Molecular developmental evidence for a subcoxal origin of pleurites in insects and identity of the subcoxa in the gnathal appendages

Pleurites are chitinous plates in the body wall of insects and myriapods. They are believed to be an adaptation to locomotion on land but their developmental and evolutionary origins are unclear. A widely endorsed explanation for their origin is through toughening pre-existing parts of the body wall; in contrast, the subcoxal theory suggests pleurites derive from a redeployment of the proximal-most section of the leg, the subcoxa. Here, by studying expression of appendage patterning genes in embryos and larvae of the beetle Tribolium castaneum, we provide the first molecular evidence for the existence of a cryptic subcoxal segment in developing legs. We follow this structure during development and show that the embryonic subcoxa later forms the pleurites of the larva as predicted by the subcoxal theory. Our data also demonstrate that subcoxal segments are present in all post-antennal appendages, including the first molecular evidence of a two-segmented mandible with a subcoxal segment in insects.

by different forms of terrestrial locomotion, including flight, and by activities such as burrowing through soil, enrolment and crawling through rock crevices 2,4 .
The subcoxal theory, in contrast, originates in the work of nineteenth century biologists 5 . Heymons was the first to hypothesize the existence, in a hemipteran, of a proximal subdivision of the embryonic leg which he labelled the subcoxa. Heymons proposed that pleurites develop from this subcoxal leg segment (see Fig. 1d) 6 . Snodgrass extended Heymons' ideas of subcoxal development and argued that the proposed ontogenetic origin of pleurites recapitulated evolutionary history and that pleurites evolved from an ancestral subcoxal segment (see Fig. 1c,d) 1,7 . Recently there has been revived interested in the subcoxal theory, with supporting morphological evidence from embryology and from muscle insertion patterns 3,8-12 . In the modern form of the subcoxal theory, pleurite rings surrounding the leg base (which may be fragmented or incomplete in different hexapod and myriapod lineages) are identified as a trochantinopleurite and eupleurite 3 . They are either considered both to be derivates of the subcoxa or to represent a subcoxa and precoxa, respectively 13 .
To investigate the evidence for the existence of a subcoxal leg segment in hexapods and its potential involvement in forming pleurites, we studied the expression of molecular markers of appendage development in developing embryos of the red flour beetle Tribolium castaneum. Tribolium has pleurites typical of holometabolous insects and is a good animal in which to study their development as the pleurites are unfragmented, relatively large and more segment-like in the larval stages than in some other hexapods 3 . To identify the subcoxa, we required markers that could define subdivisions within the developing legs and markers that could then identify the subdivisions identified in this way. Joint formation in both Drosophila and Tribolium appendages is controlled by Notch signalling and involves multiple genes including serrate (ser) and Delta [14][15][16] ; the process appears to be conserved across Arthropoda 17 . We chose to study the expression of the Tribolium homologue of the gene serrate (Tc-ser) because Tc-ser expression proved significantly easier to detect by in situ hybridization than Notch (data not shown). ser is a transmembrane ligand of the Notch receptor and regulates Notch activation in adjacent cells. It is expressed in a ring of cells on the distal part of each appendage segment and is therefore expressed slightly proximal to where the segment boundary will form 15 . Notch signalling is regulated by 'leg gap genes' in Drosophila that pattern the proximo-distal (PD) axis, such as homothorax (hth), dachshund (dac) and Distal-less (Dll). The expression and function of leg gap genes are conserved, with slight differences, across arthropod leg development 15,16,18,19 .

Results and Discussion
We found five domains of Tc-ser expression in the embryonic leg rudiment (see Fig. 2a), corresponding to five joints separating six segments. As the Tribolium larval leg has five 'true' leg segments (the coxa, trochanter, femur, tibiotarsus and pretarsus 16 ) the presence of an extra embryonic segment led us to infer that the proximal-most segment might be the subcoxa. To prove this, we aimed to determine which segments these domains of Tc-ser expression correspond to in the larval leg.
The proximal-most Tc-ser domain is co-expressed with Tc-hth, the second with both Tc-hth and the proximal domain of Tc-dac 20 , whereas the third domain has both Tc-hth and Tc-Dll co-expression. The fourth Tc-ser domain is co-expressed with the distal domain of Tc-dac and Tc-Dll and finally, the fifth is co-expressed only with Tc-Dll (see Fig. 2b-e). The proximal domain of Tc-dac is faintly expressed at this stage, and may also be expressed in the proximal-most Tc-ser domain (see below).
As the coxa is both the first typical leg segment and is easy to identify by virtue of its large size and characteristic conical shape, we first tried to identify the domain of Tc-ser that corresponds to this segment. To do this, we followed Tc-ser expression and Tc-Dll expression through to later stage embryos in which leg morphology is more similar to larval leg morphology (see Fig. 3).
Comparison of the morphology of the larval leg to that of the morphology of late stage embryos shows that the coxa (indicated with a star in Fig. 3b,f) corresponds to the second segment as identified by Tc-ser expression in the embryonic leg: the coxa lacks Tc-Dll expression and has the second domain of Tc-ser expressed at its distal border whereas Tc-Dll is co-expressed with the third Tc-ser domain, which thus identifies the latter as the third leg segment, the trochanter. Having established the identity of the coxa and trochanter, the five typical leg segments (coxa, trochanter, femur, tibiotarsus and pretarsus) are seen to correspond to the second, third, fourth, fifth and sixth segments of the embryonic leg (marked by Tc-ser expression).
Significantly, a Tc-ser domain is identified proximal to the coxal domain (arrow in Fig. 3e,f). Therefore we define the Tc-ser domains in proximo-distal order as the subcoxa-1 domain, the coxa-2 domain, the trochanter-3 domain, the femur-4 domain, and the tibiotarsus-5 domain, respectively (see Fig. 3d). The pretarsal segment is distal to the tibia Tc-ser domain-5. In Tribolium first instar larvae, the pleurites are flattened radially on the body wall and support the base of the coxa (the largest segment of the leg) (see star in Fig. 3b,f). These sclerites are partly fused to the body wall and form two plates, the episternum and the epimeron, separated by the pleural suture, which is typical of holometabolous insects (see Fig. 3c). This structure develops proximally, adjacent to the coxa, and in the same position as the first domain of Tc-ser expression in the embryonic leg (see Fig. 3f). This demonstrates that the subcoxal segment becomes the pleurites in the larva.
Having shown the existence of a subcoxa in the walking appendages of Tribolium, we looked for homologs of this segment in other limbs by comparing expression of Tc-ser and the leg gap genes between the proximal segments of the legs and the gnathal appendages of Tribolium (see Fig. 4, Supplementary Figs S1-S5). This revealed that Tc-ser is expressed in ring domains along the proximal-distal axis of all appendage types (see Fig. 4a-c and Supplementary Figs S1 and S4).
The hexapod mandible is an unsegmented appendage in both the larva and the adult, however, recent morphological studies have suggested the presence of both a subcoxa and coxa in the developing mandible, and that this may be a general character of Hexapoda 9,11,21-23 as well as in Myriapoda 24 . The Tribolium mandible limb bud shows two domains of Tc-ser expression -a ring and spot domain (see Fig. 4a and Supplementary Fig. S1). The spot domain is present on the distal tip of the mandibular limb bud, more specifically on the outer lobe of the developing endite that relates to the developing incisor process 25 . However, the expression of the subcoxa-1 domain of Tc-ser in the Tribolium mandible (see Fig. 4a, Supplementary Fig. S5) provides the first molecular evidence for the existence of a mandibular subcoxa.
Comparison of leg gap gene expression between the mandible, maxilla and legs shows similar expression patterns relative to Tc-ser expression domains (see Fig. 4m for a summary figure). For the mandible, maxilla and leg, the second segment is characterized by both Tc-hth and Tc-dac co-expression (see Fig. 4d-f,j-l). The first Tc-ser domain is co-expressed with Tc-hth (see Fig. 4j-l) and Tc-dac in the mandible and maxilla. Tc-dac is co-expressed in the lateral part of the Tc-ser subcoxal-1 domain in the mandible and maxilla (see arrowhead in Supplementary Figs S2 and S3), and in the leg subcoxal domain during early leg development (see Supplementary Figs S4 and S5).
We found additional support for serial homology of the subcoxa and coxa in the simultaneous activation of Tc-ser and Tc-dac in all post-antennal appendages (see Supplementary Figs S4 and S5). More evidence for the serial homology of the subcoxa of the mandible, maxilla and labium is provided by the expression of Tc-paired (Tc-prd), which marks the position of the developing endites 25 . Tc-prd is expressed in the coxal segment and is excluded from the subcoxal segment in the mandible, maxilla and labium (see Supplementary Fig. S6).
Taking all this evidence as a whole, we conclude that the leg subcoxa is serially homologous to the proximal-most segments of the post-antennal gnathal appendages, specifically to the cardo of the maxilla, the mandibular subcoxa and the labial postmentum.
Our results show that the pleurites likely develop from a leg segment in Tribolium in accordance with the subcoxal theory. Regarding pleurites in myriapods, it has been postulated that they are homologous with those of hexapods because these taxa were thought to be closely related in a group called the Atelocerata 3 . It is now widely accepted that the hexapods derive instead from within a wider pancrustacean group that does not include the myriapods. Most crustaceans lack pleurites, and the exceptions involve groups that have been proposed as likely hexapod relatives, remipedes 26 and malacostracans. Remipedes have a detached fragment of the coxa that forms an anterior pleurite 3 and the pleopods of malacostracans have a pleurite-like coxal sclerotization 27 . Irrespective of whether or not these are homologous with pleurites in hexapods, the absence of pleurites in other crustaceans implies convergent evolution of pleurites of hexapods and myriapods. Whether this is an adaptation to terrestrial locomotion in both is contingent on the affinities of the aquatic Remipedia. It remains to be demonstrated whether pleurites develop from a subcoxal segment in myriapods.
Understanding their precise segmental composition is necessary to understand the evolution of insect appendages from the common ancestor of hexapods and crustaceans. Regarding the evolution of the pleurites of hexapods, their absence in most crustaceans renders it likely that the insect subcoxa is homologous to one of the proximal crustacean leg segments, which typically do not integrate into the body wall. As there is significant diversity of crustacean limbs in terms of segment number and morphology, together with the fact that crustacean phylogenetic relationships relative to the hexapods remain unresolved 28 , it is currently difficult to postulate a homologous segment to the insect subcoxa and coxa. Comparative morphology argues that the insect subcoxa may be homologous to the crustacean 'coxa' , and that the insect coxa is therefore homologous to the crustacean 'basis' 3,12 . The correspondences noted above in remipede and malacostracan coxal fragments and subcoxally-derived insect pleurites are consistent with this hypothesis.
Despite the great diversity of insect limb morphology, the evidence for serially homologous limb segments suggests a shared regulatory logic for patterning limb segments within the same species. This similarity of gene regulation probably reflects common ancestry of these segments from an ancestral limb; so that, for example, the subcoxa of the mandible, maxilla, and legs could have evolved from the same leg segment in identical ancestral serially homonomous limbs. The Cambrian fossil record reveals that many arthropod taxa, such as trilobites, had a series of morphologically scarcely undifferentiated post-antennal biramous limbs [29][30][31] . Based on these facts, a scenario of insect appendage evolution from ancestral serially homologous limbs is depicted in Supplementary Fig. S7.
Our observations of development of the subcoxal segment may also contribute to understanding of the evolution of insect wings. Two main hypotheses for the origin of wings are gill theory and paranotal theory 12,32,33 . In gill theory, the wings evolved from gills (or exites), which were present on a proximal (possibly subcoxal) segment. As we have identified the precise position of the developing subcoxa in the Tribolium embryo, this could help to identify cells from the subcoxal segment that might contribute to the development of the wing, and show whether genes important for patterning wings (such as vestigial) are expressed in cells of the subcoxal segment 34,35 .

Materials and Methods
Tribolium castaneum culture and experimental methods. All experiments, including details of Tribolium castaneum culture and methods for studying gene expression in Tribolium embryos (wholemount in situ hybridisation), techniques for studying embryo and larval morphology (scanning electron microscopy and larval cuticle preparation) were performed as previously described 25 .
Specifically regarding in situ hybridization, both single stainings with nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) and double stainings (NBT/BCIP and FastRed) were performed as previously described 36 but with slight changes to the frequency and duration of washes introduced from alternative in situ hybridisation protocols 37 to reduce background staining.
An alkaline phosphatase conjugated anti-DIG antibody was incubated with the embryos either overnight at 4 °C or for one hour at room temperature. The first staining was performed with NBT/BCIP. After the staining was completed, the alkaline phosphatase conjugated anti-DIG antibody was inactivated by denaturing the protein for 15 minutes at 65 °C in hybridization solution with 0.3% (w/v) Sodium Dodecyl Sulphate (SDS). A second staining was performed by first incubating the embryos with alkaline phosphatase conjugated anti-FITC antibody and then subsequently staining with FastRed (Sigma).