Electrophysiological and Histologic Assessment of Retinal Ganglion Cell Fate in a Mouse Model for OPA1-Associated Autosomal Dominant Optic Atrophy

Abstract

Purpose.: The main disease features of autosomal dominant optic atrophy (ADOA) are a bilateral reduction of visual acuity, cecocentral scotoma, and frequently tritanopia, which have been ascribed to a progressive loss of retinal ganglion cells (RGCs) and subsequent degeneration of the optic nerve. The main disease-causing gene is OPA1. Here, we examine a mouse carrying a pathogenic mutation in Opa1 by electrophysiological measurements and assess the fate of RGCs.

Methods.: Two-year-old animals underwent a full examination by electroretinography (ERG) and visually evoked potential (VEP) measurements to assess the function of the outer and inner retina and the optic nerve. Retrograde Fluorogold labeling was performed to determine the number of surviving RGCs and to assess axonal transport by neurofilament counterstaining. Phagocytosis-dependent labeled microglial cells were identified by an Iba-1 staining.

Results.: ERG responses were normal in aged Opa1 mice. VEP measurements revealed significantly reduced amplitudes but no change in the latencies in contrast to extended latencies found in glaucoma. Retrograde labeling of RGCs showed a significant reduction in the number of RGCs in Opa1 mice. Long-term experiments revealed the presence of microglial cells with ingested fluorescent dye.

Conclusions.: This is the first electrophysiological demonstration of a visual function deficit in aged Opa1 mice. VEP measurements and retrograde labeling experiments show that the number of RGCs is reduced whereas the remaining RGCs and axons function normally. Taken together, these findings support an ascending progress of degeneration from the soma toward the axon.

Autosomal dominant optic atrophy (ADOA; Online Mendelian Inheritance in Man [OMIM] 165500; http://www.ncbi.nlm.nih.gov/Omim/ provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) has an estimated prevalence of up to 1:12,000 in Denmark. Therefore it is one of the most frequent forms of Mendelian inherited optic neuropathies. ^1,2 The clinical hallmarks of this disease are a bilateral, progressive loss of visual acuity, often associated with visual field defects, color discrimination disturbances, and optic disc pallor. Histopathologic postmortem examinations of donor eyes associated these visual impairments with a variable loss of retinal ganglion cells (RGCs) and an atrophy of the optic nerve. ^3,4 A major gene locus for ADOA was mapped on chromosome region 3q28-q29 by linkage-analyses ⁵ and mutations in the optic atrophy gene 1 (OPA1) were identified as a cause for ADOA in succession. ^6,7 To date, more than 200 different pathogenic mutations in OPA1 are known. ⁸ This demonstrates that OPA1 is the main disease gene in ADOA as suggested recently. ⁹ The OPA1 gene is expressed ubiquitously ^6,10 with the highest expression in retina and brain. ¹⁰ OPA1 is a nuclear encoded, dynamin-related GTPase that is imported into mitochondria and localizes to the inner membrane. ¹¹ OPA1 has been assigned to two functions. In concert with mitofusins, it plays a major role in mitochondrial fusion and therefore is important for the maintenance of the mitochondrial network and morphology. ^12,13 Furthermore, OPA1 is linked to mitochondrial cristae remodeling in apoptosis. ^14,15 Although OPA1 is expressed ubiquitously, it is associated with an apparently organ-specific phenotype that is restricted to the visual system. Haploinsufficiency is believed to be the major pathomechanism in OPA1 gene–related non-syndromic ADOA, in contrast to syndromic forms of ADOA that are associated with mitochondrial deletions and presumably originate from a dominant effect of OPA1 mutations, as proposed recently. ^16–18 Previously, we reported on a first mouse model which carries a splice site mutation in the Opa1 gene (c.1065+5G>A; referred to as Opa1 mouse). Similar mutations could be identified in different patients with ADOA (c.1065+3A>C, ¹⁹ c.1065+2T>C, ²⁰ and c.1065+2T>G ²¹ ). This Opa1 mouse displays symptoms of human ADOA pathology, including a progressive loss of RGCs and optic nerve axons. ²² Histologic sections of old Opa1 mice resembled the human ADOA phenotype. ²² A second ADOA mouse model with a different mutation in Opa1 (p.Q285X) showed reduced visual acuity at 12 months of age. ²³ The study presented here extends these previous findings with respect to electrophysiological assessment of visual function and the analysis of the degeneration process. Visually evoked potential (VEP) measurements presented in this study provide first electrophysiological in vivo evidences for functional deficits in Opa1 mice. Patients with ADOA present with reduced amplitudes and single patients are reported with prolonged latencies in pattern visually evoked potentials (PVEP), suggesting a ganglion cell origin for ADOA. ^19,24–26 However, these findings were not uniform presumably because of the heterogeneous patient group and/or the absence of VEPs in the majority of patients. ²⁵ Retrograde labeling together with counterstaining against neurofilament and long-term examination presented herein demonstrate that RGC loss is in fact the primary assignable pathology for the clinical impairments in ADOA. Moreover, our data show that the underlying disease mechanism in ADOA is clearly different compared with that in glaucoma, where an increase in latency is observed. ²⁷ Taken together, our experiments demonstrate that RGCs are primarily affected in ADOA; something that has been hypothesized based on post mortem examinations already more than 25 years ago, but has never been explicitly proven so far. ^3,4  

Materials and Methods

Animals

Mice were kept in a 12-hour light (10 lux)/12-hour dark cycle with food and water available ad libitum in full-barrier facilities free of specific pathogens. Mouse breeding and all experimental procedures were done according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and approved by legal authorities. The Opa1^enu/+ strain (abbreviated Opa1 mouse) has been described and characterized in many aspects, including the genotyping, previously. ²⁰ All examined mice were at least 20 months old. The control group consisted of littermates of the examined mice without any Opa1 mutation. Animals were dark-adapted for 24 hours before all electrophysiological measurements, the cornea de-sensitized with a drop of local anesthetic (Novesine; Novartis Ophthalmics, Basel, Swizerland), upper eyelids retracted slightly by a surgical silk thread, and pupils dilated (Tropicamide; Novartis Ophthalmics). Cooling down of the body temperature during anesthesia was prevented by a heated platform (37°C). A gold wire electrode placed in the mouth served as reference and a stainless steel needle inserted into the tail as ground. 

Electroretinography (ERG)

Mice were anesthetized with ketamine (120 mg/kg) and xylazine (10 mg/kg) administered intraperitoneally (n = 11 for both wild-type and Opa1 mice). Gold wire rings placed onto the corneas of both eyes served as working electrodes. Standard ERGs were recorded simultaneously on both eyes, with scotopic flash ERGs from 0.0003 to 100 cd · s/m² (filter setting: 0.5 to 200 Hz; interstimulus interval between 15 seconds and 170 seconds) and an additional run for scotopic oscillatory potentials (filter setting: 50 to 500 Hz) on a commercial ERG system (RetiPort32; Roland Consult Systems, Bradenburg, Germany). After 10 minutes of light adaptation (25 cd/m²), photopic 30 Hz Flicker (filter setting: 10 to 50 Hz), photopic flash ERGs and photopic oscillatory potentials (all with 100 cd · s/m²) were measured. The time of measurement was 160 ms, always 512 data points per measured waveform. In addition, the waveforms of the oscillatory potentials were digitally filtered off-line using a DSP filter included in the software of the ERG device (−15 dB for f<10 Hz). Amplitudes of a-waves were measured from the baseline to the bottom of the a-wave; b-wave amplitudes were measured from the bottom of the a-wave to the peak of the b-wave. 

Visually Evoked Potentials

Animals (n = 10 and n = 9 for wild-type and Opa1 mice, respectively) were anesthetized with an intraperitoneal injection of chloral hydrate in physiologic saline (0.42 mg chloral hydrate/kg body weight). A self-tapping stainless steel screw (shaft diameter: 1.12 mm; length: 3.18 mm; Y-TX00-2; Small Parts, Inc., Miramar, FL) inserted more than one week before the measurement through the skull into the left visual cortex (1.5 mm laterally to the midline, 1.5 mm anterior to the lambda), penetrating the cortex to approximately 1 mm, served as a measuring electrode. For measurements, the skin was opened, the screw head cleaned of connecting tissue and blood, and connected to the amplifier. Both eyes were stimulated for the measurement. Scotopic and photopic flash VEPs were recorded using the same parameters as for corresponding ERGs, at a frequency range from 0.5 to 200 Hz. After finishing the measurements, the wound was stitched and antibiotic ointment applied. The averaged signals of each group were compared with respect to both latency and amplitude. 

Numerical Evaluation and Statistics

Parameters of light responses were computed by using the well-known relationship by Hood and Birch, ²⁸ based on the theory by Lamb and Pugh ²⁷ :  

 

with i = stimulus intensity; t = time after flash onset; P3 = sum of individual photoreceptor responses; Rm _P3 = saturated response; t _d = delay that allows for biochemical and other recording latencies; S = sensitivity parameter. 

The parameters for equation 1 were computed using a self-written nonlinear least-square routine (Turbo Pascal; Borland, Cupertino, CA). We performed individual fitting by using the data of each single eye, and averaged the values of the parameters obtained this way. The Rm _P3 value obtained for each animal at the highest light intensity was used for the calculation of the sensitivity parameters at the lower light intensities. ^28,30 The parameter t _d was left free to obtain the best fit, with a variation ranging between 2.7 and 5.0 seconds. 

We approximated measured a- and b-wave amplitudes to the parameters R _max,1 and R _max,2 (maximal response amplitudes), i _50,1 and i _50,2 (half-saturating light intensities), and n,1 and n,2—exponents describing the slope of the Hill function using the biphasic equation  

 

firstly introduced (without the exponent n) by Naka and Rushton ³¹ in the monophasic shape, with R = response amplitude and i = light intensity. All six parameters of the modified Naka-Rushton equation 2 were left free during the calculation using a self-written non-linear least-square routine in Borland Pascal. Eight data points were used for the fit, according to the eight different light intensities used for stimulation. 

Scotopic and photopic oscillatory indices (amplitudes or latencies) were calculated by adding amplitude and latency values of the first four oscillations. Values of the parameters obtained from both eyes were used and averaged with those from the corresponding group. All electrophysiological measurements were performed blind (ie, without the experimentator's knowledge of the animals' genotype). Comparisons of the values of the different parameters were performed using the two-tailed Student's t-test. 

Retrograde Labeling and Immunohistochemistry

RGCs were labeled retrogradely by injecting a solution (Fluorogold [FG]; Fluorochrome, LLC, Denver, CO; in PBS and 2% DMSO), or Dil (Molecular Probes, Eugene, OR, in Incomplete Freund's Adjuvant; Sigma-Aldrich, Munich, Germany) into the colliculus superior. One week (counting of RGC) or six weeks (labeling of microglia) postinjection, animals were enucleated, and retinas prepared and fixed in 4% paraformaldehyde in 0.1 M PBS (pH 7.4). Pictures were taken of randomly chosen areas of flat wholemounts on black nitro-cellulose filter paper (Sartorius, Goettingen, Germany) by a fluorescence microscope (Axioplan2 imaging; Zeiss, Jena, Germany) with the appropriate software (Openlab; Improvision, Tübingen, Germany). RGCs were counted in photographs of Opa1 wholemounts (n = 13) and control wholemounts (n = 6) after a random choice of an area of 0.025 mm², with discrimination between the peripheral and the mid-central parts of the retina. 

For staining of microglial cells, retinas from Opa1 and control mice (n = 3 for each group) were fixed for three days, washed in PBS and treated with Proteinase K for 20 minutes at 37°C. The enzyme reaction was stopped by 3% H₂O₂ for five minutes, followed by three washing steps with PBS for 10 minutes. Blocking was performed with 2% BSA, 0.1% NaN₃, and 1% Triton X-100 in PBS. Retinas were incubated in a 1:50 dilution of an anti–Iba-1 antibody (Wako Pure Chemical Industries, Ltd., Osaka, Japan) for three days at 4°C, then washed five times in PBS for one hour, blocked again for one hour, and incubated overnight at 4°C with an Alexa488 donkey anti-rabbit antibody (Invitrogen, Carlsbad, CA; 1:500). 

For neurofilament staining, whole eyes from Opa1 and control mice (n = 3 for each group) were fixed overnight in formalin, embedded in paraffin wax, cut to 5-μm sections, and deparaffinized according to standard procedures. Sections were pretreated with Proteinase K, blocked with 1x blocker (Power Block; BioGenex Laboratories, San Ramon, CA), and incubated with a monoclonal mouse anti-neurofilament 200 antibody (Sigma-Aldrich) following standard procedures (dilution of 1:40 in Dako antibody diluent; Dako, Glostrup, Denmark). After washing, sections were incubated with the secondary antibody for 40 minutes (Cy3-labeled goat anti-mouse; Invitrogen). Stained wholemounts and retinal sections were embedded (FluorSave; Calbiochem, Darmstadt, Germany) and microscoped. 

Results

ERGs in Aged Opa1 Mice

The responses of various cell types in the retina, including the photoreceptors and inner retinal cells (bipolar and amacrine cells), to light stimuli can be detected by electroretinography. Therefore, we used this technique to elucidate the nature of the visual impairments in the Opa1 mice. Previous studies on 2- and 9-month-old Opa1 mice did not reveal any abnormalities. ²² We now extended these studies to 2-year-old Opa1 mice. ERGs could be well recorded in the 20-month-old mice. The ERG waveforms of the wild-type and Opa1 mice are fairly similar, though not identical, as shown by two typical examples (Fig. 1A). There might be a weak trend toward smaller amplitudes of scotopic a-waves and b-waves and photopic b-waves in the Opa1 mice (Fig. 1B). However, there is no significant difference between Opa1 mice and wild-type mice in any of the calculated parameters of scotopic a-waves and b-waves, photopic b-waves, scotopic and photopic oscillatory potentials, or photopic 30-Hertz Flicker (not shown). There were also no significant differences in the parameters calculated by equations 1 and 2 between Opa1 mice and wild-type mice (not shown). This demonstrates that the functions of photoreceptors and first order neurons are not impaired as expected and shown previously. ²² The altered contribution of the RGCs to the ERG in Opa1 mice is not detectable with the ERG methods applied in this study, if there is any in the mouse at all. 

Figure 1.

 

A. Typical electroretinographic waveforms obtained from Opa1 (solid lines) and wild-type mice (dotted lines) for the indicated measurements. Light intensities used in scotopic ERGs are indicated nearby the waveforms. Please note the different scales of the amplitudes (black bars) close to the waveforms in the diagrams. B. Amplitudes of a-waves and b-waves are shown for Opa1 and wild-type mice (mean values ± SD).

VEPs in Aged Opa1 Mice

To assess the integrity of the visual pathway including RGCs, we measured VEPs in Opa1 mice and wild-type controls. Comparison of the VEP waveforms in response to different light stimuli revealed smaller amplitudes in the Opa1 animals (Fig. 2A). The amplitudes of scotopic VEPs in Opa1 mice were reduced significantly at light intensities above 0.03 cd · s/m² (Fig. 3A). The amplitudes of the photopic VEPs were not significantly reduced (P = 0.16) although there was a trend of reduced amplitudes in the Opa1 animals (Fig. 2B). N1 latencies of the photopic and scotopic VEP did not differ between the two groups (Fig. 3B), and also P2 amplitudes showed no difference (not shown). We calculated the Naka-Rushton parameters, which describe the dependency of the amplitude on the intensity of the light stimulus for P1-N1 amplitudes (Table 1). As with the amplitude-light intensity relation in the ERG data, a second term had to be added to the original Naka-Rushton equation to achieve a good approximation also with the VEP data, because the amplitude-light intensity relation showed a clear biphasic behavior (compare diagrams in Figs. 1B, 3A). One could speculate that the first summand in equation 2 represents the activity of the rods together with their post-receptoral systems, whereas the second summand is determined by the cones and their post-receptoral systems. The R _max,1 and the i _50,1 parameters differ significantly between wild-type mice and Opa1 mice, with R _max,1 values reduced to approximately 50% and i _50,1 values reduced to approximately 25% in the latter. No significant differences were found in the parameters of the second part of equation 2 that can be attributed to the photopic response. The calculated curves (first and second parts of equation 2 and sum of both parts) are displayed as lines in Figure 3A. 

Figure 2.

 

Comparison of typical VEP waveforms obtained in Opa1 (solid lines) and wild-type mice (dotted lines) for scotopic VEPs (A) and photopic VEPs (B). Light intensities used in scotopic VEPs are indicated.

Figure 3.

 

Diagrams showing parameters of VEP measurements in Opa1 (solid lines) and wild-type mice (dotted lines; mean values ± SD). In the diagram showing scotopic amplitudes (A), curves were inserted that were calculated using the biphasic equation 2 and the parameters shown in Table 1. The numbered curves represent the first and the second part of equation 2. The asterisks denote significance of the difference between Opa1 and wild-type mice as follows: *P < 0.05, **P < 0.01.

Table 1.

 

Calculated Naka-Rushton Parameters from VEP Amplitudes Obtained in Wild-type and Opal Mice

	Wild-type Mice	Opal Mice	P
R max,1	94.64 ± 33.12	47.59 ± 21.46	0.0032
n,1	0.922 ± 0.309	1.101 ± 0.924	n.s. (0.592)
i 50,1	0.0112 ± 0.0045	0.00345 ± 0.00329	0.0009
R max,2	26.57 ± 18.71	23.40 ± 19.12	n.s. (0.734)
n,2	4.324 ± 2.647	3.604 ± 2.371	n.s. (0.564)
i 50,2	7.98 ± 4.80	7.70 ± 5.32	n.s. (0.908)

 Values are mean ± SD. n.s., difference not significant.

RGC Fate in Labeling Experiments

RGCs can be labeled in a retrograde fashion by the administration of a fluorescent dye into the colliculus superior. We have used this technique previously to demonstrate the progressive loss of RGCs in Opa1 mice. ²⁰ Here, we correlate the results of the VEPs with the actual situation of the RGCs in the retina by FG labeling. Typical appearance of mid-central and peripheral regions of retinas from wild-type and Opa1 mice is shown in Figure 4A. Counting revealed a significantly reduced number of retrograde-labeled RGCs in Opa1 mice of 2754 RGCs per mm² (SD, 751) versus 4213 RGCs per mm² (SD, 226) for wild-type in the mid-central region of the retina. In the periphery, the numbers were 1735 RGCs per mm² (SD, 567) for Opa1 mice and 2901 RGCs per mm² (SD, 152) for wild-type (Fig. 4B). Notably, there is a considerable variability among the different Opa1 animals, which can be seen by the increased SD (Fig. 4B). 

Figure 4.

 

(A) Representative fluorescent micrographs of mid-central and peripheral regions of retinal wholemounts obtained after retrograde labeling of the RGC using Fluorogold. (B) Results of RGC counting are shown (mean values ± SD). Significance, ***P < 0.001.

To answer the question whether the reduced number of labeled RGCs results from an actual loss of RGCs or an impairment of the retrograde transport of FG, retrogradely labeled retinas were counterstained with an antibody directed against neurofilament (NF). A uniform NF immunoreactivity could be detected in the ganglion cell layer and the inner retina (arrowhead and asterisk in Fig. 5, respectively). No cells that were positive for NF without FG staining could be identified in the ganglion cell layer. This result indicates that reduction of the number of FG-labeled RGCs in fact reflects RGCs loss. We further tested whether dying RGCs were phagocytosed by the retinal microglia. For this purpose, we labeled the RGCs of 18-month-old Opa1 mice retrogradely with the lipophilic dye DiI, which persists inside cells over longer time periods. If a microglial cell would phagocytose a RGC, it should contain at least traces of DiI. This concept of a phagocytosis-dependent labeling was presented for the first time by Thanos et al. ³² Six weeks after retrograde labeling, we identified microglial cells by Iba-1–staining that showed vesicle-like inclusions of DiI fluorescence (Fig. 6). This demonstrates that RGCs are phagocytosed by microglial cells. 

Figure 5.

 

Fluorescence images of retinal sections of a 20-month-old Opa1 mouse with RGC retrogradely labeled with Fluorogold (FG; green). Red fluorescence indicates neurofilament (NF) immunoreactivity. In the ganglion cell layer, numerous cells can be seen showing NF immunoreactivity and FG fluorescence simultaneously (layer indicated by two arrowheads). The arrow points to a displaced RGC. NF immunoreactivity is also visible in the inner nuclear layer (asterisk).

Figure 6.

 

Fluorescence images showing examples of microglial cells in a retinal whole-mount of 20-month-old Opa1 mice (A–F) and controls (G–I) stained with Iba-1 antibody to identify microglial cells (green fluorescence). The arrows point to reddish vesicle-like inclusions within the microglial cells representing DiI originating from phagocytosed RGC retrogradely labeled with DiI. Scale bar, 30 μm.

Discussion

It has been shown that the Opa1 mouse is a bona fide model for OPA1-associated ADOA, although the development of the phenotype is late in age and not uniform in all animals. ²² We also found a considerable variability among the different Opa1 animals by retrograde labeling in this follow-up study of 20-month-old animals (Fig. 4). Subtle neurological abnormalities vary between the individual animals, too. ³³ The Opa1 mice were initially purebred C3HeB/FeJ animals that carry the Pdeb^rd1 allele leading to retinal degeneration. A genotype-assisted breeding scheme with one outcross on C57/Bl6 and subsequent mating of these F1 animals was applied to eliminate this allele. ²² Therefore, the Opa1 mice used in this study were not purebred. Taking this into account, the phenotypic variability may be explained by neuroprotective genetic modifiers, as described previously for a different model of RGC death. ³⁴ The action of genetic modifiers has also been attributed to the pronounced differences in expression of the disease symptoms in ADOA ranging from asymptomatic mutation carriers to severely affected patients within the same family. ^9,35  

A main question on the dissection of the degeneration of RGCs and their axons (ascending or descending) in OPA1-associated optic atrophy has not yet been resolved. In fact, retrograde labeling of RGCs, which has been used in prior studies, may not necessarily reflect loss of RGCs but could also result from a decreased or impaired axonal transport. Our findings of strict correlation between FG labeling and NF staining clearly demonstrate that the number of retrograde-labeled RGCs is reduced due to actual loss of RGCs (Fig. 5). Moreover, we could show that RGCs are phagocytosed by retinal microglia cells (Fig. 6). 

As shown previously, there are no morphologic signs of damage in other retinal layers than the ganglion cell layer. ²² Preservation of the other retinal layers is also reflected by the almost perfect analogy of the ERGs between Opa1 mice and controls presented here (Fig. 1). RGC responses to light stimuli are not reflected in standard ERGs as applied here. Therefore, we assessed the function of the RGCs and hence the integrity of the visual pathway directly by measuring VEPs. 

In scotopic VEPs, we observed a significant reduction of the amplitudes in Opa1 mice to 50% at higher light intensities, while we found only a trend toward smaller amplitudes in photopic VEPs (Figs. 2, 3). This was consistent with the calculated Naka-Rushton parameters (Table 1) and may suggest that mainly RGCs receiving their input from rods undergo degeneration, whereas RGCs connected to cones remain functioning and/or alive longer. But since the mouse retina comprises only 3% cones, the reduction of the signal from these cells just might not be picked up by VEP measurements. Of note, color discrimination is often impaired within the tritan axis in patients with ADOA (blue blindness). ^9,36 The blue cone photoreceptor is the least frequent in the human retina. So one could speculate that the personal visual perception—in contrast to the VEP measurements—might be more prone to alterations in these photoreceptors' responses. 

Patients with ADOA present with abnormal latencies and reduced amplitudes in the PVEP when recordable. ^24,25 In this study, we show that the VEP amplitudes are reduced in aged Opa1 mice but the latencies are not altered and that the ERG is normal. This implies that not the output from the eye (normal ERG) but the input into the brain (reduced VEP amplitudes) is reduced; in other words, the number of RGCs that transmit the signal has to be reduced. The retrograde labeling experiments demonstrate that the axons of the remaining RGCs are not retracted or disconnected. These results suggest an ascending progress of RGC degeneration from the soma toward the axon. The VEP latencies determined in wild-type and Opa1 mice are similar (Fig. 3B). This implies that the remaining RGCs function normally (ie, the collection of input from the photoreceptors via the interneurons and the transmission of action potentials via their axons to the brain). 

These findings are in line with a study in a Swedish family with ADOA that reported reduced multifocal VEP amplitudes together with a normal Ganzfeld ERG and a normal multifocal ERG. ²⁴ Another study in a British family with ADOA described prolonged latencies and reduced amplitudes in the PVEP measurements in contrast to this study. ^25,36 More sophisticated ERG measurements revealed also deficiencies in patients with ADOA by using pattern stimulation ²⁵ or checking oscillatory potentials and the photopic negative response. ³⁷ In our Ganzfeld ERG measurements, we did not find any reduction of a-wave or b-wave amplitudes. Amplitudes of oscillatory potentials were also not changed in Opa1 mice compared with wild-type mice. The amplitudes of photopic negative responses could not be determined because these responses were not visible in the waveforms obtained in our study. However, pattern stimulation as performed in humans and single flash stimulation as performed herein are limited in their comparability, not only due to the different technique, but also because the human retina is—especially with its macula—cone-dominated and therefore the transmission of nerve signals certainly differ between humans and mice. For example, it has been postulated that photopic negative responses in mice could be elicited by amacrine cells instead of retinal ganglion cells. ³⁸  

Our findings are also in perfect agreement with the results obtained for a different mouse model for optic neuropathies that has been created by an intravitreal injection of small interference RNA specific to the OPA1 RNA in mice. ³⁹ Whereas no difference was found in the ERG response between the treated versus the nontreated eye, a clear reduction of flash VEP amplitudes was found, without noteworthy changes in the latency. ³⁹  

Taken together, our results of reduced VEP amplitudes together with the unaltered ERGs suggest that a reduced number of RGCs transmit the light stimulus to the visual cortex and that the remaining RGCs are not impaired in their function. Our retrograde labeling experiments support this finding. This is different from other models of RGC degeneration. For example in glaucoma, the VEP latency is significantly increased. ²⁷ Aged DBA/2J mice, a model for pigment dispersion glaucoma, ⁴⁰ clearly present with decreased ERG amplitudes in addition to altered VEPs (Heiduschka et al., in preparation). This suggests that loss of RGCs in glaucoma is associated with a mechanical stress or inflammatory processes that affect the function of the RGCs and subsequently cause their death. Our results suggest that in ADOA, it is vice versa; the degeneration of RGCs is the initial event that leads to a secondary impairment of visual function by their loss. 

Footnotes

 Supported in part by the Fritz Thyssen Stiftung für Wissenschaftsförderung, Köln, Germany.

Footnotes

 Disclosure: P. Heiduschka, None; S. Schnichels, None; N. Fuhrmann, None; S. Hofmeister, None; U. Schraermeyer, None; B. Wissinger, None; M.V. Alavi, None

The authors thank Beate Leo-Kottler (University of Tübingen, Tübingen, Germany) for helpful discussion.