Perpetuation by sexual reproduction is the rule within the animal Kingdom. A plethora of sex determination mechanisms exist which commit the embryo to following either the male or female developmental pathway. The mechanism underlying this process has been thoroughly analysed in Drosophila melanogaster. In this species, sex determination is under the control of the gene Sex lethal (Sxl). The epistatic relationships between Sxl and the other sex determination genes transformer (tra), transformer-2 (tra-2) and doublesex (dsx) have revealed a hierarchical interaction to exist among them. Their characterisation has shown that the sex-specific splicing of their primary transcripts controls their expression during development, the product of one gene controlling the sex-specific splicing of the pre-mRNA of the downstream gene in the sex determination cascade (reviewed in [1]).

The gene Sxl, which is at the top of this cascade, acts as the memory device for female sexual development via its auto-regulatory function: the Sxl protein participates in the female-specific splicing of its own pre-mRNA [2], [3]. The downstream target of Sxl is the gene transformer (tra). A transcript found in both males and females encodes a non-functional truncated Tra protein, and a female-specific transcript encodes the functional Tra protein [4], [5], [6], [7], [8]. The Tra product and the product of the constitutive gene transformer-2 (tra-2) [9], [10] control the sex-specific splicing of the pre-mRNA of the gene doublesex (dsx), the last gene in the genetic cascade, and which is transcribed in both sexes [11], [12], [13], [14], [15], [16]. In females, the Tra-Tra2 complex directs the splicing of the dsx pre-mRNA according to the female mode, giving rise to the female DsxF protein that promotes female sexual development. In males, where no functional Tra protein is available, the dsx pre-mRNA follows the default, male mode of splicing, which produces male DsxM protein. This promotes male sexual development.

Sex determination mechanisms have long been of major interest from both developmental and evolutionary points of view. The search in different insects for genes homologous to the sex determination genes of D. melanogaster is underway. The aim is to determine how much of the sex determination genetic cascade has been modified between the more ancient dipteran phylogenetic lineages and the drosophilid lineage. The gene tra of Drosophila simulans, D. erecta, D. hydei and D. virilis has been characterised [17] and an unusually high degree of divergence found with that of D. melanogaster. Even so, D. melanogaster tra mutants can all be rescued by the tra of these species [17].

Outside the genus Drosophila, tra was first characterised in Ceratitis capitata [18] and recently in Bactrocera oleae [19]. As in Drosophila, in Ceratitis and Bactrocera alternative splicing also regulates the expression of tra so that only females contain the full-length protein. The Ceratitis capitata (Cctra) and Bactrocera oleae (Botra) tra genes have male-specific exons, which contain translation stop codons; their inclusion in the mature mRNA produces truncated and most probably non-functional Tra protein. In females, these exons are spliced out, and a mature mRNA is made that produces a functional Tra protein. Surprisingly, putative Tra-Tra2 binding sites have been found in the male-specific exons of Cctra [18] and Botra [19]. The injection of the respective tra-dsRNA in Ceratitis [20] and in Bactrocera [19] results in the destruction of endogenous tra function in both species and the subsequent male-specific splicing of the endogenous tra pre-mRNA, ensuing in the complete transformation of females into fertile XX pseudomales.

The early results in Ceratitis led to Pane et al. [18] to propose a novel auto-regulatory function for the Cctra gene with respect to its Drosophila homologue. In C. capitata, the gene tra would play the key regulatory role, acting as the memory device for sex determination through its auto-regulatory function. The Tra protein would act as a splicing inhibitor of its own pre-mRNA splicing. This was based on the idea that the Tra-Tra2 complex binds to the Tra-Tra2 binding sites in the tra pre-mRNA−because of the sequence conservation of these sites−inhibiting the incorporation of the male-specific exons into the mature tra mRNA. This would also apply to the gene tra of B. oleae [19].

To better analyse the evolution of tra gene, in the present work its characterisation was undertaken in tephritid species belonging to the less extensively analysed genus Anastrepha. We chose those species which, according to morphological [21] and molecular data [22], belong to distinct intrageneric taxonomic groups. The present analysis therefore included Anastrepha obliqua, A. ludens, A. amita and A. sororcula, plus the four closely related species of the so-called Anastrepha fraterculus complex−A. sp.1 aff. fraterculus, A. sp.2 aff. fraterculus, A. sp.3 aff. fraterculus and A. sp.4 aff. fraterculus [23], [24], all of which belong to the fraterculus group [21]–along with A. serpentina, A. striata and A. bistrigata (of the serpentina group, see 25]) and A. grandis (of the grandis group).

The gene tra in the reference species A. obliqua was first isolated and its molecular organisation, expression pattern and encoded product studied. The tra ORFs in the other Anastrepha species were then identified, and a comparative analysis of all the known insect Tra proteins undertaken. In this way, a comparison of the Tra protein at different phylogenetic levels was made. Within-genus comparisons of the members of Drosophila and Anastrepha; distinct genera of the same family, such as Ceratitis, Bactrocera and Anastrepha (Tephritidae); and two different families of the same order (Diptera), such as Drosophilidae and Tephritidae. Next, the genomic tra region that controls the sex-specific splicing of its primary transcript was characterised and compared in C. capitata, B. oleae and the above Anastrepha species. This should indicate whether the molecular organisation of gene tra in Ceratitis and in Bactrocera arose before or after the splitting off of the frugivorous Tephritidae lineage. Finally, the phylogeny of gene tra in these different insects was investigated.

Figures 1A and B shows the molecular organisation of the Aotra gene. The transcription unit is made up of 7665 bp, and the transcription start site located at –197 bp from the initial ATG of the ORF. It is composed of 4 exons (1–4) common to both sexes, and three male-specific exons, ms1, ms2 and ms3, located between exons 1 and 2. The Aotra gene produces three mRNAs in females formed by exons 1–4. These differ in the length of their 3′UTR, yet they encode the same female Tra protein. In males, 5 distinct mRNAs (M1–M5) are produced depending on the male-specific exons included (see Fig. 1B). These latter exons contain translation stop codons. The comparison of male and female Aotra mRNAs indicates that they arise by sex-specific splicing of the tra pre-mRNA, with the male-specific exons being skipped in the female mRNA.

Figure 1 Comparison of the molecular organisation of the gene tra of C. capitata, B. oleae and A. obliqua (A) and the transcripts encoded by the A. obliqua tra gene (B).

Figure 1A compares the molecular organisation of Aotra, Cctra and Botra. In all three cases the female mRNA contains 4 exons. Exons 2–4 are homologous across all three species, their sizes differing only slightly. Exon 1 of Aotra corresponds to exon 1 of Cctra and to exons 1A and 1B of Botra. In Cctra and Aotra, exons 1 and ms1 are contiguous. However, in Botra, exon 1 is split and exon 1A precedes and is contiguous with ms1. Both are separated by an intron from the downstream exon 1B with which exon ms2 is contiguous. This latter exon is flanked by introns in Aotra and Cctra. Both Botra and Aotra have an additional male-specific exon, ms3.

The conceptual translation of the female Aotra mRNA shows it to encode a polypeptide of 417 amino acids. Male Aotra mRNA, however, encodes a truncated, presumably non-functional polypeptide of either 55 or 67 amino acids depending on the male-specific splicing pathway followed (see Fig. 1B). The female Tra protein contains the SR dipeptides that characterise the family of SR proteins.

In D. melanogaster [26]], in other drosophilids [17], in C. capitata [18] and in B. oleae [19], the gene tra is closely linked to the well-conserved gene l(3)73Ah. In the three named species, these two genes are transcribed in opposite directions, and in D. melanogaster and C. capitata the genes tra and l(3)73Ah overlap at their 3′-UTR regions. In B. oleae such overlapping seems not to be present and the stop codons of both genes are 830 bp from each other. In A. obliqua the genes tra and l(3)73Ah do not overlap either, and are separated by about 3.5 kb (data not shown).

Figure 2 Expression of A. obliqua tra.

The gene tra is transcribed in both sexes to produce two different spliced mRNAs−one in each sex−during development and adult life. In females, the mRNA encoding the full-length Tra protein is produced, whereas in males mRNA encoding a truncated, non-functional Tra protein is made. Importantly, tra is also expressed in the ovaries where it produces female mRNA. This indicates that, as in C. capitata and B. oleae, the mother provides the zygote with female tra mRNA and/or female Tra protein.

The Tra protein of the 12 Anastrepha species is composed of 417 amino acids, except that of A. grandis, which contains 416 amino acids (Fig. S1 in Supporting material). Their degree of similarity (i.e., identical plus conservative amino acids) ranges from 88 to 99% (upper half of Table 1). The Tra protein of the tephritids is larger than that of the drosophilids due to its bigger amino terminal end. This is composed of about 103 amino acids in the Anastrepha species and of 105 amino acids in Ceratitis and Bactrocera. The comparison of the specific amino-terminal region in all Anastrepha species revealed an extraordinary high degree of similarity (between 89–100%) (lower half of Table 1).

Table 1 Percentage of similarity among the Anastrepha Tra proteins.

Table 2 and Fig. 3 compare the Tra protein of the tephritids C. capitata, B. oleae and A. obliqua (the reference for the genus Anastrepha). They differed slightly in the number of amino acids and the degree of similarity ranged between 54 and 56% (upper half of Table 2). A similar degree of conservation was seen when the specific amino-terminal region of the tephritid Tra protein were compared among the tephritid species (lower half of Table 2). In Fig. 3, the most conserved regions are shaded. Notice the two large domains in the amino-terminal end. Finally, the Tra proteins of the three tephritids Anastrepha, Ceratitis and Bactrocera contain SR dipeptides like the Tra proteins of the drosophilids, a feature that shows the SR proteins to be involved in splicing regulation. These SR dipeptides are also found in the specific amino terminal region of the three tephritids.

Table 2 Percentage of similarity among the tephritid Tra proteins.

Figure 3 Comparison of the predicted Tra polypeptides of C. capitata [18], B. oleae [19] and A. obliqua (taken as the reference for the twelve Anastrepha species here studied).

The auto-regulation model proposed by Pane et al. [18] for C. capitata tra consider that the Tra protein acts as a splicing inhibitor of its own pre-mRNA splicing (see Introduction). A similar model seems to be applicable to the other tephritids Bactrocera [19] and Anastrepha [this work] tra genes, since the molecular organisation of tra pre-mRNA of these tephritids were similar to that of Ceratitis. Nothing is presently known about the mechanism through which the tephritid Tra protein controls the splicing of its own transcript. We undertook a comparison of the tra genomic region (encompassing the male-specific exons and their flanking introns, where the regulation of sex-specific splicing occurs) of the twelve Anastrepha species studied here, of C. capitata, and of B. oleae. Given the possible involvement of Tra2 in the auto-regulation of tra, we also looked for the presence and location of putative Tra-Tra2 and RBP1 binding sites, as well as Tra2-ISS binding sites [30], in this tra genomic region. The tra genomic region corresponding to the Anastrepha species was amplified by PCR processing of genomic DNA using the primer pair TRA39 and TRA41 (see Materials and Methods and Table S1 in Supporting material).

In the Anastrepha species, the size of intron is1 varied between 1024 and 3296 bp, and that of is2 between 337 and 472 bp. Intron is3 was composed of 85 bp in all twelve species (Fig. S2 in Supporting Material). Six putative Tra-Tra2 binding sites were found in all the Anastrepha species except for A. striata and A. bistrigata, which had five sites (Fig. 4). These elements were located at similar positions: two sites in exon ms1, one site in intron is1 (missing in A. striata and A. bistrigata), and three sites in exon ms3. Between 5 and 8 putative RBP1 binding sites were found, all located in intron is1. All the RBP1 binding sites were of type B but for two of type A (a and b in Fig. 4). Between one and six Tra2-ISS binding sites were found in intron is1 in all species (Fig. 5). The two CAAGG and CAAGA Tra2-ISS binding sequences reported in the Drosophila tra-2 gene [30] were also found.

Figure 4 Comparison of the molecular organisation of the tra genomic region (encompassing the male-specific exons and their flanking introns) involved in sex-specific splicing regulation of the tra pre-mRNA, in the twelve Anastrepha species, in C. capitata (unpublished (more ...)

Figure 5 (A) Phylogenetic tree reconstructed from 22 protein TRA sequences belonging to different insect species (taxonomic groups indicated on the right hand side of the tree).

In C. capitata, eight repeats of Tra-Tra2 binding sites have previously been described [18]. In the present work, five type B RBP1 binding sites were recorded: one in exon ms1 and in exon ms2, one in intron is1, and three in intron is2 (Fig. 4). Finally, two types of Tra2-ISS binding sites were located, one in intron is1, and one in exon ms2 (Fig. 4).

In B. oleae, eight repeats of Tra-Tra2 binding sites have previously been reported [19]. Five type B RBP1 binding sites were seen in this work: two in exon ms2, another two in intron is2, and one in intron is3 (Fig. 4). Finally, two Tra2-ISS binding sites (both CAAGG-type) were identified; one in exon ms2 and the other in intron is2 (Fig. 4).

The comparison of the respective consensus sequences for all these binding sites in the Anastrepha species, C. capitata and B. oleae, revealed a high degree of similarity among the tephritids and Drosophila (Fig. 5). The complete sequences of all these binding sites are given in the Supporting Material (Fig. S3). Another shared feature is the mixing and grouping of these binding sites.

Table 3 provides information showing the variation of the tra gene. Within the pool of three tephritid genera (Anastrepha, Ceratitis and Bactrocera species), the degree of amino acid variation is significantly smaller than that seen within the genus Drosophila. When variation of the Tra protein within the genus Anastrepha was evaluated, a still more reduced degree of Tra protein variation was found. The observed levels of protein variation agreed well with those observed at the nucleotide level, finding high levels of silent variation.

Table 3 Average number of amino acid and nucleotide differences per site among tra genes in different species, plus standard errors, calculated using the bootstrap method (1000 replicates).

The Tra protein belongs to the family of SR-proteins. These are characterised by having serine-arginine (SR) dipeptides, which are involved in protein–protein interactions (reviewed in [32]). The percentage of SR dipeptides in the different species was determined, and means of 7.072% for the Anastrepha species and 5.644% for the other tephritid species C. capitata and B. oleae recorded. These values are low compared to the estimated 16.5% in Drosophila species [17]. Figure 5B shows the proportion of nucleotide sites at which two sequences being compared are different (p) against the ratio between the numbers of non-synonymous and synonymous nucleotide substitutions per site (pN/pS). The results show a high degree of coincidence between peaks of p and those of pN/pS, matching the gene regions encoding the SR domains, indicating a higher degree of nucleotide diversity in these particular areas.

These results support the model of Wilkins [34], who proposed that the evolution of sex-determining cascades was bottom up, with the genes at the bottom being more conserved than those further upstream genes (for a theoretical analysis of this model see [35]). Indeed, the tra/tra2>dsx elements at the bottom of the cascade, and their relationships, have been found conserved in all the dipterans analysed so far. This suggests that they represent the ancestral state (which still exists in the Tephritidae and Muscidae lineages) with respect to the extant cascade found in the more evolved Drosophilidae lineage (in which tra is just another component of the sex determination gene cascade regulated by Sxl). Thus, in the phylogenetic lineage that gave rise to the drosophilids, evolution co-opted for the Sxl gene, modified it, and converted it into the key gene controlling sex determination, thus substituting for the loss of tra auto-regulation.

This prompted us to look for Tra2 ISS and RBP1 binding sites in the tra genomic region, which controls the sex-specific splicing of its primary transcript in C. capitata, B. oleae and the Anastrepha species. In addition to the previously described putative Tra-Tra2 binding sequences, putative Tra2-ISS and RBP1binding sites were found–an important discovery. These sequences are highly conserved in the tephritids and in Drosophila. Moreover, RBP1−but not Tra2-ISS−binding sites was found in the region of Anastrepha, Ceratitis and Bactrocera dsx pre-mRNA involved in sex-specific splicing regulation (data not shown). It is suggested here that the Tra2-ISS binding sites provide a discriminative feature for the tra and dsx pre-mRNAs regions involved in sex-specific splicing regulation.

Questions naturally arise regarding the molecular basis underlying the putative dual splicing role of the tephritid Tra protein. The presence and molecular organisation of Tra-Tra2 and RBP1 binding sites in the dsx gene of tephritid flies−which is similar to that seen in Drosophila−suggests that Tra, Tra2 and RBP1 bind co-operatively to form a splicing activator complex at the female-specific exon of dsx pre-mRNA. This would allow the exon to be incorporated into the mature mRNA. The tephritid tra pre-mRNA also contains Tra-Tra2 and RBP1 binding sites so that the Tra-Tra2-RBP1 complex can bind the male-specific exons−but in this case it would prevent the incorporation of these exons into the mature tra mRNA. It is proposed here that this is so because the presence of Tra2-ISS binding sites in the tra pre-mRNA determines the additional binding of Tra2 to these sequences, and that this overrides the activator effect of the Tra-Tra2-RBP1 complex.

As mentioned above, the interaction of Tra2 with the ISS sites in Drosophila tra-2 pre-mRNA requires an as yet unknown factor. Thus, it is possible that such a factor is also required for the interaction of Tra-2 at the ISS sites in tephritid tra pre-mRNA, and that the complex formed by Tra2 and the proposed factor blocks, either by itself or by recruiting another SR protein, the splicing activation effect of the Tra-Tra2-RBP1 complex. While in Drosophila the Tra protein cannot be this unknown factor, it cannot be ruled out that this is not the case in the tephritids. The tephritid Tra protein has an amino terminal region not found in the Drosophila Tra protein. This region is very strongly conserved among the tephritid Ceratitis, Bactrocera and Anastrepha species. In addition it contains SR-dipeptides involved in protein-protein interactions. Thus, the possibility exists that the tephritid Tra protein has a double interaction with the Tra2 protein−one mediated through the region of the protein homologous in the tephritids and drosophilids, and a second through the specific amino terminal region associated with the binding of Tra2 to ISS sites. This second interaction would convert the Tra-Tra2-RBP1 splicing activator complex into a Tra2-Tra-Tra2-RBP1 splicing inhibitor complex. Alternatively, the double interaction of Tra and Tra2 may recruit an SR protein, which would inhibit splicing. In this scenario, the amino terminal region of the tephritid Tra protein would be essential for this protein to exert its auto-regulation. Although both tra and dsx are transcribed within the same cell, and therefore their primary transcripts are exposed to the same splicing machinery, the splicing inhibitor complex (howsoever composed) would only be formed at tra pre-mRNA since only this contains Tra2-ISS sites (see a simplified diagram in Fig. 6).

Figure 6 Proposed explanation of the dual role-played by the tephritid Tra protein in sex determination.

Under this scenario, it is hypothesised here that in the phylogenetic lineage that gave rise to the drosophilids, the tra gene lost the Tra2-ISS binding sites and the Tra protein lost the amino terminal region that characterises the tephritid Tra protein; hence, its female-specific auto-regulatory function disappeared. This protein was, however, still feasible in the drosophilids because the tra gene gained Sxl-binding sequences so that the female-specific splicing regulation of the tra pre-mRNA came under the control of Sxl, which is only present in females.

Particularly interesting is the comparison of tra variation between the Drosophila and Anastrepha species. McAllister and McVean [36] reported high rates of neutral evolution when comparing D. americana and D. virilis (separated 60 Myr ago), while Kulathinal et al. [37] reported a high level of divergence among the sibling species of the Melanogaster complex D. melanogaster, D. simulans, D. mauritiana and D. sechellia (separated 2.5 Myr ago). In this work, the tra gene of the twelve Anastrepha species showed extraordinarily reduced variation, and no insertions were detected, unlike in some Drosophila species. Indeed, silent variation appears to be significantly more common than non-silent variation when considering the complete coding region of tra, suggesting that this gene is subject to strong purifying selection in order to preserve the mechanism of action of Tra proteins.

Although the SR dipeptide content of the tephritids and the drosophilids is also different, the distribution of these regions over the Tra protein is very much conserved within both groups. This supports the proposal that during the evolution the Tra proteins these maintained enough SR dipeptides regions to bestow functionality on these proteins [32]. The Tra proteins seem to lack an RNA binding domain so that its role in splicing regulation is exerted at the level of its interaction (through their SR domains) with other proteins carrying RNA-binding domains, such as Transformer-2 (reviewed in [32]).

To determine the molecular organisation of Aotra the following strategy was followed. Firstly, 3′- and 5′-RACE analyses were performed. To this end, specific primers from the amplified sequenced were synthesised: AZ1+ and AZ2+ for 3′-RACE, and B41- and B42- for 5′-RACE. These primers were used in nested PCR reactions, the products of which were cloned and sequenced. The 5′-RACE generated three overlapping fragments of about 300, 450 and 500 bp. The largest one contained the start ATG codon of the tra ORF. Thus, exon 1 and the 5′UTR were identified. The 3′-RACE, which produced three overlapping fragments of about 243, 288 and 528 bp, allowed the identification of the end of exon 4 and the 3′UTR region containing the three poly-A(+) signals corresponding to the three amplified fragments. Consequently, the gene Aotra encodes three female mRNAs of 1894, 1654 and 1603 bp that differ in the length of the 3′-UTR depending on the poly-A(+) signal used. Next, RACE overlapping PCR was performed on cDNA synthesised from total RNA of adult males. The amplified fragments were cloned and their sequences compared to that of the female cDNA. Five different isoforms of male mRNA of 2326 (M1), 2285 (M2), 2108 (M3), 2412 (M4), and 2145 (M5) bp were found, depending on the male-specific exons included (see Fig. 1B).

Secondly, the GenomeWalker kit was used to perform PCR on genomic DNA of A. obliqua in order to determine the exon/intron junctions via genomic walking. The sequences of the genomic fragments thus generated were compared with the A. obliqua male and female cDNA sequences previously determined. In this way, the exon/intron junctions were unambiguously identified.

For identification of the Tra protein of other Anastrepha species, RT-PCR analyses of total RNA from female adults were performed. Reverse transcription was performed with the oligo-dT primer. PCR amplification of the cDNA was undertaken using the pair of primers TRA23 (Table S1) plus Ao25 (Fig. 2) corresponding to sequences of the Aotra gene. The first primer represents part of the exon 1 sequence, while the second represents part of the exon 4 sequence – respectively before and after the start and stop codons of the ORF. Thus, the amplicon expands the whole ORF. All amplicons were cloned in the TOPO-TA vector and subsequently sequenced.