Phylogenetic analysis and embryonic expression of panarthropod Dmrt genes

Background One set of the developmentally important Doublesex and Male-abnormal-3 Related Transcription factors (Dmrt) is subject of intense research, because of their role in sex-determination and sexual differentiation. This likely non-monophyletic group of Dmrt genes is represented by the Drosophila melanogaster gene Doublesex (Dsx), the Caenorhabditis elegans Male-abnormal-3 (Mab-3) gene, and vertebrate Dmrt1 genes. However, other members of the Dmrt family are much less well studied, and in arthropods, including the model organism Drosophila melanogaster, data on these genes are virtually absent with respect to their embryonic expression and function. Results Here we investigate the complete set of Dmrt genes in members of all main groups of Arthropoda and a member of Onychophora, extending our data to Panarthropoda as a whole. We confirm the presence of at least four families of Dmrt genes (including Dsx-like genes) in Panarthropoda and study their expression profiles during embryogenesis. Our work shows that the expression patterns of Dmrt11E, Dmrt93B, and Dmrt99B orthologs are highly conserved among panarthropods. Embryonic expression of Dsx-like genes, however, is more derived, likely as a result of neo-functionalization after duplication. Conclusions Our data suggest deep homology of most of the panarthropod Dmrt genes with respect to their function that likely dates back to their last common ancestor. The function of Dsx and Dsx-like genes which are critical for sexual differentiation in animals, however, appears to be much less conserved. Electronic supplementary material The online version of this article (10.1186/s12983-019-0322-0) contains supplementary material, which is available to authorized users.


Background
Dmrt (Doublesex and Male-abnormal-3 Related Transcription factor) genes represent a group of transcription factors that are characterized by the presence of an unusual zinc-finger motif called the DM domain (Doublesex/Male-abnormal-3 domain) ( [1] Erdman and Burtis 1993). The first Dmrt gene to be proposed and identified was Drosophila melanogaster Doublesex (Dsx), a gene that is involved in sex determination in the fly ( [2] Hildreth 1965, [3] Burtis and Baker 1989), but Dmrt genes represent an ancestral class of developmental genes that must have evolved before the appearance of bilaterian animals: they are present in cnidarians, placozoans and ctenophores ( Dmrt genes have been intensively investigated because of their various functions in sex determination and differentiation, and have been identified in various animal groups such as vertebrates (e.g. [ [28] Siehr et al. 2011). However, despite the great amount of research undertaken on Dmrt genes and their function(s) in sex determination and differentiation, data on arthropods are mostly restricted to insects (and a few crustaceans) and the sex-determining factor Doublesex (Dsx). In fact, neither expression nor function of the other Dmrt genes has been investigated in detail even in Drosophila, and research on panarthropod (i.e. arthropods, tardigrades and onychophorans) Dmrt genes (including Dsx) outside Pancrustacea (i.e. crustaceans and insects together) is almost completely lacking. Furthermore, studies on Dsx and other Dmrt genes in Pancrustacea mostly focus on their role in adults or sub-adults, and virtually no data exist on their expression patterns or potential functions during embryogenesis, although there are some studies investigating transcript levels in embryos and embryonic tissues, but without providing any detailed data on transcript location (e.g. [29] Morrow et al. 2014).
We have therefore studied the embryonic expression patterns of the full complement of Dmrt genes in three arthropod species representing Insecta (the red flour beetle Tribolium castaneum); Myriapoda (the pill millipede Glomeris marginata) and Chelicerata (the common house spider Parasteatoda tepidariorum); and the onychophoran Euperipatoides kanangrensis. Our data thus represent the first comprehensive study of embryonic Dmrt gene expression patterns in Panarthropoda as a whole. Our phylogenetic analysis clearly groups panarthropod Dmrt genes into three families, Dmrt11E/Dmrt2/terra, Dmrt93B/ Dmrt3 and Dmrt99B/Dmrt4,5 (Table 1), and identifies possible Dsx orthologs in at least the spider. We find that most of the identified Dmrt genes are often expressed in a tissue-or structure-specific pattern. For orthologs of Dmrt11E, Dmrt93B and Dmrt99B, these patterns are highly conserved in all panarthropod species including Drosophila suggesting ancestral function(s) in panarthropod development that likely dates back to their last common ancestor. In contrast, Doublesex-like genes are either not expressed during embryogenesis, or show lineagespecific expression patterns, likely due to neofunctionalization after duplication. A hallmark of insect Dsx genes is their alternative splicing. We detected splice variants of several Dmrt genes, including Dsx genes, and found that at least some of these are expressed in different embryonic structures.

Animal husbandry and fixation of embryos
Embryos were obtained and fixed for in-situ hybridization experiments as described in [30] Table S1. All investigated gene fragments were cloned into the PCRII vector (Invitrogen) and sequenced on an ABI3730XL automatic sequencer (Macrogen, Seoul, South Korea). Gene identification-numbers are listed in Additional file 2: Table S2. Colorimetric in-situ hybridizations for all investigated species were performed as described in [38] Janssen et al. (2018). For confocal microscopy, embryos were stained with SIGMAFAST Fast Red TR/Naphto-lAS-MX (SIGMA) instead of BM Purple (ROCHE). Cell nuclei were visualized by either incubation of the embryos in 3-5 μg/ml of 4-6-diamidino-2-phenylindole (DAPI) or SYBR Green in phosphate buffered saline with 0.1% Tween-20 (PBST-0.1%).

Phylogenetic analysis
Reciprocal BLAST searches against sequenced embryonic transcriptomes of Glomeris (SRA accession: PRJNA525752) and Euperipatoides (SRA accession: PRJNA525753) (applying tblastn), against published protein sequences from Tribolium and Parasteatoda (applying both blastp and blastx), and the sequenced transcriptome of the priapulid worm Priapulus caudatus (SRX507009) were run with the Drosophila melanogaster sequences of Dsx, Dmrt11E, Dmrt93b and Dmrt99B, and with a Dmrt gene from the Chinese mitten crab Table 1 Alternative names of DMRT genes as used in the fly Drosophila melanogaster, the nematode worm Caenorhabditis elegans, and in verterbrates  Figure  S3 and Additional file 8: Figure S5) or the Dmrtdomains (DM domains) (Additional file 4: Figure S1, Additional file 5: Figure S2 and Additional file 7: Figure  S4) were aligned using

Sequence analysis
We identified three Dmrt genes in Tribolium (cf. [5] Wexler et al. 2014), three in Glomeris, six in Parasteatoda, and four in Euperipatoides ( Fig. 1a/b). A phylogenetic analysis based on the sequence of the DM domain did not resolve very well, especially with respect to Dmrt99B orthologs and Euperipatoides Dsx_like (Additional file 4: Figure S1 and Additional file 5: Figure S2 Figure S3) (note that Pt-Dmrt_like2 was not used in this phylogenetic analysis because the sequence did not align properly due to the presence of two DM domains). Each of the species investigated here possesses one Dmrt93B gene, one Dmrt99B gene, and one Dmrt11E gene (except for Tribolium). We identified one Tribolium Doublesex (Dsx) gene (cf. [47] Shukla and Palli 2012), but two Parasteatoda Dsx genes (Dsx1 and Dsx2). In Euperipatoides, we identified one Dmrt gene that clusters with arthropod Dsx genes; but support for this relationship is relatively low and its structure is significantly different from that of arthropod Dsx genes (Figs. 1a and 2). However, because of its position in the phylogenetic tree we named this gene Euperipatoides Dsx_like. We could not detect a Glomeris Dsx gene, but this may represent an artefact of incomplete transcriptome data, or it could be that Glomeris Dsx is expressed at later developmental stages that are not covered by the sequenced transcriptome.
Glomeris Dmrt11E is expressed in two isoforms that we confirmed by RT-PCR, cloning and sequencing  (Fig. 3b). Parasteatoda Dsx2 is expressed in four isoforms, of which three (isoforms A, C and D) contain the same DM domain, but one, isoform B, contains an alternative DM domain (Fig. 3c). Prediction of these isoforms was based on published genome and transcriptome data. We confirmed all isoforms except for isoform C by RT-PCR, cloning and sequencing. Possibly, isoform C is not represented by the embryonic cDNAs we used for RT-PCR, or is expressed at very low levels.  Whenever the core motif (SAF) of the SCM is derived, this is indicated with the different sequence in purple letters next to the SCM-domain box. Position of primers of cloned and sequenced gene fragments is indicated by red arrows. The region marked by the red arrows was used to synthesize anti-sense mRNA probes for in-situ hybridization experiments. Blue arrows mark the region of Pt-Dmrt_like that we cloned to verify the presence of the RBD-like motif. Red filled circles in front of gene names indicate genes that have been cloned in this study. The black filled circle indicates that we attempted to amplify this gene by means of RT-PCR (position of primers marked by black arrows), but did not succeed. Locations of primers for the different splice variants of Tribolium Dsx, Parasteatoda Dsx2 and Glomeris Dmrt11E are indicated in Fig. 3 Kato et al. 2008). Another conserved domain is the "short conserved motif" Fig. 3 Splice variants of here investigated Dmrt genes. Each coloured box represents one exon. For GmDmrt11E, the position of exon 2 and 3 is arbitrary, since genomic data are not available. The position of stop codons is marked by a red 'X'. The DM domains and the oligomerization domain 2 (OD2a and OD2b) are indicated. Length of exons is given in base pairs. Relative position of primers used to verify splice variants and for subsequent probe synthesis are indicated: 1) long isoform of Gm-Dmrt11E; 2) probe for Gm-Dmrt11E; 3) short isoform of Gm-Dmrt11E; 4) universal probe of Tc-Dsx; 5) female-specific probe of Tc-Dsx; 6) primers used to confirm presence of isoforms A and D (*note that we could not isolate isoform C (as predicted from genomic data) from our cDNA samples); 7) probe for isoforms A/C/D; 8) probe for isoform B; 9) primers used to confirm presence of isoform B. All PCR fragments were cloned and sequenced For all investigated genes, we found at least one DM domain (Fig. 2). The Parasteatoda Dsx2 gene contains two DM domains each of which appear in different isoforms. One of the Parasteatoda orphans (Pt-Dmrt_like2), however, possesses two DM domains that appear in the same transcript and the other orphan, Pt-Dmrt_like, possesses a C-terminal DM domain and an unusual second conserved motif that is similar to the reticulocyte binding domain (RBD) that has only been reported from the parasite Plasmodium (reviewed in e.g. [56] Gunalan et al. 2013). The DMA domain is present in all Dmrt99B and Dmrt93B genes. However, we also found a DMA domain in Priapulus Dmrt11E suggesting that the ancestral Dmrt11E gene may have possessed a DMA domain that was later lost in the lineage leading to Panarthropoda (Fig. 2). The SCM is present in all Dmrt93B genes as well as most of the Dmrt99B genes, for each gene group with its conserved core sequence (SAF) (e.g. [ (Fig. 2). In Dmrt11E genes, however, the SCM is divergent in arthropods, but in the onychophoran and the priapulid, this domain is conserved, with its core sequence (SAF) suggesting that the domain has diverged in the lineage leading to Arthropoda (Fig. 2 (Fig. 2).

Embryonic expression patterns of panarthropod Dmrt genes Dmrt11E
The genome of Tribolium does not contain a Dmrt11E gene (see also [5] Wexler et al. 2014).
In Glomeris, Dmrt11E is expressed in two isoforms. A probe targeting the specific sequence of the longer transcript was used (Fig. 3a). At stage 2, isoform_1 of Dmrt11E is expressed in the anterior mesoderm of the mandibles (Fig. 4B-D, F-H), the mesoderm of the anal valves ( Fig. 4A-D, H), and in the outer lining of the developing hindgut ( Fig. 4A-C, H). At stage 5, expression appears in the mesoderm of the labrum (Fig. 4D/E). Unfortunately, it was not possible to design a specific probe for the shorter isoform, as the specific sequence is too short to use in in-situ hybridization experiments.
Parasteatoda Dmrt11E is first expressed at stage 10.1 in the form of a dot in each of the most anterior walking legs (Fig. 4I). Later, a dotted pattern along the proximaldistal axis appears in the mesoderm of all limbs except for the chelicerae (Fig. 4J-M).
In Euperipatoides, Dmrt11E is first expressed at stage 8 in the form of a weak expression in the jaw, the slime papilla and the first pair of legs (Fig. 4N). As development progresses, more stripes of expression appear successively in differentiating posterior segments ( Fig. 4O/ P). This expression is in the mesoderm of the limbs, as confocal microscopy reveals (Additional file 9: Figure  S6). At later developmental stages, Dmrt11E is also expressed in the ventral lining of the head lobes anterior to the position of the mouth (Fig. 4Q).

Dmrt93B
In all species, Tribolium, Glomeris, Parasteatoda and Euperipatoides, Dmrt93B is exclusively expressed in tissue in and around the developing mouth (Fig. 5).

Dmrt99B
In all species, Dmrt99B is predominantly expressed in the developing brain (Fig. 6). In Tribolium this is first in the form of four domains in the head lobes (two in each hemisphere), that later transform into six domains (or two domains appear de novo) (Fig. 6A/B). In Glomeris, Dmrt99B is expressed as two domains in the ocular region ( Fig. 6C-F). In Parasteatoda, first four domains of expression form in the head lobes (Fig. 6G), that shortly after become six (by splitting of the most posterior of the initial domains) (Fig. 6H-J). In Euperipatoides, two broad domains of expression are detectable in the head lobes ( Fig. 6K-N). In all species, Dmrt99B is also expressed in the mouth at later developmental stages (asterisks in panels 6B, D, F, J, N). In the onychophoran, segmental expression appears at later stages in an anterior to posterior order that is likely associated with the formation of the openings of the nephridia (Fig. 6L-N

Doublesex (Dsx)
From around 12 h after gastrulation onwards, Tribolium Dsx is first expressed in the form of a single domain in the tenth abdominal segment (Additional file 10: Figure  S7A). Later this domain transforms into two dots (Fig. 7A/B and Additional file 10: Figure S7B/C). This pattern is present in all embryos hybridized with a av., anal valve; ch, chelicera; fap, frontal appendage; j, jaw; hl, head lobe; Lx, walking-leg bearing segment number X; lr, labrum; m, mouth; md, mandible; oc, ocular region; pp, pedipalp; sp, slime papilla universal probe detecting all isoforms of Dsx (Fig. 3b). The female-specific probe (targeting all female isoforms) (Figs. 3b and 7B) detected the same signal as the universal probe (Fig. 7A). However, in only approximately 50% of the embryos (20/37) the signal appears fast (within one hour) and equally strong as for the universal probe. In the other embryos, the same signal appears after an elongated staining period of at least 16 h, and the signal is significantly lower as for the general probe (Fig. 7B). We assume that these latter embryos represent males and that at least one of the "female-specific" isoforms of Dsx is expressed at low levels in male embryos as well.
We did not detect any specific signal of Parasteatoda Dsx1 (Pt-Dsx1). The isoforms of Pt-Dsx2, however, are expressed in at least two unique patterns. A probe targeting isoforms A, C and D detects expression in a saltand-pepper like pattern in the dorsal field surrounding the head lobes ( Fig. 7C/D). At later stages, cells in the complete dorsal field express this isoform (Fig. 7E). In late developmental stages, the probe detects expression  Fig. 4; T, trunk segment in a subset of cells on either side of the now closed dorsal midline (Fig. 7F). The Pt-Dsx2 isoform B is exclusively expressed in the developing spinnerets, the silkproducing and processing organs of the spider (Fig. 7G-J). First, expression is in the form of single dots in the fourth and fifth opisthosomal segments (O4 and O5) (Fig. 7G). Later, additional small dot-like domains of expression appear in the morphologically differentiating spinnerets. At stage 12, this additional dot appears lateral to the expression in O5 (Fig. 7H). At stage 13.1, the initial expression in O5 splits into two (Fig. 7I). We did not detect any specific signal of Euperipatoides Dsx_like (Ek-Dsx_like).

The orphan genes PtDmrt_like and PtDmrt_like2
Parasteatoda Dmrt_like is expressed in the dorsal field from approximately stage 10.2 onwards and throughout further development (Fig. 8). At later stages, a metameric pattern is seen within the dorsal field (Fig. 8C/D).
We were not able to amplify PtDmrt_like2 from cDNA synthesized from embryonic RNA.

Gene expression suggests highly-conserved function of Dmrt genes Dmrt11E
In both the onychophoran Euperipatoides and the spider Parasteatoda, Dmrt11E is expressed in the mesoderm of  Figure 7B represents such latter case embryo.) and Parasteatoda (C-F, Dsx2 isoformsA,C,D; G-J, Dsx2 isoformB). Developmental stages are indicated. Panels A´-J´represent DAPI stained embryos as seen in panels A-J. In all panels, anterior is to the left and ventral views, except panels C and E (lateral views) and F (dorsal view). Arrows in panel F point to single cells expressing the gene. Asterisks in panels H-J mark splitting expression in the fourth opisthosomal segment. Dashed lines in panel E/E´mark the dorsal field. Abbreviations as in Fig. 4; A10, tenth abdominal segment; br, brain; O(x), opisthosomal segments. Note that the primordia of spider spinnerets are located on opisthosomal segments four (O4) and five (O5) (panels G-J) most of the appendages except for the frontal appendages of the former and the chelicerae of the latter (Fig. 4). In the myriapod Glomeris, however, Dmrt11E is only expressed in the mesoderm of the labrum and the mandibles, but not the other appendages. This raises the question if the (almost) universal function of Dmrt11E in the development of the mesoderm in the appendages has been lost in Glomeris marginata, myriapods in general, or even all Mandibulata. In Tribolium, this gene has been lost indicating that the function of Dmrt11E in mandibulates may be less critical than in other panarthropods. This is supported by the fact that Dmrt11E appears to be lost in the isopod Armadillidium vulgare as well ( Likewise, there are no data on the embryonic expression of Dmrt11E in any spiralian species. The assumption that Dmrt11E genes play conserved function(s) in mesoderm patterning in Bilateria is thus solely based on data from panarthropods and vertebrates, two distantly related groups of animals.

Dmrt93B
Data on Dmrt93B genes is very limited, and functional investigations have only been undertaken in vertebrates (Dmrt3 gene group). There, the gene is inter alia expressed in the central nervous system (CNS), nasal placodes, Müllerian ducts, forming somites, and the developing gonads (e.g. [ Embryonic gene expression data or functional studies are not available from any protostomian animal, except for the model arthropod Drosophila melanogaster, and even here published data are restricted to the mere presentation of embryonic gene expression data on the BDGP in-situ homepage (Berkeley Drosophila Genome Project (see Additional file 3: Table S3 for the link)). Apart from that, the gene has also been investigated in the crustacean Daphnia magna, but in-situ gene expression data have not been provided ( In all the panarthropods investigated here, Dmrt93B orthologs are expressed in tissue associated with the developing mouth. In Drosophila, expression is found in the frontal ganglion, a structure that is associated with the stomatogastric nervous system (and thus with mouth development) ( [71] Hartenstein 1997, [72] Ayali 2004). It is  Table S3 for the link). Interestingly the hmc is located in the pseudo-coelom on the dorsal side of the pharynx ( [73] Altun and Hall 2009). It is tempting to speculate that expression in the pharynx and stomatogastric nervous system in panarthropods and the expression in the hmc in the nematode could be evolutionary related. This would place the origin of the function of Dmrt93B/dmd-4 at the base of the Ecdysozoa (e.g. . We assume that this Dmrt gene is a likely ortholog of Dmrt93B which could place the origin of its function in mouth/foregut development at the base of the Bilateria. However, additional data from additional groups of animals are needed to corroborate this assumption.

Dmrt99B
Dmrt99B is expressed in the developing brain of all investigated panarthropod species (Fig. 6). Interestingly, the number of domains in which Dmrt99B is expressed in the brain seems to correspond with the number of eyes of the investigated organisms: two in Euperipatoides and Glomeris, and eight in Parasteatoda (and this correlation also appears to be conserved in Drosophila (Berkeley Drosophila Genome Project (see Additional file 3: Table S3 for the link))) . In Tribolium, however, this correlation is absent. As for most beetles, Tribolium lacks ocelli and only possesses one pair of lateral eyes, but Dmrt99B is expressed in two additional domains in the brain (one per hemisphere). However, the ancestor of Tribolium likely possessed ocelli ([75] Leschen and Beutel 2004), and the additional domains of Dmrt99B expression in the head lobes of Tribolium may represent rudimentary structures associated with these ocelli. Interestingly, the bona fide arthropod eye markers glass (gl), Pax6, dachshund (dac), eyes absent (eya) and sine oculis (so) are expressed in similar (gl, Pax6, eya, so) or identical (dac) patterns as We assume that the expression domain marked with roman numeral 1 (Fig. 6G-J) may be connected to the development of the single pair of median eyes, and that the expression domains marked 2, 3a and 3b may be correlated to the formation of the three pairs of lateral eyes (cf. [83] Schomburg et al. 2015). However, a closer look at the correlation of optical systems and the expression of Dmrt99B suggests that its expression may not always be in the exact place where the eyes develop as it is the case in for example the onychophoran (Fig. 6M´). Expression of Dmrt99B may thus indeed rather be correlated with the regions of the brain that process vision than to the optical organs itself. Comparison with eye or brain-marker genes in species of the genus Euperipatoides reveals that the expression of Dmrt99B overlaps with many of these factors such as orthodenticle (otd), scro/nkx2. In the nematode worm Caenorhabditis, the Dmrt99B ortholog dmd-5 is expressed in the nervous system and the intestine (WormBase, see Additional file 3: Table S3 for the link).
Together our data strongly suggest that Dmrt99B orthologs are involved in the development of the eyes (and associated brain structures) in at least arthropods, but that its function in onychophorans is only in patterning brain structures that may be involved in processing visual information.
Another conserved aspect of Dmrt99B in Panarthropoda is its expression in the putative stomatogastric nervous system, similarly to the conserved expression of Dmrt93B/Dmrt3 in these structures (discussed above). Interestingly, the vertebrate orthologs of . Differential gene expression of such isoforms has largely been investigated in adult tissues, rather than during development. In this study, we found one non-Dsx Dmrt gene that is spliced into (at least) two isoforms; this gene is Glomeris Dmrt11E (Fig. 3a). In Glomeris, the germ cell marker-gene vasa and Sox3 genes are all expressed in the mesoderm of the anal valves and in tissue anterior to that ([38] Janssen et al. 2018), and this is where Dmrt11E is (inter alia) expressed too ( Fig. 4A-D, H). Additionally, Gm-Dmrt11E is expressed in the outer lining of the developing hindgut, tissue that potentially contributes to the developing gonads. Unfortunately, it was not possible to perform an isoform-specific in-situ hybridization for the shorter of the two Dmrt11E isoforms. Possibly, such insitu experiment could uncover tissue-or tissue/sex-specific expression patterns (cf. expression of Pt-Dsx2 isoforms (Fig. 7)).
For the other panarthropod non-Dsx Dmrt genes, there is no obvious correlation with gonadal development and/ or differentiation of any other sex-specific trait, nor is any such gene expressed in a subset of embryos (a 50:50, or similar ratio) that would suggest sex-specific function already at such early stages of development. This, however, is not unexpected because sex-specific traits of panarthropods regularly develop after embryogenesis during post-embryonic or even post-larval developmental stages, or they are difficult to spot on morphological grounds. Examples include the initiation of sex-specific sexual behavior, the Drosophila melanogaster sex combs, the sex brushes of related drosophilids, the bulbi of male spiders, the development of the male phenotype in water fleas, the gonopods in male myriapods, and gonadal differentiation in general ( [103] Rice et al. 2018). In the future, it would be interesting to study expression of Dmrt genes in different sex-specific tissues of larval, juvenile and adult sex-specific structures in order to get a better understanding of the general role these genes may play in sexual differentiation.

Doublesex
We could not detect expression in the investigated developmental stages for Euperipatoides Dsx_like and Parasteatoda Dsx1, and we did not detect a Dsx gene in the sequenced embryonic transcriptome of Glomeris (although Dsx genes have been identified in another myriapod, Strigamia maritima ([106] Chipman et al. 2014) (Fig. 1)), suggesting that these genes act later during development, and may then be involved in for example gonad development and differentiation.
At least one of the female-specific isoforms ([47] Shukla and Palli 2012, Fig. 3) of the single Tribolium Dsx gene, however, is expressed in a pattern that suggests a role in gonad development and/or differentiation (note that we could not use specific probes for the different female isoforms). At late embryonic developmental stages, Tc-Dsx is exclusively expressed in the form of two distinct dots in the tenth abdominal segment (Fig. 7A/B). Unfortunately, the universal germ line marker vasa is not expressed at these late developmental stages ( [107] Schröder 2006), but another potential gonadal marker, SoxE, is expressed in the same segment as Dsx ([38] Janssen et al. 2018). Given that Tribolium Dsx is known to be involved in gonad development and sexual differentiation ( [47] Shukla and Palli 2012), and that it is expressed in close proximity of SoxE, it is likely that the detected embryonic expression of Dsx is indeed in the developing gonads.
Interestingly, we found that at least one of the femalespecific isoforms of Tribolium is expressed considerably stronger in about half of the investigated embryos, and we assume that these embryos represent females. In the other half of the embryos, the expression is much weaker. We assume that those embryos represent males in which at least one of the "female-specific" isoforms appear to be expressed albeit at a very low level (Additional file 10: Figure S7).
Sex-specific splicing of the sex-determining factor Dsx is common among insects, but not found in the crustacean Daphnia magna (and related cladocerans) implying that sex-specific splicing of Dsx could be an insectspecific trait ( The fact that at least one of the two spider Dsx paralogs is also expressed in several isoforms raises the question if one (or more) of these isoforms could be sex-specific, and thus if sex-specific splicing of Dsx could be an ancestral feature of arthropod sex-determination. Interestingly, the different spider Dsx isoforms do not only differ outside their DM domain, but also contain different DM domains (Fig. 3c), which to our knowledge represents a unique feature of Dsx genes; at least in this detail the splice variants of insect Dsx and spider Dsx are fundamentally different.
At least one of the isoforms (A, C, D) of Pt-Dsx2 is exclusively expressed in the dorsal field (DF) (also called extraembryonic area, layer, tissue or field (  (Fig. 7C-F). Similar expression in the DF has been reported for the GATA transcription factor serpent (srp) and hepatocyte-nuclear-factor-4 (hnf4), and it has been suggested that these cells may contribute to yolk consumption and/or midgut development ( [113] Feitosa et al. 2017 and references therein).
The isoform B of Pt-Dsx2 is exclusively expressed in the developing spinnerets. This is interesting for several reasons. First, spinnerets clearly represent an evolutionary novelty of spiders, and second because the set of spinnerets and their silk-producing glands is partially sexspecific ( [114] Correa-Garhwal et al. 2017). Pt-Dsx2 isoform B could thus, like so many other Dmrt genes in other animals, have been recruited for the establishment of sex-specific morphological differences in spiders. Evidence for this hypothesis comes from the work of [115] Schomburg (2017) who found that the same transcript of Dsx2 (Dmrt1B in his study) is indeed upregulated in developing male pedipalps (but not female pedipalps). The male pedipalps carry the spider copulation organs (bulbi) and thus display clear sex-specific morphological differences ([115] Schomburg 2017). Finally, the unique and distinct expression patterns of Pt-Dsx2 isoform B vs the isoform(s) A, C and D strongly suggest that this gene has undergone multiple neo-functionalization processes. After the duplication into Dsx1 and Dsx2, the latter likely acquired a new function during development (either in the DF or the spinnerets), and after that differential splicing allowed the gene to act independently in the development/differentiation of the other structure. Availability of data and materials All data generated or analysed during this study are included in this published article and its Additional files.

Ethics approval
The use of the invertebrate species investigated in this study does not cause ethical issues. Therefore, approval from ethics committees is not required. The investigated species have been obtained from either established laboratory cultures (Tribolium castaneum and Parasteatoda tepidariorum), or have been collected from the wild; the latter (Euperipatoides kanangrensis and Glomeris marginata) do not represent endangered (as to our knowledge) or protected species. We possess permission for collection of onychophorans (Collection permit SL100159).