Homology-based search of sex-determining genes from Athalia rosae genome
Genetic analysis demonstrates that sexual fate in A. rosae is determined by the single-locus CSD system (Naito and Suzuki 1991). To investigate whether the molecular basis for the sex-determining mechanism of this insect is the same as those reported in honeybee species, a tblastn search of the NCBI database was performed, specifying an A. rosae dataset containing 22,130 of gene models, using the full amino acid sequence of CSD, FEM, and DSX in Apis mellifera as a query sequence. As a result, we were not able to find any sequences with significant homology to CSD and FEM. On the other hand, the tblastn search retrieved one predicted gene (NCBI accession number XM_012406840.1) with significant higher similarity (hit score >200, E value <e−60), when either female (AmDSXF1) or male-specific isoform (AmDSXM) of Apis mellifera DSX served as a query sequence (Fig. 1c, d). Comparison of the predicted protein encoded by the retrieved gene with proteins in the NCBI database using the blastp program indicated that XM_012406840.1 encodes a protein showing high homology to the female-specific DSX isoform (AmDSXF2) of A. mellifera with 54% identity and 69% similarity and also to the female-specific DSX isoform (TcDSXF3) of Tribolium castaneum with 49% identity and 59% similarity. Prediction of conserved protein domains using Pfam demonstrated that the protein encoded by XM_012406840.1 has a DM domain characteristic of the DM superfamily genes containing dsx, mab3, and Dmrt and the oligomerization domain that is characteristic for insect DSX. Taken together, we concluded that the predicted gene (XM_012406840.1) is an A. rosae ortholog of dsx, and hereafter we label this gene “Ardsx”.
Molecular cloning of full-length Ardsx cDNAs from males and females
To examine whether the aforementioned genes retrieved by the tblastn searches are transcribed in vivo, we first performed RT-PCR analysis with primers that were designed on the basis of the nucleotide sequence of the predicted gene using cDNAs prepared from male and female pupae. The RT-PCR with primers dsx1 and dsx2 (amplicon size 1025 bp) amplified a DNA fragment whose size was almost similar to the predicted sizes in females (Fig. 2a, lane 2). On the other hand, in males, the same PCR resulted in an amplified product of approximately 1200 bp, which was larger than the expected size (Fig. 2a, lane 1). Similar results were obtained when the RT-PCR was performed with primers dsx1 and dsx7, which were designed to amplify a 1018-bp cDNA fragment (Fig. 2a, lanes 3 and 4). These RT-PCR products were cloned and sequenced. The sequences of the DNA fragments amplified from females were identical to that of XM_012406840.1, while the PCR products obtained from males were found to contain an insertion of 119 bp sequence (Fig. 2b). A blastn search of the A. rosae genome sequence (Version GCA_000344095.1 Aros_1.0) revealed that scaffold 11 contained the region encoding the whole sequence of the aforementioned cDNAs, which spanned at least 15 kb. Comparison between the cDNAs and the genomic DNA sequence demonstrated that the 119-bp fragment was derived from a single intronic sequence. Thus we concluded that inclusion of this intron in the pre-mRNA processing yields male-specific products. This strongly suggests that Ardsx in the sawfly may be sex-differentially spliced like dsx orthologs identified in other insects.
Next, we carried out RACE using total RNA samples purified from whole bodies of female and male pupae to determine the full-length coding and 5′ and 3′ sequences of the Ardsx gene. The 5′ ends of the Ardsx cDNAs determined by RACE were consistent with the 5′ end of the predicted gene XM_012406833.1. In addition, our 5′RACE amplified another two cDNA fragments, which respectively started at 698 and 380 nt upstream of the 5′ end of the predicted gene XM_012406833.1. Because no open reading frame of significant length was identified in these new fragments, we regarded them as 5′ UTRs. The 3′ RACE resulted in a single amplified product with a nucleotide sequence identical to the XM_012406840.1. The nucleotide sequences obtained by the aforementioned RACE showed no difference between males and females. Sex-specific difference was restricted to either presence or absence of the 119-bp sequence described in Fig. 2b. Consequently, Ardsx appeared to yield three male-specific variants (Ardsx
M1, Ardsx
M2, and Ardsx
M3) and three female-specific variants (Ardsx
F1, Ardsx
F2, and Ardsx
F3) (Fig. 3a). The difference among these variants rests entirely in the 5′ UTR, owing to alternative transcription start sites. Comparison between the full-length Ardsx cDNAs and the genomic sequence revealed that exon 5 was differentially spliced between male and female-specific isoforms (Fig. 3a). In the female isoforms, the male-specific 119-bp sequence, which contains a stop codon, was spliced out, causing amino acid sequence difference in the C-terminal region between male and female ArDSX isoforms.
Ardsx
M1, Ardsx
M2, and Ardsx
M3 encode the same protein (ArDSXM) of 233 amino acids. Ardsx
F1, Ardsx
F2, and Ardsx
F3 also encode the protein with an identical sequence of 337 amino acids (ArDSXF). ArDSXM and ArDSXF share an N-terminal region of 222 amino acids that encodes a conserved DNA-binding domain called DM domain (or OD1 domain) and part of a conserved dimeriztion domain known as OD2 domain, which is specifically conserved among DSX proteins (Price et al. 2015) (Fig. 4a). Amino acid sequence of the C-terminal part shows difference between ArDSXM and ArDSXF (Fig. 4a).
Phylogenetic analysis using the full amino acid sequence of ArDSXF indicated that the identified sequence from A. rosae was grouped with other hymenopteran DSX proteins (Fig. 4b). These results strongly suggest that Ardsx is a dsx ortholog of A. rosae.
Expression analysis of Ardsx
As shown above, exon 5 of Ardsx was differentially spliced between two sexes, resulting in the male-specific insertion of the 119-bp fragment (Fig. 3a). To investigate whether sexual difference in the expression pattern of Ardsx mRNA changes according to the developmental stages, RT-PCR analysis was performed using cDNAs prepared from males and females at the different stages with primers that specifically anneal to the regions flanking the 119-bp fragment (Fig. 3b).
As a result, the amplification product of 269 bp was observed in males at all the examined stages (Fig. 3b). On the other hand, the same RT-PCR amplified the 150-bp DNA fragment in females throughout the examined stages. In females, the 269-bp DNA fragment was weakly detectable until 4 days after hatching. After that, the DNA band gradually disappeared with a concomitant increase in the amplification level of the 150-bp DNA fragment. After cloning of these amplified products and sequencing the cloned DNA, the amplified product of 269 bp corresponded to the male-specific Ardsx isoforms identified by RACE, while the RT-PCR product of 150 bp had a nucleotide sequence identical to that of the corresponding region in the female-specific Ardsx isoforms. These results clearly demonstrate that Ardsx pre-mRNA is spliced alternatively in a sex-dependent manner. However, in females, sex-specificity of the splicing is less accurate until middle larval stages, yielding the male-specific Ardsx in addition to the female-specific Ardsx isoforms.
Functional analysis of Ardsx
In order to analyze the function of Ardsx, we investigated the effects of Ardsx knockdown on sexual developments. Two different dsRNAs (Fig. 5a, Ardsx1 and Ardsx2), both of which targeted to a region common between Ardsx isoforms, were injected into larvae at the wandering stage. qRT-PCR analysis confirmed a significant reduction in Ardsx mRNA level in males and females injected with Ardsx1 (Fig. 5b). Similar results were observed when males and females were injected with Ardsx2 (data not shown). Morphological analysis of adult phenotypes demonstrated that negative control males and females, which were injected with dsRNA targeting the EGFP gene, had normal genital organs as observed in wild-type adult males and females (Fig. 5c, d, g). On the other hand, Ardsx knockdown males (Ardsx KD males) injected with Ardsx1 had external genital organs whose shapes were very similar to those observed in the control females (Fig. 5c). Normal females have external genitalia that contain an ovipository apparatus consisting of two pairs of valvifers, which give rise to the other parts: a saw formed by two pairs of blades (a ventral pair of blades derived from the first valvifers and a dorsal pair of blades derived from the second valvifers); and a sheath composed of a pair of appressed end segments of the second valvifers (Ross 1945) as represented by Fig. 5g. The genitalia observed in the Ardsx KD males contained several imcomplete parts of the ovipository apparatus including the dorsal pair of blades and the sheath, both of which are derived from the second valvifers (Fig. 5e). Differing from the normal female genitalia, Ardsx KD male genitalia lacked the ventral blades but instead contained an abnormal tissue with a saw tooth-like structure. Morphological analysis of internal reproductive organs revealed that these Ardsx KD males showed abnormal (Fig. 5k). Moreover, the seminal ducts looked thicker than those of the control male. The seminal vesicle filled with mature sperms in the control male (Fig. 5j) was not observed in the Ardsx knockdown males (Fig. 5k). Similar morphological abnormalities were observed in Ardsx KD males injected with Ardsx2 (Fig. 5f, l).
Ardsx knockdown females injected either with Ardsx1 or Ardsx2 were also subjected to the same analyses but their external genitalia showed the same phenotype as those observed in the control females (Fig. 5h, i). All the examined Ardsx KD females had normal ovaries and internal reproductive organs and fertile (data not shown).
These results suggest that expression of Ardsx during the pupal stage is essential for normal sexual development in the male external genitalia and that its expression is also important for testis and seminal vesicle development.