The mammalian photoreceptor mosaic-adaptive design
Introduction
Most multicellular animals have evolved with light sensitive cells or organs enabling them to navigate towards light at the simplest or to perceive chromatic, spatial and temporal variations of light at the most complex. The photoreceptor layer of the vertebrate retina is a highly specialized light sensor, made up of millions of individual photoreceptors for the sampling and computation of visual stimuli. Understanding the design of the sensory cells and their mosaic arrangement in the retina allows us a basis for understanding the organization of visual pathways in general and the degree of visual perception in particular for each species.
The photoreceptor mosaic of the primate retina, including our own, is one of the most highly evolved of all the mammalian families, thus giving us a wide range of visual stimuli to be appreciated and adaptation of our species to almost all ecosystems and light conditions. As in other fields of comparative research, we tend to project our own sensory capabilities and limitations onto other groups. Comparative studies could help us to revise our view of the world and appreciate that ours is only one of many possible ways that seeing may occur. For example many cold blooded vertebrates have photoreceptors with spectral capabilities beyond our own. During the course of this review, it will become evident that photoreceptor mosaic architecture is remarkably specialized to the particular environment in which the animal lives and optimally designed for survival of the species.
Some of the classic monographs on the vertebrate retina and its photoreceptors have provided us important information in the past, particularly about the design of non-mammalian, cold-blooded species’ retinas (Detwiler, 1939, Kolmer, 1930, Kolmer, 1936, Polyak, 1941, Walls, 1942). Compared to the diversity of photoreceptors in groups such as teleost fish or reptilian families such as geckoes (Crescitelli, 1972), mammalian photoreceptors are uninterestingly uniform and difficult to study. Besides being smaller and thinner than in non-mammalian species, mammalian photoreceptors cannot readily be identified into spectral types (Fig. 1(a)). In fact, in some retinas (e.g. bats, deer, squirrels) even rods and cones are not easily told apart, and until only a few years ago these retinas were thought to contain only one type of photoreceptor cell (Fig. 1(c) and (d)).
During the last two decades, several new methods have been developed which now allow us to more readily distinguish the different mammalian photoreceptor types, and analyze their photopigments, optical properties and mosaic organization. These efforts have of course mostly focused on our own species and closest relatives but gradually the horizon is widening. The present review attempts to provide a picture of recent advances in our understanding of the basic organization and ecologically adapted variation of photoreceptor mosaics across a broad spectrum of mammalian retinas.
Müller (1872), in the 19th century, established that there were two different types of photoreceptors. Based mostly on their shape and prevalence in diurnal and nocturnal mammals, the two classes were named rods and cones, “Stäbchen” and “Zapfen” (Schultze, 1866). They were associated with scotopic and photopic vision, respectively (Fig. 1(a) and (b); see (Polyak, 1941) for historical review). This dualism of photoreceptors was considered to be a general feature of all vertebrate retinas at that time, but it later became evident that the situation was not so simple. Based solely on morphological features, many species’ photoreceptor classifications turned out to be difficult. The diameters and shapes of inner and outer segments varied considerably, and in some groups one of the photoreceptor classes appeared to be missing altogether.
Yet the distinction between rods and cones is still largely based on morphology (Fig. 1). The distinction is clear in many primitive and ancient groups such as hagfish (Holmberg, 1970) but blurred in many more recently evolved groups such as those that have transformed their photoreceptors while switching their activity patterns once (e.g. geckos) or even twice (certain snakes). Consequently, a “Transmutation theory” was proposed by Walls (1942). Walls noticed that photoreceptors when in nocturnal animals lost the intensely colored oil droplets or even lacked oil droplets altogether in their photoreceptors when compared with diurnal species. Furthermore, he described the photoreceptors to become elongated in shape with extremely long outer segments in nocturnal species. It seemed that evolving families transformed their photoreceptors when switching their activity patterns.
It was not until the development of semi- and ultra-thin sectioning and newer histological staining techniques that we could really distinguish between rods and cones, in retinas where they had overall similar morphologies (Fig. 1(b)). Electron microscopy (deRobertis, 1956, Missotten, 1965, Sjöstrand, 1961) demonstrated the existence of two categories of photoreceptors in mammals based on a consistent set of additional features at the fine structural level (Fig. 1). We do now identify rod and cone photoreceptors primarily on outer segment disc morphology, and complexity of the synaptic terminal. Microspectrophotometry (Bowmaker, 1984, Crescitelli, 1972, Goldsmith, 1991), electrophysiology (Jacobs, 1993), and most recently, immunohistochemistry have extended our characterization of photoreceptors beyond the morphology. Most importantly, the genetic code for the vertebrate photoreceptor opsins, including human (Nathans et al., 1986) have been sequenced and classified (Kawamura and Yokoyama, 1998) so providing a tremendous advancement in our understanding.
Recent evidence has demonstrated that there are five opsin types in the vertebrate retina (not counting opsins expressed in the “third eye” the pineal, in certain species). It seems that the visual pigment classes are the conservative element during vertebrate evolution, irrespective of the epigenetic “rod-like” or “cone-like” morphology of their carriers. In some species, the spectral range of (rhod)opsin classes may be very finely tuned to their ecological niche (Bowmaker, 1998). For example, cottoid fish species living at different depths in Lake Baikal have pigments closely associated with the spectral shifts they experience in their environment. In fish, reptiles and birds, but also mouse, short wave sensitive opsins allow sensitivity in the violet or even ultraviolet range (Chiu et al., 1994, Jacobs et al., 1991). Closely related pigments such as the human red and green opsins (Nathans et al., 1986) or Coelacanth rhodopsins (Yokoyama et al., 1999) are cued to different spectral sensitivities by the alteration of only a few amino-acids. By comparing the sequences of the different visual pigments, it is possible to trace evolutionary adaptations among closely related species (Hunt et al., 1997). It will even become possible to genetically engineer visual pigments and reintroduce them into a species retina some day in the future (Yokoyama et al., 1998).
Thus, the classic “dualism’ of only two photoreceptor types in mammalian retinas, has to be expanded now to include the new visual pigment types being discovered. Fortunately, these new biochemical approaches can be combined with anatomical approaches, so allowing us to reevaluate the photoreceptor mosaic, not only in mammals but in all vertebrate retinas.
We now have an evidence that mammals evolved from a group of therapsid reptiles in the Jurassic period, the same geological period (ca. 200 million years ago) that saw the beginning of the dinosaur radiation (Fig. 2). Mammals remained inconspicuous for at least a further 50 million years and it was not until about 70 million years ago (the Triassic/Tertiary transition) that a diversity of mammals began to appear.
The cones of sauropsids (reptiles and birds) show a variety of morphologies (e.g. double cones and several types of single cones), different colors and sizes of oil droplets (Bowmaker, 1977, Cserháti et al., 1989, Kawata et al., 1992, Lipetz, 1985, Ohtsuka, 1985a), and extended ranges of photopigments (Yokoyama et al., 1993) (Fig. 2). This variety probably reflects a long history of diurnal lifestyles and photopic visual systems. The multiple pigment types and oil droplet filters enabled these groups to even have tetrachromatic vision extending beyond what is violet for human vision (Goldsmith, 1980, Ventura et al., 1999).
At first glance it is unclear how the design of the modern mammalian retina is related to that of its reptilian ancestors. Features such as oil droplets are found in non-placental marsupials so indicating a link to the past reptilian ancestry (Fig. 2). On the other hand, though, marsupial retinas have many features of mammalian retinas, particularly in their ganglion cell classification. Recent work in sequencing of vertebrate opsins has begun to aid us with this lineage problem.
Five major pigment classes are found among present day sauropsids (Okano et al., 1992) and many fishes (Vihtelic et al., 1999). Two of these — a medium wave sensitive (MWS, green), and a short wave sensitive (SWS-II, blue) type, were either absent in pre-mammalian-era reptiles or were lost during mammalian evolution because they are not now present in mammals. Besides rod opsin, there are two opsin classes in most mammals today: one belongs to the long wave sensitive (LWS, red cone) class and one belongs to a second short wave sensitive (SWS-I) cone class (Yokoyama, 1994) which can peak either in the ultraviolet (common among sauropsids) or more towards blue (common in diurnal mammals). Thus, there is presently a general agreement that the basic design of the mammalian retina is dichromatic (Fig. 2). Conditions differing from this pattern, i.e. tri-chromacy, is a further specialization during evolution.
The data from morphological, spectrophotometric and electrophysiology studies will have to be reorganized as the importance of species specific factors becomes appreciated. By studying primitive and specialized mammals, we may see phylogenetic trends and the constraints put upon mammalian retinal organization. The importance of factors that change during a species life history (e.g. hibernation, migration) may also determine species specific expression (Crescitelli, 1972, Crescitelli, 1990, Jacobs, 1993, Walls, 1942).
There seems to be a basic ontogeny of all mammalian retinas. Differentiation generally starts in the center of the retina and spreads concentrically towards the periphery. In spite of the apparent dominance of rods in most mammals, cones are generated first (LaVail et al., 1991 reviewed in Curcio and Hendrickson, 1991). This could reflect a phylogenetic scheme. Cone opsin sequences appear closer to a hypothetical common vertebrate photopigment than rod rhodopsins are (Eckmiller, 1997), and among the cones, it is thought that S-cones differentiate before L-cones (Bumsted et al., 1997, Wikler and Rakic, 1991, Wikler et al., 1997, Wikler and Stull, 1998). However, recent studies in monkey retina suggest that there are transitory stages of cone pigment phenotyping with expression of both L- and M-opsin in a primordial set of cones (Wikler and Rakic, 1994). Furthermore in some species (mice), there is a transdifferentiation from S-opsin to L-opsin expression (Szél, 1998, Wikler and Rakic, 1991, Wikler and Stull, 1998).
Thus, expression of a particular opsin in a particular cone may not be irreversibly determined. It may be a subject to genetic programming and local factors. A recent paper indicates that transgenic mice may not only express but also actually use an inserted human photopigment in their cones (Jacobs et al., 1999). Thus, while cases for “transmutation (between rods and cones) in the classical sense of Walls have largely found other explanations, “transdifferentiation” may occur regularly during ontogenesis. Studying these amazing properties may provide insight into the mechanisms of chromatic differentiation and may even direct us to future treatments of some forms of color blindness.
Section snippets
Identification of mammalian photoreceptors
The early mammals were well adapted to terrestrial lifestyles but during the course of evolution differences between species occurred to accommodate different environments. Some mammalian species adopted nocturnal or crepuscular activity patterns. This was accompanied by a major transformation of their visual system from cone- to rod-dominated, both in qualitative and quantitative terms. The rod-based scotopic visual system started to use cone pathways and developed specific bipolars and
Early mammalian history
Apparently the first mammals evolved in parallel to dinosaurs but played a marginal role during most of the Mesozoic. (Novacek, 1997). Such a prolonged initialization period for new groups is not unusual in evolutionary history but arising as it does simultaneously with the successful dinosaur radiations, it makes one speculate about a direct or indirect correlation between classes.
Optimization of cone photoreceptor morphology
The morphologies of the photoreceptors are constrained by the optics of the eye, and the direction, intensity and spectral bandwidth of the light entering the photoreceptor layer. Throughout the mammalian orders three main groups of morphological types of cone photoreceptors are seen. These are grouped by inner segment length and diameter in Fig. 13(a). Cluster I has the typical “rod-like” cone inner segments as found in the majority of mammals; cluster II is comprised of relatively shorter
Cone dominance, an adaptation for optimized motion detection?
Three mammalian groups have drastically reduced the rod mosaic and established cone dominant, or pure cone retinas or pure cone regions (foveas) of their retinas (Fig. 15, Fig. 16). They are the: simian primates, squirrels and tree shrews. Clearly, cone based vision provides higher spatial and temporal acuity and also, providing more than one spectral subtype of cone, color vision. However, edging out the rod system means giving up the ability to function well at very low light levels. The
Was there a prototype mammalian retina?
It seems clear, especially from comparative studies of photopigment amino acid sequences (Yokoyama et al., 1993), that mammalian ancestors possessed only three types of opsins, a rhodopsin, a short wavelength and a long wavelength cone opsin. Compared to the wealth of opsins in cold blooded vertebrates (Vihtelic et al., 1999) this is clearly a severely restricted set. Probably, the early mammals were nocturnal as we discussed earlier, and so an emphasis on the rod opsin, rhodopsin, would have
Future directions
In Walls’ classic monograph (Walls, 1942) only 15 of 134 pages are dedicated to the retinas of placental mammals. The lack of appropriate tools had left the impression that the mammalian retina (after the initial nocturnal specialization) has stayed relatively uniform, with few options for diversification. This static view was maintained until the recent development of the techniques for identifying spectral cone types and the resultant detailed analyses of the photoreceptor mosaics across many
Acknowledgements
We thank the following friends and colleagues for their contributions and support to this work over the years: E. Ahnelt, M. Blumer, L. DeKorver, E. Fernández; M. Glösmann, P. Goede, N. Hokoç, A. Kafka-Lützow, C. Keri, T. Klepal, A. Kübber-Heiss, L.Peichl, K. Moutairou, R. Normann, D. Pum, R. Pflug, H. Reitsamer, P. Röhlich, C. Schubert, A. Veres, L. Wündsch and Ken Linberg for critical comments on the manuscript.
References (395)
Characterization of the color related receptor mosaic in the ground squirrel retina
Vision Res.
(1985)- et al.
A morphological study of the retinal ganglion cells of the Afghan pika (Ochotona rufescens)
Neurosci. Res.
(1995) - et al.
Stiles-Crawford effect in garter snake
Vision Res.
(1989) The visual pigments, oil droplets and spectral sensitivitiy of the pigeon
Vision Res.
(1977)Microspectrophotometry of vertebrate photoreceptors. A brief review
Vision Res.
(1984)The visual pigments of fish
Prog. Ret. Eye Res.
(1995)Comparative studies of 3H-fucose incorporation into vertebrate photoreceptor outer segments
Vision Res.
(1980)- et al.
Evidence that circuits for spatial and color vision segregate at the first retinal synapse
Neuron
(1999) - et al.
Foveal cones form basal as well as invaginating junctions with diffuse on bipolar cells
Vision Res.
(1996) - et al.
Murine and bovine blue cone pigment genes: cloning and characterization of two new members of the S family of visual pigments
Genomics
(1994)