Abnormalities of the Photoreceptor-Bipolar Cell Synapse in a Substrain of C57BL/10 Mice

Abstract

purpose. A preferential loss of ERG b-wave was detected in a substrain of C57BL/10 mice. Electroretinographic and histologic techniques were used to investigate this hitherto unknown retinopathy.

methods. ERGs were obtained from normal and affected C57BL/10. C57BL/6 mice served also as controls. a-Wave and c-wave analyses were performed. Microscopic investigations were done at two different ages.

results. In the scotopic ERG, a severe reduction of the b-wave amplitude could be observed, whereas the a-wave was only moderately attenuated (“negative ERG”). With age, the a-wave amplitude further decreased, but the rate of reduction was comparable to normals. Oscillatory potentials were severely altered, and the photopic ERG was absent. The ERG c-wave was comparable to normal. ERG a-wave analysis also revealed a reduced maximum amplitude, but no significant difference of receptor sensitivity. Light microscopy revealed a thinning of all retinal layers but mostly of the outer plexiform layer. The number of photoreceptor nuclei was reduced by one third. Electron microscopy revealed a profound loss of ribbon-shaped synapses between rod and rod-bipolar cells and severely abnormal ribbons in the case of cones.

conclusions. The so-called negative ERG was associated with alterations in the synaptic layer between rods and rod bipolars. The absent cone ERG may be due to the altered cone-on-bipolar synapses. The overall thinning of the retina as well as the moderately reduced scotopic a-wave amplitude remain unexplained.

Mouse models of retinal degenerations provide a valuable means to investigate pathophysiology and therapeutic strategy. For some naturally occurring mouse strains the mutation is homologous to mutations found in human disease. ^{1 2} Recently, Pardue and coworkers ³ published a naturally occurring mouse model of X-linked congenital stationary night blindness (CSNB). One of the characteristic traits of this model was the missing b-wave in the scotopic electroretinogram (ERG), indicating a compromised postreceptoral neuronal activity. Because of the X-linked mode of inheritance and the normal retinal cytoarchitecture, as revealed by light microscopy, it was speculated that this mouse strain might be a model of complete X-linked CSNB. Other mouse models with known gene defects had shown that a missing or reduced ERG b-wave may be linked to a postreceptoral alteration. ^{4 5 6}  

Recording normal values for the ERG device, we by chance found a pronounced loss of scotopic ERG b-wave in commercially available C57BL/10 mice. These mice were considered to be normal, that is, there is no information available regarding defects of the visual system in this particular strain. ⁷ To elucidate the pathophysiology of this retinal abnormality, functional and morphologic investigations were performed. 

Methods

Mice

The mice used in this study were purchased from regular dealers of laboratory animals. The identification of affected C57BL/10 mice was possible with the first ERG, recorded at the age of 4 weeks. It showed unequivocally the presence or absence of the defect. Current findings imply that the trait is inherited in an autosomal recessive way (Fig. 1) . Males and females are affected in equal proportion. C57BL/6 and unaffected C57BL/10 mice were used as control animals. The mice were held in an animal laboratory with a 12-hour/12-hour dark-light cycle. All experiments were approved by the local Animal Use Committee and were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The number of animals investigated is indicated at each subexperiment. 

Basic Electroretinography

The methodology used to record the ERG has been described previously. ⁸ The mice were kept in darkness for at least 2 hours before examination. A longer adaptation time (12 hours) had no influence on the ERG characteristic (data not shown). Pupils were dilated by tropicamide 0.5% and atropine 1%. Xylazine, 20 mg/kg body weight, and ketamine, 40 mg/kg, were injected subcutanously (SC) for anesthesia. A monopolar contact lens electrode served as a recording electrode. The contact between the electrode and the eye was achieved by methylcellulose gel. However, for recording the a-wave, fiber electrodes were used to prevent any damage to the cornea that would have been caused by the extended recording time. A disadvantage of theses electrodes was a lower amplitude level compared with contact lenses. Silver needle electrodes fixed SC served as reference and neutral electrodes. While the ERG was being recorded, the mouse was placed into a commercially available Ganzfeld bowl (Toennies Multiliner Vision, Höchberg, Germany), the examined eye facing the back of the globe. The signal was amplified by 10,000 with a bandpass filter, including the range of 1 to 300 Hz. Noise level was 1.0μ V_eff from 0 to 10 kHz. The signals were digitized at a rate of 1.7 kHz. Oscillatory potentials were obtained by bandpass filtering from 100 to 300 Hz. The Ganzfeld stimulus was characterized by a duration of approximately 50 μsec and a color temperature of the white flash of 6000 K. The background light was calibrated by a Minolta (Ramsey, NJ) Spot–Luminance Meter and the flash by a Flash-Photometer Dk0295. In the dark-adapted state, a flash series consisting of six steps started at −2.4 log cdsm⁻² and reached 2.5 cdsm⁻². The first two responses were averaged five times (flash interval: 2 seconds) and the final responses two times (flash interval: 5 seconds). Subsequently, the oscillatory potentials were recorded (2.5 cdsm⁻², average of 3 responses; flash interval: 15 seconds). After 10 minutes of light adaptation (30 cdm⁻²), the photopic ERG was recorded (15 cdsm⁻², average of 20 recordings at 1.5 Hz). To control the intraindividual variability, three measurements were performed at each anesthetic session and averaged. The a-wave amplitude was measured from baseline to a-wave trough, and the b-wave amplitude was determined from a-wave trough to b-wave peak, behind the last prominent oscillatory potential. 

ERG c-Wave

Eight 4-month-old C57BL/10 mice with a negative ERG and eight 5-month-old C57BL/10 mice with a normal ERG (siblings) were examined. The recordings were taken in the dark-adapted state. A red light-emitting diode mounted in a Kooijman electrode (Roland Consult, Brandenburg, Germany) served as stimulus. It was located directly in front of the recording electrode. Stimulus duration was 250 msec, and three stimulus strengths were chosen (1.3, 1.6, and 1.8 log cdm⁻²). Because DC-recording produced too many artifacts, a bandpass filter between 0.08 and 20 Hz was chosen. 

ERG a-Wave Analysis

The high-intensity stimuli necessary for the analysis of the dark-adapted a-wave were generated in the Ganzfeld globe, which was equipped with a photoflash for this purpose (Ganzfeld manufactured by Toennies). This equipment provides flashes ranging in strength from 0.4 to 2.1 log cdsm⁻². Five normal and five affected animals (mean age, 5.8 and 5.6 months, respectively) were dark-adapted for at least 12 hours. Seven stimulus strengths were applied ranging from 0.4 to 1.9 log cdsm⁻² (1.27–2.74 log scotopic trolands). Above these energies the scotopic a-wave amplitudes did not change. At the two lowest energies two recordings were averaged. To prevent rod adaptation, no averaging was performed for the rest. The interstimulus interval was at least 2 minutes in duration. 

The analysis of a-wave data are related to the Lamb–Pugh model of phototransduction ^{9 10} :  

\[\mathrm{P3}(i,\ t){=}{\{}1-\mathrm{exp{[}}-iS(t-t_{\mathrm{d}})^{2}{]}{\}}\ Rm_{\mathrm{P3}},\ t{>}td\]

where P3 represents the mass response of rod photoreceptors. ^{11 12} The amplitude of this response is a function of flash energy (i) and time (t) after flash onset. S is the sensitivity and is a parameter of the gain of phototransduction, Rm _P3 is the maximum response, and t _d is a delay. This model helps to discern between mechanisms impairing the phototransduction process and a reduction of the rod disc area. A curve-fitting program, based on a program kindly provided by Donald Hood (Columbia University, NY) was used to fit the equation to the leading edge of the dark adapted a-wave. Data obtained at all flash energies were used to estimate Rm _P3. ¹³ For the estimation of S, a fit was made to each of seven flash intensities, holding Rm _P3 constant at the value derived from the ensemble fit. An elimination of a potential cone intrusion was not performed. The reason for this was that the smaller proportion of cones in the mouse retina, compared with humans, leads to a relatively small photopic a-wave in the mouse ERG. Lyubarsky and coworkers ¹⁴ estimated the cone a-wave amplitude of the mouse to be smaller than 10% of the rod a-wave. 

Microscopic Morphology

For light microscopy three C57BL/10 mice with and without the ERG characteristic were killed at the ages of 3 and 11 months. Electron microscopy was performed in one animal of each group at the same time point. The eyes were enucleated from the anesthetized animals and hemisected. The vitreous was removed, and the eyecups were placed in ice-cold paraformaldehyde (2%) and glutaraldehyde (2.5%) in PBS for 1 hour. The eyes were postfixed in osmiumtetroxide (1%) for 30 minutes and embedded in Epon. Semithin sections were stained with toluidine blue. Ultrathin sections were contrasted with uranyl acetate and lead citrate. 

Statistical Analysis

In the case of multiple testing at different stimulus energies, the analysis of variance (ANOVA) for repeated measures was performed. The two-factor ANOVA was applied when two factors simultaneously influenced the variables. If single ERG data had to be compared, the unpaired t-test or the nonparametric Mann–Whitney U test was applied; the respective method has been indicated in each case. The level of significance was set to P < 0.05. 

Results

Basic Electroretinography

Comparison between Affected and Not Affected C57BL/10 Mice.

Figure 2A 2B and 2C show scotopic and photopic ERGs of 5-week-old not affected, presumed heterozygous, and affected C57BL/10 mice, respectively. These ERGs are representative of the kinds of responses that were used to distinguish affected from not affected animals at 4 to 5 weeks of age. Although the a-wave in the scotopic ERG (top) was moderately reduced in affected animals compared with those not affected, there was a more pronounced attenuation of the b-wave leading to the picture of a“ negative” ERG. The oscillatory potentials (middle panel) were also severely changed in affected mice. There was no detectable cone response in the presence of a rod desensitizing background of 30 cdm⁻² (bottom panel). 

Comparing the data of 13 affected and 17 not affected 3-month-old C57BL/10 animals by repeated-measures ANOVA, there was not only a statistically significant difference between scotopic b-wave amplitudes (P = 0.0001), but also between a-wave amplitudes (P = 0.0004). The preferential loss of b-wave amplitude was also reflected by the b/a-wave ratio. At the stimulus strength of 1 cdsm⁻² in the dark-adapted ERG, the ratios were 0.55 in affected BL/10 mice and 2.1 in BL/6 mice (age, 3 months). No significant difference between wild-type and heterozygous mice could be detected from any of the ERG paramters (P > 0.05, ANOVA [scotopic ERG], t-test and Mann–Whitney U test for the remaining recordings). 

Comparison between Affected C57BL/10 Mice and Control C57BL/6 Mice over Time.

Unaffected C57BL/10 and C57BL/6 mice were compared by ERG up to the age of 14 months. No significant difference between the two groups could be detected (data not shown). To determine whether the affected C57BL/10 mice show a retinal degeneration over time or whether the condition is stationary, we looked at scotopic a- and b-wave amplitudes at the ages of 3 and 6 months (10 animals) and at 14 months (4 animals). These ERGs were compared with data obtained from C57BL/6 mice with normal retinal function at the age of 3 months (10 animals) and 14 months (7 animals; there were only four intensities available in these mice). Figure 3 (top) shows the results for the scotopic a-wave amplitude. There was no significant difference between the a-wave amplitudes of affected BL/10 mice at the ages of 3 and 6 months (ANOVA, P = 0.26), but there was a considerable loss of amplitude between the ages of 6 and 14 months (ANOVA; P < 0.0001). However, BL/6 mice as controls also showed a decline of a-wave amplitude at the age of 14 months (P < 0.0001). 

Analysis of the b-wave data (Fig. 3 , bottom) showed that the amplitude level of the affected BL/10 mice was much lower. In contrast to the a-wave, the b-wave amplitudes were different at 3 and 6 months of age (ANOVA, P = 0.0086). The b-wave amplitudes were found to be significantly lower in 14-month-old BL/10 mice (ANOVA; P = 0.0003) compared with 6-month-old mice. As with the a-wave, a significant amplitude decline could also be observed in BL/6 mice at the age of 14 months (ANOVA, P = 0.0001). From the age of 3 months to 14 months, the a-wave amplitudes at 1.0 cdsm⁻² had dropped to 43% in BL/10 and 47% in BL/6 mice. In contrast, the b-wave amplitudes were attenuated to 34% in BL/10 and 54% in BL/6 mice. Using the two-factor ANOVA, only the b-wave amplitudes appeared to be influenced by an interaction of age and strain (a-wave: P = 0.24; b-wave: P = 0.019). 

a-Wave and b-wave amplitudes of C57BL/6 and affected C57BL/10 showed a decline, with the highest stimulus strength being 2.5 cdsm⁻². This effect was most probably due to adaptation. At the highest stimulus energies (1.0 and 2.5 cdsm⁻²) averaging should be avoided, and the interval between the recordings should be more than 5 seconds. Although this effect was not investigated systematically, it seemed to be comparable in normal and affected animals. 

ERG c-Wave

Figure 4 (top) displays c-wave recordings of 6-month-old normal and affected C57BL/10 mice at three different stimulus strengths. The most notable difference in the recording was the presence of a b-wave in the wild-type mice. There were no significant differences the sampled data of the eight affected and eight normal mice, either between the amplitudes (ANOVA; P = 0.12) or between the implicit times (ANOVA; P = 0.055; Fig. 4 , bottom). 

ERG a-Wave

ERG a-wave analysis was performed in five affected and five not affected C57BL/10 mice. In Figure 5 representative examples of the a-wave recording and the fit are shown. The values of Rm _P3 and t _d derived from the ensemble fit (all stimulus strengths integrated) are listed in Table 1 . The value of log S is a mean of single fits performed at seven stimulus strengths. The indicated P values refer to the nonparametric Mann–Whitney U test. There was no significant difference between the time delays (t _d) and the parameter of sensitivity, log S. In contrast, the maximum response amplitudes were significantly higher in normals compared with affected mice, which is in accordance to the basic ERG results. 

Microscopic Morphology

The fundus visualized by a scanning laser ophthalmoscope was unremarkable in affected animals. When the morphology of normal and affected C57BL/10 mice (Fig. 6) was compared, all retinal layers of the latter were reduced in thickness but not all layers were equally affected. The question arose whether this reduction was due to a decline of the number of photoreceptors or of outer segment discs. There was no detectable abnormality in the ultrastructure of the discs (data not shown). In the outer nuclear layer (ONL) the number of pericarya was reduced by about one third, whereas the changes of the inner nuclear layer (INL), inner plexiform layer (IPL), and the ganglion cell layer were less conspicuous at the light microscopic level. The most striking thickness reduction could be seen in the outer plexiform layer (OPL) (Fig. 7) . In general, the OPL is divided into a more intensely stained outer layer and a lighter stained inner layer. On an ultrastructural level, the former corresponds to cone pedicles and rod spherules (Fig. 7C) . This subdivision of the OPL was lost in the affected mice. In addition, some photoreceptor pericarya protruded into the OPL (Fig. 6) . 

When viewed by electron microscopy, the rod spherules in a normal retina are located in the external periphery of the OPL. Cone pedicles are more proximal, and adjacent to the INL the processes of horizontal and bipolar cells become visible. This highly ordered arrangement was disturbed in the affected retinas. The location of the cone pedicles varied. They were sometimes displaced to the external periphery of the OPL, thus lying between the rod spherules. The synapses of both receptor types, with their appropriate invaginating bipolar cells are shaped presynaptically by the ribbon and postsynaptically by the triade of two horizontal and one or two invaginating bipolar cells (Fig. 7C) . This specific arrangement was absent in the affected retinas. The cone pedicle ribbons showed a pronounced morphologic variability of their distribution and had a tendency to cluster in small groups of two or three ribbons. The characteristic invagination and the postsynaptic triade were not observed; instead, the ribbons showed only a loose connection to the membrane. The rod spherules never formed any ribbon (Figs. 7A 7B) . 

In normal retinas, ribbon synapses of the IPL are characterized postsynaptically by a dyade of a ganglion and an amacrine cell. In contrast, ribbons of 3-month-old affected retinas completely lacked dyades, whereas conventional synapses of the IPL were not affected (data not shown). Interestingly, in retinas of 11-months-old animals, some round and electron-dense structures occurred just at the sites where ribbons of dyades are expected to be (Figs. 7D 7E) . The comparison of 3- and 11-month-old affected animals did not show any additional changes beyond the normal aging process also observed in controls. The three animals investigated by light microscopy showed virtually identical results. 

Discussion

The mouse model presented here has some features in common with the incomplete Schubert-Bornschein type of human CSNB, for which the gene has recently been identified. ^{15 16} These features are as follows: there is the so-called negative ERG, that is, a predominant loss of b-wave amplitude in the scotopic ERG, an altered cone ERG, and a normal fundus. ^{17 18 19} Although there are reports about autosomal inheritance of the Schubert-Bornschein type of CSNB, ²⁰ in most cases it is believed to be X-linked. However, for the mouse strain described here, there is strong evidence that the trait is inherited in an autosomal recessive way (Fig. 1) . A morphologic study of CSNB retinas by Yamaguchi and coworkers rendered results that were comparable to our findings, with the eye donor being a woman (i.e., probably an autosomal disease). However, most of the human CSNB eyes examined morphologically had additional features such as old age and infiltrating carcinoma ²¹ or old age and glaucoma. ²² All these facts indicate that there is not sufficient evidence to consider this newly described mouse strain as a model for human incomplete CSNB. The recently published mouse model for the complete form of X-linked CSNB ³ has several different features compared with the model described here: (i) the reduction of the scotopic b-wave is more pronounced, (ii) the cone ERG is also negative compared with an absent cone ERG, (iii) the light microscopy is not remarkable, and (iv) the mode of inheritance is X-linked. 

Beside CSNB there are other human retinal diseases leading to a negative ERG (i.e., Aland Island Eye Disease, ²³ Unilateral Cone Dystrophy ²⁴ ). The effects of dystrophin gene mutations on the ERG in humans ^{25 26} also resemble the alterations found in the mouse model presented here. The presence of an ERG alteration associated with dystrophin mutations depends on the site of the gene alteration. Recently, evidence has been provided that the mutations at the 3′ end of the gene tend to cause a negative ERG. ²⁷ Dystrophin isoform Dp260 obviously plays the most important role. Interestingly, in the case of the rat retina, Dp260 is almost exclusively localized in photoreceptor cells. ²⁸  

Apart from playing a potential role as a model for a human disorder, the mouse strain described here should be regarded as an opportunity for a better understanding of retinal physiology. The most striking morphologic alterations occurred in the OPL. The ribbons of the ERG-negative animals showed an abnormal morphology and arrangement. In the IPL they were completely absent in 3-month-old mice, whereas in 11-month-old mice some rod- or club-shaped structures occurred. Those structures were probably degradation or depolymerization products of synaptic ribbons (reviewed in ^{Ref. 29)} . The most widely accepted view of synaptic ribbon function is the “conveyor belt” concept. According to this hypothesis, synaptic vesicles are transported to their docking sites at the presynaptic membrane. After exocytosis of a synaptic vesicle, rapid docking of the next vesicle may be necessary to avoid a noisy signal in starlight vision due to a pause in exocytosis. ³⁰ The ribbon could slide tethered vesicles to the docking site, which must then be reloaded within 40 msec. ³¹  

Recently, an alternative pathway for rod signals in the rodent retina has been suggested. ^{32 33} Classically, rod signals are known to travel via rod bipolar and AII amacrine cells to cone bipolar and ganglion cells. In addition, there is a direct signal transmission from rods to cones via gap junctions. The recent findings indicate that rod photoreceptor signals may bypass the rod bipolar cell and directly connect to OFF-cone bipolar cells through an AMPA glutamate receptor. In the study presented here, neither in normal C57BL/10 mice nor in the affected mice could any contact between rods and OFF-cone bipolar cells be detected. However, the existence of alternative pathways for rod signals may be an indication that the ERG abnormalities found in this substrain of C57BL/10 mice are not necessarily due to abnormalities of the classical synapse. 

The morphologic investigation showed that there were not only changes of the OPL. The ONL was less densely packed, whereas the outer segments appeared to be normal in shape. The reduced number of cells in the ONL might be the reason for the attenuated scotopic ERG a-wave. Notably, affected C57BL/10 as well as controls (C57BL/6) showed a decline of a-wave amplitude with time, which may be a consequence of aging. This has also been described by others. ³⁴ This decline was comparable (43% in affected C57BL/10, 46% in C57BL/6), so that there is no indication of a photoreceptor dystrophy in the ERG-negative mice. The b-wave amplitude decline with age seemed to be more pronounced in affected C57BL/10 than in C57BL/6 mice. However, the relatively low number of animals examined so far does not allow the assumption of a faster progression of b-wave amplitude loss than in normal mice. 

The ERG c-wave is believed to be a compound of the transient hyperpolarization of the apical RPE membrane (cornea positive) and the hyperpolarization of Müller cells (the so-called cornea negative slow P-III). ³⁵ Normal c-wave amplitudes in the negative ERG mice confirm the hypothesis that this dysfunction is mainly located at a postreceptoral layer. The reduction of the scotopic a-wave amplitude of affected mice in the basic ERG prompted us to use the Lamb and Pugh model of phototransduction to further analyze the a-wave. According to the basic ERG results and according to histology, we found a reduction of the amplitude parameter Rm _P3. However, the values of S in the affected mice were not significantly different from those of normals. As S is a parameter of the activation steps in the phototransduction cascade, although not specific, this result implicates that the transduction process in the outer segments is not handicapped in these ERG-negative animals. 

The location of the gene defect leading to the abnormalities found in this C57BL/10 mouse model is yet unknown. Studies concerning this issue are currently under way. As soon as the protein is known, it might be possible to learn more about the function of the retinal ribbon synapses. It will be important to know whether a structural gene at the synapse layer is involved or whether the alterations of the synapse are secondary to changes of the photoreceptor layer. A key question will be whether the excitatory transmitter is involved in the process. To answer this and other questions, further studies, especially biochemical and immunohistochemical, are certainly necessary. 

 Supported by Deutsche Forschungsgemeinschaft (Bonn, Germany), Grant Ru 457/6.

 Submitted for publication June 28, 1999; revised December 27, 1999, February 29, 2000, and June 23, 2000; accepted July 11, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Three generation pedigree. The not affected animals in F₀ are C57BL/6 mice and have normal ERGs. In generation F₁, the ERG was not recorded, but should be normal in the heterozygous mice. In generation F₂, 30 (24.8%; 15 females, 15 males) of 121 mice are affected, strongly suggesting autosomal recessive inheritance.

Figure 2.

 

Conventional ERGs of 5-week-old normal (A), assumed heterozygous (B), and affected C57BL/10 mice (C). Top: scotopic ERG; middle: oscillatory potentials; bottom: photopic ERG.

Figure 3.

 

Scotopic a-wave and b-wave V-log I-functions of affected C57BL/10 and C57BL/6 control mice at different ages.

Figure 4.

 

Smoothed c-wave recordings (top) at three stimulus strengths of affected (left) and normal (right) C57BL/10 mice. Bottom: c-wave amplitude and implicit time bar charts.

Figure 5.

 

ERG a-wave recording and analysis in normal (top) and affected (bottom) C57BL/10 mice. The noise ripples indicate the original recordings, whereas the smooth lines are the fit. The stimulus ranges from 0.4 to 1.9 log cdsm⁻².

Table 1.

 

Analysis of the ERG Dark-Adapted a-Wave

Normals					Affected
log Rm P3 (μV)	log S [s−2(t d − s)−1]	t d (ms)	log Rm P3 (μV)	log S [s−2(t d − s)−1]	t d (ms)
2.61	2.47	3.73	2.35	2.60	3.63
2.58	2.63	3.59	2.41	2.59	3.84
2.53	2.65	3.36	2.54	2.56	3.69
2.51	2.61	3.68	2.31	2.62	3.1
2.55	2.60	3.31	2.41	2.66	3.4
Σ/n 2.55	Σ/n 2.59	Σ/n 3.53	Σ/n 2.40	Σ/n 2.61	Σ/n 3.53
			P = 0.03	P = 0.83	P = 0.75

Figure 6.

 

Semithin sections of control (A) and ERG-negative Bl/10 mice (B), both 3 months old. The retina in (B) is reduced in its thickness, mainly because of the decrease of the photoreceptor layer, ONL, OPL, and IPL. The subdivision of the OPL in an intensely stained external and a lighter internal layer (A) is not to be seen in (B). Additionally some photoreceptor perikarya are displaced into the OPL (arrowhead in B). Scale bar, 40 μm.

Figure 7.

 

Ribbon synapses of the OPL of 11- (A) and 3-month-old (B) ERG-negative Bl/10 mice. The typical invaginations enclosing the processes of horizontal and bipolar processes are absent in cone pedicles, although they contain ribbons loosely associated with the plasma membrane (arrows). Rod spherules (arrowheads) neither contain ribbons nor do they form invaginations. (C) OPL of a 11-month-old Bl/10 mouse with a normal ERG. Ribbon synapses of cone pedicles (arrow) and of rod spherules (arrowheads) show the normal morphology with the invaginations of horizontal and bipolar cells. IPL of 11-month-old Bl/10 mice with normal (D) and altered ERG (E). In (D) the normal morphology of the dyadic ribbon synapse is shown (arrowheads). Instead of the ribbons, two rod- or spherule-like structures are formed in (E) (arrowheads). All scale bars, 1 μm; scale bars are the same in (A) and (B) and (D) and (E).

The authors thank Donald Hood (New York) for providing a software program for the a-wave analysis and continuous efforts to support us and Wolfgang Berger (Max-Planck Institut fuer Molekulare Genetik, Berlin) for his valuable advice concerning the genetic basis of this animal model.