A Comment On the Origin of the Vertebrate Eye


In her recent article in The New Anatomist, “Chordate Evolution and the Origin of Craniates,” Butler (2000) reviews the elegant work of N.D. and L.Z. Holland (Glaron et al., 1998) and Lacalli et al. (1994) which suggests that the cerebral vesicle of amphioxus is homologous to the diencephalon of the craniates, using in situ hybridization and serial section reconstruction methods on larval amphioxus. The question remains as to how the modest organ of the lancelet evolved into the elaborate diencephalic sensory derivatives. As Butler states: “The elaboration of the brain across bilaterally symmetrical animals often includes elaboration of paired, enlarged eyes. In craniates the retinas of the lateral eyes derive from the diencephalon itself. …An elaborated visual system with descending pathways…could bestow a major selective advantage for an organism”, making the transition from cephalocordate to craniates highly plausible (Butler, 2000 p.120).

In the course of study of amphioxus since the early 20th Century, a number of structures in the cerebral vesicle have been considered to be the precursors of the lateral eyes of the craniates. These include: the anterior pigment spot and its surrounding cells, which Lacalli et al. (1994) call the frontal eye (discussed below); the sensory cells of the Hesse's organs; the dorsal cells of Joseph; the cells of the infundibular organ; and certain cells that I discovered in 1958 in the adult cerebral vesicle of amphioxus (see Miller, 1960). At that time, I had begun to work in Keith Porter's laboratory, and I called those cells “modified ependymal cells” in the spirit of Keith Porter, Harvey Lecture (Porter, 1957). Lacalli et al. (1994) refer to these cells as “the lamellar body.”

In my study of adult Branchiostoma caribbaeum, the modified ependymal cells lie just below the Joseph's cells with their apical surfaces forming part of the lateral sides of the neurocoel (Fig.1). Like other ependymal cells, they bear cilia, but in these cilia, the ciliary membrane has been greatly expanded into a series of whorls; in other words, they resemble the outer segment of a vertebrate photoreceptor cell. Eakin and Westfall (1962) confirmed my observations, and made additional observations to show that the cilium was 9+0, missing the central pair and dyneins arms of the usual 9+2 pattern of motile cilia, (Satir, 1961) and that the 'whorls' were branches of the ciliary membrane, as expected in a vertebrate photoreceptor. I was unable to determine the basal extent of these cells, but they obviously lie above a rich neuropil that could easily become part of the first order multipolar neuron sensory pathway necessary for vision described by Butler. Eakin (1968) suggests, without further specification (but probably for the orientational reasons mentioned below), that the modified ependymal cells might be the evolutionary precursor of the median eye, a suggestion endorsed by Lacalli et al. (1994). Lacalli et al. (1994) make the case, which Butler echoes, that the “frontal eye” is the homolog of the vertebrate paired eyes, but the cilia of frontal eye are 9+2, and, if not simply motile cilia, are more structurally similar to olfactory cilia (see for example, Plate 47 in Porter and Bonneville, 1973) than to the 9+0 connecting cilium of the rod outer segments. (Porter, 1957; Satir, 1961).

Figure 1.

Cross section through the posterior cerebral vesicle of amphioxus. A) The large Joseph cells (Jc) are evident in the dorsal portion of the nerve cord. Below the Joseph cells, the modified ependymal cells are found with their 9+0 cilia-based outer segments (os) projecting into the neurocoel (nc). Bar = 100μm. B) High power view. Bar = 20 μm (from Satir, 1958. unpublished micrographs)

Regarding olfactory cilia, Lacalli and Huo (1999) have identified certain ciliated type II rostal epithelial cells in amphioxus that could be olfactory receptor cell precursors. Lacalli's 3D reconstructions of neural pathways in the amphioxus nerve cord are important evidence for the frontal eye hypothesis, but they are far from definitive (Lacalli, 1996). As the reconstructions show, in the posterior cerebral vesicle, two main functions of the neural activity “would appear to the concerned with photoreception and response to touch. The former involves the lamellar body and the ventral commissure, but the ultimate targets or fibres from the lamellar body have yet to be determined.”

Several papers from the Holland laboratory provide additional information on this subject. Stokes and Holland (1995) show that larval amphioxus orient themselves with respect to a light source so that the mouth points away from the light in such a manner that the pigment spot provides maximum shielding to the putative photoreceptive cilia of the anterior neurons of the frontal eye (or alternatively, so that the modified ependymal cilia are exposed to the maximum light intensity). In any event, this observation seems to confirm that there is a physiologically important anterior photoreceptor in amphioxus, a subject not without controversy. Most significantly, in the same vein as discussed by Butler for amphioxus AmphiOtx and Hox3 genes, Glardon et al. (1998) examined the expression of Pax-6 genes during amphioxus development. Pax-6 genes are expressed in photoreceptors as a master gene directing eye development and are also expressed in olfactory receptors, vertebrate anterior pituitary and pancreatic α and β cells (see also Fernald, 2000). Glardon et al. (1998) found that the amphioxus Pax-6 gene shows domains with high sequence identities to vertebrate and invertebrate Pax-6 genes, and is orthologous to vertebrate Pax-6 genes. Expression of Amphi Pax-6 in the central nervous system was detectable by in situ hybridization in the cerebral vesicle, but not more posteriorly. In the cerebral vesicle, there is limited expression in a few cells of the developing frontal eye, “but expression in the posterior half is ubiquitous”, encompassing the lamellar body (modified ependymal cells) and most of the neural cells.

An unusual aspect of the vertebrate lateral eyes, of course, is the fact that the retina is inverted — that is light passes through the neural layers of the retina and the cell bodies of the photoreceptors before striking and reacting with the photosensitive pigment of the outer segments of the photoreceptors. Figure 2 shows how inversion might take place, with concurrent transformation of the photoreceptive modified ependymal cells into a retina. This sort of scenario is actually a modest re-interpretation of Studnicka's theory from 1912 as quoted by Walls (1942). An interesting point is that the median (pineal) eye is not inverted, but fits into such scenarios as a transitional form, as discussed and illustrated in Walls (1942).

Figure 2.

Proposal for transformation of the modified ependymal cells of amphioxus into the inverted retina of the paired lateral eyes, based on Walls (1942). A) Cross section of the cerebral vesicle of amphioxus. Each line projecting into the neurocoel represents a connecting cilium exaggerated for emphasis; the squares are the expanded membrane segments of the photoreceptors. B) Transitional form (optic vesicle stage); The photoreceptive region has formed outpocketings on either side of the expanding brain. After Walls and Studnicka, who also propose a similar outpocketing at the roof of the ventricle that, without invagination, becomes the median eye. C) The first eye (beginning of invagination of the optic vesicle). The photoreceptive surface abuts the pigment epithelium narrowing the ventricle, and the optic stalk develops. In the larval lamprey, optic stalk formation and ventricular closure precede the invagination.(Walls, 1942). Changes in photoreceptor number and disc morphology (filled squares) are schematized. D) Schematic of the vertebrate eye emphasizing the inversion of the retina. Abbreviations: Jc Joseph cells; os modified ciliary outer segments of photoreptors; nc neurocoel; v ventricle; n neural retina; l lens

To summarize, the modified ependymal cells meet the criteria of a precursor of the vertebrate photoreceptor in four ways:

  • 1Ciliary ultrastructure — These are the only 9+0 sensory type modified cilia in the cerebral vesicle
  • 2Pax-6 expression
  • 3Photoreception (although the evidence for photoreception is weak)
  • 4Orientation and position consistent with development of an inverted vertebrate retina

Despite the orientational differences, the median eyes and lateral eyes of vertebrates may be serially homologous structures that evolved from a single common precursor.

I favor the hypotheses that the modified ependymal cells of amphioxus evolved into the medial eye and, more significantly, into the lateral eyes of vertebrates, the latter specifically because of the disposition of the modified outer segments with respect to the neurocoel of amphioxus.


I belatedly thank E. Lowe Pierce who supplied me with the living amphioxus used in my original studies. I very much appreciate the time spent last spring with Nicholas and Linda Holland at the Scripps Institution of Oceanography, where they kindly brought me up to date on their exciting new investigations on amphioxus referred to in this letter. Yuuko Wada provided vital help with the illustrations, for which I am grateful.

Biographical Information

Dr. Satir is a professor and chairman in the Department of Anatomy and Structural Biology at the Albert Einstein College of Medicine of Yeshiva University. His research focus is currently on cell biology neuroanatomy, development and evolution.