Transcription factors in eye development: a gorgeous mosaic?

Abstract

The dramatic physical differences between compound eyes of insects (such as flies) and camera eyes of vertebrates (like mice and men) support the orthodox view that these two eye types have evolved independently. This view is based not only on the obvious anatomical differences between these two eyes but also on striking physiological and cell biological differences between the photoreceptor neurons they contain. Vertebrate eyes are comprised of a single optical unit, with one lens projecting an image onto a continuous neural retina. The vertebrate retina not only includes photoreceptor neurons but also four types of interneurons (bipolar, horizontal, amacrine, and ganglion cells), all of which are considered to be part of the central nervous system (for review, see Saha et al. 1992). In contrast, insect compound eyes consist of many repeated unit eyes or ommatidia, each of which (in Drosophila) contain 12 accessory cells and eight photoreceptor neurons. The fly’s photoreceptors project directly to the optic lobes of the brain without any retinal interneurons, and the insect neural retina is considered part of the peripheral nervous system (Ready et al. 1976). Vertebrate photoreceptor cells carry their visual pigment (opsin) on intracellular membrane sacs in the rod or cone cell outer segments that are supported by a ciliary body, whereas insect photoreceptor cells carry their opsin on extracellular microvillae (forming an aggregate known as the rhabdomere) and have no ciliary bodies. Both types of photoreceptors use a G-proteinbased system of phototransduction but differ in their response to light. Vertebrate photoreceptor cells use a cyclic GMP-linked phosphodiesterase and hyperpolarize upon light reception, whereas insect photoreceptor cells use phospholipase C and depolarize upon stimulation by light (for review, see Zuker 1996). This extensive catalog of differences has led some to suggest that the eye has evolved at least 40 independent times and to use the eye as the archetype of convergent evolution (for review, see Land and Fernald 1992). Recently this view of the independent evolution of vertebrate and arthropod eyes has come under some doubt. The first (inferential) argument for homology followed from the spectacular series of discoveries of the functionally conserved homeobox gene clusters (for review, see McGinnis and Krumlauf 1992). This led to a view of the last common ancestor as far more complex than was previously supposed and that such a complex animal, complete with a head, was blind is difficult to believe (Slack et al. 1993). Although there have been other data in support of homology, such as the conservation of some transcription factor binding sites in opsin genes (Sheshberadaran and Takahashi 1994; Sheng et al. 1997), the most dramatic discovery was that the Drosophila eyeless gene encodes a homeodomain protein that is the functional homolog of the vertebrate small eye and Aniridia genes (also known as Pax6; for review, see Zuker 1994). The expression of Pax6 protein in Drosophila imaginal discs is necessary and sufficient to induce ectopic eyes, and Pax6 has been proposed as a candidate ‘‘master gene’’ for the visual system. It is likely, however, that things are more complex than that. Recently, another gene has been shown to produce ectopic eyes in flies (Shen and Mardon 1997). Moreover, the expression of Pax6 is not restricted to the developing visual system in mice, nor is Pax6 ectopic expression sufficient to induce ectopic eyes in transgenic mice (for review, see Hanson and Van Heyningen 1995). Thus, although the homology of vertebrate and arthropod eyes is not yet proven, there is certainly an active debate taking place and further data supporting that hypothesis are now available. In this issue, Fu and Noll present a paper on the isolation of the Drosophila gene sparkling, which has a role in the development of nonphotoreceptor accessory cells in the compound eye. Furthermore they show that the Sparkling protein is a homolog of the murine Pax2, which plays a role in the development of optic nerve glial (nonphotoreceptor) cells in the mouse eye (MacDonald and Wilson 1996). In the mouse, Pax2 is required for the development of non-neuronal optic stalk glial cells and also for the development of the inner ear (Torres et al. 1996). Fu and Noll (this issue) find that Sparkling is expressed in several nonphotoreceptor accessory cell types in the developing Drosophila eye and is required for their development. Despite the dissimilarity of these cells to the mouse glia, which require Pax2 function, and the lack of definitive data that sparkling is a functional hoCorresponding author. E-MAIL kmoses@mizar.usc.edu; FAX (213) 740-4787.