We are all familiar with the idea that there are strikingly different kinds of eyes in animals: insects have compound eyes with multiple facets, while we vertebrates have simple lens eyes. It seems like a simple evolutionary distinction, with arthropods exhibiting one pattern and vertebrates another, but the story isn't as clean and simple as all that. Protostomes exhibit a variety of different kinds of eyes, leading to the suggestion that eyes have evolved independently many times; in addition, eyes differ in more than just their apparent organization, and there are some significant differences at the molecular level between our photoreceptors and arthropod photoreceptors. It's all very confusing.
There has been some recent press (see also this press release from the EMBL) about research on a particular animal model, the polychaete marine worm, Platynereis dumerilii, that is resolving the confusion. The short answer is that there are fundamentally two different kinds of eyes based on the biology of the cell types, and our common bilaterian ancestor had both—and the diversity arose in elaborations on those two types.
Common features of metazoan eyes
Molecular, developmental, and morphological studies have revealed some common ground in the eyes of virtually all multicellular animals. Their formation is regulated by a common homeobox gene, a pax6 homolog. All photoreceptors use a light-sensitive pigment derived from vitamin A, and this pigment is bound to a protein called opsin. Light activates opsin by causing a conformation change in the photopigment, and opsin then binds to a G-protein, a common and versatile molecule used in many signal transduction cascades. These similarities suggest that all eyes have a common evolutionary ancestor.
There are also significant differences, though. Beyond those similarities in developmental signaling and the general outline of how they turn light into a chemical signal, there are two different kinds of photoreceptors, rhabdomeric and ciliary. They differ in their strategy for increasing membranous surface area (the photoreceptor molecule are imbedded in the membrane, so the more membrane, the more opsins they can pack in), and the steps of signal transduction after the G-protein is bound.
Rhabdomeric photoreceptors are found in the compound eyes of arthropods. They increase their surface area by throwing up their apical surfaces into numerous folds—in some forms, the cell looks like it has had a flat-top crewcut, with a crowning bristle of fine membranous bristles, although the cell itself can have many different shapes in different species.
Signal transduction in rhabdomeric photoreceptors involves activation of phospholipase C (PLC) and the inositol phosphate (IP3) pathway.
The increase in membrane surface area in ciliary photoreceptors, the kind of receptor we vertebrates use, is by modification of the cilium, a process that extends from the cell. The ciliary membrane is expanded and thrown into deep folds, so that the actual receptor region of the cell looks like a stack of discs.
Ciliary photoreceptors use a different signalling pathway, activating a phosphodiesterase (PDE) that changes the concentration of cyclic GMP in the cell. Both the IP3 and the PDE pathways exist in all animals; the difference is in which pathway is used in the different photoreceptors. The diagram below illustrates the two different pathways, and also shows the phylogenetic relationships between their different molecular components (beware of tiny print! Click on the image for a more readable verson).
So, there are two distinct kinds of photoreceptors, with different deep molecular pathways. The evolutionary question is how they arose—when did this distinction first appear? The diagram below illustrates the problem: did the urbilaterian, the last common ancestor of all bilateral animals, a) have just one kind of photoreceptor that later diverged into the two types, or b) did it possess both kinds, and we vertebrates simply lost the rhabdomeric form?
The diagram itself suggests that (b) is probably the best answer, since diverse animals have both types, and it's us vertebrates that are the oddballs in lacking one of the forms. Arendt et al. have found additional evidence for (b) in the polychaete worm, Platynereis.
That's the lovely little worm above, and at it's head end you can see some big dark eyes. Platynereis has multiple sets of eyes, and in particular, it first develops a very simple larval eye (in A, below) that consists of a single photoreceptor cell (in yellow) with a single pigment cell. The adult eye starts out similarly simple (B), but as it matures acquires a simple spherical lens and a larger array of photoreceptors. These are all of the rhabdomeric type.
One can never have enough eyes, though. In addition to the larval eyes (le) and adult eyes (ae), Platynereis has another interesting pair of organs imbedded in its brain, marked by the arrows below. These have turned out to be another pair of simple "eyes"…but of the ciliary type. They have ciliary extensions, and contain opsin; a specific form of opsin, c-opsin, of the type found in ciliary photoreceptors. The other eyes all contain r-opsin, the kind expected to be found in rhabdomeric photoreceptors, and do not express c-opsin. Phylogenetic analysis shows that Platynereis c-opsin clusters with the vertebrate opsins. As the authors state, "This result indicates that two distinct opsin orthology groups exist in Bilateria: the ciliary opsins (c-opsins, active in ciliary PRCs in vertebrates and polychaetes) and the rhabdomeric opsins (r-opsins, active in rhabdomeric PRCs)."
The authors identified another marker, Pdu-rx, a homolog of the vertebrate rx (retinal homeobox) genes. The expression of vertebrate rx genes is restricted to just the ciliary photoreceptor cells of the retina (and a few other interesting places, such as the ciliary receptor cells of the pineal), and what do you know, the ciliary receptors of Platynereis also express this gene.
A couple of questions: what do these c-opsin cells do in the polychaete? They aren't for vision. They also contain proteins that vary with a circadian rhythm, so what these cells are almost certainly involved in is detecting ambient light for resetting the circadian clock.
What happened to the r-opsin cells in the vertebrate lineage? And there we see an interesting and complicated answer: they seem to have been subsumed into various functions in the vertebrate eye other than direct photoreception. Arendt also examined various proteins known to be expressed in other cells in the retina, the bipolar, horizontal, amacrine, and retinal ganglion cells (RGCs), and compared those to related proteins expressed in r-opsin and c-opsin cells in the worm. Surprise: while not traditionally considered receptor cells, several of these other retinal cells seem to cluster in their molecular properties with the r-opsin cells from the invertebrate.
What's also persuasive here is that vertebrate retinal ganglion cells have been recently discovered to contain a photopigment, melanopsin, and to function in resetting the vertebrate circadian rhythm—and melanopsin is an r-opsin homolog.
It's a solid story that ties visual system history in protostomes and deuterostomes together, resolving the differences between them into a convincing evolutionary account. It's a bit like finding the germ of a vertebrate eye imbedded insided the brain of a worm; it's an additional link between two remote branches on the tree of life, and at the same time it clarifies our understanding of the relationships between the different kinds of eyes we see in the animal world.
We propose the following scenario for the evolution of animal PRCs and eyes. Early metazoans possessed a single type of precursor PRC [photoreceptor cell] that used an ancestral opsin for light detection and was involved in photoperiodicity control and possibly in phototaxis. In prebilaterian ancestors, the opsin gene then duplicated into two paralogs, c-opsin and r-opsin, allowing the diversification of the precursor PRC into ciliary and rhabdomeric sister cell types. The rhabdomeric PRCs associated with pigment cells to form simple eyes, whereas the ciliary PRCs formed part of the evolving brain, active in nondirectional photoresponse. This ancestral setting of Bilateria is still present in extant invertebrates such as Platynereis. In the evolutionary line leading to vertebrates, both photoreceptor types were incorporated into the evolving retina. The rhabdomeric PRCs transformed into ganglion cells, acquiring a new role in image processing. A distinctive feature of vertebrate eye evolution is that the ciliary (not rhabdomeric) PRCs became the main visual PRCs, the rods and cones. The vertebrate eye thus represents a composite structure, combining distinct types of lightsensitive cells with independent evolutionary histories.
Arendt D, Tessmar-Raible K, Snyman H, Dorresteijn AW, Wittbrodt J (2004) Ciliary photoreceptors with vertebrate-type opsins in an invertebrate brain. Science 306:869-871.
Arendt D (2003) Evolution of eyes and photoreceptor cell types. Int. J. Dev. Biol. 47:563-571.
One of my favourite posts from the Classic Pharyngula - glad to see it moved here.
What makes the different sorts of eyes that we see in the world recognizeably eyes, despite their differences?
Great post - love this stuff. Just curious, where do the cubozoan eyes fit in?
Dr. Pretorious asked:
What makes the different sorts of eyes that we see in the world recognizeably eyes, despite their differences?
To me, an eye is any metazoan organ devoted to the capture of photons with the potential to mediate a behavioral response to light. Alternatively, one could restrict that definition to only those organs that form images.
Does anyone know how this relates to photosensitive cells in the pineal gland? They also are tied to circadian rhythms.
But, but... I thought eyes were irreducibly complex! They could not have evolved.
God! What a masterful bit of engineering ... and design!
Vertebrate retina and pineal are both developmentally outgrowths of Diencephalon. Pinealocytes in vertebrates other than mammals and snakes look like photoreceptor cells of the retina, with disks and all. The photopigment is pinopsin which is similar to rhodoposin.
Frontal organ in frogs and parapineal organ in lizards and tuataras is even more similar to the eyes - it has an organization like retina and can detect outlines of shadows as well as some color (blue and green).
Clarification requested for those who might not know details of biology quite so well:
In the vertebrates that have them, are we talking about cone cells here, or rods, or both? Do our cones contain a slight variation of those vitamin A derivatives, or of r-opsin, to respond differently to different wavelengths? Or are there pigments on top of the cones to filter the light that gets to them?
in the text above, rods and cones are referred to collectively as photoreceptors. The result which is surprising is that cells which are neither rods nor cones seem to have the cellular machinery to detect and respond to light.
Anatomically, these cells are in layers in the retina which anatomy texts usually describe a bit like:
'The function of these layers is poorly understood but is thought to involve precortical processing of visual information.'
The fact that they can respond to light casts some light on their function.
The pigments are inside the rods and the cones, and there are three pigments; red, green and blue. The brain turns the different responses of these pigments into the different colours we percieve.
Interstingly, if counterintuitively, the photoreceptors (viz: rods and cones) are not at the front of the retina, and only at the centre of the visual field where the most accurate vision is required do multiple layers of nerve fibres move out of the way.
The activation of this same cellular machinery in the pineal gland suggests interesting avenues of enquiry into its evolution and function.
Might I in turn enquire of someone more across the topic than me: where are the other places where the vertebrate retinal homeobox is expressed?
Things have become a bit more complex since this article was originally written. Some melanopsin containing retinal ganglion cells are now known to project to the visual cortex and there is even evidence this year of a small subset of cones in the human retina that contain melanopsin. There has clearly been some intermingling between visual and non-visual roles. Melanopsin is also in the pineal gland.
please...ca you tell me one agens of animals with ciliary and one with rhabdomeric eyes??