Cyclic nucleotides of cone-dominant retinas. Reduction of cyclic AMP levels by light and by cone degeneration

Transcription

1 Cyclic nucleotides of cone-dominant retinas Reduction of cyclic AMP levels by light and by cone degeneration Debora B. Farber, Dennis W. Souza, David G. Chase, and Richard N. Lolley Dark-adapted retinas or whole eyes of 13-line ground squirrels (Citellus tridecemlineatus) and western fence lizards (Sceloporus occidentalis) contain higher levels of cyclic AMP than of cyclic GMP. In these cone-dominant retinas, light reduces cyclic AMP content selectively. Freezing of dark- or light-adapted retinas or eyes also reduces cyclic AMP content, with only minimal changes in cyclic GMP levels. In addition, exposure of frozen retinas of dark-adapted ground squirrel to light results in a significant decrease in cyclic AMP content. The destruction of cone visual cells of ground squirrel retina by iodoacetic acid injection decreases the cyclic nucleotide content of the dark-adapted retina. Considering the relative loss of cyclic nucleotides from cone degeneration, we estimate that the content of cyclic AMP in visual cells of ground squirrel retina is about four times greater than that of cyclic GMP. Key words: cone visual cells, ground squirrel retina, iodoacetic acid-induced degeneration, cyclic AMP, cyclic GMP V,isual responses originate in retinal rod or cone photoreceptor cells of the vertebrate eye. In most animals, the mature retina contains a mixture of rods and cones, but some species have evolved so that one type of visual cell is in great excess. For example, mice and rats have retinas which are dominated morphologically by rods, whereas ground squirrels and lizards have cone-dominant ret- From the Jules Stein Eye Institute, University of California School of Medicine, Los Angeles, and the Developmental Neurology Laboratory and the Cell Biology Research Laboratory, Veterans Administration Medical Center, Sepulveda, Calif. Supported by NIH grants EY 2651 and RCDA 5 K04 EY144 to D. B. F. and by the Medical Research Service of the Veterans Administration. Submitted for publication March 4, Reprint requests: D. B. Farber, Developmental Neurology Laboratory (151B9), Veterans Administration Medical Center, Sepulveda, Calif inas. By careful choice of animal, the biochemistry of either visual cell type can be investigated in relative isolation. The biochemistry, morphology, and physiology of vertebrate retinal rods have been studied extensively. Cyclic nucleotides are distributed unequally throughout the layers of the rod-dominant retina, guanosine 3',5'- monophosphate (cyclic GMP) being more abundant by severalfold than adenosine 3',5'-monophosphate (cyclic AMP). 1 Studies of the developing retinas of mice and rats suggested that cyclic GMP was localized primarily in the rod visual cells. 2 This was confirmed by work with dystrophic animals, in which a genetic defect causes photoreceptor cell degeneration and blindness. 3 Moreover, the microdissection of retinal layers showed that 90% of the total cyclic GMP present in rod-dominant retinas is found in the photoreceptor cells and that most of it is 24

2 Volume 20 Number 1 Cyclic nucleotides of cone-dominant retinas 25 Table I. Cyclic nucleotides of dark-adapted cone- and rod-dominant retinas camp A (pmol/mg protein) cgmp A (pmol/mg protein) camp/cgmp Cone-dominant retina: Ground squirrel 8 (16) Western fence lizard (6) Rod-dominant retina: Mouse c (53) Rat (18) Toad D (8) 91.0 ± ± ± ± ± ± ± ± ± ± A Values represent the mean ± S.E.; number of retinas analyzed in parentheses. B Of total visual cells, 96% are cones. 13 c Of total visual cells, 97% are rods. 14 D Bufo marinus. concentrated in the light-sensitive outer segments. In contrast, cyclic AMP is distributed quite uniformly within all layers of the retina except the outer segment layer, which contains minimal levels of cyclic AMP. 1 The content of cyclic GMP in rod-dominant retinas as well as that in isolated rod outer segments 2 ' 4 ~ 6 is reduced by light, suggesting a role for cyclic GMP in the function or metabolism of rod photoreceptors. In contrast to rods, the biochemistry of cone visual cells has received relatively little attention despite the fact that these photoreceptors play a major role in human vision. Cones differ from rods in many respects, including the outer segment and synaptic morphology, time of shedding, and the visual pigments. In a preliminary communication, 7 we presented evidence that cone-dominant retinas differ substantially from those dominated by rods in the relative concentrations and responsiveness to light of cyclic AMP and cyclic GMP. Recently, DeVries et al. 8 reported that the distribution of cyclic AMP and cyclic GMP in layers of ground squirrel retina is similar to that in rod-dominant retinas. Moreover, these authors failed to detect an effect of light on the cyclic nucleotide content of cone-dominant retina. In this paper, we extend our previous observations, present data which explain the findings of DeVries et al., and show that cone visual cells contain about four times more cyclic AMP than cyclic GMP. Materials and methods Wild ground squirrels (Citellus tridecemlineatus) were trapped by commercial vendors in Illinois and Wisconsin, shipped to our laboratory, and maintained on a grain/nut diet. Western fence lizards (Sceloporus occidentalis) were obtained in California and maintained in our laboratory on a cricket diet. In order to minimize stress, ground squirrels were anesthetized with carbon dioxide several hours prior to an experiment, fitted into a body harness which permitted full movement of the head, and allowed to recover. Animals were either dark-adapted for at least 2 hr or exposed to laboratory illumination before sacrifice by decapitation. All operations in the darkroom were carried out under infrared illumination (FJW Find-R-Scope and infrared microscope converters) or dim red light (Kodak Safelight, Model C, and microscope illuminated with red light). Enucleation and dissection of the retina were done as quickly as possible, requiring routinely 20 sec for enucleation and about 1 min for retinal dissection. A whole eye or an isolated retina was homogenized in 600 or 300 /xl of 0. IN HC1, respectively. After centrifugation, the supernatant fractions were taken to 1.5 to 2 ml with 50 nim sodium acetate, ph 6.2, and duplicate aliquots were serially diluted with the same buffer prior to acetylation of the cyclic nucleotides. Cyclic AMP and cyclic GMP concentrations were measured by radioimmunoassay. 9 Protein was determined by the method of Lowry et al. 10 Variations in the protocol above were carried out in order to evaluate the effects of light and freezing or freeze-drying on the cyclic nucleotide content of retinas or whole eyes. To ensure that a bias from postsurgical trauma was eliminated, we

3 26 Farber et at. Invest. Ophthalmol. Vis. Sci. January 1981 Table II. Cyclic nucleotide content of dark- and light-adapted, cone-dominent retinas and eyes Retina Ground squirrel: Dark Light % reduction by light Western fence lizard: Dark (6) Light (4) % reduction by light Intact eye Ground squirrel: Dark (15) Light (6) % reduction by light Western fence lizard: Dark (6) Light (4) % reduction by light Cyclic AMP (pmollmg of protein) 91.0 ± 5.0 (25) 40.0 ±2.3 (18) 56.1 (p < 0.001) 21.6 ± ± (p < 0.005) Cyclic AMP (pmol/eye) ± ± (p < 0.001) 20.4 ± ± (p < 0.001) Cyclic GMP (pmollmg of protein) 11.1 ± 0.8(17) 9.9 ± 0.8 (10) 10.9 (N.S.) 9.6 ± ± (p <0.2) Cyclic GMP (pmol/eye) 41.0 ± ± (N.S.) 5.7 ± ± 0.5 N.S. = not significant. Values represent the mean ± S.E.; number of retinas analyzed in parentheses. alternated the order in which the enucleated eyes or dissected retinas were subjected to different experimental conditions. For example, in some cases the first eye was kept in the dark, and the second eye was exposed to light; in other cases the order was reversed. The effect of light was determined on samples prepared in the following manner. Dark-adapted eyes (ground squirrel or western fence lizard) were enucleated and, if desired, the retinas were dissected under infrared illumination or red light. (The levels of cyclic AMP and cyclic GMP did not differ significantly with either condition of lighting.) The cyclic nucleotides in these specimens were compared to those in dark-adapted eyes or retinas that had been exposed to laboratory illumination for 30 sec or to those of ocular tissues from animals that had been light-adapted for 30 min. Alternatively, dark-adapted retinas of ground squirrel were divided into three pieces: two were frozen, and the third piece was homogenized directly in 0. IN HC1. One dark-adapted piece of frozen retina was exposed to 30 sec of light before both frozen pieces were extracted with HC1 in the dark. Although the procedure above gave some information about the effect of freezing on cyclic nucleotide content of dark- or light-exposed ground squirrel retina, a separate series of freezing experiments was carried out with whole eyes or isolated retinas. For example, one eye or retina was immediately homogenized in HC1, and the other was frozen in liquid nitrogen before extraction with HC1. Some frozen eyes were freezedried after they were fractured into small pieces; samples of retina were dissected manually from these pieces of freeze-dried eyes. To obtain the retinal layers, frozen eyes were sectioned at -20 C in the light; samples of photoreceptor or combined inner layers were dissected manually by cutting the freeze-dried retina in the region of the outer plexiform layer. Pigment epithelium-choroid samples were removed from the eyecups after the retinas had been dissected and were processed for cyclic nucleotide measurements as described for the retina. The effect of iodoacetic acid on cone morphology and retinal cyclic nucleotide content was evaluated in a series of experiments with six animals. Ground squirrels were anesthesized with CO 2 and injected intracardially with iodoacetic acid (45 mg/kg body weight in physiological saline, neutralized to ph 7.0 with sodium hydroxide). Some animals tolerated a single injection of the drug, whereas others received the same dose divided into two injections that were given 24 hr

4 Volume 20 Number 1 Cyclic nucleotides of cone-dominant retinas 27 Table III. Effect of freezing on cyclic nucleotide content of ground squirrels retinas Dark-adapted retina: Fresh Frozen % reduction by freezing Frozen in darkness and exposed to light % reduction by freezing + light Freeze-dried in light % reduction by freeze-drying in light Light-adapted retina: Fresh Freeze-dried % reduction by freeze-drying Cyclic AMP (pmol/mg of protein) 91.0 ±5.0 (25) 69.1 ± 5.5(14) 24.1 (p < 0.02) 40.6 ± 1.5 (12) 55.5 (p < 0.001) 16.4 ± 2.9 (6) 82.0 ± (0 < 0.001) 40.0 ± 2.3 (18) 17.7 ±1.1 (17) 55.8 (p < 0.001) N.S. = not significant. Values represent the mean ± S.E.; number of retinas analyzed in parentheses. Cyclic GMP (pmol/mg of protein) 11.1 ± 0.8(17) 9.3 ± 0.3 (12) 16.2 (p < 0.1) 8.9 ± 0.5 (8) 19.8 (p < 0.1) 9.0 ± 0.4 (6) 18.9 (p < 0.2) 9.9 ± 0.8 (10) 9.2 ± 1.0 (17) 7.1 (N.S.) apart. Eleven days after injection of iodoacetic acid, dark-adapted ground squirrels were sacrificed by decapitation, and the retinas were dissected. A quadrant of the eye was prepared for histological analysis byfixationin glutaraldehydeosmium and by embedding in epoxy resin. The remaining retina was dissected from the eyecup and was homogenized in 0.1N HC1 before determination of cyclic nucleotide and protein content. Mice, rats, and toads were dark-adapted for at least 2 hr, and the cyclic nucleotide content of their retinas determined by radioimmunoassay. Results Cone-dominant retinas had higher levels of cyclic AMP than of cyclic GMP (Table I). The cyclic AMP/cyclic GMP ratios in the retina of the two species studied, ground squirrel and western fence lizard (8.2 and 2.2, respectively), are in contrast to those of rod-rich retinas of mice, rats, and toads, which are less than 1. Light reduced the cyclic AMP levels of dark-adapted retinas of the ground squirrel (Table II). We did not find any difference between the cyclic AMP content of retinas exposed to light for 30 sec in vitro and that of retinas from light-adapted animals. Cyclic GMP levels of the ground squirrel retina were not changed significantly by illumination. A similar response to light was observed in the cyclic nucleotide content of the retinas of the western fence lizard (Table II). In the isolation of ground squirrel retina, there is always the possibility of losing many of the outer segments since the cone cells break very easily at the base of the outer segments and these adhere firmly to the pigment epithelium cells. When cyclic nucleotide levels were measured in retinal pigment epithelium of dark-adapted ground squirrels, a considerable concentration of cyclic AMP was found, probably contributed by the outer segments. This cyclic AMP (24.3 pmol/mg of protein) was reduced by 35% after exposure to light. In contrast, the level of cyclic GMP in the pigment epithelium preparation was minimal, and it was not changed by illumination. In order to increase the speed of dissection and reduce the trauma imposed on the retina, the cyclic nucleotide content from the intact eyes of dark- and light-adapted ground squirrels has been measured (Table II). Similar to what has been observed in the dissected retina, light reduced by about 50% the cyclic AMP content of the dark-adapted ground squirrel eye, whereas the levels of cyclic GMP were not significantly affected. The cyclic AMP content of western fence lizard eyes was reduced selectively also by light (Table II). Freezing reduced the cyclic AMP levels of dark- and light-adapted retinas of ground squirrel. It is shown in Table III that when

5 28 Farber et at. Invest. Ophtlialmol. Vis. Sci. January 1981 Table IV. Effect of freezing on cyclic nucleotide content of ground squirrel eyes Dark-adapted eye: Fresh (15) Frozen (4) % reduction by freezing Light-adapted eye: Fresh (6) Frozen (4) % reduction by freezing Freeze-dried (3) % reduction by freeze-drying % reduction from dark-adapted, fresh eye by light and freeze-drying Cyclic AMP (pmol/eye) ± ± (p < 0.001) ± ± (p < 0.001) 51.9 ± (p < 0.005) 84.2 (p < 0.005) Cyclic GMP (pmol/eye) 41.0 ± ± (N.S.) 38.2 ± ± ± (p < 0.2) 27.6 (p <0.1) N.S. = not significant. Values represent the mean ± S.E.; number of retinas analyzed in parentheses. the dark-adapted retinas were frozen in liquid N 2 in the dark, the content of cyclic AMP decreased about 39%. Exposure of the frozen retinas to light after removal from liquid N 2 resulted in a further reduction (24%) in cyclic AMP content. This level of camp was essentially identical to that observed in lightadapted freshly dissected retinas (Table II). An additional depletion (19%) was observed after freeze-drying. Freeze-dried, light-adapted retinas also showed a considerable decrease in cyclic AMP levels (56%). In fact, retinas that had been frozen in the dark before freeze-drying in the light contained levels of cyclic AMP which were comparable to those of retinas exposed to light from the onset of enucleation (16.4 vs pmol/mg of protein). Retinal cyclic GMP content was minimally reduced by freezing, by exposing frozen retinas to light, or by freeze-drying. We estimate that dark-adapted as well as illuminated retinas contained, after freezing and exposure to light, about 18% of the cyclic AMP and 80% of the cyclic GMP content of freshly dissected retinas of dark-adapted ground squirrels. The stability of cyclic nucleotide levels has to be taken into account when one measures the distribution of cyclic nucleotides in microdissected layers of freeze-dried retinas of ground squirrel. In our hands, the photoreceptor layer from light-adapted, freeze-dried retina had slightly more cyclic AMP than the combined inner layers (22.7 ± 4.0 vs ± 2.5 pmol/mg of protein). Cyclic GMP was also more concentrated in the photoreceptor than in the inner layers (12.1 ± 1.2 vs. 5.4 ± 2.9 pmol/mg of protein). No correction for loss in cyclic nucleotide content during processing has been applied to these data. Freezing or freeze-drying also lowered the cyclic AMP levels of dark- or light-adapted eyes of ground squirrels (Table IV), whereas cyclic GMP levels might be reduced only by freeze-drying (p < 0.2). Overall, eyes that were freeze-dried in the light contained only about 16% of the cyclic AMP and 73% of the cyclic GMP that were present in the freshly enucleated, dark-adapted eye. The localization of cyclic AMP and cyclic GMP within the ground squirrel retina was studied by destroying the cone visual cells with iodoacetic acid. The loss of photo receptors was confirmed by light and electron microscopy. The morphology of retinas from control and iodoacetate-treated animals is compared in Fig. 1. Eleven days after an intracardiac injection of iodoacetic acid, the intact cone cells were replaced by a layer of macrophages containing large, irregular masses of dense material (Fig. 1, B). In the electron microscope, these masses were identified as the partially degraded residue of photoreceptors (not shown). This pattern was found in most areas of the retina, but photo-

6 Volume 20 Number 1 Cyclic nucleotides of cone-dominant retinas 29 10pm B I Fig. 1. Light micrographs of 13-line ground squirrel (Citellus tridecemlineatus) retinas. A, Control retina showing an area extending from just above the pigment epithelium to the inner synaptic layer. B, Full thickness of a retina fixed 11 days after an intracardiac injection of iodoacetic acid (45 mg/kg body weight in physiological saline, neutralized to ph 7.0 with sodium hydroxide). P, Pigment epithelium; C, cone cell layer; S, outer (above) and inner (below) synaptic or plexiform layers; N, inner nuclear layer; Q, ganglion cell layer; M, an irregular layer of macrophages. receptors, debris, and macrophages were absent altogether in sparse patches, so that pigment epithelium and inner nuclear layers were in direct contact. Although there were some changes in iodoacetate-treated retinas, such as the greater thickness of the pigment epithelium and the enlarged nuclei of the pigment epithelial cells, the nonphotoreceptor cell layers of the retina remained intact. Small sections of each dark-adapted retina were separated for verification of the pathology, and cyclic nucleotides were measured in extracts of the remaining retina. It was found that the levels of cyclic AMP were reduced from 91.0 ± 5.0 to 49.2 ± 0.2 pmol/mg of protein (approximately 46%) and that those of cyclic GMP were decreased from 11.1 ± 0.7 to less than 1.0 pmol/mg of protein. Apparently, most of the cyclic GMP in the ground squirrel retina is localized in the visual cells, similar to what is observed in rod-dominant retinas. After destruction of the photoreceptors, the inner layers of the ground squirrel retina contained about 54% of the cyclic AMP content of normal retinas. This suggests that there are significant levels (perhaps 50 pmol/ mg of protein) of cyclic AMP in the inner retinal layers and that cone visual cells may contain about 41 pmol of cyclic AMP per milligram of protein.

7 30 Farber et al. Invest. Ophthalmol. Vis. Sci. January 1981 Discussion Many aspects of cone morphology and biochemistry are different from those of rod visual cells. Our study suggests that still further differences might exist at the level of biological control systems. We find that freshly dissected retinas of dark-adapted ground squirrel have cyclic AMP levels severalfold greater than those of cyclic GMP. The fact that the cyclic nucleotide ratio observed in freshly dissected retina of western fence lizard is also higher than 1 supports the idea that cone-dominant retinas possess more cyclic AMP than cyclic GMP. It is surprising that freezing decreases selectively the cyclic AMP content of ground squirrel ocular tissue since, in brain tissue, freezing is used to prevent an anoxia-induced increase of cyclic AMP. 11 The loss of cyclic AMP during freezing is not understood. However, cyclic AMP might become accessible to degradative enzymes if freezing disrupts intracellular compartments, or perhaps freezing deactivates the synthetic reaction before that catalyzing cyclic AMP hydrolysis. The observation that exposure of frozen retinas to light results in a selective decrease in cyclic AMP content apparently supports the latter possibility. The reduction by light of the cyclic AMP levels in cone-dominant retinas (ground squirrel and western fence lizard) is perhaps analogous to the fall in cyclic GMP levels which occurs upon exposure of rod-dominant retinas to light. 2 4> ' 6 The ability of light to reduce cyclic AMP levels in fresh as well as frozen dark-adapted retinas of ground squirrel identifies a need for better control of illumination in quantitative histochemical studies. For example, microdissected retinal samples which have been frozen in the dark and freezedried in the light contain cyclic AMP levels which are diminished from those of darkadapted retinas by the effects both of freezing and light. We observe this to be true in our experiments, and we believe that this was the case in the work of DeVries et al. 8 In both studies, cyclic AMP levels in freeze-dried specimens represent less than 20% of the cyclic AMP in dark-adapted retinas. Moreover, dark-adapted eyes that are processed for quantitative histochemistry in the light are sensitive to this illumination. Thus, in the final stages of analysis, all retinal tissues have to be considered "light adapted." Therefore it is not unreasonable that DeVries et al. 8 observed no difference in cyclic nucleotide content between eyes frozen in the dark before processing in the light and those exposed to light from the onset of enucleation, because they were working only with lightadapted layers of the ground squirrel retina. Our findings reveal the conditions which mask the effect of light and those conditions in which light sensitivity can be observed. The final conclusion is clearly that conedominant retinas possess the capacity for light to trigger the selective hydrolysis of cyclic AMP. Moreover, our present data suggest but do not prove that the levels of cyclic AMP reduced by light are restricted to cone visual cells. In order to estimate how much of the retinal cyclic nucleotide content resides in the photoreceptors, we destroyed the visual cells. The earlier work of Noell 12 provided a mechanism by which cones could be destroyed without affecting the neurons of the inner retina. Iodoacetic acid injections caused selective degeneration of cones in ground squirrel retina. Biochemical analysis of the same retina from which pieces were taken for morphology showed that cyclic AMP and cyclic GMP both were decreased by degeneration of the visual cells. This observation indicates that both cyclic nucleotides are located in cone visual cells. The relative loss of cyclic AMP after iodoacetic acid treatment is about four times greater than that of cyclic GMP (41.8 vs pmol/mg of protein). Therefore a reasonable estimate is that cone photoreceptors contain about four times more cyclic AMP than cyclic GMP. In summary, cone-dominant retinas contain both cyclic AMP and cyclic GMP. However, only cyclic AMP levels are decreased by light. About 50% of the total cyclic AMP is present in the inner layers and may be associated with neurotransmitter function, e.g.,

8 Volume 20 Number 1 Cyclic nucleotides of cone-dominant retinas 31 dopamine, in the integrative neuronal circuits. The other 50% of retinal cyclic AMP and most of the cyclic GMP are localized in cone visual cells. The light modulation of cyclic AMP levels in cone-dominant retinas is perhaps analogous to the light-triggered changes in cyclic GMP levels that occur in rod photoreceptors. Therefore we suggest that cyclic AMP may serve as a regulator of metabolism or function in cone visual cells. We recognize the excellent technical assistance of Mrs. Elisabeth Racz and Mrs. Suni Kloss and thank Miss Louise Eaton for thoughtful assistance in the preparation of the manuscript. REFERENCES 1. Orr HT, Lowry OH, Cohen AI, and Ferrendelli JA: Distribution of 3',5'-cyclic AMP and 3',5'-cyclic GMP in rabbit retina in vivo: selective effects of dark and light adaptation and ischemia. Proc Natl Acad Sci USA 73:4442, Farber DB and Lolley RN: Light-induced reduction in cyclic GMP of retinal photoreceptor cells in vivo: abnormalities in the degenerative diseases of RCS rats and rd mice. J Neurochem 28:1089, Farber DB and Lolley RN: Cyclic guanosine monophosphate: elevation in degenerating photoreceptor cells of the C3H mouse retina. Science 186:449, Goridis C, Virmaux N, Cailla HL, and DeLaage MA: Rapid, light-induced changes of retinal cgmp levels. FEBS Lett 49:167, Fletcher RT and Chader GJ: Cyclic GMP: control of concentration by light in retinal photoreceptors. Biochem Biophys Res Commun 70:1297, DeVries GW, Cohen AI, Hall IA, and Ferrendelli JA: Cyclic nucleotide levels in normal and biologically fractionated mouse retina: effects of light and dark adaptation. J Neurochem 31:1345, Farber DB and Lolley RN: camp and cgmp content of cone-dominant retinas of ground squirrel, INVEST OPHTHALMOL VIS SCI 17(ARVO Suppl.):255, DeVries GW, Cohen AI, Lowery OH, and Ferrendelli JA: Cyclic nucleotides in the cone-dominant ground squirrel retina. Exp Eye Res 29:315, Harper JF and Brooker G: Femtomole sensitive radioimmunoassay for cyclic AMP and cyclic GMP after 2'-O-acetylation by acetic anhydride in aqueous solution. J Cyclic Nucleotide Res 1:207, Lowry OH, Rosebrough NJ, Farr AL, and Randall RJ: Protein measurement with the Folin phenol reagent. J Biol Chem 193:265, Steiner AL, Ferrendelli JA, and Kipnis DM: Radioimmunoassay for cyclic nucleotides. J Biol Chem 247:1121, Noell WK: Differentiation, metabolic organization, and viability of the visual cell. Arch Ophthalmol 60:702, West RW and Dowling JE: Anatomical evidence for cone and rod-like receptors in the gray squirrel, ground squirrel and prairie dog retinas. J Comp Neurol 159:439, Carter-Dawson LD, LaVail MM, and Sidman RL: Differential effect of the rd mutation on rods and cones in the mouse retina, INVEST OPHTHALMOL VIS SCI 17:489, 1978.