Chelifores, chelicerae, and invertebrate evolution

One of the most evocative creatures of the Cambrian is Anomalocaris, an arthropod with a pair of prominent, articulated appendages at the front of its head. Those things are called great appendages, and they were thought to be unique to certain groups of arthropods that are now extinct. A while back, I reported on a study of pycnogonids, the sea spiders, that appeared to show that that might not be the case: on the basis of neural organization and innervation, that study showed that the way pycnogonid chelifores (a pair of large, fang-like structures at the front of the head) were innervated suggested that they were homologous to great appendages. I thought that was pretty darned cool; a relic of a grand Cambrian clade was swimming around in our modern oceans.

However, a new report by Jager et al. suggests that that interpretation may be flawed, and that sea spider chelifores are actually homologous to the chelicerae of spiders.

The original study by Maxmen et al. was based on apparent neuronal homologies. The arthropod brain is made up of several lobes, in particular an anteriormost protocerebral ganglion and a deuterocerebral ganglion behind it, with a tritocerebrum as a third section. The great appendages of those Anomalocaris-like creatures was innervated by the protocerebrum; the chelicerae of modern spiders are innervated by the deuterocerebrum, as are the first antennae of crustaceans. This strongly suggests that chelicerae and great appendages are not homologous. In sea spiders, on the other hand, Maxmen et al. observed that the anteriormost ganglia, the protocerebrum, had nerves that innervated their chelifores, which suggest that chelifores are homologous to great appendages, and are not homologous to chelicerae.

What other kind of evidence would be useful to make this idea rock solid? The definitive markers for the different segments of arthropods are the Hox genes. The Hox genes that are expressed in the arthropod head are labial (lab), proboscipedia (pb), and Deformed (Dfd), and their boundaries of expression are excellent markers for the different segments of the brain. They don't actually extend up into the protocerebrum, but we know that lab and pb are active in the tritocerebrum and are turned off at the deuterocerebral border, while Dfd is active farther back, and is turned off at the tritocerebral border. So where are the Hox genes turned on in sea spider larvae?

This is a little complex to interpret, because no matter whether the chelifores are protocerebral or deuterocerebral, these Hox genes are not expected to be active in them. The question is whether the anterior boundary of lab/pb expression ends just before the chelifores (which makes them deuterocerebral), or one segment back (which makes them protocerebral.) The answer is shown below: in particular, look at b, where you can clearly see that expression of lab ends flush with the posterior end of the prominent chelifores.

Anterior is to the left in all pictures. a, Scanning electron microscopy (SEM) of the protonymphon larva, showing its external morphology in dorsal view. Cp, chelifore; A2, second larval appendage; A3, third larval appendage. b, Expression of labial. The signal is located in the ectoderm of the second and third larval appendage pairs, and in the corresponding trunk region. c, Expression of proboscipedia occurs in the same body region as the expression of labial, except that in the legs (larval appendage pairs 2 and 3) the signal is restricted to the basal article. d–f, Expression of Deformed (in three different larvae). Arrows in d and e indicate the anterior boundary of Deformed expression in the trunk. Panel e shows the trunk and the base of the appendages (labels as in a) at a higher magnification. The round structure is the larval proboscis (Pro). A detailed view of the legs is provided in panel f. Although there is a strong epidermal staining in the third appendage pair, the second appendage pair shows no staining. A probe targeting a fourth Hox gene, Sex comb reduced (Scr) (orthologous to Hox5), produced no staining. Scale bars, 50 µm (a), 17 µm (e), 12 µm (f); magnification in b–d is as in a.

Those photos are of in situ staining of the Hox gene expression products, and like all of biology, are messy and complicated. If you prefer, here's a simpler diagrammatic summary of the expected results if chelifores were protocerebral, and the observed results in crustaceans and spiders and sea spiders. Sea spider larvae have patterns of gene expression that line up perfectly with what we see in more ordinary spiders—chelifores and chelicerae are homologous.

The dashed vertical line indicates the limit between the protocerebral and deutocerebral segments. A2, second pair of larval appendages; A3, third pair of larval appendages; An1, first antennae; An2, second antennae; Ch, chelicerae; Cp, chelifores; L1, first pair of walking legs; Md, mandibles; Oc, ocular segment; Pp, pedipalps.

Wait, wait! What about the Maxmen et al. results? Were they wrong? No. They were looking at an anatomical feature a step removed from the primary determination of segment identity. One thing we can see in arthropod evolution is an increasing compression of head structures—there are several segments worth of stuff squished into the highly derived head, and tissues can get confusingly rearranged. As I mentioned in the earlier article, sorting out patterns of tagmosis isn't easy, and is a source of much argument among scientists. It takes multiple approaches to puzzle out how brains were modified over evolutionary time.

At first glance, our results conflict with those of Maxmen et al., but they can be reconciled through a re-interpretation of their morphological observations. The location of chelifore ganglia—just in front of, and in close proximity to, the protocerebrum (including optic ganglia and nerves)—is probably a derived character. This implies that pycnogonid ancestors underwent both anterior condensation of the central nervous system, and a rotation of the brain, which brought the chelifore ganglia forward to occupy a position more anterior than the protocerebrum. This derived condition is observed also in arachnids, with a comparable forward migration of the cheliceral ganglia during embryonic development. In contrast, a straighter brain, with cheliceral ganglia standing in a more posterior location, is found in adult horseshoe crabs, and probably represents the primitive condition in chelicerates. This condensation and rotation process during ontogeny is not unusual for arthropods, and is found convergently: for example, a condensed and rotated brain with the deutocerebrum located in front of the protocerebrum occurs in spiny lobsters among crustaceans.

While we may not have an Anomalocarid-like beasties left, this is still a lovely story of pattern and complexity in the extant arthropod lineages.

Jager M, Murienne J, Clabaut C, Deutsch J, Le Guyader H, Manual M (2006) Homology of arthropod anterior appendages revealed by Hox gene expression in a sea spider. Nature 441:506-508.