Insects bear a stereotyped set of limbs, or ventral body appendages. In the highly derived dipteran Drosophila melanogaster, the homeodomain transcription factor encoded by the Distal-less (Dll) gene plays a major role in establishing distal limb structures. The Dll ortholog (TcDll) has been isolated from the beetle Tribolium castaneum, which, unlike Drosophila, develops well-formed limbs during embryogenesis. TcDll is initially expressed at the sites of limb primordia formation in the young embryo and subsequently in the distal region of developing legs, antennae and mouthparts, except the mandibles. Mutations in the Short antennae (Sa) gene of Tribolium delete distal limb structures, closely resembling the Dll phenotype in Drosophila. TcDll expression is severely reduced or absent in strong Sa alleles. Genetic mapping and molecular analysis of Sa alleles also support the conclusion that TcDll corresponds to the Sa gene. These data indicate functional conservation of the Dll gene in evolutionarily distant insect species (Beermann, 2001).

Across the insects, with the exception of the Diptera, limb primordia are established during embryogenesis, giving rise to well-formed larval limbs within which the founder cells of the imaginal limbs proliferate. By contrast, the dipteran larvae are limbless. In lower dipterans, the imaginal limb primordia of both head and thorax originate in the embryo in close association with distal sense organs. In higher dipterans such as Drosophila, this association is not obvious for the primordium of the eye-antennal disc and the distal sense organs of the antenna and the maxilla. These evolutionary changes, culminating in head involution, result in a derived condition that is characterized by divergent pathways of larval and imaginal limb development. However, the initial expression of Dll at specific distal positions within the segments of head and thorax closely resembles the TcDll expression pattern before limb primordia are morphologically discernible. Subsequently, larval limb development is suppressed in Drosophila. The small imaginal primordia are involuted toward the end of embryogenesis: they grow into large disc-shaped epithelia within the larva and are transformed into imaginal limbs by evagination during the pupal stage. Thus, a pupal leg in Drosophila appears to correspond to an embryonic leg in Tribolium. This view is supported by the equivalent expression of Dll in the distal region corresponding to the tarsus and the distal tibia, and in a proximal ring corresponding to the proximal part of the femur. Similar patterns of Dll expression have been observed in the embryos of diverse insect species, the butterfly Precis, the grasshopper Schistocerca and the cricket Gryllus, using a crossreacting antibody. Although it is not known when during disc development the mature pattern of Dll expression is established in Drosophila this is likely to occur after the imaginal disc has reached a certain cell number. Thus, the Dll expression pattern appears to reflect a heterochronic change in limb development (Beermann, 2001).

Limbs are segmentally repeated ventral body appendages that are subject to segment-specific developmental modifications, including changes in shape and function as well as complete loss. A large body of evidence, mainly from Drosophila, indicates that Dll integrates the developmental signals that initiate and modify limb development. Among the positive regulators of Dll expression are the segment-polarity genes such as wingless and engrailed, which provide the anterior-posterior coordinate, and genes like dpp, which provide the dorsoventral coordinate. Superimposed is a negative control of Dll expression that is exerted by the homeotic genes. In Drosophila, a negative cis-regulatory element of Dll has been shown to bind Bithorax complex gene products. Similarly, larval prolegs in abdominal segments 2-6 of caterpillars are formed by cell groups that express Dll but not the abd-A gene. This regulatory network appears to be largely conserved between Drosophila and Tribolium. Not only are the initial patterns of Dll expression very similar but also the requirement for Dll action, as evidenced by the Sa and Dll mutant phenotypes, is essentially the same. This parallel is best illustrated by the mandible whose limb status has been controversial. In both Drosophila and Tribolium, the mandible does not express Dll nor does it require Dll function for normal development. These results support the notion that the mandible is not a limb but corresponds to a basal structure derived from the body wall. Snodgrass (1935) proposed a hypothetical ground state of the arthropod leg (R. E. Snodgrass: Principles of Insect Morphogenesis. New York, London: McGraw Hill). The leg was divided into the proximal part consisting of the coxa, a simple outpocketing of the body wall called the coxopodite, and into the distal telopodite representing the limb proper. The idea that the coxopodite and the telopodite are distinct is supported by the finding that genes regulating the proximal-distal axis in Drosophila define a sharp boundary at the coxa-trochanter junction. The proximal region corresponding to the coxa is characterised by the expression of Homothorax (hth) which facilitates the nuclear localisation of Extradenticle (Exd). In the distal region, Wingless (Wg) and Decapentaplegic (Dpp) act through their targets Dll and dachshund (dac) to restrict hth expression and thereby nuclear localisation of exd. Thus, the domain of the 'coxopodite genes' hth and exd and that of the 'telopodite genes' of the Wg/Dpp pathway are mutually exclusive. Comparable analyses have not been possible in other insect species. Analysis of the lack-of-function phenotype of the Sa/TcDll gene, however, provides additional support for Snodgrass' original idea. As in Drosophila, lack of Dll function abolishes the limb proper, or telopodite, but leaves the coxa and body wall unaffected, thus demonstrating that a basic subdivision of the leg into proximal and distal regions also holds true in another, evolutionarily distant insect species (Beermann, 2001).

Antibodies were used to examine the expression patterns of Antennapedia (Antp), Ultrabithorax (Ubx), Ubx and abdominal-A combined (Ubx/abd-A), and Distalless (Dll) in the embryos of the moth Manduca sexta. The spatial and temporal pattern of Antp expression in Manduca is correlated with the anterior migration of two patches of epithelium that include the anterior-most tracheal pits, and with the development of functional spiracles. Ubx expression shows an intricate pattern that suggests complex regulation during development. Throughout Manduca embryogenesis, the expression of Ubx/Abd-A and Dll is similar to that reported for other insects. However, there is no apparent reduction in Ubx/Abd-A expression in the Manduca abdominal proleg primordium that expresses Dll. The expression of these four proteins was also examined in embryos of the Manduca homozygous homeotic mutant Octopod (Octo). The Octo mutation results in the transformation of A1 and A2 in the anterior direction, with homeotic legs appearing on A1 and occasionally A2. These results suggest that in Octo animals there is a reduction in the level of Ubx protein expression throughout its domain (Zheng, 1999).

Insects show a dramatic diversity in the number and segmental distribution of abdominal appendages. For example, among Lepidoptera larvae, the number of abdominal appendages varies from none to seven pairs. The number of appendages also varies during the course of embryonic development. During early stages of embryogenesis of Pieris rapae and Bombyx mori ventral appendages are present on all abdominal segments. Later some of these appendages regress, leaving prolegs on A3-A6 and on the terminal segment. In Drosophila both Ubx and Abd-A act to repress Dll and the development of abdominal appendages. However, such repression is absent in the crustacean, indicating that the repressive function of Ubx and Abd-A evolved in insects. The presence of homeotic legs in Octo Manduca demonstrates that Manduca Ubx also represses appendage development. However, in the beetle Tribolium and the grasshopper Schistocerca an A1 appendage, the pleuropodia, develops despite high level of Ubx in A1 appendage primordium. Thus the repressive function of Ubx on A1 appendage development evolved late in insect evolution, in the Diptera/lepidoptera lineage. These findings indicate that the repressive function of Abd-A evolved even later than that of Ubx. The expression of Dll and the emergence of prolegs in A3-A6 apparently initiated in the presence of strong Abd-A expression, suggests that in Manduca the repressive function of Abd-A on A3-A6 appendages has not yet evolved. This conclusion for Manduca Abd-A differs from that which has been suggested for the butterfly Precis, where Abd-A is believed to suppress appendages development. In order to develop prolegs both Ubx and Abd-A are locally repressed in the proleg primordia (Zheng, 1999).

Insects are easily distinguishable by the absence of legs on the adult abdomen. Studies performed on the Dipteran, Drosophila melanogaster, indicate that this is because of the repressive effects of the homeotic genes Ultrabithorax and abdominal-A on the limb promoting gene Distal-less during embryonic development. However, in many species appendage-like structures are present on abdominal segments in embryonic and juvenile stages. By using classical genetics and double-stranded RNA-mediated gene silencing in the red flour beetle, Tribolium castaneum, a species that develops an appendage on the first abdominal segment, it was possible to examine the roles of Ubx and Abd-A in abdominal limb development. In Tribolium, Abd-A, but not Ubx, represses early expression of Dll in the embryonic abdomen. Ubx appears to modify the A1 appendage. This difference in the activities of Abd-A and Ubx is critical for proper development of this appendage. It is suggested that an ancestral role of Abd-A in insect abdominal appendage development was in the repression of Dll initiation and that an ancestral role of Ubx was in modulation of abdominal appendage morphology (Lewis, 2000).

By examining TcDll and TcEn expression in TcUbx and Tcabd-A mutant embryos, a better understanding of the role of each in suppressing and modifying limb programs in the beetle abdomen was obtained. In TcUbx mutant embryos, TcDll expression in the abdomen remains restricted to anterior A1, whereas, in Tcabd-A, mutant embryo TcDll is ectopically expressed in each abdominal segment, resulting in abdominal appendage development. These results clearly support the role for TcAbd-A as a primary TcDll repressor (and therefore appendage repressor) in the Tribolium abdomen. The role of TcUbx in regulating Dll expression appears to be more complex. Although TcDll and TcUbx are initially coexpressed during early pleuropod development, later TcUbx is absent in the TcDll-expressing cells, leaving open the possibility that TcUbx represses TcDll late in development. Whether or not late expression of TcUbx represses TcDll expression in these cells, it is evident from mutant analysis that TcUbx is required for the proper differentiation of these cells. In TcUbx mutants, the nuclei of TcDll-expressing cells in the pleuropod never become morphologically distinct as they do in the wild type. It is therefore believed that TcUbx acts as a modifier rather than a repressor of abdominal appendage development (Lewis 2000).

The dynamic relationship between TcUbx and TcDll expression in the pleuropod and the effect of TcUbx expression on the differentiation of TcDll-expressing cells suggests that TcUbx acts to modify the way cells in the anterior A1 compartment interpret signaling cues. In the absence of TcUbx, cells respond to signaling cues as if they were no longer pleuropodial. The failure of the appendage to invaginate and the presence of the subterminal tarsal claw in TcUbx mutant larvae support this view. In addition, the position of the subterminal tarsal claw appears to correspond to the boundary of TcEn expression and the cluster of TcDll-expressing cells in the developing appendage of the embryo. This is interpreted as evidence that these cells now respond to signaling cues as if they were leg, with the distal-most tip, the tarsal claw in the leg, at the intersection of the anterior-posterior boundary (Lewis 2000).

Comparing the data obtained in this study on beetle abdominal appendage development with that obtained from other holometabolous insects, it is suggested that abdominal limb repression through direct Abd-A repression of Dll expression evolved at the latest in the last common ancestor of the holometabola. This is the most parsimonious interpretation, given that the repressive activity of Abd-A is evident in species from all of the holometabolous orders examined. However, one holometabolous insect species, the Lepidopteran Manduca sexta, appears to be an exception. In the developing abdominal prolegs in this species, Dll is expressed despite the coexpression of Ubx/Abd-A. It is interesting to note that the ability to express Dll in developing prolegs has arisen using at least two different mechanisms within the Lepidoptera. In the butterfly Precis coenia, activation of Dll expression in the abdomen is correlated with regional repression of Ubx/Abd-A, whereas, in the moth Manduca sexta, Dll expression occurs through a different mechanism, presumably involving the escape of Dll from the repressive effects of Abd-A. These data suggest that the release of the repressive effect of Abd-A on abdominal limbs in higher holometabolous insects occurred convergently through changes at different levels of the limb regulatory hierarchy. Alternatively, it is possible, although less likely, that the regional repression/expression of Ubx/Abd-A has no causative effect on proleg outgrowth, leaving open the possibility that the presence of prolegs in these two Lepidopteran species is not convergent (Lewis 2000).

Mandibles are feeding appendages functioning as 'jaws' in the arthropod groups in which they occur. The specific part of this appendage involved in food manipulation (limb tip versus limb base) has been thought to suggest phylogenetic relationships among some of the major taxa of arthropods (myriapods, crustaceans, and insects). As a way to independently verify the conclusions drawn from previous morphological analyses, the expression pattern of the gene Distal-less (Dll), which specifies the distal part of appendages, has been studied. In contrast to the traditional view, both insect and crustacean adult mandibles are gnathobasic, handling food with the basal portion of the appendage. Furthermore, as is evident by the reduction in the number of Dll-expressing cells in the later developmental stages, adult diplopod jaws are also gnathobasic. Thus, jaws of all mandibulates (myriapods, crustaceans, and insects) seem to have a similar gnathobasic structure. Dll is also expressed in the labra of all arthropod taxa examined, suggesting that this structure is of appendicular derivation. Additionally, the spinnerets and book lungs of spiders, long considered on other grounds to be modified appendages, express Dll, confirming this interpretation. This study shows that, in addition to their use in phylogenetic and population genetic studies, molecular markers can be very useful for inferring the origins of a particular morphological feature (Popadic, 1998).

Arthropod diversity is apparent in the variations in limb number, type, and position along the body axis. Among the insects, for example, butterflies and moths (Lepidoptera) develop larval abdominal and caudal appendages ('prolegs'), whereas flies (Diptera) do not. A DLL homolog from the butterfly Precis coenia has been cloned. It is expressed in all developing limbs (except the mandible), including the prolegs. The relationship between Dll and wingless expression observed in Drosophila is conserved in Precis among all limbs. These data suggest that DLL function, suppressed in the abdomen early in insect evolution, has been derepressed in Lepidoptera (Panganiban, 1994).

Studies of the genes involved in patterning the appendages of Drosophila melanogaster have revealed a system of signaling and transcriptional regulation that is responsible for specifying the proximo-distal limb axis. The expression patterns of presumptive homologs of the Drosophila genes extradenticle, dachshund, nubbin, ventral veins lacking, and Dll in the limbs of the woodlouse Porcellio scaber and the spider Steatoda triangulosa is reported. Although the expression domains of the appendage genes roughly correspond to those of Drosophila, their relative positions and segmental affiliation are distinct. In addition, the expression patterns of the appendage genes allows a resolution of the segmental composition of different appendages within crustacean and spider embryos. It is concluded that certain limb types, e.g., mouthparts, appear to be derived from a leg-like ground-plan via the elimination/fusion of the intermediate and distal podomeres. Moreover, just such a modification is observed during the transformation of the anterior legs into mouthparts in P. scaber. Although these data do not unequivocally resolve the question of homology of the arthropod leg segments, they do provide evidence for a single conserved proximo-distal patterning system in the development of noninsect arthropod limbs (Abzhanov, 2000).

Most of the knowledge about appendage development comes from developmental genetic studies of D. melanogaster. Although the legs of adult flies and crickets are very similar in terms of metameric origin and segment number, the developmental processes by which their morphology is achieved are quite distinct. Cricket legs develop directly from the body wall and are patterned and segmented during subsequent outward growth in the embryo. Drosophila has a more derived system for adult appendage formation involving the imaginal discs, which are specified and patterned during later larval stages. The development of an appendage, such as a leg, from an imaginal disc is a multistep process involving initiation of the disc itself via signaling from neighbor tissues, prepatterning, establishment of limb field primordia, dorsal-ventral patterning, proximo-distal patterning culminating in growth, and evagination and differentiation of adult structures. It is possible that much of leg development in Drosophila is characteristic only of the higher Diptera and cannot be compared directly to that of other arthropods. Therefore, a relatively basal insect, the cricket A. domesticus (Orthoptera), which has a direct limb development, has been included in this comparative analysis. The morphology of cricket limbs, such as antennae, mouthparts, and legs, is more typical of the lower insects. The expression patterns and dynamics of the leg patterning genes, with regard to the specific podomeres, are conserved between Drosophila and Acheta. Nuclear Exd is detected in the basal part (coxa and trochanter in the legs) of the appendages in Drosophila and Acheta. In both species, the early Exd domain is immediately adjacent to that of Dll. In Acheta, Dll is detected throughout the telopodite during early development but then fades in the intermediate portion of the leg and forms two separate domains: proximally in the trochanter/proximal femur and distally in the tarsus/distal half of tibia. In Drosophila, this change in Dll pattern is associated with the growth of the imaginal disc and is accompanied by initiation and establishment of the Dac domain in the intermediate portion of the leg. It is notable that the dynamics of the Exd and Dll expression patterns in the maxillary and labial palps are very similar to those in the walking legs in Acheta. This suggests that the segmented maxillary and labial mouthparts, which belong to the basal mandibulate type, are patterned very similarly to the legs, in striking contrast to the adult proboscis in Drosophila. Another observation in Acheta is the restriction of Exd, which is required for antennal identity in Drosophila, to the two most basal antennal segments, which are sometimes homologized with the coxa and trochanter. This observation indirectly supports the hypothesis that the insect antenna is basically a three-segmented structure with the third, distal segment subdivided into subsections, forming a flagellum. However, little or no Dll and Exd coexpression is observed despite the fact that both genes have been shown to be required for the development of antennal structures in D. melanogaster. Thus it would appear that the cricket may pattern its antenna differently from flies (Abzhanov, 2000).

The homolog of the Drosophila gene dachshund (dac) was isolated from the beetle Tribolium castaneum. Tc'dac is expressed in all appendages except urogomphi and pleuropodia. Tc'dac is also active in the head lobes, in the ventral nervous system, in the primordia of the Malpighian tubules and in bilateral stripes corresponding to the presumptive dorsal midline. Expression of Tc'dac in the labrum lends support to the interpretation that the insect labrum is derived from a metameric appendage. The legs of Tribolium accommodate two Tc'dac domains, of which the more distal corresponds to the single dac domain described for Drosophila leg discs. In contrast to Drosophila, where this domain is thought to intercalate between the homothorax (hth) and the Distal-less (Dll) domains, in Tribolium it arises from within the Dll domain. Both the distal Tc'dac domain in the legs as well as the expression in the labrum are deleted in embryos mutant for the Tc'Dll gene, while the proximal leg domain and the mandibular expression are unaffected. Based on Tc'dac expression in wild-type and mutant embryos, serial homology of the complete mandible with the coxa of the thoracic legs has been demonstrated. This homology affirms the gnathobasic nature of the insect mandible (Prpic, 2001).

dac expression in the leg primordia of Tribolium and the imaginal discs of Drosophila is similar. In both species, strong expression of dac appears relatively late during leg development, and at intermediate proximal-distal positions. In mature leg primordia, this dac domain is flanked proximally and distally by the 'ring' and the 'sock' of Dll expression, respectively. Also, a region of overlap between dac and the Dll 'sock' is present in both species. One difference between the Tribolium patterns and those in Drosophila is that distal regions appear to be underrepresented in the beetle leg primordia. While the region of Dll/dac overlap is large in Drosophila,including most of the primordia of tibia and tarsus, in Tribolium this overlap is restricted to a narrow ring. However, this correlates well with the fact that certain distal structures of the adult beetle leg are only formed during metamorphosis. The tibia and the tarsomeres of the imaginal leg are represented in the larva only as a single rather short podomere, termed tibiotarsus. The final number and size of adult podomeres arise in a second growth phase during metamorphosis (Prpic, 2001).

In addition to the difference just mentioned, there are three more discrepancies in dac expression in the embryonic thoracic leg primordia of Tribolium versus the postembryonic imaginal discs of Drosophila . (1) A surprising finding is the presence of an additional proximal domain in the thoracic appendages of Tribolium. In Drosophila, dac is reported to form a single domain in the leg discs. In crustaceans and chelicerates, too, only one dac domain has been observed. Despite its low abundance and although it is not yet known if a proximal dac domain is typical for insect leg primordia in general, it is believed this domain is relevant since its existence provides a rationale for understanding expression and regulation of Tc'dac in the gnathal segments (Prpic, 2001).

(2) A second apparent difference between the expression patterns of Tc'dac and Dm'dac in the leg concerns the position where the (distal) dac domain arises relative to the Dll domain. In Drosophila,it is believed that dac expression intercalates between the hth and Dll domains. The proximal 'ring' of Dll in mature discs is thought to arise as a secondary expression domain once the dac domain has already established itself. In Tribolium, however, it is evident that the distal Tc'dac domain arises within the still unbroken Dll domain, and that increasing abundance of Tc'dac coincides with loss of Tc'Dll expression in these cells, which results in the division of the initial Tc'Dll domain into the proximal 'ring' and distal 'sock' observed at later stages. In other words, the Dll ring in Tribolium is a remnant of the original Tc'Dll domain and not a newly formed, secondary domain. Future work will show if the two species really differ in this expression detail or if persistence of Dll expression proximal to the dac domain has been missed in Drosophila due to the folded architecture of leg imaginal discs. Emergence of the ring-and-sock pattern by downregulation of Dll in intermediate positions has also been described for Schistocerca, suggesting that this mode is ancestral in insects. Since antagonism between dac and Dll occurs in the Drosophila leg, dac may well be required for the generation of the ring-and-sock pattern (Prpic, 2001).

(3) The third difference between Tc'dac and Dm'dac expression in the leg provides an example of how limb patterning during dipteran evolution has adapted to the specific condition of imaginal disc development. In Drosophila,the dac domain is asymmetric along the dorso-ventral axis of the leg imaginal disc. This expression domain can be described as a ring which is much broader dorsally than ventrally. This dorsal expansion correlates with the way femur and tibia form in Drosophila,and how the leg disc evaginates during metamorphosis. Most leg podomeres are derived from ring-shaped primordia in the imaginal disc. During disc eversion, these rings are transformed into podomeres along the elongated leg in a process often compared to the extension of a telescope. However, the first podomeres that exit the disc cavity through its connection to the body wall epidermis are not the distal-most tarsus, but a dorsal lobe of the leg disc which comprises distal femur and proximal tibia -- which both express dac. Only after disc eversion do these two podomers become separated by longitudinal fission. The asymmetric expression of dac in Drosophila appears, therefore, to be an adaptation to the mechanical needs of disc eversion, while the symmetric pattern in Tribolium represents a more simple ancestral situation. It is intriguing to speculate that the need for an asymmetric dac domain may be the reason behind the altered expression of dpp in Drosophila, where the dpp expression domain is greatly expanded dorsally, while it is confined to the distal tip of the appendages in more ancestral arthropods like Tribolium, Schistocerca and Cupiennius (Prpic, 2001).

The labrum expresses Dll in all insects investigated; nevertheless, its metameric nature is still disputed since Dll can be seen as a general marker for structures growing out from the body wall rather than a specific marker for segmental limbs. Based on a homeotic phenotype in adult Tribolium, previous studies have concluded that the labrum was indeed of appendicular origin. Additional molecular evidence for this interpretation is provided by demonstrating that another gene with an essential function in Drosophila leg development, dac, is expressed in the Tribolium labrum. Even though the Tc'dac expression domain in the labrum is incomplete in the sense that it does not form a full ring, its intermediate proximal-distal position -- and its dependence on Dll function -- provides additional support for the labrum indeed being serially homologous to the other metameric arthropod appendages (Prpic, 2001).

It is generally believed that the maxillary and labial palps represent the telopodite of these appendages, while the maxillary stipes, cardo, galea and lacinia, and the labial mentum, prementum and glossae represent the coxopodite. In the labium, Tc'dac is expressed exclusively in the proximal part (which after metamorphosis will form mentum, prementum and glossae), while the labial palp remains free of expression. In Tc'Dll mutants, the labial palp is deleted, but Tc'dac expression remains unaffected. In this respect, the labial Tc'dac expression resembles the proximal expression domain in the thoracic legs. These data strongly suggest that the labial Tc'dac expression domain is actually homologous to the proximal rather than to the distal leg domain (Prpic, 2001).

Tc'dac expression in the maxilla is stronger than in the labium, but also for the maxilla the expression in Tc'Dll mutants makes it clear that most of the expression is in the coxopodite. However, in wild-type animals a small Tc'dac domain forms in the maxillary palp. This domain is missing in Tc'Dll mutant embryos, similar to the distal domain in the legs. This suggests that the palp domain is serially homologous to the distal leg domain, while the major maxillary Tc'dac expression again is homologous to the proximal leg domain. That no distal Tc'dac domain is present in the labial palps concurs with the fact that positional values present in the maxilla are missing in the labium: the maxillary palp in Tribolium larvae has three podomeres, while the labial palp has only two (Prpic, 2001).

It is argued that the expression pattern of dac in Tribolium actually provides evidence for the gnathobasic origin of the mandible, i. e. formed from the limb base. Like the coxopodite domains in leg, maxilla and labium, the Tc'dac expression in the mandible also is independent of Tc'Dll, and the similarity in size and expression of all three gnathal appendages in embryos mutant for Tc'Dll strongly suggests serial homology of the mandible to the coxopodite of maxilla and labium. This line of reasoning also makes it clear that the mandibular Tc'dac expression corresponds to the proximal leg domain, not to the distal domain -- even though it has been upregulated such that it is as strongly active as the distal leg domain. Since the proximal leg domain is located in the coxa, serial homology of coxa, mandible and the coxopodites of maxilla and labium can be inferred. This lends further support to the notion that the insect trochanter is part of the telopodite. Accordingly, the maxillary galea and the labial paraglossa are probably not serial homologs of the trochanter, but represent coxal endites (Prpic, 2001).

In summary, the expression of Tc'dac is strong confirmative evidence in favor of the gnathobasic origin of the mandibles. This interpretation of gnathal Tc'dac expression is based on the presence of a proximal dac domain in the developing Tribolium legs, a domain which apparently is missing in Drosophila. This shows once again the importance of comparative analyses for the correct interpretation of morphological structures and developmental processes. It remains to be seen how the proximal Tc'dac domains in the legs and in the gnathal appendages are regulated, whether they serve similar functions, and to which degree differential regulation of Tc'dac is responsible for the morphological differences among these appendages (Prpic, 2001).

The expression patterns of Gryllus (cricket) homothorax (Gbhth) and dachshund (Gbdac) are described, together with localization of Distal-less or Extradenticle protein during leg development. Their expression patterns have been correlated with the morphological segmentation of the leg bud. The boundary of Gbhth/GbDll subdivision is correlated with the segment boundary of the future trochanter/femur at early stages. Gbdac expression subdivides the leg bud into the presumptive femur and more distal region. During the leg proximodistal formation, although the early expression patterns of GbDll, Gbdac, and Gbhth significantly differ from those of Drosophila imaginal disc, their expression patterns in the fully segmented Gryllus leg are similar to those in the Drosophila late third instar disc (Inoue, 2002).

Although the legs of adult flies and crickets are very similar in their segmental compositions, the developmental processes producing their morphologies are quite distinct. In the Drosophila imaginal disc, leg formation occurs through the concentric folding and the subsequent segmentation of monolayered epithelia in the leg disc during later larval stages, although the early leg disc remains as a flattened two-dimensional structure. In contrast, the cricket leg bud is formed directly from the body wall and segmented during the subsequent outward growth in the early embryogenesis period. The leg segmentation in Gryllus occurs intercalatively step by step until stage 11 (6 days after EL). The changes have been classified into the five following stages: (1) formation of the leg bud at stages 6-7 (2.5 days after EL); (2) the first morphological segmentation at the future trochanter/femur boundary at stage 8 (3 days after EL); (3) the second segmentation in the femur/distal telopodite boundary at stage 9 (4 days after EL); (4) the third segmentation in the tibia/tarsus boundary at stage 10 (5 days after EL); (5) the fourth segmentation at the coxa/trochanter boundary. Then, the elongation of each segment takes place to form the legs of the nymph until hatching (14 days after EL) (Inoue, 2002).

Before the onset of the leg bud formation (stage 5), Gbhth is expressed uniformly throughout the embryos. Just before the onset of leg bud formation (stage 5, 40 h after egg laying), GbDll starts to be expressed in the presumptive thoracic leg region. Subsequently, Gbhth expression was downregulated in the same regions (stage 6, 48 h after egg laying). Thus, the Gbhth/GbDll antagonistic subdomains seem to be established in the early cricket embryos, as observed in Drosophila. In the early stages of leg bud formation (stages 7-9), Gbhth is expressed in the proximal region of the leg bud. In embryos at stages 8-10, when the leg segments are visible, the distal boundary of the Gbhth expression domain is localized at the proximal region of the femur segment (Inoue, 2002).

Expression patterns of Gbdac during leg development are dynamic, although Dmdac is expressed as a single ring in the Drosophila leg imaginal disc throughout its development. Gbdac expression is first detected in an anterior patch of cells in the leg bud at stage 7. At stage 8, the Gbdac expression domain becomes a narrow circumferential ring in the middle of the leg bud. At stage 9, the expression domain divides into two rings. Gbdac expression is not observed in the presumptive intersegmental border between the femur segment and more distal segments. Between stages 10 and 11, the proximal ring of expression corresponds to the distal part of the femur segment, while the distal ring of expression covers a region from the distal tibia to the proximal tarsus segment. In sagittal sections of the metathoracic femur, Gbdac expression is detected in the invaginating apodeme, as well as in the epithelial tissue of the distal femur (Inoue, 2002).

To compare expression patterns of Gbhth and Gbdac with localization of GbDll or GbExd, double staining was performed, using the corresponding RNA probes and an antibody against Dll or Exd, which are used as distal and proximal markers, respectively, in Drosophila, Acheta (Orthoptera, cricket), and Schistocerca (Orthoptera, grasshopper).Prior to the leg bud formation (stage 5), due to weak expression of GbDll, no double stainings could be observed. Prior to the first leg segmentation, Gbhth is expressed in the proximal region of the leg bud, while GbDll is detected in the distal region of the leg bud (stages 6-7). A merged panel reveals that the leg bud is stained with red and green without significant overlap, indicating that the bud is divided into two domains: a distal GbDll-localized domain and a proximal Gbhth-expressing domain. These domains are complementary at this stage, but by the end of stage 7 they partially overlap. At the same stage, GbExd accumulation, caused by nuclear localization of GbExd, is strongly detected in the proximal region of the leg bud, though low levels of GbExd expression can be seen in the distal region. The GbExd accumulation domain in the proximal region overlaps the Gbhth expression domain, as is observed in Drosophila (Inoue, 2002).

The subdivision of the leg bud into proximal and distal domains becomes morphologically discernible as a circumferential constriction, which is the boundary between the trochanter and femur (stage 8). The proximal limit of the GbDll domain corresponds to this boundary at this stage. However, the GbDll domain includes the distal region of the trochanter in later stages (Inoue, 2002).

At stages 8-9, Gbhth is expressed continuously in the proximal domain of the leg bud, overlapping the GbExd accumulation domain. By early stage 8, GbDll is localized throughout the distal tip of the leg bud, including the telopodite. Gbdac is expressed in the middle of the leg bud as a narrow ring, overlapping the GbDll expression domain. GbDll expression starts to become downregulated in the central region at the end of stage 8, in a region where Gbdac is expressed. The domain of GbDll and Gbhth co-expression continues to be observed at stage 9. The proximal ring of GbDll expression, corresponding to the proximal domain of the femur segment, remains unchanged from stage 9 to stage 11 (Inoue, 2002).

By stage 9, the Gbdac expression domain observed at stage 8 has divided into two domains: the proximal and distal domains. Meanwhile GbDll expression becomes undetectable in the middle region. The distal Gbdac domain therefore overlaps a proximal region of the distal GbDll domain, while no such co-expression is observed in the proximal domain. A narrow domain showing no expression of Gbhth, Gbdac, or GbDll can be observed between the two Gbdac domains. Consequently, six proximodistal expression domains appear in the Gryllus leg (Inoue, 2002).

At stage 10, the third morphological segmentation is observed in the distal leg bud, which divides into the tibia and tarsus. Subsequently, the fourth segmentation is discernible in the proximal leg bud, which divides into the coxa and trochanter. It has not been possible to find any of the boundaries of expression of the known appendage genes corresponding of the tibia/tarsus and coxa/trochanter boundaries (Inoue, 2002).

At stages 11-12, following the establishment of the leg segments, structures that connect adjacent segments such as articulates, muscle patterns, etc., are constructed at the segment boundaries. These stages are therefore designated here as the articulation phases. In Gryllus, since the major leg segments can be identified morphologically, the leg segments and the six domains determined by expression patterns of Gbhth, Gbdac, and GbDll can be easily correlated. The proximal-most domain in which only Gbhth is expressed, corresponds to the body wall, coxa segment, and proximal trochanter. The proximal boundary of the GbDll/Gbhth domain lies in the trochanter, and the overlapping GbDll/Gbhth domain extends into the proximal femur. The Gbdac domain corresponds to the middle of the femur segment, including the articulation between the femur and tibia. At stages 11-12, a narrow domain can also be found in which neither GbDll nor Gbdac is expressed, corresponding to the proximal tibia. The Gbdac/GbDll domain includes the articulate between the tibia and tarsus, and extends to the presumptive boundary between tarsal segments 1 and 2. The distal-most GbDll domain corresponds to presumptive tarsal segments 2 and 3 and the pretarsus. The intermediate three expression domains, i.e. the GbDll/Gbhth, Gbdac, and GbDll/Gbdac domains, include the segmental boundaries of the trochanter/femur, the femur/tibia, and the tibia/tarsus, respectively, but do not correspond to the segments (Inoue, 2002).

Expression patterns of Gryllus hth, dac, and Dll in the leg bud with those in the Drosophila leg imaginal disc were compared. The results reveal that these expression domains resolve into similar patterns. In contrast, some differences in the elaborating processes of the expression patterns of the three genes can be seen between Gryllus and Drosophila (Inoue, 2002).

The position of the proximal limit of the GbDll domain has a very sharp boundary between stages 6-8, corresponding to the morphological segmental boundary between the femur and trochanter. Thus, the most proximal segment of the telopodite, i.e. the trochanter, is not included in the GbDll domain in early stages. In contrast, in Drosophila, genetic evidence has demonstrated the subdivision of the leg disc by Dmexd and DmDll into the coxopodite and telopodite. In addition, it has been reported that Dll is detected throughout the telopodite during early development in Acheta (Orthoptera, cricket). Despite these discrepancies, the observations of the expression patterns of GbDll and Gbhth or GbExd in later stages are basically consistent with the results obtained for Acheta, Schistocerca (Orthoptera, grasshopper, and Drosophila (Inoue, 2002).

Following the primary subdivision by GbDll and Gbhth, their expression domains overlap at the boundary at later stages. The domain of Gbhth expression does not strictly correspond to the coxopodite, but expands distally into the femur. This is the same for Drosophila leg, in which the limit of expression of Dmhth expands into the proximal femur. In Drosophila, the first intercalated region between the Dmhth and DmDll domains is the intermediate region expressing Dmdac, and as a result, three discrete domains are established in the leg imaginal disc. In contrast, in Gryllus, three discrete expression domains of the three genes are not observed during leg development (Inoue, 2002).

The expression domain of Gbdac transiently overlaps with that of GbDll at early stages. In Drosophila, Dmdac expression is asymmetrically turned on in dorsal cells that still express DmDll, in the early third instar leg disc. Thus, in both insects, the overlapping Dll/dac domain is observed transiently and subsequently resolves into two domains; proximally Gbdac and distally Gbdac/GbDll in Gryllus, and proximally Dmdac and distally DmDll in Drosophila. At stages 8-9, the GbDll expression fades in the intermediate portion of the leg, and this leads to a new proximal boundary of GbDll. This boundary is correlated with the second segmentation of the femur/distal telopodite. No corresponding expression boundary has been reported in the Drosophila leg imaginal disc. In Drosophila, the expression boundary between the DmDll/Dmdac domain and distal DmDll domain corresponds to that between tarsal segments 1 and 2. Future tarsal segment 1 may be generated in the distal-most region of the Dmdac expression domain. In Gryllus, the distal limit of the Gbdac domain is likely to correspond to the boundary between tarsal segments 1 and 2 (Inoue, 2002).

From these results it was found that the expression patterns of the three genes are essentially conserved between Drosophila and Gryllus, although the time course of the pattern varies according to the developmental mode. The following conclusions were reached: (1) at early stages, expressions of Gbhth and GbDll do not correspond to the future coxopodite and telopodite, but rather to the presumptive trochanter/femur boundary; (2) Gbdac expression subdivides the leg bud into the presumptive femur and more distal region; (3) the expression patterns of GbDll, Gbdac, and Gbhth in the fully segmented Gryllus leg are similar to those in the Drosophila late third instar disc (Inoue, 2002).

Evolutional changes in homeotic gene functions have contributed to segmental diversification of arthropodan limbs, but crucial molecular changes have not been identified to date. The first leg of the crustacean Daphnia lacks a prominent ventral branch found in the second to fourth legs. This phenotype correlates with the loss of Distal-less and concomitant expression of Antennapedia in the limb primordium. Unlike its Drosophila counterpart, Daphnia Antennapedia represses Distal-less in Drosophila assays, and the protein region conferring this activity was mapped to the N terminal region of the protein. The results imply that Dapnia Antennapedia specifies leg morphology by repressing Distal-less, and this activity was acquired through a change in protein structure after separation of crustaceans and insects (Shiga, 2002).

The genes Distal-less, dachshund, extradenticle, and homothorax have been shown in Drosophila to be among the earliest genes that define positional values along the proximal-distal (PD) axis of the developing legs. In order to study PD axis formation in the appendages of the pill millipede Glomeris marginata, homologs of these four genes were isolated and their expression patterns examined. In the trunk legs, there are several differences from Drosophila, but the patterns are nevertheless compatible with a conserved role in defining positional values along the PD axis. However, their role in the head appendages is apparently more complex. Distal-less in the mandible and maxilla is expressed in the forming sensory organs and, thus, does not seem to be involved in PD axis patterning. No components of mouthparts could be identified that are homologous to the distal parts of the trunk legs and antennnae. Interestingly, there is also a transient premorphogenetic expression of Distal-less in the second antennal and second maxillary segment, although no appendages are eventually formed in these segments. The dachshund gene is apparently involved both in PD patterning as well as in sensory organ development in the antenna, maxilla, and mandible. Strong dachshund expression is specifically correlated with the tooth-like part of the mandible, a feature that is shared with other mandibulate arthropods. homothorax is expressed in the proximal and medial parts of the legs, while extradenticle RNA is only seen in the proximal region. This overlap of expression corresponds to the functional overlap between extradenticle and homothorax in Drosophila (Prpic, 2003a).

In order to reconstruct the evolution of the gene expression pattern, the expression of Dll homologs has been studied in a great number of different arthropod species. This study is the first detailed account of Dll expression in a myriapod and the first report on expression of dac, hth, and exd in a myriapod. The data on Dll presented here show that, in contrast to the antenna and trunk legs, in the two gnathal appendages, mandible and maxilla, no Gm-Dll expression can be found that can be confidently correlated with PD axis formation. This implies that the Glomeris mouthparts lack elements serially homologous to the distal Gm-Dll-expressing parts of the antenna and legs. Thus, both mandible and maxilla are gnathobasic in nature and lack distal parts (telopodite, palp). A gnathobasic mandible traditionally is considered as a synapomorphy of crustaceans, insects, and myriapods, uniting them into a monophyletic Mandibulata. A gnathobasic maxilla, however, is found only in myriapods [apart from Glomeris (diplopod) also the maxilla of chilopods, symphylans, and pauropods apparently lacks a telopodite, to judge from external morphology] and adults of several crustacean species. The maxilla in insects and most crustaceans is not gnathobasic and has a palp (telopodite). This demonstrates that the gnathobasic condition of an appendage can evolve independently by convergence and thus questions the homology of the mandibular gnathobasy in crustaceans, insects, and myriapods. Indeed, recent molecular phylogenies corroborate two monophyletic groups comprising chelicerates and myriapods in one case and crustaceans and insects in the other. This would argue for an independent origin of the gnathobasic mandible in myriapods and the so-called Tetraconata (insects and crustaceans). However, the results with Gm-dac presented in this study support the homology of the gnathobasy in the mandibles of crustaceans, insects, and myriapods. It has been shown previously that expression of dac is very strong in the mandibles of crustaceans and insects. Similarly, strong expression of Gm-dac is seen in the mandible of Glomeris. The role of the strong dac expression in the development of the mandible is unclear, but it seems possible to correlate it with the specific tooth-like morphology of this appendage. Thus, a common genetic mechanism appears to underlie the gnathobasic nature of the mandibles in myriapods, crustaceans, and insects. In contrast to this, no comparable dac expression has been found in the first locomotory leg (L1) in chelicerates. The L1 segment in chelicerates corresponds to the mandibular segment in the other arthropods (Prpic, 2003a).

In summary, the results presented here support earlier reports that the mandible in myriapods is indeed gnathobasic. Moreover, the results indicate that the maxilla is also gnathobasic, which is different from insects and most crustaceans and also from chelicerates, which do not possess a single gnathobasic appendage. Strong expression of dac in the mandible in crustaceans, insects, and myriapods suggests that the mandibular morphology is produced by a homologous mechanism and supports the homology of the mandible in all mandibulate arthropods. Although a homologous mandible in crustaceans, insects, and myriapods would support the monophyly of the Mandibulata, it has to be pointed out that a homologous mandible is also compatible with recent molecular phylogenies on the assumption that a true mandible already existed in the last common ancestor of all extant arthropod classes and has been lost secondarily in the chelicerates (Prpic, 2003a).

Leg development in Drosophila has been studied in much detail. However, Drosophila limbs form in the larva as imaginal discs and not during embryogenesis as in most other arthropods. Appendage genes have been analyzed in the spider Cupiennius salei and the beetle Tribolium castaneum. Differences in decapentaplegic expression suggest a different mode of distal morphogen signaling suitable for the specific geometry of growing limb buds. Also, expression of the proximal genes homothorax and extradenticle (exd) is significantly altered: in the spider, exd is restricted to the proximal leg and hth expression extends distally, while in insects, exd is expressed in the entire leg and hth is restricted to proximal parts. This reversal of spatial specificity demonstrates an evolutionary shift, which is nevertheless compatible with a conserved role for this gene pair as instructor of proximal fate. Different expression dynamics of dachshund and Distal-less point to modifications in the regulation of the leg gap gene system. The significance of this finding is discussed in terms of attempts to homologize leg segments in different arthropod classes. Comparison of the expression profiles of H15 and optomotor-blind to the Drosophila patterns suggests modifications also in the dorsal-ventral patterning system of the legs. Together, these results suggest alterations in many components of the leg developmental system, namely proximal-distal and dorsal-ventral patterning, and leg segmentation. Thus, the leg developmental system exhibits a propensity to evolutionary change, which probably forms the basis for the impressive diversity of arthropod leg morphologies (Prpic, 2003b).

The Drosophila genes wingless and decapentaplegic comprise the top level of a hierarchical gene cascade involved in proximal-distal (PD) patterning of the legs. It remains unclear, whether this cascade is common to the appendages of all arthropods. Here, wg and dpp are studied in the millipede Glomeris marginata, a representative of the Myriapoda. Glomeris wg (Gm-wg) is expressed along the ventral side of the appendages compatible with functioning during the patterning of both the PD and dorsal-ventral (DV) axes. Gm-wg may also be involved in sensory organ formation in the gnathal appendages by inducing the expression of Distal-less (Dll) and H15 in the organ primordia. Expression of Glomeris dpp (Gm-dpp) is found at the tip of the trunk legs as well as weakly along the dorsal side of the legs in early stages. Taking data from other arthropods into account, these results may be interpreted in favor of a conserved mode of WG/DPP signaling. Apart from the main PD axis, many arthropod appendages have additional branches (e.g., endites). It is debated whether these extra branches develop their PD axis via the same mechanism as the main PD axis, or whether branch-specific mechanisms exist. Gene expression in possible endite homologs in Glomeris argues for the latter alternative. All available data argue in favor of a conserved role of WG/DPP morphogen gradients in guiding the development of the main PD axis. Additional branches in multibranched (multiramous) appendage types apparently do not utilize the WG/DPP signaling system for their PD development. This further supports recent work on crustaceans and insects, that lead to similar conclusions (Prpic, 2004).

One important pattern element on butterfly wings, the eyespot, arises through the extablishment of an inductive organizing center (the focus) at a specific location on the developing wing. Transplantation of the focus can induce surrounding host cells to form an eyespot in an ectopic site. The focus is proposed to be a signaling source for a morphogen, the levels of which determine the pigmentation of surrounding cells. Distal-less is specifically expressed in the eyespot focus. In the imaginal discs of early fifth-instar larvae, Dll is expressed in a broad distal band and at high levels in stripes running perpendicular from the distal band, down the middle of each subdivision. At the proximal tips of those stripes, well separated in distance from the distal band but attached to it by a stripe of Dll expressing cells, the Dll pattern begins to enlarge. By the late fifth-instar stage in each subdivision of the hindwing imaginal disc, the proximal end of the stripes have resolved into stable circular spots of Dll expression. In the early pupa, strong Dll expression expands to encompass a broader circular field of the scale-forming cells, which are now aligned in rows and protrude above the other epithelial cells. It is thought that this expanded domain reflects a response to signaling from the epithelial focus, which remains visible at the center of the field of Dll-expressing cells. Dll spots must include the inductive focus, because grafts containing them induce eyespots. The positions of these spots of Dll expression correspond to the central regions of the eyespots that form several days later when scale color differentiation occurs (Brakefield, 1996).

The morphological and functional evolution of appendages has played a critical role in animal evolution, but the developmental genetic mechanisms underlying appendage diversity are not understood. Given that homologous appendage development is controlled by the same Hox gene in different organisms, and that Hox genes are transcription factors, diversity may evolve from changes in the regulation of Hox target genes. Two impediments to understanding the role of Hox genes in morphological evolution have been the limited number of organisms in which Hox gene function can be studied and the paucity of known Hox-regulated target genes. An analysis was carried out of Hindsight, a butterfly homeotic mutant in which portions of the ventral hindwing pattern are transformed to ventral forewing identity, and the regulation of target genes by the Ultrabithorax (Ubx) gene product was compared in Lepidopteran and Dipteran hindwings. Ubx gene expression is lost from patches of cells in developing Hindsight hindwings, which correlates with changes in wing pigmentation, color pattern elements, and scale morphology. This mutant was used to study how regulation of target genes by Ubx protein differs between species. Several Ubx-regulated genes in the Drosophila haltere are not repressed by Ubx in butterfly hindwings, but Distal-less (Dll) expression is regulated by Ubx in a unique manner in butterflies. In Hindsight hindwings, in which patches of cells that lack Ubx protein expression encompass a portion of the eyespot focus, Dll expression clearly increases, as compared with that found in wild-type hindwings. Outside of these patches, where Ubx expression is 'normal' in the eyespot field, Dll is expressed at very low levels in a cell-autonomous fashion. These results suggest that hindwing eyespot size may be controlled by Ubx at two steps in the eyespot developmental pathway: (1) Ubx depresses the production of the focal signal, which is relieved when a portion of the focus loses Ubx expression; (2) Ubx affects the response of genes that are downstream of the focal signal, as for example, Dll. Because the eyespot pattern element has no counterpart in other insect orders, it is deduced that Ubx regulation of eyespot patterning genes must have evolved within the Lepidoptera. It is concluded that the morphological diversification of insect hindwings has involved the acquisition of different sets of target genes by Ubx in different lineages (Weatherbee, 1999).

Drosophila Serum response factor (blistered), Achaete-Scute Complex, and wingless are repressed in Drosophila halteres. Portions of the expression pattern of Lepidopteran homologs of these genes are not repressed in butterfly hindwings. Unlike the expression patterns of the homologous genes in halteres, butterfly wg is not repressed along the posterior margin in the hindwing, nor is butterfly SRF repressed in intervein regions, and the AS-C homologs are not repressed in cells flanking the dorsal-ventral boundary. These differences in the regulation of wg, SRF and AS-C between Drosophila halteres and butterfly hindwings suggest that these genes became repressed by Ubx when an ancestral hindwing evolved into a haltere in the dipteran lineage, with a concomitant reduction of appendage size, loss of margin bristles, and changes in shape. Two additional exampes of Ubx-regulated differences in gene expression between fly and butterfly flight appendages were found. (1) wg is expressed in two stripes in butterfly forewings that roughly correspond to the future location of the proximal band elements. This protein of the wg pattern is absent from butterfly hindwings and has not counterpart in flies and represents a novel feature regulated by Ubx in butterflies. (2) Dll is expressed along the margin of both butterfly wings and the Drosophila forewing, but this expression is modified in halteres and may be regulated by Ubx. Changes in Hox-regulated target gene sets are, in general, likely to underlie the morphological divergence of homologous structures between animals (Weatherbee, 1999).

The color patterns decorating butterfly wings provide ideal material to study the reciprocal interactions between evolution and development. They are visually compelling products of selection, often with a clear adaptive value, and are amenable to a detailed developmental characterization. Research on wing-pattern evolution and development has focused on the eyespots of the tropical butterfly Bicyclus anynana. There is quantitative variation for several features of eyespot morphology but the actual genes contributing to such variation are unknown. However, studies of gene expression patterns in wing primordia have implicated different developmental pathways in eyespot formation. To link these two sets of information it is necessary to identify which genes within the implicated pathways contribute to the quantitative variation accessible to natural selection. This study begins to bridge this gap by demonstrating linkage between DNA polymorphisms in the candidate gene Distal-less (Dll) and eyespot size in B. anynana (Beldade, 2002).

Artificial selection was used to establish lines that differed in the size of the two dorsal forewing eyespots of B. anynana. After nine generations of selection, the lines show markedly different phenotypes; the 'high' line has large eyespots and the 'low' line, small eyespots. The rapid and gradual response to selection indicates that there is substantial additive genetic variance for eyespot size. Realized heritabilities of around 0.6-0.7 are comparable to previous estimates for the size of the posterior eyespot. The selected butterflies also show quantitative differences in Dll expression: wing discs of the high line have considerably larger areas of Dll expression around the location of the center of the presumptive eyespots. These differences are already apparent in late final instar larvae, and more marked in pupal wing discs. The changes in Dll expression might be caused by DNA poly-morphisms in Dll itself or in upstream regulators of Dll. To distinguish between the hypotheses of cis versus trans polymorphisms affecting Dll expression, a series of crosses were made using the high and low selection lines. The progeny from a backcross between a hybrid butterfly and each of the parental lines have either both Dll alleles from the same origin or one allele from each parental line. To distinguish Dll alleles from high and low origin a B. anynana Dll homolog was cloned; this showed 99% amino acid identity with the published Precis coenia Dll. 1,250 base pairs (bp) of this gene were sequenced in individuals from the selection lines, and 27 segregating sites were identified (corresponding to levels of DNA polymorphism comparable to Drosophila). Two of these polymorphisms were used to distinguish high (H) and low (L) Dll alleles and to genotype individuals from the crosses. The high backcross produced progeny of genotype HH or HL, and the low backcross produced HL or LL. Eyespot sizes of butterflies from distinct genotypic classes were compared to test for an association between Dll genotype and that phenotype (Beldade, 2002).

High and low Dll alleles segregate with eyespot size in the backcrosses. Significant differences were found between the size of the posterior eyespot of HH versus HL backcross females and for both the anterior and posterior eyespots in HL versus LL males. The results show that there are differences in the estimated allelic effects between males and females and between the two target eyespots. Work in Drosophila and other organisms has frequently detected this type of sex-specific effect and correlated character-specific effects. Since the identified Dll-linked factor explains only part of the difference between selection lines, other genes must have been involved in the response to selection (for ventral eyespot size at least five loci were implicated). These genes may include upstream trans-regulators of Dll (Beldade, 2002).

Segregation of a molecular marker with eyespot size is evidence of linkage but does not necessarily implicate the marker site as directly contributing to the trait. The phenotypic differences between back-cross genotypic marker classes depend on the effect of the quantitative trait locus (QTL) linked to the marker site and on the recombination fraction between the two. Although an interval mapping approach is necessary to distinguish between the effect of a QTL and its location, the log odds (LOD) score for a putative QTL can be calculated some recombination fraction away from the markers in Dll. For both the posterior and anterior eyespots in males from the low backcross, the LOD score was maximized at Dll, although a 1 LOD support interval indicates that the QTL could be as far as 40 cM from Dll. The effect detected for the posterior eyespot in females from the high back-cross might be due to the same QTL that affects males (its LOD score did not peak at Dll but the value at Dll was not significantly lower than its maximum value) or to a second QTL linked to, but not at Dll. A QTL at Dll is the most parsimonious explanation for the observed data, although a linked QTL cannot be ruled out. Since B. anynana is thought to have approximately 26 linkage groups, linkage to Dll rules out a large number of other candidate genes. Mapping of the QTL to one of 26-29 linkage groups localizes it to about 4% of the genome. To improve this resolution and locate the causative sites within Dll, future work will use population-based association studies (Beldade, 2002).

In butterflies there is a class of 'intervein' wing patterns that have lines of symmetry halfway between wing veins. These patterns occur in a range of shapes, including eyespots, ellipses, and midlines, and were proposed to have evolved through developmental shifts along a midline-to-eyespot continuum. Notch (N) upregulation, followed by activation of the transcription factor Distal-less (Dll), is an early event in the development of eyespot and intervein midline patterns across multiple species of butterflies. A relationship between eyespot phenotype and N and Dll expression is demonstrated in a loss-of-eyespot mutant in which N and Dll expression is reduced at missing eyespot sites. A phylogenetic comparison of expression time series from eight moth and butterfly species suggests that intervein N and Dll patterns are a derived characteristic of the butterfly lineage. Furthermore, prior to eyespot determination in eyespot-bearing butterflies, N and Dll are transiently expressed in a pattern that resembles ancestral intervein midline patterns. In this study N upregulation is established as the earliest known event in eyespot determination, gene expression associated with intervein midline color patterns is demostrated, and molecular evidence is provided that wing patterns evolved through addition to and truncation of a conserved midline-to-eyespot pattern formation sequence (Reed, 2004).

Butterfly eyespots provide a prime example of how a novel character system may arise through the evolutionary recruitment of developmental genes and then diversify under the influence of natural selection. During development, eyespot pigment patterns are induced by a long-range signal that originates from a group of focal cells at the center of the eyespot. In late last-instar wing discs, several molecules normally associated with axis specification are expressed in focal cells, and it is proposed that these molecules are involved with activating the focal signal. Of these focal molecules, the transcription factor Dll is of particular interest because the geneencoding it is genetically linked to eyespot size. While gene expressi on studies have provided insight into eyespot development, little is known about how gene expression may be associated with the evolution and development of noneyespot patterns (Reed, 2004).

Intervein pattern elements, those with centers of symmetry halfway between wing veins, occur in a range of shapes, including eyespots, tapered ellipses, and midlines, with gradients of intermediate shapes occurring both within and between species. Based on these adult phenotypes, it has been proposed that midline patterns are developmental precursors of circular eyespot patterns and that the observed gradient of intervein pattern morphologies can be explained by evolutionary changes in the timing of a common underlying developmental process. The observation that Dll expression passes through an intervein midline stage before forming an eyespot focus increases interest in this idea; however, there has been little comparative work done to test the molecular basis of this model across species. In this study the expression of Dll and the receptor molecule N in a variety of butterflies and moths were compared in order to explore the relationship between prepattern regulation and the evolution of wing patterns (Reed, 2004).

It was hypothesized that the N signaling pathway may be an upstream component of the focal determination process because ectopic expression of activated N in Drosophila melanogaster imaginal discs is sufficient to cause expression of Dll. N is a membrane bound receptor that plays several roles during Drosophila wing development. Its functions that are best understood in this context include defining the dorsoventral boundary and defining intervein tissue via a lateral inhibition interaction with its ligand Delta. In pupal butterfly wings, there is evidence that N-mediated lateral inhibition may be involved with organizing wing scales. During a lateral inhibition process, N expression tends to increase over time as a result of a local positive-feedback mechanism (Reed, 2004).

To test for an association between N and Dll expression and eyespots, the localization of N and Dll was examined in late last-instar wing imaginal discs of three species of eyespot-bearing nymphalid butterflies: Vanessa cardui, Junonia (Precis) coenia, and Bicyclus anynana. In all three of these species there is a perfect correlation between presence of forewing and hindwing eyespots and late last-instar N and Dll focal expression. To further characterize the relationship between N and Dll expression and eyespot phenotype, the effects of the B. anynana eyespot mutant missing were ascertained on levels of N and Dll. missing greatly reduces or eliminates two specific eyespots from the hindwing. In missing mutants, focal N and Dll expression is observed at lower levels than at corresponding positions in wild-type wings. N and Dll accumulation must therefore be regulated directly or indirectly by the missing locus, and localized inductions of N and Dll may serve as markers for the process of eyespot focus determination at a point downstream of missing activity (Reed, 2004).

To determine the spatiotemporal relationship between N and Dll in focal determination, a time series of N/Dll double stains was produced from V. cardui, J. coenia, and B. anynana. Focal N upregulation was found to precede Dll activation with a lag time of approximately 1.5 stages (equivalent to 12 to 24 hr, depending on species, temperature, and individual variation). In all three eyespot-bearing species, stages of gene expression were observed where N was upregulated in discrete focal patterns, while Dll was upregulated only in intervein midlines. The spatiotemporal relationship between N and Dll may be outlined in the four following primary phases. (1) Margin and intervein expression. N expression occurs at moderate levels across the wing disc, except for in the presumptive veins where N expression is relatively low. Early during this phase, Dll expression occurs only along the wing margin but progressively moves proximally after the upregulation of N along the intervein midline. (2) Midline prepatterning. Between the developing wing veins, N is upregulated in an intervein midline pattern along with an accompanying midline of Dll expression. In most species Dll expression in the midline tends to be more discretely focused than N. (3) Focal determination. N expression is increased in foci, which is subsequently mirrored by Dll. (4) Focal maturation. N and Dll express strongly in foci while fading sequentially from the intervein midline. During the focal maturation phase, expression of genes in the hedgehog pathway have been observed in foci of J. coenia and B. anynana (Reed, 2004).

Eyespot patterns have been gained and/or lost multiple times throughout butterfly evolution, and eyespots are even seen in some moths. While a rigorous phylogenetic treatment is required to infer the specific pattern of eyespot evolution throughout the Lepidoptera, some questions many nevertheless be answered about eyespot evolution by using selected exemplar taxa. Specifically, it was of interest to determining if a secondary loss of eyespots in a lineage is associated with a change in the N/Dll prepatterning process. To address this two species were examined from the nymphalid subfamily Heliconiinae: Agraulis vanillae and Heliconius melpomene. These species belong to a monophyletic subtribe called the Heliconiiti, in which there are no obvious eyespot-bearing species, although intervein midline patterns are common throughout the group. Eyespots are found in non-Heliconiiti heliconiines, as well as throughout the rest of the Nymphalidae, suggesting that the Heliconiiti represent a secondary loss of eyespots (Reed, 2004).

In A. vanillae the margin and intervein expression and midline prepattern phases appear similar to those in other butterflies. In this species, however, midline definition occurs relatively slowly, and development only reaches the midline prepattern phase by pupation. Furthermore, intervein midline gene expression was not observed to fade as it does during focal maturation in eyespot-bearing species. The midline expression patterns correspond with orange midline pigment patterns on both the hindwing and forewing (Reed, 2004).

In last-instar H. melpomene, N and Dll are expressed in an intervein midline pattern similar to A. vanillae, although they are extended proximally. There is an association between gene expression and pigment pattern in H. melpomene, where a recessive gene from the Ecuadorian race plesseni reveals a melanic midline pattern in the forewing that matches N and Dll expression. It is notable that even though expressivity of the intervein midline pigment pattern varies throughout the genus Heliconius, the N/Dll expression pattern appears to be similar between species both bearing and lacking these patterns (including H. cydno, H. erato, and H. hecale). These observations suggest that N and Dll expression is, in itself, not sufficient for midline pigment patterns in Heliconius (Reed, 2004).

In order to gain insight into the origin of the N/Dll prepatterning process, the expression patterns of these proteins were determined in the outgroup pierid butterfly Pieris rapae and two 'higher' (ditrysian) moths: the sphingid Manduca sexta, and the gelichiid Pectinophora gossypiella (Reed, 2004).

A time series of N/Dll stains in P. rapae resembles the time series from A. vanillae in that N and Dll form persisting intervein midline patterns. Interestingly, however, P. rapae does not display a midline pigment pattern in the adult. Midline pigment patterns are found in many pierid species, suggesting that as with Heliconius, N/Dll midline expression may be associated with, but is not sufficient for, pigment midlines in adults. No gene expression was observed associated with the black chevrons on the P. rapae forewing (Reed, 2004).

In the moths M. sexta and P. gossypiella, early N and Dll expression resembles initial margin and intervein expression in butterflies. In late stage M. sexta wing discs, Dll forms vaguely defined proximal extensions along the wing veins themselves. In P. gossypiella no expression of Dll was observed in intervein tissue. Given the species sampling in this study, it is most parsimonius to infer that the N/Dll intervein midline originated sometime after the divergence of the sphingid lineage and before the divergence of the pierid lineage. Further sampling of gene expression patterns from basal butterfly families and moth outgroups would help clarify the point of origin of the midline prepattern. It should be noted that published expression patterns for the monarch Danaus plexippus and the B. anynana Cyclops mutant do not show Dll midline expression; however, gene expression time series have not been described from either of these species, so it remains unknown if midlines may be expressed earlier or later than the sampled time points (Reed, 2004).

Gene expression data was mapped onto a phylogeny of the Lepidoptera used in this study and several conclusions were drawn regarding the evolution of intervein and focal prepatterning in butterfly wings. (1) The temporal order of gene expression states is conserved in all the taxa examined. (2) As outlined above, the formation of a discrete intervein midline appears to be a synapomorphy of the butterfly lineage. What, then, can be concluded about the origin of focal gene expression? Given the species sampling, there are two equally parsimonious hypotheses for the evolution of focal expression patterns: (1) there were two independent origins of foci in the lineages leading to the Satyrinae and Nymphalinae, respectively, or (2) there was a gain of foci in the lineage leading to the Nymphalidae and a loss of foci in the lineage leading to the Heliconiiti. Although a greater species sampling is required to rigorously distinguish between these possibilities, at this point the latter model is favored because of (1) the rarity or absence of eyespots (i.e., concentric circular or oval pigment patterns consistent with a focal induction model) among the basal butterfly families Pieridae, Papilionidae, and Hesperiidae, and (2) the occurrence of putative inductive eyespots in basal heliconiine genera such as Vindula and possibly Cethosia. It would be of great interest to determine the expression patterns of N and Dll in eyespot-bearing lepidopteran lineages not closely related to the Nymphalidae, such as the papilionid genus Parnassius or the eyespot-bearing saturniid and sphingid moths. These lineages potentially represent origins of inductive eyespot patterns evolutionarily independent from the nymphalid clade, and studying them could provide insight into the developmental basis of parallel pattern evolution (Reed, 2004).

The gene expression patterns reported provide useful markers for the wing pattern-formation process; however, it remains unknown what the developmental significance of the observed prepatterning sequence is. It is striking that in the eyespot-bearing butterflies examined, midline gene expression occurs prior to focal determination and that midlines always terminate proximally at the eyespot foci. These observations suggest a noncoincidental relationship between formation of midlines and foci; however, with the current data it cannot be determined if the midline/focus relationship is causal or if these prepatterns are both downstream of an as-of-yet unknown coordinate system (Reed, 2004).

The data presented in this study establish N upregulation as the earliest known event in the development of butterfly eyespots. Furthermore, finding that eyespots and midlines share a similar prepatterning process supports earlier modelsthat these intervein pattern elements are developmentally related. The observation in eyespot-bearing species that N and Dll pass through a transient, and apparently ancestral, phase of midline expression prior to focal determination raises the possibility that this developmental sequence represents a kind of evolutionary heterochrony at the level of molecular pattern formation. In sum, the data illustrate how a discrete morphological character may evolve through temporal changes in a conserved molecular pattern-formation process (Reed, 2004).

Specialized insect mouthparts, such as those of Drosophila, are derived from an ancestral mandibulate state, but little is known about the developmental genetics of mandibulate mouthparts. The metamorphic patterning of mandibulate mouthparts of the beetle Tribolium castaneum was studied RNA interference to deplete the expression of 13 genes involved in mouthpart patterning. These data were used to test three hypotheses related to mouthpart development and evolution. First, the prediction was tested that maxillary and labial palps are patterned using conserved components of the leg-patterning network. This hypothesis was strongly supported: depletion of Distal-less and dachshund led to distal and intermediate deletions of these structures while depletion of homothorax led to homeotic transformation of the proximal maxilla and labium, joint formation required the action of Notch signaling components and odd-skipped paralogs, and distal growth and patterning required epidermal growth factor (EGF) signaling. Additionally, depletion of abrupt or pdm/nubbin caused fusions of palp segments. Second, the hypotheses was tested for how adult endites, the inner branches of the maxillary and labial appendages, are formed at metamorphosis. The data reveal that Distal-less, Notch signaling components, and odd-skipped paralogs, but not dachshund, are required for metamorphosis of the maxillary endites. Endite development thus requires components of the limb proximal-distal axis patterning and joint segmentation networks. Finally, adult mandible development is considered in light of the gnathobasic hypothesis. Interestingly, while EGF activity is required for distal, but not proximal, patterning of other appendages, it is required for normal metamorphic growth of the mandibles (Angelini, 2012).

In D. melanogaster, Dll mutants lack maxillary structures and portions of the proboscis (i.e., labium), although Dll expression in the maxillary anlagen is weaker than in the leg or antennal discs. Paralleling the results for T. castaneum, in the horned beetle Onthophagus taurus distal regions of the adult mouthparts were deleted with larval Dll RNAi (Simonnet 2011). The embryonic and metamorphic functions of Dll in T. castaneum are also similar: the gene is required for the development of distal structures at both stages, and during embryogenesis Dll is expressed throughout the developing palps. Interestingly, removal of T. castaneum Dll expression earlier during larval life led to delayed metamorphosis, as well as changes in appendage morphology (Suzuki, 2009). Many insects delay molting after appendage loss to allow time for regeneration, and this dual role of Dll suggests a mechanism linking these processes (Angelini, 2012).

The data from T. castaneum provide evidence for a conserved gap gene role of dac during patterning of mouthparts and legs of this species. dachshund is not expressed in or required for development of the labial and maxillary anlagen of D. melanogaster. In T. castaneum embryos dac is expressed strongly in the proximal maxilla and part of the developing endite. Embryonic dac expression is weaker in the distal maxillary palp and the labium. The current data show a clear metamorphic requirement for dac in the intermediate regions of the maxillary and labial palps, as does a recent study of O. taurus (Simonnet, 2011). A function for dac in the development of an intermediate portion of the maxillary and labial appendages has so far only been observed in these two beetles, while data from two species with specialized mouthparts (the milkweed bug O. fasciatus and D. melanogaster) found that dac is not required for PD patterning of the mouthparts. Thus, comparative data from other species do not support the hypothesis that this mouthpart patterning role is ancestral. However, if mandibulate mouthparts evolved from leg-like structures similarities in the expression and function of genes patterning both legs and mouthparts are expected to be plesiomorphic. This hypothesis can be further tested by examining the role of dac in mouthpart development in additional insect orders, particularly those that retain mandibulate mouthparts, and in other arthropods (Angelini, 2012).

The effects of hth depletion are distinct in different species, but typically involve some degree of homeotic transformation. In D. melanogaster, hth is expressed in the labial discs, but without nuclear expression of its cofactor Extradenticle. Maxillary palps are retained in hth loss-of-function flies, but they may possess bristles typical of legs, indicating a partial proboscis-to-leg transformation. In the cricket Gryllus bimaculatus, which has mandibulate mouthparts, hth depletion causes transformation of proximal mouthpart structures towards antennal identity, with a loss of endites, while distal structures are transformed towards leg identity (Ronco, 2008). hth RNAi in T. castaneum transformed intermediate regions of the maxilla and labium towards distal mouthpart identity. Proximal regions also appeared transformed, but their identity could not be established, while distal regions appeared wild type. In the beetle O. taurus, proximal regions of the labium are transformed towards maxillary endite identity, but distal regions of the labium and the entire maxilla remain relatively unaffected (Angelini, 2012).

These results highlight the similarity between patterning of the maxilla, labium and legs in T. castaneum. Functional data from two species with highly derived mouthpart morphologies, D. melanogaster and the milkweed bug Oncopeltus fasciatus, suggest only limited similarity between mouthpart and leg patterning. One explanation for this low degree of conservation is that evolution of the ancestral patterning mechanism has occurred in concert with the functional and morphological diversification of these mouthparts. A correlation between generative mechanisms and structural morphology has been used as a common null hypothesis, although exceptions in which similar morphologies result from different developmental pathways are documented. Nevertheless, this hypothesis predicts that developmental patterning should be more highly conserved across appendage types in species that retain the ancestral mandibulate mouthpart morphology (Angelini, 2012).

The maxillary and labial palps are an interesting case of serial homology. Despite a difference in overall size, their shape and arrangement of sensillae are similar. The intermediate segments of each palp type are also similar, but differ in number, which suggests that segment number is regulated independently from other morphological traits. The RNAi depletion of pdm in T. castaneum caused the reduction and deletion of the third maxillary palp segment, producing a phenotype closely resembling the wildtype morphology of the labial palps. While a role for pdm in the labium cannot be excluded, the absence of observed labial phenotypes was significant compared to maxillary results. Therefore, it is hypothesized that the difference in the number of palp segments results from specific activation of pdm in the maxillary palp. Loss of function in the Hox gene Deformed during T. castaneum embryogenesis causes a transformation of the larval maxillae towards labial identity. Since Hox genes are the primary determinants of body segment identity, it is proposed that pdm is activated by Deformed, and repressed by the labial Hox gene Sex combs reduced. RNAi targeting pdm in another mandibulate insect, the cricket Acheta domesticus, generated defects in the antenna and legs, but no defects in the mouthparts, despite similar pdm expression in these appendages (Turchyn; 2011; Angelini, 2012).

Endites are a primitive component of arthropod appendages, and they are retained in insect mouthparts, as well as in the mouthparts and thoracic appendages of many crustaceans (Boxshall 2004). At least three hypotheses have been put forward for how endites are patterned, and these hypotheses are not mutually exclusive. The first hypothesis states that multiple PD axes result from redeployment of a PD axis patterning mechanism shared by palps and endites. A second hypothesis posits that endites and appendage segments form by the same mechanism, Notch-mediated in-folding of the cuticle. A third hypothesis states that dac expression initiates endite branching from the main appendage axis. The axis redeployment hypothesis predicts that depletion of genes involved in PD axis patterning will have similar effects on the development of palps and endites. Some support for this hypothesis comes from studies of endite morphogenesis and the expression and function of leg gap genes in the embryos of T. castaneum and the orthopteran Schistocerca americana, but not all data are consistent with it. The segmentation hypothesis predicts that endites will fail to differentiate if genes required for joints are depleted. This hypothesis was posed based on a comparative developmental study of segmented and phyllopodous crustacean limbs. Finally, the dac-mediated hypothesis predicts that depletion of dac will lead to reduced endites. This hypothesis emerged from the observation that dac expression is reiterated along the medial edges of larval endites in the crustacean Triops longicaudatus. Comparative expression data from the isopod Porcellio scaber are also consistent with the dac-mediated hypothesis (Angelini, 2012).

The current data are consistent with predictions of the axis redeployment and segmentation hypotheses but do not support a role for dac in endite metamorphosis. Adult endites were disrupted by depletion of Dll, Krn, the odd-related genes, and Notch signaling, and to a lesser degree hth. In the maxilla depletion of most of these genes led to the failure of the single larval endite to divide into two distinct branches, while in the labium, their depletion caused reduction of the ligula. Their requirement in the endites is consistent with the hypothesis that these structures are generated by redeploying appendage PD axis determinants. Depletion of Notch signaling components and the odd paralogs produced reductions and fusions between palp segments, between the palps and endites, and between the lacinia and galea. Thus, these data are compatible with both the hypothesis that a reiterated PD axis is used to pattern the endites and the hypothesis that endite formation is linked to joint formation. Normal endite development in dac-depleted specimens is inconsistent with the dac-mediated hypothesis (Angelini, 2012).

It is noteworthy that endite specification and the division of the single larval endite into the adult galea and lacinia appear to be separable functions. For example, Ser RNAi resulted in a single endite lobe with lacinia identity medially and galea identity laterally. In contrast, severe Dll RNAi individuals had a single endite that lacked also obvious lacinia identity (Angelini, 2012).

The mandibulate structure of Tribolium mouthparts is the pleisomorphic state for insects and is shared by a majority of insect orders. These mouthparts are characterized by robust mandibles, lacking segmentation. A classic debate in arthropod morphology concerns whether the mandibles of insects and myriapods are derived from a whole appendage or only from proximal appendage regions; the latter are called gnathobasic mandibles. Palps are retained on the mandibles of many crustaceans, making it clear that the biting regions of their mandibles are gnathobasic. Phylogenetic support for the gnathobasic hypothesis comes from phlyogenetic studies that place insects nested within crustaceans (Regier, 2010). The first developmental genetic support for the gnathobasic hypothesis came from the discovery that insect mandibles lack Dll expression. Furthermore, neither mutations in Dll nor its depletion through RNAi have been observed to alter mandible development in insects, including T. castaneum. This evidence has led to widespread acceptance of the gnathobasic hypothesis. Of the 13 genes depleted in this study, two (Krn and hth) produced results that would not be predicted by the most straightforward form of the gnathobasic hypothesis for mandible origins (Angelini, 2012).

Loss of EGF function in insects leads to distal appendage defects, including pretarsal or tarsal deletions. The role of EGF signaling in distal appendage regions is conserved in T. castaneum metamorphosis, since depletion of the EGF ligand Krn leads to reduction of the antennal flagellum, and maxillary and labial palps, as well as to deletion of the pretarsus and malformation of the tarsus. In light of the restriction of Krn’s role to distal appendage regions and regulation of distal EGF ligand expression by Dll in D. melanogaster, the gnathobasic hypothesis predicts that Krn should not be required for normal development of the mandible in T. castaneum. In contrast to this prediction, depletion of Krn produced a significant reduction in mandible length (Angelini, 2012).

The hypothesis of a gnathobasic mandible also predicts that hth depletion should produce effects in the mandible similar to those in the proximal regions of other appendage types. In T. castaneum, hth RNAi during metamorphosis caused homeotic transformation of proximal regions of the maxilla, labium and legs. However, the mandibles were not affected by hth depletion. In the beetle O. taurus, hth depletion slightly altered mandible shape, but also without apparent homeosis. In contrast, hth RNAi in embryos of the cricket G. bimaculatus transformed the mandible towards a leg-like structure distally and an antenna-like structure proximally, paralleling the transformation observed in other appendages. Because these results come from only two lineages and from different life stages, additional data are needed to determine whether a homeotic role for hth was present ancestrally in insect mandibles (Angelini, 2012).

These data must be weighed alongside other evidence bearing on the gnathobasic hypothesis. In T. castaneum, the lack of phenotypic effects on mandible metamorphosis of other genes in this study is consistent with the gnathobasic hypothesis. In particular, it was observed that mandible metamorphosis was normal following depletion of genes involved in distal growth and patterning or joint formation. Moreover, homology at one biological level, such as anatomy, does not preclude divergence at other levels, such as development. Nevertheless, since developmental genetic studies of Dll and other appendage-patterning genes have been used as strong support for the gnathobasic homology of the insect mandible, the findings of Krn function highlight the difficulties in establishing serial homology based solely on developmental data (Angelini, 2012).

This study provides a genetic model of adult mouthpart development in Tribolium castaneum based on 13 genes. While previous studies have examined patterning in species with derived mouthpart morphologies, T. castaneum retains the pleisomorphic, mandibulate state of insect mouthparts. These results demonstrate the conservation of many gene functions in the maxilla and labium, relative to the legs, thus supporting the interpretation of novel gene functions in groups with derived mouthpart morphology as indicative of their specialized morphogenetic roles in those species. Mandibulate mouthparts such as those of T. castaneum include medial maxillary and labial endites, and the current data are consistent with hypotheses of reiteration in the PD axis and specification by Notch signaling, but rule out a direct role for dac in branch generation or patterning at metamorphosis. Additionally these results demonstrate that a regulator of distal leg development, Krn, which encodes an EGF ligand, is required for normal mandible elongation. This finding underscores the complex relationship between homology at the levels of anatomy and developmental patterning (Angelini, 2012).