The diversity of eyes throughout the bilaterians has led to an ongoing debate regarding their evolution. Based upon the observation that photoreceptive structures generally have either ciliary or rhabdomeric morphology, Eakin [1, 9] proposed two lines of photoreceptor evolution, with the rhabdomeric type having evolved from the ciliary type early in the protostome lineage. Vanfleteren and Coomans [17] concluded that rhabdomeric photoreceptors were a subset of ciliary photoreceptors, and thus photoreceptors are homologous across bilaterians. In contrast to both these interpretations, Salvini-Plawen and Mayr [10, 18] interpreted the diversity of photoreceptors as evidence of polyphyletic origins, postulating that photoreceptors have evolved many times independently in bilaterians.
The advent of molecular genetics led to the unexpected discovery that homologs of several transcription factor families are required for eye formation in both vertebrates and Drosophila, including otx/orthodenticle, Six3/sine oculis, and Pax6/eyeless. The role of Drosophila eyeless and vertebrate Pax6 genes in eye development, as well as the apparent functional equivalence of numerous orthologs from diverse bilaterian taxa, prompted Gehring and Ikeo [8] to propose a monophyletic origin for bilaterian eyes. While the homology of eyes based upon a role for Pax6 as a "master control gene" has been criticized as an oversimplification [19] it appears that Pax6 is a key player in eye development of most bilaterians.
The potential for the phylogenetic distribution of photoreceptors with ciliary and rhabdomeric morphologies to be of evolutionary significance has been bolstered by the recognition that the two types of photoreceptors are characterized by the expression of distinct classes of opsin genes. In all cases examined to date, photoreceptors with ciliary morphology have been shown to express c-opsin class genes and rhabdomeric photoreceptors express r-opsin genes (with the exception of the scallop Mizuhopecten, where the mantle eyes have ciliary morphology, but have been shown to express a Go opsin [20]).
Based upon the occurrence of rhabdomeric photoreceptors in the larval eyes of annelids, arthropods and hemichordates, Arendt and Wittbrodt [3] have suggested that while rhabdomeric and ciliary photoreceptors co-existed in the last common ancestor of bilaterians, the use of rhabdomeric photoreceptors is the ancestral condition, with ciliary photoreceptors having been co-opted for vision within the vertebrate lineage. To date, the expression of c-opsin class genes has been investigated in only two protostomes, the annelid Platynereis[6] and the honeybee Apis[16]. In both cases, expression was observed in the brain, and due to the lack of pigmentation in both these structures they have been inferred to be non-visual photoreceptors. While we observed early expression of Tt-c-opsin in the presumptive neuroectoderm at gastrula and early larval stages of Terebratalia development, we did not observe expression in the apical ganglion, which may be comparable to the brain of other protostomes.
Ciliary larval eyes in Terebratalia: novelty, substitution, or ancestral condition?
By ultrastructural analysis we have demonstrated that the larval eyespots of the brachiopod Terebratalia are composed of two photoreceptors cells, one forming the lens cell and the other the pigmented shading cell. Both cells possess elaborated ciliary membranes in the intercellular space between the lens and the pigment granules, as well as axonal projections extending to the apical ganglion. Supporting the ciliary nature of the larval photoreceptors, we observed that a c-opsin gene is expressed specifically at the position of the eyespots in the larva. Together these results evidence that the Terebratalia larvae possess cerebral eyes that are capable of directional light detection, and that are of a ciliary nature, based upon both morphological and molecular criteria.
Given that Brachiopoda group within the protostome clade Spiralia (Lophotrochozoa) in molecular phylogenies [11, 12, 21], our results provoke the question of whether ciliary photoreceptors play a more important role in protostome eye evolution than previously thought. Whereas polychaete trochophore larvae deploy rhabdomeric photoreceptors (r-opsin expression) directly connected to locomotory cilia for directional movement [22], the morphology of the Terebratalia eyespots strongly suggests that brachiopods use ciliary photoreceptors with c-opsin expression for the same purpose. We cannot rule out the possibility that an r-opsin homolog is also expressed in the larval eyes of Terebratalia, as our attempts to clone such a gene with degenerate primers were unsuccessful.
If manifold independent acquisition of photoreceptor cells in Bilateria can be ruled out by the duality of the existing phototransduction cascades and their respective photoreceptor cell types (rPRCs versus cPRCs) based on bilaterian opsin phylogeny [3], a functional switch from visual to non-visual roles (or vice versa) for the two commonly inherited photoreceptor cell types may have happened several times independently in bilaterian evolution. Such a scenario has already been proposed for the evolution of the visual rods and cones in the vertebrate retina as derivatives of non-visual ciliary deep-brain photoreceptors of invertebrates, such as those in the annelid Platynereis[6].
For the evolution of brachiopod larval eyes this suggests that brachiopods have retained ciliary photoreceptor cells from the bilaterian ancestor, and deployed these for directional light detection in their larval eye spots. Given that ciliary photoreceptor cells are a plesiomorphic trait, rather than being independently evolved in brachiopods, three alternative scenarios may account for the ciliary nature of the larval eyes in Terebratalia: I) The larval eyes of Terebratalia are evolutionary novelties, unrelated to the rhabdomeric cerebral eyes of other larvae in the clade Spiralia; II) The cerebral eyes of Terebratalia are homologous to the larval eyes of other members of the clade Spiralia (for example, polychaetes and mollusks), with ciliary photoreceptor cells having been substituted for the rhabdomeric photoreceptor cells observed in the larval eyes of other taxa; III) Larval eyes with ciliary photoreceptors are the ancestral condition for protostomes and have been inherited by Terebratalia.
Visual photoreceptors with ciliary-type morphologies have been identified in several protostomes; however, these organs have generally been regarded as evolutionary novelties due to their morphological locations (for example,the branchial crown eyes in polycheates [23, 24], and the mantle eyes of scallops [25]. The ciliary larval eyes of Terebratalia could likewise represent an evolutionary novelty that has recruited a ciliary photoreceptor to form a cerebral larval eye comparable to, but not homologous with, the rhabdomeric larval eyes of other spiralians (Figure 9A).
Although the ciliary photoreceptor cells in the larval eyes of Terebratalia seem not to be homologous to the rhabdomeric photoreceptor cells in the larval eyes of Platynereis and other protostomes, the possibility exists that there is homology at the level of the larval eye. If the regulation of eye specification is distinct from that of photoreceptor cell differentiation, then the ciliary photoreceptor cell may have been substituted for the rhabdomeric photoreceptor cell in a homologous larval eye (Figure 9B). In a variety of protostomes and deuterostomes, Pax6 and Otx (among other transcription factor genes) have been shown to be involved in eye specification and differentiation, or to be expressed in cerebral eyes or their precursors (for example, Drosophila[26, 27], Platynereis[4, 28], mouse [29, 30]). Expression of Pax6 and Otx in the precursors of the Terebratalia larval photoreceptors suggests that the genetic network underlying the formation of the larval eyes in Terebratalia shares common features with the network underlying the specification of rhabdomeric larval eyes in other protostomes. In Terebratalia, ciliary photoreceptor cells, and c-opsin expression, may have been co-opted to supplant the rhabdomeric photoreceptor cells in homologous ancestral eyes. By analogy, ontogenetic changes from one photoreceptor cell type to the other have been observed in the larval cerebral eyes of the gastropod mollusk Aporrhais pespelecani, in which the photoreceptor cell initially has a ciliary morphology, only to later develop microvilli, taking on a mixed-type morphology [31]. Although the expression of opsin genes in the Aporrhais eye is unknown, the ontogentic alterations it undergoes suggest that photoreceptor cell morphology may be decoupled from cerebral eye specification.
Finally, it should be considered that eyes with ciliary photoreceptors represent the ancestral state for Spiralia, and possibly for Bilateria (Figure 9C). Arendt and Wittbrodt [3] proposed that cerebral larval eyes with rhabdomeric photoreceptors represent the ancestral state for Bilateria, based in part upon the occurrence of eyes with this morphology in annelids, mollusks, platyhelminthes, crustaceans and hemichordates. As stated above, eyes with ciliary photoreceptors in protostomes have been regarded as evolutionary novelties or "phylogenetically young organs" [10]. However, it should be noted that cerebral larval eyes with ciliary morphology have been described from gastropod mollusks [32–35], and cerebral larval eyes with both ciliary and rhabdomeric photoreceptor cells have been described in both platyhelminthes [36, 37], and the hemichordate Ptychodera flava[38]. Likewise, photoreceptors with ciliary morphology, which may be cerebral eyes, have been described from the larvae of ectoprocts [39, 40] and an entoproct [41], both of which are members of the protostome clade Spiralia, along with brachiopods, annelids, mollusks, and platyhelminthes [12].
While ciliary photoreceptors are not the predominant form in the larval cerebral eyes of protostomes, they are found in a phylogenetically diverse range of taxa. It should, therefore, be considered that the use of ciliary photoreceptors in eyes may be an ancestral condition for Spiralia, and possibly Bilateria. In contrast to this hypothesis, Arendt et al. [6] proposed that localization of unpigmented ciliary receptors to the deep-brain, as seen in Platynereis, represents the ancestral state for bilaterians. However, Nilsson [42] has recently suggested that such photoreceptors might historically been associated with shading pigments for use in directional photoperception (that is, pigmented eyes). If this is the case, then the ciliary eyes of Terebratalia and other spiralians may represent an ancestral condition, rather than being evolutionary novelties.
A minimal photoreceptor mediating early photoresponse behavior
The photoresponse behavior of the gastrula stage embryo is a somewhat surprising result. This photoresponse may be attributed to one of two alternative mechanisms, phototaxis or photokinesis. Phototaxis, movement to or away from light, is associated with directional photoreceptors, which are generally viewed as requiring an associated shading pigment to block off-axis light [3, 43]. Nilsson [41] has proposed that scanning photoperception, wherein movement of the photoreceptor allows detection of differential light intensities, may have provided a primitive mechanism detecting the directionality of light; however, relatively few examples of such a photoreceptor have been described. Although photoreceptors without associated shading pigments have been described from a variety of metazoans, they have almost always been attributed to having non-visual roles in monitoring ambient luminance, such as for detection of diurnal or lunar cycles of illumination. We propose that in middle gastrula stage Terebratalia embryos, the yolk of the lecithotrophic embryo acts as a partial shading pigment to block off axis light, while the spiral swimming patterns serves to generate a scanning movement for directional sampling of illumination intensities. Alternatively, accumulation of embryos on the brightly illuminated side of the chamber may represent a photokinetic response; that is, a change in movement in response to light, independent of photoreceptor orientation. In this scenario, a slowing in the speed of ciliary beating of c-opsin expressing cells in response to increased illumination could cause the distribution of embryos to shift towards the light source. Further experiments are required to resolve whether phototaxis of photokinesis is responsible for the observed positive photoresponse behavior of the middle gastrula stage embryos.
Irrespective of the mechanism of the photoresponse, it is of particular interest that the positive photoresponse behavior of the middle gastrula stage embryos occurs before the onset of neuronal differentiation. At this stage the c-opsin expressing cells and their neighbors constitute a ciliated columnar epithelium without axonal connections. This suggests that the photoresponsive cells may also serve as direct behavior effectors, either through alteration of ciliary beating patterns or through mediation of changes in the orientation of the elongate cilia of the apical tuft to alter the direction of swimming in response to light. An analogous case is that of the parenchymella larvae of certain sponges, which show positive and/or negative phototaxis behavior in the absence of a nervous system or gap junctions for cell-to-cell communication. In sponge larvae, phototactic behavior is thought to be mediated by a posterior ring of pigmented cells [44–46]. However, the mechanism by which changes in ciliary behavior in response to variable illumination may affect larval swimming behavior is not fully understood [45, 46]. In addition, no definitive opsin ortholog has yet been isolated from sponges, although over 200 rhodopsin-related GPCR genes have been identified in the recently published genome of Amphimedon queenslandica[47]. It has been hypothesized that sponge larvae may use a non-homologous mechanism for photoreception, such as flavin [48], carotenoid [48], or cytochrome c oxidase [49]. A minimal photoreceptor cell has also been proposed to occur in the larva of the box jellyfish Tripedalia cystophora, based upon morphology [50]. Pigmented cells in these larvae have rhabdomeric microvilli and a motor cilium, and occur in the absence of a nervous system. However, opsin expression has not been shown for these cells, nor has phototactic behavior been demonstrated for the larvae.
Our results suggest that in Terebratalia middle gastrula stage embryos, c-opsin expressing cells at the anterior of the embryo may be mediating a positive phototactic response in the absence of discrete shading pigments or axonal connections between cells. As such, the Terebratalia gastrula may utilize one of the simplest systems of directional photoperception and effector behavior described to date in bilaterians. Additional studies will be required to understand the details of this phototactic behavior, including the effect of changes in light intensity on rates of ciliary beating, and the potential role of c-ospin expression in mediating this behavior.