UC Berkeley

Arthropod appendages are the most diverse in the animal kingdom. They vary in terms of number of segments, number of secondary axes branching off of them, and type of secondary branches. In terms of the number of leg segments, the ancestor of arthropods likely had 8 leg segments, while chelicerates have a maximum of 7 or 8 leg segments, myriapods 6 or 7, crustaceans 7 or 8, and insects 6. While they all evolved from a single Onychophoran-like leg type, the homologies between extant arthropod leg segments is uncertain. For example, the proximal leg segment of all arthropods is called the coxa, but whether the spider coxa is homologous to the myriapod coxa, the crustacean coxa, or the insect coxa has remained unanswered since the late 1800s. The answer to this question is relevant to other hotly debated topics, including the origin of insect wings and the evolution of novel appendages.

I hypothesize that the way to homologize hexapod and crustacean legs is the insect subcoxa theory. It was first articulated 125 years ago by Hansen in 1893. Hansen proposed that the sclerites (hardened exoskeletal plates) around the base of each thoracic leg are remnants of two leg segments proximal to the coxa, which became fused to the body wall in the ancestor of hexapods. Hansen homologized these two fused leg segments to the precoxa and coxa of crustaceans. Since Hansen, morphologists have accumulated a wealth of data in hexapods and crustaceans supporting the subcoxa theory, including evidence from external and muscular morphology, paleontology, and embryology.

However, the insect subcoxa is highly reduced and modified compared to the free crustacean leg segments from which it is hypothesized to be derived. Therefore, the comparative morphological and embryological data in insects and crustaceans cannot unambiguously determine whether these segments are homologous. In fact, morphologists have quite exhausted the issue without arriving at a clear answer. In order to demonstrate unambiguously that the insect subcoxa is indeed appendicular, and that it is homologous to the crustacean precoxa and coxa, molecular expression and functional data must be examined in a crustacean where these segments exist as free, unambiguous leg segments.

Malacostracan crustaceans are a good comparative model, because they are relatively closely related to hexapods, and have uniramous walking legs like insects with readily identifiable segments. Parhyale hawaiensis is an ideal malacostracan crustacean system for this work, with many available molecular and functional tools, including in situ hybridization and CRISPR-Cas9 knockout. Like many malacostracans, Parhyale has 7 leg segments, the coxa (1), basis (2), ischium (3), merus (4), carpus (5), propodus (6), and dactyl (7). Insects have 6 leg segments, the coxa (1), trochanter (2), femur (3), tibia (4), tarsus (5), and pretarsus (6, terminal claw).

Leg gap genes pattern the arthropod proximal-distal axis, and therefore represent a logical set of tools to determine crustacean-insect leg segment homologies. I have examined the expression patterns and CRISPR-Cas9 knockout phenotypes of five leg gap genes in the amphipod crustacean Parhyale hawaiensis: Distalless, Sp6-9, Dachshund, Extradenticle, and Homothorax. In insects, Dll is expressed and functions in the telopod, in leg segments 2 – 6. There are three Parhyale Distalless paralogs, but the canonical one is Distalless early (Dll-e). In Parhyale, Dll-e is expressed and functions in leg segments 3 – 7. In insects, Sp6-9 is expressed and functions in leg segments 1 – 6. In Drosophila, the leg is occasionally transformed towards wing and notum identity. In Parhyale, Sp6-9 is expressed and functions in leg segments 2 – 7. Additionally, the truncated stumps of thoracic legs 6-8 were occasionally transformed towards a proximal leg segment identity. This shared transformation of leg segment identity following Sp6-9 knockout in insects and Parhyale is evidence that the insect wing and notum are indeed a proximal leg segment. Dachshund patterns the medial leg segments. In Drosophila, it is expressed and functions in distal leg segment 2 (trochanter) through proximal leg segment 5 (tarsus 1). Parhyale has two Dac paralogs. I do not have expression data for either of these paralogs, but CRISPR-Cas9 knockout of Dac2 resulted in deletion of leg segments 3 – 5. Extradenticle and Homothorax are obligate heterodimers, and appear to function only where their expression overlaps. In all arthropod examined, they are expressed in the body wall and proximal leg segments. In insects, Exd is expressed in leg segments 1 – 4. In Parhyale, Exd is expressed in leg segments 1 – 5. In insects, Hth expression is in the coxa through proximal femur (leg segments 1 – 2.5). Knockout phenotypes for Exd or Hth appear to be qualitatively indistinguishable, as would be expected for obligate cofactors. In insects, Exd or Hth knockout results in deletions/fusions of body segments, transformation of head appendages towards thoracic identities, and deletions/fusions of the coxa through proximal tibia (leg segments 1 – 3.5). This is 1 segment farther than its apparent expression domain. In Parhyale, Exd/Hth are expressed in the body wall and proximal leg segments 1 – 3. CRISPR-Cas9 knockout results in deletions/fusions of body segments, transformation of head appendages towards thoracic identities, and deletions/fusions of leg segments 1 – 4. Like in insects, this is 1 segment farther than its apparent expression domain.

In summary, the expression and function of Exd, Sp6-9, and Dll in insects has shifted proximally by precisely one segment relative to Parhyale. The expression and function of Hth and Dac in insects have also shifted by approximately one segment relative to Parhyale. Notably, the peculiarities of expression and function in Sp6-9 and Exd/Hth also reflect this shift. This shift is accounted for if the insect subcoxa is presumed to be a leg segment, which brings the expression and function of all five leg gap genes in insects into register with crustaceans. Thus, the crustacean basis is homologous to the insect coxa, the crustacean ischium is the insect femur, and so on for all leg segments. These results represent the first molecular and functional evidence in a crustacean brought to bear on the question of the insect subcoxa. These data are also the first unambiguous demonstration that the insect subcoxa is indeed a leg segment, and that it is homologous to the crustacean coxa (and perhaps precoxa).

These results also provide clues about other appendage-like structures in insect. For example, embryonic abdominal appendages are known in all but the most derived insect clades. They form prominent protuberances during development, but fuse to the abdominal body wall to form the sternites before hatching. My results in Parhyale argue that these appendages are comprised of the insect subcoxa and coxa. These embryonic insect abdominal appendages are probably the source of insect appendages that have been presumed to evolve de novo, such as butterfly prolegs, larval gills, and male sepsid fly abdominal appendages. This is informative when thinking about the evolution of novel structures. Perhaps cases of loss and re-evolution are in fact not a complete loss of the entire pathway, but merely a repression of the distal or downstream network, leaving the proximal or upstream pathway intact, such as, for example, the repression of Dll by Ubx and AbdA in insects. When this downstream repression is relieved, such as in caterpillar prolegs, the structure or network forms what appear to be a novel structure. Thus, in order for these insect groups to form apparently novel appendages, there is no need to speculate about how an entire new leg field could have re-evolved de-novo, precisely where legs used to be in the ancestor. These leg fields were never lost in the first place. Indeed, it is more parsimonious to posit that apparently novel appendages are merely the result of de-repressing the distal part of the existing leg field.

Similarly, cases of appendage-like outgrowths forming in novel locations, like beetle head horns and arthropod eye stalks and horns, may in fact form from pre-existing appendage fields. In fact, all arthropods express all five leg gap genes in various locations around the ocular segment, as well as in the eye. I show that leg gap genes are also expressed around the ocular lobe and in the eye of Parhyale. Knockout of Exd or Hth caused ectopic, appendage-like growths in various locations of Parhyale head lobes, including one that appears to emerge from the eye. This argues that the leg gap gene expression domains around the ocular lobes of arthropods are in fact cryptic appendage-like fields, and also that the arthropod eye forms on an appendage field. Exd and Hth are known to act as cofactors for other proteins, like the Hox genes, and loss of Exd and Hth is known to cause homeotic mutations. Perhaps Exd and Hth act as cofactors to select truncated, head-specific fates for these cryptic ocular appendage fields. Thus, when Exd or Hth is reduced, these cryptic ocular appendage fields are homeotically transformed towards a more appendage-like outgrowth.

Derived malacostracans, like amphipods and decapods, appear to have lost the ancestral crustacean precoxa. This precoxa may have become fused to the body wall, similar to the insect subcoxa. In amphipods, the exite of the precoxa might still be retained in the form of the coxal plate, while in decapods, the gills of the body wall may be the fused remnant of the precoxa. This hypothesis can only be unambiguously demonstrated by examining the expression of leg patterning genes in a crustacean with a precoxa, such as a mantis shrimp. I am currently working on such experiments.

Finally, my results in Parhyale may provide the link for how to homologize leg segments across all arthropods. The horseshoe crab Limulus may have fused a leg segment, represented by a moveable endite proximal to the Limulus coxa. This is similar to the situation in Parhyale, which also must have lost a proximal leg segment (the precoxa). Notably, Dll expression in Limulus and Parhyale is in register: in both animals, Dll is expressed in leg segments 3 – 7. Thus, I hypothesize that both Parhyale and Limulus fused one leg segment to the body, and that there is a one-to-one homology between the remaining leg segments: the Parhyale coxa is homologous to the Limulus coxa, the Parhyale basis to the Limulus trochanter, and so on. The fusion of proximal leg segments to the body across multiple arthropod taxa appears to be associated with a transition from an aquatic/swimming to a terrestrial/ambulatory lifestyle.