiation [86] (Figure 2E). However, depletion of ix in the silkworm does not affect gonad development or splicing of the dsx pre-mRNA in either sex, suggesting that ix functions differently in silkworms than in flies [87]. The transcription factor Fruitless (Fru) plays roles in sexual behaviors and is regulated by the sex determination cascade in flies [7]. A recent study in the silkworm demonstrated that Fru is also affected by the upstream sex-determining factors and acts together with DSX to regulate both mating and courtship behavior [88].

Although DSX exists in the majority of insect species, the mechanisms by which it functions in sex determination and subsequent developmental processes are poorly understood in most insect species. Only a few genes directly controlled by DSX have been identified, although numerous targets of DSX have been predicted and analyzed on the genome-wide scale in Drosophila [89,90]. It remains to be determined whether the events downstream of DSX were conserved among insects during evolution [91].