Info

Figure 20.1 Diagram illustrating a small portion of the diversity of arthropod appendages. The upper row shows exemplary appendages for sensory perception, the centre row shows gnathal appendages and the lower row shows walking appendages from the arthropod classes indicated at the top. Note that the pedipalps in spiders (top row left) are used for perception, feeding and sperm transfer. The diagrams in the top row show simplified and generalised drawings of (from left to right): spider pedipalp, amphipod first antenna, bee antenna, pauropod antenna. The diagrams in the centre row show simplified and generalised drawings of (from left to right): scorpion chelicera (upper drawing) and spider chelicera (lower drawing), syncarid mandible, first maxilla of Remipedia, ectognathan mandible, ectognathan maxilla, millipede mandible, chilopod second maxilla. The bottom row shows simplified and generalised drawings of (from left to right): spider prosomal walking leg, amphipod pereiopod, dipteran thoracic leg, millipede trunk leg. Please note that the alignment of appendages in the diagram is based on similar functions and explicitly not on homology. Drawings after Westheide and Rieger (1996), Prpic and Damen (2004) and Prpic and Tautz (2003).

was the starting point for the evolution of this unique diversity? What is the 'ground state' of appendage that has served as the basis for further evolution? (2) What evolutionary changes have occurred to transform one appendage type into another? What are the underlying genetic mechanisms and how did they change during evolution to produce morphological innovations?

the search for the ground-state appendage

It is generally assumed that the diversity of extant appendages has evolved from a limited number of ancestral appendage forms or even from a single prototype appendage. The search for this 'ground-state' appendage has attracted considerable attention. Snodgrass (1935), for example, suggested that the leg segments (podomeres) of all extant arthropods are divided into a group of proximal podomeres and a group of distal podomeres. Snodgrass suggested that this common ground reflects an ancestral state of all arthropod appendages and that the ground-state appendage therefore consisted of two podomeres that he called coxopodite (the proximal one) and telopodite (the distal one) (Figure 20.2A). Interestingly, the early leg discs in Drosophila are subdivided into a proximal domain co-expressing the genes extradenticle (exd) and homothorax (hth), and a distal domain expressing the gene Distal-less (Dll) (e.g. Gonzalez-Crespo and Morata 1996, Gonzalez-Crespo et al. 1998, Wu and Cohen 1999). A different approach towards the ground-state appendage has been taken more recently. This approach is based on the finding that appendage morphology is also determined by

Figure 20.2 Different concepts of the ancestral or 'ground-state' appendage. A, The bipartite ancestral appendage according to Snodgrass (1935). B, The 'ground-state' appendage that results after the depletion of all known selector genes (after Casares and Mann 2001). C, A generalised trilobite leg (combined after Whittington and Almond 1987 and Westheide and Rieger 1996). According to the majority of authors, the ancestral appendage was relatively complex, similar to the appendages of the fossil arthropod group Trilobita (e.g. Walossek and Müller 1997, Boxshall 2004, Giorgianni and Patel 2005). See text for details.

Figure 20.2 Different concepts of the ancestral or 'ground-state' appendage. A, The bipartite ancestral appendage according to Snodgrass (1935). B, The 'ground-state' appendage that results after the depletion of all known selector genes (after Casares and Mann 2001). C, A generalised trilobite leg (combined after Whittington and Almond 1987 and Westheide and Rieger 1996). According to the majority of authors, the ancestral appendage was relatively complex, similar to the appendages of the fossil arthropod group Trilobita (e.g. Walossek and Müller 1997, Boxshall 2004, Giorgianni and Patel 2005). See text for details.

selector genes, for example by the Hox genes (see below for details). Eliminating their influence, therefore, might reveal the ancestral, undiversi-fied state. An experiment of that sort has been performed in Drosophila, where such genetic manipulations are possible. The appendage that develops in the absence of the influence of the known selector genes consists of a large proximal segment and a largely normal tarsus (Figure 20.2B) (Casares and Mann 2001). Experiments like this are intriguing, but selector genes like the Hox genes and their role in identity determination are phylogenetically much older than the arthropods. It is certainly wrong to assume that the ancestral arthropod did not have Hox genes or other selector genes. Thus the appendage obtained in Drosophila most probably does not correspond to any real stage of appendage evolution in the arthropods.

The question is not what the appendages look like without the influence of selector genes. Rather we have to ask what selector genes were present in the ancestral appendage and what functions they performed (see also Hughes and Kaufman 2002). Questions of this kind can only be answered by a comparative approach. Features common to all species are likely to be derived from a common ancestor. Such features, called symplesiomorphies, thus reveal details about the ground-state appendage. Apart from this indirect approach there is one alternative approach to learn about historical conditions: direct evidence in the form of fossils. There are a number of arthropod fossils that preserve the morphology of the appendages. Trilobites, for example, are among the earliest arthropod fossils, some dating from the Early Cambrian. They show very little appendage specialisation: their anterior-most appendage is a long antenna, but all following appendages are more or less identical (Whittington 1997). It is interesting to note that apart from the antenna the appendages of the trilobites unite functions that are relegated to different appendage types on separate body regions in extant arthropods: the trilobite post-antennal appendages were used for feeding, locomotion and breathing (Figure 20.2C). Thus, the trilobites do indeed appear to preserve a body plan from before the time of appendage specification and diversification, providing a direct image of the prototype arthropod appendage (see also Walossek and Müller 1997, Bitsch 2001, Boxshall 2004, Giorgianni and Patel 2005).

from ground state to diversity: insights from drosophila

Whatever the design of the prototype appendage, what possibilities are there to develop new appendage types from it? Principally, there are three different ways of change: (1) loss of an existing feature; (2) gain of a novel feature; or (3) modification of an existing feature.

There are many examples for each of these three modes or combinations of them. The mandible of insects is an example, where the entire distal portion of the appendage is lost. An example of a novel feature would be the brush legs of moths, as already mentioned in the introduction. Finally, modification of the number of leg joints in the tarsi of insects is an example of the third kind of change. These examples are all fine, but the interesting question is: how did it happen? What genes are involved in the changes? How did developmental pathways change during evolution? How can we explain new morphologies in terms of the underlying genetic and developmental mechanisms?

The data on leg development in the fruit fly Drosophila melanogaster is a starting point that can help address these questions. In Drosophila, the legs develop from so-called imaginal discs, which is unusual for arthropods. These discs start developing in the embryo as a small group of cells, and during the larval stages this group of cells folds into the body, forming the leg by ingrowth, rather than outgrowth as in most other arthropods (Cohen 1993). During metamorphosis this 'inward' leg turns outward, thus becoming a rather normal insect leg (Fristrom and Fristrom 1993). Despite this peculiar mode of development, the genetic mechanisms operating during Drosophila leg development can serve as a first guide to study the developmental mechanisms in other appendage types and other species. Drosophila leg development is governed by a hierarchic gene cascade (Figure 20.3) (Rauskolb and Irvine 1999). At the top level (Figure 20.3, top) two morphogens, Wingless (Wg) and Decapentaplegic (Dpp) generate a grid of morphogen concentrations and this information is read out by the next level in the cascade, the leg gap genes (Figure 20.3, centre) (e.g. Lecuit and Cohen 1997). These genes include dachshund (dac) and Distal-less (Dll). The leg gap genes have two functions. First, they identify broad domains along the proximal-distal axis (e.g. Cohen and Jürgens 1989a,b, Mardon et al. 1994, Abu-Shaar and Mann 1998). Second, soon after the expression of the leg gap genes is initiated, their expression domains expand and thus partially overlap. These overlaps, together with overlaps with already expressed genes like homothorax (hth) and its co-factor extradenticle (exd), create areas of combinatorial gene expression that serve as a kind of 'address code' for the genes at the next level of the gene cascade (Rauskolb 2001). These genes, including mostly members of the Notch signalling pathway or its target genes, are activated in narrow rings along the leg by a specific combination

Figure 20.3 Genes and gene cascades involved in walking leg development. In Drosophila leg discs a hierarchic gene cascade guides proximal-distal axis development. At the top level are the two genes dpp and wg that are expressed in a dorsal and ventral sector, respectively. However, the Dpp and Wg proteins spread throughout the disc and activate the genes at the next level (e.g. dac, Dll). These genes are expressed in broad concentric but of leg gap genes (Figure 20.3, bottom) (Rauskolb 2001). These rings define the locations where the boundaries between the segments are made (DeCelis et al 1998, Bishop et al 1999, Rauskolb and Irvine 1999). Of course the small number of leg gap genes are not enough to generate sufficient differential overlap zones to determine all nine segment borders (including the tarsal segments and the claw). Recently a number of additional genes have been identified that might be called 'tarsal gap genes'. These seem to supplement the already known leg gap genes in the tarsus to generate the additional segment borders there (genes such as Bar and dlim1; see Galindo et al 2002 and references therein, reviewed in Kojima 2004). This entire process of leg patterning appears to be organised by the Hox gene Antennapedia (Antp), because, if misexpressed in the antennal segment, this gene is capable of generating a relatively normal leg, and if Antp function is lost in the legs, the tissue is transformed into antennae (Struhl 1982, Abbot and Kaufman 1986, Emerald and Cohen 2004).

How do the mechanisms in other appendage types in Drosophila differ from the one in the legs and how does this correlate with the differences in morphology between the appendage types? This question is rather difficult to answer, again because of the rather peculiar mode of appendage development in Drosophila. Not only do all appendages develop from imaginal discs, but also the mouthparts form the so-called proboscis, a structure peculiar to flies that is formed from parts of several highly modified appendages. Instead of Antp, it is the Hox genes proboscipedia (pb) and Sex-combs-reduced (Scr) that appear to trigger the crucial developmental steps necessary to generate the labial morphology in Drosophila (Abzhanov et al. 2001, Joulia et al. 2005, 2006). Abzhanov et al. (2001) were able to show that both Hox genes together repress several of the genes known to be important factors in leg development, such as exd, hth, Dll and dac. However, morphologically the proboscis is far from being simply a 'repressed leg' and it seems clear that additional unknown factors must also be involved in its development.

The development of the Drosophila antenna is independent from Hox gene function, because no Hox gene is expressed in the antennal

Fig. 20.3 (Cont.) overlapping domains, which establish a first crude subdivision of the developing leg and then activate, in a combinatorial fashion, the genes at the next level of the cascade, which comprise mainly members of the Notch signalling pathway or its target genes. These genes are expressed in concentric rings and define the location where the joints (i.e. the borders between the leg segments) are made. Simplified after Rauskolb and Irvine (1999).

segment. The role of an antennal selector gene is taken up by hth, which in the leg specifies proximal regions (Casares and Mann 1998). This substantial difference from the leg already hints at quite different genetic mechanisms operating in the developing antenna. Recent work has shown that in the antenna the subdivision into a proximal and a distal domain, which is so fundamental in the leg, does not exist (Dong et al

2000, 2001); rather, hth and Dll overlap broadly. This co-expression of hth and Dll activates several antenna specific factors that have not been identified in any other appendage type, such as cut, distal-antenna (dan) or spalt (sal) (Chu et al. 2002, Dong et al. 2002, Emerald et al. 2003, Suzanne et al 2003). Thus, unlike in the labial disc, in the antennal disc several appendage-specific factors have already been identified, but their exact correlation with the specific morphology of the antenna is still unclear.

from drosophila melanogaster to millions of arthropod species

As already noted, the diversity of arthropod appendage morphology is immense. What about the genetic mechanisms generating this diversity? In the following, we will focus on the appendage types 'walking leg', 'mouth-part' and 'antenna', because for these types some details are known from Drosophila for comparison.

Walking legs in the form of unbranched, segmented appendages are present in representatives of all four arthropod classes. Comparative studies in a number of arthropod species have focused on the middle level of the leg genes cascade (e.g. Panganiban et al 1994, Niwa et al. 1997, Abzhanov and Kaufman 2000, Jockusch et al. 2000, Prpic et al.

2001, 2003, Inoue et al. 2002, Prpic and Tautz 2003). These works revealed that the tripartite structure is conserved in all species. In addition, where functional analysis has been done, the role of these genes is also very similar to the Drosophila homologues (Beermann et al 2001, Schoppmeier and Damen 2001, Angelini and Kaufman 2004). Details in the relative expression and expression dynamics of the genes, however, differ between the species. Since the areas of overlap of these genes in Drosophila determine the location where the genes at the lowest level are activated and the leg segment boundaries are made (Rauskolb 2001), the differences in the relative expression in other species might explain the differences in leg segment number.

Given the relatively high degree of conservation at the middle level of the cascade, it is surprising that not much seems to be conserved above that level. The dpp gene that in Drosophila is expressed along the dorsal side is expressed in the leg tip in most other species and later in one or several rings and dots, depending on the species (e.g. Sanchez-Salazar et al 1996, Jockusch et al. 2000, Niwa et al. 2000, Prpic et al. 2003, Prpic 2004, Yamamoto et al. 2004). This suggests that there might be no cooperation between dpp and wg like in Drosophila. Indeed, recent studies were not able to confirm a similar role for wg in leg development in species other than Drosophila (Angelini and Kaufman 2005a,b). Also the role of dpp remains entirely unclear. Its role as a partner for wg has been questioned (Ober and Jockusch 2006), but the later expression in rings and dots has been linked to smaller differences in walking leg morphology (Niwa et al. 2000). Severe pheno-types of Antp RNAi in Oncopeltus fasciatus lead to the transformation of leg into antenna like in Drosophila (Angelini et al. 2005). Data from the beetle Tribolium castaneum also demonstrate a role of Antp in specifying leg identity (Beeman et al. 1989). Thus, Antp is obviously the selector gene for leg morphology in insects. However, other than in insects, the role of Antp in specifying walking leg morphology has been questioned. For example, it has been shown that in the crustacean Daphnia magna, Antp has the opposite function to Drosophila Antp: it represses Dll expression (Shiga et al. 2002). And in chelicerates Antp is not expressed in the walking legs at all (Figure 20.4D) (Damen et al. 1998, Telford and Thomas 1998, Abzha-nov et al 1999). The chelicerate walking legs are quite similar to the walking legs in crustaceans, myriapods and insects, but develop on segments that do not express Antp at all, except for the posterior portion of the fourth walking leg.

These data begin to illustrate an unexpected diversity of genetic mechanisms above a relatively conserved middle level with genes like dac and Dll (nothing is known about the lower level so far). Intriguingly, this introduces another level of diversity: the somewhat paradoxical phenomenon that there is diversity of genes and gene regulation on the top level, but relatively little differences in terms of morphological output, namely a walking leg.

Other appendage types such as mouthparts or antennae are not yet studied in such detail in other arthropod species. The studies on the development of the different mouthparts in crustaceans and insects seem to indicate a global role for the anterior Hox genes. Several crustacean species for example have additional mouthparts (so-called maxil-lipeds). Maxillipeds develop in segments where Ubx and Abd-A are not expressed, thus leaving space for the anterior Hox genes pb, Deformed (Dfd) and Scr (Averof and Patel 1997, Abzhanov and Kaufman 1999,

Figure 20.4 Expression of Hox genes in the spider Cupiennius salei. In Dro-sophila, walking leg morphology is determined by the Hox gene Antp, whereas the anterior Hox genes like Dfd and Scr specify mouthpart identity. This is not the case in chelicerates: first, the anterior Hox genes like Dfd-1 (A), Dfd-2 (B), and Scr (C) are expressed in all or some walking legs, but not in the mouthparts of C. salei (chelicera, pedipalp). Second, the Antp gene is not expressed in the walking legs at all (D), except for a small portion of the last walking leg (L4) (not visible in the figure). Note: C. salei has two Dfd paralogues (1, 2) that show differential expression in the neuroectoderm (arrows in A and B) and also some differences in detail in the walking legs, but the overall expression domain of both genes covers all four segments bearing walking legs. All embryos are oriented with anterior to the left. The embryos in A and B are late inversion stages, C is a mid inversion stage, and D is younger still (mid germband extension stage). Abbreviations: L1-L4, walking legs 1 to 4.

Figure 20.4 Expression of Hox genes in the spider Cupiennius salei. In Dro-sophila, walking leg morphology is determined by the Hox gene Antp, whereas the anterior Hox genes like Dfd and Scr specify mouthpart identity. This is not the case in chelicerates: first, the anterior Hox genes like Dfd-1 (A), Dfd-2 (B), and Scr (C) are expressed in all or some walking legs, but not in the mouthparts of C. salei (chelicera, pedipalp). Second, the Antp gene is not expressed in the walking legs at all (D), except for a small portion of the last walking leg (L4) (not visible in the figure). Note: C. salei has two Dfd paralogues (1, 2) that show differential expression in the neuroectoderm (arrows in A and B) and also some differences in detail in the walking legs, but the overall expression domain of both genes covers all four segments bearing walking legs. All embryos are oriented with anterior to the left. The embryos in A and B are late inversion stages, C is a mid inversion stage, and D is younger still (mid germband extension stage). Abbreviations: L1-L4, walking legs 1 to 4.

2004). Also in insects the anterior Hox genes are selector genes for mouthpart identity. Lack of the pb gene and expression of the Dfd gene have been connected with the specific stylet-shaped mouthparts in true bugs (Rogers et al. 2002). In addition, Scr is involved in labial identity and Dfd is involved in mandibular identity in Oncopeltus fasciatus (Hughes and Kaufman 2000). In Tribolium castaneum pb, Dfd and Scr appear to be involved in the specification of the pincer-shaped mouthparts (Beeman et al. 1989, Shippy et al. 2000, DeCamillis et al. 2001). The Hox targets in the mouthparts are unknown, although in Tribolium pb appears to repress Dll in the mouthparts (DeCamillis and ffrench-Constant 2003). It is intriguing, however, that again in chelicerates the anterior Hox genes pb, Dfd and Scr obviously do not specify mouthpart morphology, as the segments expressing these genes develop normal walking legs (Figure 20.4A-C) (Damen et al. 1998, Telford and Thomas 1998, Abzhanov et al. 1999).

With respect to the expression of other leg genes, such as dac, Dll, hth or exd, the patterns in maxilla and labium of (for example) Tribolium castaneum and Schistocerca americana are not dramatically different from the patterns in the leg (e.g. Prpic et al. 2001, Giorgianni and Patel 2004, Jockusch et al. 2004), indicating that there must be other factors as yet unidentified that are responsible for the specific morphology of the mouthparts. Significant differences exist, for example, in the maxillary stylet of the bug Oncopeltus fasciatus where several leg genes are broadly co-expressed (Angelini and Kaufman 2004, 2005a), and in the insect mandible where Dll expression is entirely missing (e.g. Popadic et al. 1996, 1998, Scholtz et al. 1998). Endites, finally, are proximal and ventral protrusions that are specific to mouthparts, but their specification is unclear. There seem to be mechanisms guiding the outgrowth of these structures that are different from the mechanisms in the main appendage axis (e.g. Giorgianni and Patel 2004, Jockusch et al. 2004). The dac gene is expressed in all endites of insect, myriapod and crustacean mouthparts that will adopt a biting or chewing tooth morphology later on (Abzhanov and Kaufman 2000, Prpic et al. 2001, Prpic and Tautz 2003), but is not expressed in endites that will be just weak shovel-like outgrowths like the gnathendite of spiders (Prpic and Damen 2004) or that will be brush-like like the pectinate lamella in millipede myriapods (Prpic and Tautz 2003). There might thus be a correlation of dac expression and tooth-shaped endite morphology.

Finally, we would like to mention a peculiar case of mouthpart: the chelicera of spiders. This is a stout appendage comprising a basal segment and a movable venom fang. The cheliceral body segment does not express any Hox genes and thus the chelicera develops without Hox input. The patterns of hth, exd and Dll largely overlap (Prpic and Damen 2004). This is quite different from the more typical gnathal appendages in other arthropod classes, but it is very reminiscent of

Figure 20.5 Is there a 'walking-leg-typic stage' at the junction between the diversity of early proximal-distal patterning mechanisms and the diversity of morphological form? It has been shown for Drosophila that the tripartite stage comprising a proximal domain with co-expression of hth and exd, a medial domain expressing dac and a distal domain expressing Dll is necessary for the development of the walking legs, but not of the antenna (Dong et al. 2000). This tripartite structure is present in the walking legs of other arthropods as well, but is absent from other appendage types like the chelicera in spiders (Prpic and Damen 2004) or the maxillary stylet in true bugs (Angelini and Kaufman 2004). It has been suggested that the tripartite structure is a necessary constraint to be passed through by all appendages that are to develop a walking-leg-like appearance (Prpic and Damen 2004). It has been shown that the diversity of genetic mechanisms before this tripartite stage is high, including the diversity of dpp expression patterns and the debated role of wg, the role of Antp and the role of the anterior Hox

Figure 20.5 Is there a 'walking-leg-typic stage' at the junction between the diversity of early proximal-distal patterning mechanisms and the diversity of morphological form? It has been shown for Drosophila that the tripartite stage comprising a proximal domain with co-expression of hth and exd, a medial domain expressing dac and a distal domain expressing Dll is necessary for the development of the walking legs, but not of the antenna (Dong et al. 2000). This tripartite structure is present in the walking legs of other arthropods as well, but is absent from other appendage types like the chelicera in spiders (Prpic and Damen 2004) or the maxillary stylet in true bugs (Angelini and Kaufman 2004). It has been suggested that the tripartite structure is a necessary constraint to be passed through by all appendages that are to develop a walking-leg-like appearance (Prpic and Damen 2004). It has been shown that the diversity of genetic mechanisms before this tripartite stage is high, including the diversity of dpp expression patterns and the debated role of wg, the role of Antp and the role of the anterior Hox the expression patterns in the antenna of Drosophila. In fact, the cheli-cera, although functionally a gnathal appendage, is homologous to the antenna (Damen et al. 1998, Telford and Thomas 1998, Mittmann and Scholtz 2003). The similarities in gene expression between the chelicera and the Drosophila antenna may therefore be based on a common origin of the appendage rather than on common function. However, an alternative and probably more likely explanation is that changing the leg-like sequence of the leg genes hth/exd, dac, Dll by broadly co-expressing some or all of these genes assists in the activation of different target genes that lead to novel morphologies such as the fang-like chelicera or the short and stubby antenna in Drosophila (Prpic and Damen 2004). This is further supported by the largely overlapping expression of hth and dac (and also Dll) in the maxillary stylet of Oncopeltus fasciatus (Angelini and Kaufman 2004, 2005a).

a kind of phylotypic stage for the different types of arthropod appendage?

The following sentence from Angelini and Kaufman (2005a) sums up our current ignorance with respect to the origin and evolution of one of the greatest diversifications of morphological form on Earth: 'We are still far from an explanation of biological diversity in which morphology may be unambiguously described by our knowledge of ontogenetic mechanisms.'

Research into the evolution of arthropod appendage development is still very much at its beginning. The broad comparative approach can identify features that are conserved among all species and thus reveal information about the ancestral appendage. This approach can also identify genes and gene regulations that are not conserved and thus may be responsible for morphological novelties. Unfortunately, the species studied so far cover only a tiny fraction of the diversity of the arthropods. What we also need are more detailed studies of gene function in selected species to achieve a more complete understanding of the relationship between gene expression and regulation and morphology.

Fig. 20.5 (Cont.) genes, like pb, Dfd and Scr (see text for details). Also the diversity of adult morphology of walking legs is high. The tripartite stage seems to channel the diversity of early patterning mechanisms into a path towards 'walking leg morphology' but still allows for a high amount of morphological diversity of adult structures. In this way it is similar to the phylotypic stage (Sander 1983, Raff 1996).

The current knowledge about the evolution of genetic mechanisms in appendage types like mouthparts or antennae is too fragmentary yet to allow for sound conclusions. But if the data from comparative studies of walking leg development are any indication, then it looks like there is not one level of appendage diversity but two: one level, as noted, is the diversity of adult walking leg morphology. But a second level is the diversity of developmental mechanisms above the level of genes like dac or Dll that, despite their significant differences, all converge again on creating a walking leg type of appendage (Figure 20.5). This, in a way, is reminiscent of the hour-glass model (Raff 1996) describing the diversity of early embryonic and adult body plans against the limited diversity of phenotypes at so-called phylotypic stage (Sander 1983). It seems that what is valid for animals as a whole, is also valid for single parts of them: one might construct an hour-glass model also for the walking legs of arthropods, with diversity in early development and adult morphology, but a 'podotypic' stage involving the genes exd, hth, dac and Dll connecting them in the middle (Figure 20.5).

Was this article helpful?

0 0

Post a comment