The evolution of segmentation in bilaterian animals is an ongoing area of investigation and debate (reviewed in [1–4]). Three major animal clades contain segmented representatives, the Annelida, the Arthropoda and the Chordata. According to molecular phylogenies [5, 6], each of these clades is more closely related to animals that lack a segmented body than to each other. Thus, the question remains whether segmentation arose independently in distinct lineages, or whether most extant clades lost the segmented body plan that was present in a common segmented ancestor. One approach to address this question is to compare the molecular mechanisms controlling segment generation across taxa, with the assumption that shared molecular mechanisms reflect a common evolutionary history.
Segment formation is best understood in Drosophila melanogaster, in which a detailed understanding of the genetic control of segmentation has served as a framework for comparative studies in other arthropods. Briefly, a segmentation gene cascade sequentially divides the embryo into smaller units through the hierarchical action of the gap, pair-rule and segment polarity genes (reviewed in ). The pair-rule genes are the first to be expressed in a periodic pattern in D. melanogaster, and include the genes paired (Pax3/7), even-skipped (eve) and runt. These genes are expressed with a two-segment periodicity in ectodermal stripes along the anterior-posterior axis. Mutants of pair-rule genes lack alternating segmental structures such as denticle bands in the larval cuticle . In contrast, mutants in segment polarity genes exhibit defects in the pattern of every segment.
The mode of segmentation in D. melanogaster is derived compared to other arthropods (reviewed in ). In D. melanogaster, all segments form at virtually the same time (long germ band development) and much of the patterning critical for segment formation occurs prior to cellularization. In most other arthropods, segments form in a temporal progression from anterior to posterior and the earliest segments form before the tissue for additional segments is present (short germ band development). Although many insects also develop via an early syncytial stage, this is a derived feature within arthropods, and many arthropods form all or most segments within a cellular environment.
The expression patterns of pair-rule and segment polarity gene orthologs have been examined across arthropods, including in insects, and also in chelicerates, myriapods and crustaceans. In general, pair-rule genes show more variability in their expression patterns than do the highly conserved segment polarity genes, and there are examples lacking periodic expression patterns even within insects . In several arthropods, pair-rule gene orthologs are expressed in stripes in every segment; in some cases the onset of expression is clearly prior to segment generation. This pattern is more consistent with a segment polarity role than a pair-rule function. For example, in the centipede Lithobius atkinsoni and the spider Cupiennius salei, Pax3/7 is expressed in a portion of every segment. In the spider mite Tetranychus urticae and the millipede Glomeris marginata, runt has a segmental pattern. Likewise, eve is expressed with a segmental periodicity in L. atkinsoni, C. salei and in the insect Oncopeltus fasciatus. In D. melanogaster it is notable that several pair-rule genes, including even-skipped, runt and paired, have an initial pair-rule expression pattern of seven alternating stripes that later matures into a segmental 14-stripe pattern [18–20]. These data suggest that the pair-rule orthologs have a general, and likely ancestral, function in arthropod segment formation. The extent to which pair-rule patterning is conserved across arthropods, and whether pair-rule patterning was utilized primarily in holometabolous insects (flies, bees, beetles and moths) or ancestrally at the base of arthropods, remains an open question [7, 11, 18]. Regardless, the pair-rule genes are useful markers for inter-taxonomic comparisons to reconstruct the evolution of segmentation.
The mechanistic understanding of the molecular control of segment formation in annelids is poor compared to that in chordates and arthropods. Efforts to identify genes involved in annelid segmentation have largely utilized a candidate gene approach. Examining the expression of genes involved in arthropod segmentation and vertebrate somitogenesis and identification of shared components of a common genetic program for segment formation would support a shared evolutionary origin of segmentation. Expression patterns of segment polarity gene orthologs have been examined in a number of annelid species; the most well-characterized is the segment polarity gene engrailed. In contrast to the highly conserved expression pattern of en across arthropods [15, 21–23], en expression patterns among annelids exhibit substantial variability. Although en is expressed in ectodermal stripes in the polychaete annelid Platynereis dumerilii, this pattern is not apparent in any other annelid examined, including in the polychaetes Chaetopterus sp. , Hydroides elegans and Capitella teleta Blake, Grassle & Eckelbarger, 2009  (previously known as Capitella sp. I) and in the leech Helobdella triserialis. Thus, there is currently a discrepancy in en expression patterns among annelids and the P. dumerilii pattern may represent a convergence, rather than a common origin, with the arthropod pattern. Alternatively, if the P. dumerilii and arthropod patterns represent a conservation, there has been extensive divergence in en expression among annelids.
Previously, we investigated the evolution of bilaterian segmentation by characterizing the expression of orthologs of the segment polarity genes, en, wg and hh and of the pair-rule genes hairy and odd-paired in C. teleta. Here, we report the expression of the pair-rule gene orthologs runt, paired (also called pax group III or Pax3/7) and two eve genes in C. teleta immediately prior to and during larval segment formation. Additionally, we characterize Pax3/7 and eve expression during adult segment formation. C. teleta eve, Pax3/7 and runt genes lack segmental or pair-rule stripes of expression in both the ectoderm and mesoderm, even though Ct-Pax3/7, Ct-eve1 and Ct-eve2 expression is initiated prior to overt segmentation. Each ortholog examined has a distinct expression pattern and exhibits expression domains conserved with those found in other bilaterians.