Identification of a β-Dystroglycan Immunoreactive Subcompartment in Photoreceptor Terminals

Abstract

purpose. Mutations in the dystrophin-associated glycoprotein complex (DGC) cause various forms of muscular dystrophy. These diseases are characterized by progressive loss of skeletal muscle tissue and by dysfunctions in the central nervous system (CNS). The CNS deficits include an altered electroretinogram, caused by an impaired synaptic transmission between photoreceptors and their postsynaptic target cells in the outer plexiform layer (OPL). The DGC is concentrated in the OPL but its exact distribution is controversial. Therefore, the precise distribution of β-dystroglycan, the central component of the DGC, within the OPL of the mature chick retina, was determined.

methods. Double immunolabeling with antibodies against β-dystroglycan and against Bassoon, a component of the presynaptic cytomatrix, concentrated at the insertion point of the synaptic ribbon into the active zone of the photoreceptor synapses, showed a nonoverlapping distribution of both proteins within individual rod and cone photoreceptor terminals. The three-dimensional distribution of the DGC within the photoreceptor terminals was determined by reconstruction of the β-dystroglycan immunoreactivity from serial electron microscopic sections.

results. We found that β-dystroglycan was not directly associated with the ribbon synapse but instead concentrated perisynaptically in processes extending from the photoreceptors into the OPL. The processes displayed dystroglycan immunoreactivity primarily along their lateral sides and at their tips. Processes from bipolar or horizontal cells were not labeled.

conclusions. The perisynaptic concentration of β-dystroglycan in photoreceptor terminals suggests a novel domain within photoreceptor terminals with functions in synaptic transmission

Muscular dystrophies are a heterogeneous group of diseases characterized primarily by a progressive weakening and loss of skeletal muscle tissue. The genetic basis of the different forms of muscular dystrophy are mutations in proteins of the dystrophin-associated glycoprotein complex (DGC) or in enzymes that posttranslationally modify proteins of the DGC. ^{1 2} The DGC is a large multimeric complex with transmembrane, cytoplasmic, and extracellular proteins, which is best characterized in skeletal muscle. In muscle tissue, the DGC has at least two functions: it forms a molecular bridge between the extracellular matrix and the myofiber’s cortical actin cytoskeleton, conferring mechanical stability to the muscle fiber and allowing it to withstand the mechanical forces during muscle contraction and relaxation. ^{1 3} In addition, the DGC serves as a scaffold to localize and to concentrate a number of proteins, some of which have signal transducing activity, to the subsarcolemmal plasma membrane compartment. ⁴  

The central protein of all DGCs is dystroglycan, a cellular receptor expressed in a variety of tissues, with a complex structure and biosynthesis. ^{3 5} Dystroglycan is translated from a single gene transcript but posttranslational cleavage of the precursor protein by an unknown protease during or shortly after translation generates two subunits, α-dystroglycan and β-dystroglycan. ⁶ The β-dystroglycan subunit is a 43 kDa transmembrane protein with a single membrane spanning region. In contrast, α-dystroglycan is a heavily glycosylated extracellular protein with a molecular mass varying between 120 and 180 kDa, depending on the degree of glycosylation. In skeletal muscle, β-dystroglycan binds intracellularly with its C terminus to the β-spectrin/α-actinin-like peripheral membrane proteins dystrophin or utrophin. ⁷ Dystrophin and utrophin interact via their N termini with the subsarcolemmal actin filaments. ⁸ On the extracellular side, β-dystroglycan interacts noncovalently with α-dystroglycan. ⁹ α-Dystroglycan has the potential to bind to several extracellular proteins containing the laminin-G module, including laminin-1 and -2, agrin, biglycan, perlecan, and neurexins. ^{10 11 12 13} These molecular interactions form a continuous link between the myofiber basal lamina and the intracellular cytoskeleton. ¹⁴  

The expression of the different proteins of the DGC is not restricted to skeletal muscle but also occurs in other tissues, including the central nervous system (CNS). ^{4 15} Accordingly, muscular dystrophy patients frequently have CNS defects. ¹⁶ The deficits are very heterogeneous and range from mild mental retardation to severely altered neuronal migration, heterotopias, ectopias, increased ventricle volume, and lissencephaly, ¹⁷ demonstrating that the DGC has an important role in normal brain development and function. However, the precise role of the DGC in the developing and adult CNS and the cause for the CNS pathology in patients with mutated DGC proteins are unknown. ⁴  

The DGC is also expressed in the retina, and a frequent finding with muscular dystrophy patients, as well as with corresponding mouse models, is an altered electroretinogram (ERG), indicating that the DGC is required for normal retinal physiology. ^{18 19 20} The ERG is characterized by the delayed onset (time-to-peak response) and reduced amplitude of the b-wave under conditions of dark adaptation, suggesting an impaired synaptic transmission between photoreceptors and their postsynaptic cells in the outer plexiform layer (OPL). Accordingly, previous studies have shown that the DGC is concentrated in the retinal OPL, and considerable evidence in man and mice has been presented that the altered ERG is due to a disturbed synaptic communication within the OPL of the retina. ^{21 22 23} However, conflicting results exist with respect to the labeled cell types within the OPL and to the subcellular structures where the immunoreactivity is concentrated. Some studies reported a concentration of the DGC presynaptically in the terminals of rod and cone photoreceptors, ^{24 25 26 27} while others reported a postsynaptic concentration of the DGC in bipolar and horizontal cell processes. ^{25 28 29} Moreover, within the photoreceptor terminals, a concentration of the dystroglycan immunoreactivity directly at the ribbon synapse active zone has been postulated in rabbit retina as well as in retinas from dystrophin exon 52 knock-out mice and mdx ^3Cv mice. ^{21 23 25} In contrast, other studies found dystroglycan perisynaptically in rod and cone photoreceptors, concentrated in small extensions of the terminals protruding into the OPL, and at a certain distance to the ribbon synapses. ^{24 27} To analyze these conflicting results in more detail and to determine the precise three-dimensional distribution of the DGC within the outer plexiform layer, we reconstructed a larger part of the OPL from chick retina, labeled with antibodies against β-dystroglycan. 

Materials and Methods

Animals

Fertile White Leghorn chicken (Gallus gallus domesticus) eggs were purchased from a local hatchery and incubated at 38°C in a humid atmosphere. The stage of development was expressed as days after hatching (P). All experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and with the guidelines for the welfare of experimental animals issued by the Federal Government of Germany and the University of Mainz. 

Antibodies and Immunocytochemistry

The following primary antibodies were used: a monoclonal mouse anti-β-dystroglycan antibody (43DAG/8D5; Novocastra, Newcastle-on-Tyne, UK), which recognizes 15 of the last 16 amino acids at the C terminus of dystroglycan and cross-reacts with human, mouse, rat, bovine, rabbit, and chick tissue. ³⁰ This antibody has been extensively used in immunohistochemistry and is specific for β-dystroglycan in light and electron microscopy, as well as in Western blot analysis from skeletal muscle, retina, and other tissues. ^{23 24 25 26 29 31 32} In addition, we used a sheep anti-α-dystroglycan serum to verify the distribution of dystroglycan in the retina. ³³ Two polyclonal rabbit anti-Bassoon antisera detecting different epitopes (rb-anti-sap7f and rabbit anti-bassoon BSN1) were used to label the ribbon synapses in chick photoreceptors. ^{34 35 36} Application of both Bassoon antisera resulted in identical staining patterns in the chick retina. 

Newly hatched or adult chicken (P1) were killed by cervical dislocation. For light microscopy, eyes were opened along the ora serrata, and the eyecups were immersion-fixed for 10 minutes in 4% paraformaldehyde in phosphate buffer (PB; 0.1 M, pH 7.4). The retinas were then cryoprotected in 10% and in 20%, sucrose in PB for 1 hour each, and in 30% sucrose in PB overnight at 4°C. Pieces of the central-most part of the retina were frozen in freezing medium (OCT-compound; Reichert-Jung, Bensheim, Germany), sectioned vertically at 12-μm thickness in a cryostat, collected on gelatin-coated slides, air-dried for 15 minutes, and stored at −20°C. Immunocytochemical labeling was carried out after blocking in PBS that contained 1% bovine serum albumine and 0.05% Triton-X-100. The sections were incubated with the primary antibodies overnight at 4°C, and the primary antibodies were visualized by using goat–anti-rabbit and goat–anti-mouse secondary antibodies conjugated to Alexa488 or Alexa594 fluorochromes, respectively (Molecular Probes, Eugene, OR), as described. ²⁴ Specimens were investigated by using a Leitz DMRA fluorescence microscope equipped with a digital camera (DX 200; Leica, Solms, Germany) and an appropriate filter system for complete separation of both fluorochromes. Images were processed, including contrast adjustment, using image-editing software (Adobe Photoshop, version 5.5; Adobe, Mountain View, CA). Control sections stained with the secondary antibodies alone did not show any specific immunoreactivity. Preabsorbed antisera were used to avoid cross-reactivity between the secondary antibodies. Accordingly, in sections stained with only one primary and secondary antibody, no signal was detected in the corresponding other channel, indicating the absence of cross-bleaching. 

Electron Microscopy

For electron microscopic analysis of dystroglycan distribution in photoreceptor terminals, parts of the central region of the retina dorsal to the optic nerve head from postnatal day 1 chicken were prepared and stained as described by Blank et al. ²⁴ After cutting the optic nerve, the superior part of the eye was marked and the eye was removed. The eyecups were fixed in 4% paraformaldehyde and 0.05% glutaraldehyde in phosphate buffer for 10 minutes, followed by an additional 40 minutes in 4% paraformaldehyde in phosphate buffer. After dissecting out the retina and cryoprotection in 10%, 20%, and 30% sucrose, the tissue was repeatedly frozen and thawed to enhance the penetration of the antibody. ³⁷ After washing in PBS, small pieces of the retina were embedded in agar, and vertical sections (60-μm thick) were cut with a vibratome (Vibratome 1000; The Vibratome Company, St Louis, MO) for pre-embedding electron microscopic immunocytochemistry. The sections were collected in cold PBS, immersed for blocking for 2 hours in 10% normal goat serum in PBS, and then incubated with the monoclonal antibody 43DAG/8D5 for 4 days at 4°C, as described previously. ²⁴ Thereafter, the sections were rinsed in PBS several times, and immunolabeling was detected with a biotinylated goat anti-mouse IgG (diluted at 1:100; Vector Laboratories, Burlingame, CA) and a peroxidase-based enzymatic detection system (Vector Elite ABC kit; Vector Laboratories). After washes in PBS and in 0.05 M Tris-HCl (pH 7.6), the sections were preincubated for 10 minutes in 3,3′-diaminobenzidine (DAB; 0.05% in 0.05 M Tris-HCl, pH 7.6) and then reacted in 0.05% DAB with 0.01% H₂O₂. The staining reaction was stopped by rinsing the sections in Tris-HCl. Subsequently, the sections were washed in 0.1 M cacodylate buffer (pH 7.4), post-fixed in 2.5% glutaraldehyde in cacodylate buffer (2 hours at 4°C), and washed in cacodylate buffer overnight at 4°C. After several washes in distilled water, the DAB reaction product was silver-intensified and treated with gold chloride, as described. ³⁸ The sections were then post-fixed with 0.05% osmium tetroxide in cacodylate buffer for 30 minutes, dehydrated in a graded series of ethanol (30%–100%) followed by propyleneoxide, and flat-embedded in epoxy resin (EPON 812; Serva, Heidelberg, Germany). Serial 60-nm ultrathin sections were cut and stained with uranylacetate and lead citrate. Control Vibratome sections were processed in the same way as described above, except that the first antibody was omitted, resulting in no specific staining. Note that the fixation schedule and the tissue preparation represent a compromise between the needs for adequate tissue preservation and maintenance of antigenicity. 

Three-Dimensional Reconstruction

A series of 21 sections of the outer plexiform layer of the retina of a P1 chicken was micrographed at ×8000 on an electron microscope (Zeiss EM 10; Zeiss, Oberkochen, Germany). Negatives were scanned with the film adapter of a scanner (Epson Perfection 1640Su; Seiko Epson Corporation, Long Beach, CA) at a resolution of 1.600 dpi. The stack of resulting digitized images was aligned by using image-editing software (Adobe Photoshop). The best possible overlay of each image and its consecutive one (transparency set to 50%) was applied. Hereby at least three different constant tissue structures, e.g., a nucleus or a synaptic ribbon, were superimposed, manually correcting shift and rotation differences to achieve the best overlay. Due to technical limitations, few sections were lost during serial sectioning of the tissue. One section was missing between section number 1 and 3; two sections were missing between section number 8 and 13 and between 19 and 20. The missing sections were reconstructed by interpolation (overlay of neighboring sections). 

The resulting stack of 27 aligned images was converted into a volume data set. Only the region of interest was saved for further processing. The segmentation of relevant structures was performed for four cone terminals using a reconstruction program (VoxelCruncher, version 5.3; ConVis, Mainz, Germany). Cell membranes and synaptic ribbon outlines were redrawn, their volumes were marked by a fill function, and the labeling was marked with a pen, acting only in the defined low-gray value range typical for gold particles. The image stack of segmented structures was extracted and cropped to a smaller region of interest for easier handling. Until this step, tissue distortions were not corrected. This was performed now, resulting in a more accurate overlay. The newly aligned sequence of segmented structures coded in different gray values was imported into a 3D visualization and volume reconstruction software (Amira, version 3.1; Mercury Computer Systems, San Diego, CA). For three-dimensional visualization of the cells and the synaptic ribbons, a constrained smoothing algorithm that interpolates between sections was applied. Motion pictures were generated from sequences of images obtained when taking snapshots along a virtual camera path using a movie production software (AnimationShop3; Jasc Software, Eden Prairie, MN). 

Results

Distribution of Dystroglycan and Bassoon in the Retinal Outer Plexiform Layer

As a first step to analyze the distribution of the DGC in the outer plexiform layer, we labeled chick retinas with antibodies against β-dystroglycan, the central component of the DGC, as well as with antibodies against Bassoon, a component of the presynaptic cytomatrix. Bassoon has been shown to be concentrated at the anchoring site of the ribbon in the presynaptic membrane and the adjacent active zone of photoreceptor terminals. ^{36 39} Figure 1Ashows a vertical section through an adult retina double labeled with antibodies against β-dystroglycan (mAb 43DAG/8D5; red channel) and against Bassoon (rb-anti-sap7f; green channel). Note the generally punctate distribution of Bassoon and its concentration in two broad bands in the inner plexiform layer (IPL). ³⁹ In the outer plexiform layer, we observed two structures that were labeled by the monoclonal anti-β-dystroglycan antibody: approximately 4-μm-long sickle-shaped bands in the outer part of the OPL, and puncta approximately 1 μm in diameter in the inner part of the OPL. We have previously shown that the bands correspond to the labeling of cone photoreceptor terminals and that the puncta correspond to the labeling of rod photoreceptor spherules. ^{23 24} The nonoverlapping distribution of Bassoon and β-dystroglycan in the adult retina was evident at higher magnification of the outer plexiform layer (Fig. 1B) . Although the immunoreactivity for both proteins was closely apposed, the proteins did not directly overlap or colocalize. Instead, within a single rod or cone photoreceptor, the immunoreactivity of Bassoon was always more concentrated in the outer part, while β-dystroglycan immunoreactivity was more concentrated in the inner part of the same terminal. A similar distribution was also observed in P1 chick retina (data not shown). These results show that Bassoon and β-dystroglycan are concentrated at different levels within the depth of the OPL and suggest that β-dystroglycan is not directly concentrated at the anchoring site of the ribbon to the photoreceptor plasma membrane. Interestingly, we found no β-dystroglycan immunoreactivity at ribbon synapses of bipolar cells within the inner plexiform layer, indicating that dystroglycan is not concentrated at all ribbon synapses within the retina. 

To more directly compare the distribution of dystroglycan and Bassoon, horizontal sections through the OPL of a P1 retina were double labeled with antibodies against both proteins. As shown in Figure 1C , the Bassoon immunoreactivity appeared punctate, in some instances horseshoe-shaped, corresponding to the form of the attachment site of the ribbons in the photoreceptor active zone. ³⁹ In contrast, β-dystroglycan immunoreactivity was less punctate and appeared cloudy, filling the space between the Bassoon puncta. We observed little overlap in the distribution of both proteins within the OPL, indicating that both proteins have a differential distribution and that dystroglycan, although concentrated in the OPL, is not directly associated with the synaptic region and with the active zone of the photoreceptor ribbon synapses. 

To investigate if β-dystroglycan was concentrated in the pre- or postsynaptic side of the photoreceptors and to analyze the distribution of dystroglycan at the ultrastructural level within a complete photoreceptor synapse, the β-dystroglycan distribution was three-dimensionally reconstructed from serial ultrathin sections from a P1 chick retina labeled with the 43DAG/8D5 antibody. Figure 2Ashows a typical section used for the reconstruction of the terminals. In a single section like this one, β-dystroglycan immunoreactivity was not directly concentrated at the ribbon synapse active zone but appeared clustered at perisynaptic localizations. We did not find any evidence for dystroglycan immunoreactivity associated with postsynaptic bipolar or horizontal cell processes. Instead, the labeling was concentrated in short finger-like processes (arrows in Fig. 2 ), extending from the photoreceptor terminal into the OPL. However, within a single section some immunoreactivity could not be directly attributed to a particular cell and appeared as round isolated profiles (“islands”) within the OPL (arrowheads in Fig 2 ). Moreover, in individual sections, it was not possible to determine whether postsynaptic cells were devoid of labeling throughout their entire three-dimensional extension. 

We next investigated if the perisynaptic concentration of β-dystroglycan was present throughout the entire cone photoreceptor terminal and if all of the photoreceptor extensions contained dystroglycan immunoreactivity. To this end, serial sections through the OPL of a P1 chick retina were digitized and reconstructed after pre-embedding immunocytochemical labeling of dystroglycan and subsequent silver intensification and gold toning of the DAB reaction product. Figure 2Bshows the same electron microscopic section as in Figure 2Aafter marking of various structures, including nucleus, synaptic ribbons and the photoreceptor plasma membrane, to illustrate the process of data acquisition. A total of 27 consecutive serial sections were used to reconstruct a part of the OPL that contained 4 different cone terminals. Two of these cones formed a twin terminal with their surfaces facing each other, while the others were most probably S cones due to their size, location, and lack of telodendria. However, the sections of the series covered only parts of some terminals and, therefore, a precise discrimination of the cone type was not possible. One terminal (most likely representing a red or green cone that contained 9 individual ribbon synapses) was further analyzed in detail, and all subsequent figures represent different visualizations of this particular terminal. Figure 3Ashows an en face view of this terminal from the vitreal side of the OPL. The photoreceptor plasma membrane is labeled in gray, synaptic ribbons are marked in green, and the dystroglycan immunoreactivity is shown in orange. The majority of label was found intracellularly in the photoreceptor terminal. However, some immunoreactivity could be detected directly on the photoreceptor plasma membrane. Close inspection revealed that this label did not associate with horizontal or bipolar cells but instead was closely apposed to intensive intracellular labeling and, therefore, most likely resulted from the diffusion of the horseradish peroxidase reaction product into the extracellular space. 

Reconstruction of the cone terminals from the P1 chick retina revealed a very complex structure of the plasma membrane within the OPL, containing deep cavities as well as numerous short and few longer processes extending from the cone pedicle into the OPL. The cavities were directly underneath synaptic ribbons (marked in green in Fig. 3 ) and, thus, were completely filled by the processes of horizontal and bipolar cells, which were not visualized in Figure 3for a clearer view of the labeling. The complex morphology, however, precluded the detailed analysis of an entire cone terminal. We, therefore, analyzed the distribution of the β-dystroglycan immunoreactivity in smaller areas. One such area is shown in Figure 3B . Analysis of the spatial distribution of the β-dystroglycan immunoreactivity (shown in orange) within this area demonstrated that the labeling was almost exclusively associated with protrusions emanating from the photoreceptor terminal. Within these processes, the labeling was preferentially found at the tip and the lateral side but labeling of the top of the processes was also observed. Similar results were obtained by analysis of other parts of the same terminal, as well as by analysis of the other three partially reconstructed terminals (data not shown). In all cases, little if any labeling was found in the cavities of the terminals. Because these areas were filled by processes from horizontal and bipolar cells, our results indicate that these cells do not contain dystroglycan. 

To analyze the spatial distribution of β-dystroglycan immunoreactivity with respect to the ribbon synapses, a region with 5 individual ribbons of the same cone photoreceptor terminal analyzed in Figures 2 and 3was reconstructed (Fig. 4A) . This allowed the direct comparison of the β-dystroglycan immunoreactivity and the localization of the ribbons (Fig. 4B) . Selective visualization of the immunoreactivity (orange) and the synaptic ribbons (green) showed that the entire β-dystroglycan immunoreactivity was found at a certain distance to the ribbon and that none of the ribbons had dystroglycan immunoreactivity directly associated with them (Fig. 4B) . Note the cloudy distribution of the β-dystroglycan labeling in panel 4B, in agreement with the light microscopic distribution as shown in Fig 1C . Similar results were obtained when the other photoreceptors were analyzed (data not shown). These findings are consistent with the light microscopic data, demonstrating a nonoverlapping distribution of Bassoon and β-dystroglycan. The results also support the conclusion that β-dystroglycan is concentrated in photoreceptor terminals in a perisynaptic localization. 

We then analyzed the distribution of the β-dystroglycan immunoreactivity within a part of the cone terminal at higher magnification. To this end, an area of the terminal covered by 10 consecutive sections was reconstructed. Figure 5shows this region from two different viewing angles. The plasma membrane (but none of the immunoreactivity) of horizontal and bipolar cell processes were digitally removed for an unobscured view of the presynaptic terminal. Little if any immunoreactivity was observed directly associated with the plasma membrane (asterisks in panel B) surrounding the ribbon synapse (marked in green). Instead, throughout the reconstructed area, labeling was preferentially associated with the processes. Most if not all processes were labeled, and immunoreactivity was concentrated at their tips and at their lateral sides. In Figure 5A , a process extending from the cone terminal was outlined by a dashed line to illustrate its three-dimensional extension. This process bifurcated, with one branch extending directly into the OPL and the other bending horizontally parallel to the cone terminal. Bending of the process similar to that shown in Figure 5Awas frequently observed and represents the explanation for the immunoreactivity associated with the isolated profiles (“islands”) observed in single sections, including those marked by arrowheads in Figure 2 . The horizontal branch of the process showed β-dystroglycan immunoreactivity primarily at its tip, as well as along its lateral side. The isolated spot at the bottom of each panel in Figure 5(marked by an arrowhead) represents the tip of one branch of the process outlined by the dashed line in panel A. Note that this structure was strongly labeled by the anti-β-dystroglycan antibody. Comparable patterns of β-dystroglycan immunoreactivity were obtained after reconstruction of other parts of the same cone terminal, as well as after reconstruction of the two adjacent terminals. Thus, our results show that the β-dystroglycan immunoreactivity within the mature chick retinal OPL was exclusively associated with photoreceptor terminals, in particular, with the tips and the lateral side of processes extending from the terminals into the OPL and absent from the area directly adjacent to the ribbon synapse. Stereo images and motion pictures of the terminal together with several other supplementary data, demonstrating the localization of the immunoreactivity and the synaptic ribbons can be retrieved as supplementary material from the following Web page: http://www.staff.uni-mainz.de/jastrow/dystro/. 

Discussion

There is increasing evidence in man and mice that the altered ERG observed in muscular dystrophy patients and in corresponding mouse models is due to a disturbed synaptic communication within the retinal OPL. ^{19 21 40} It is, therefore, of considerable interest to determine the precise distribution of the DGC within this retinal layer. We present three different lines of evidence that within the OPL dystroglycan is exclusively found presynaptically in photoreceptor terminals and not associated with dendrites of postsynaptic horizontal or bipolar cells and that within the photoreceptor terminals, the immunoreactivity is not concentrated directly at ribbon synapses but instead is found perisynaptically. First, at the light-microscopic level, dystroglycan immunoreactivity was adjacent to, but did not overlap with Bassoon immunoreactivity, which marks the anchoring area of the ribbon to the presynaptic photoreceptor plasma membrane, ^{35 39} suggesting that the DGC is not directly associated with the ribbon synapse. Second, at the ultrastructural level, no immunoreactivity was associated with the invaginating bipolar or horizontal cell dendrites, demonstrating that β-dystroglycan in the OPL was exclusively concentrated in photoreceptor terminals. Third, three-dimensional reconstruction of several photoreceptor terminals confirmed that the active zone and the area directly adjacent to the ribbon synapses were devoid of label throughout their three-dimensional extension. Instead, β-dystroglycan was concentrated in processes, extending from the photoreceptor terminals into the OPL. Because the distribution of β-dystroglycan immunoreactivity in single sections of chick, mouse, and rat retinas was very similar, ^{23 25} it appears likely that the results from this study apply to all three species. The expression of β-dystroglycan mRNA by photoreceptors, ³² the appearance of dystrophin and β-dystroglycan immunoreactivity concomitant with the formation of presynaptic specializations within the photoreceptors, ³¹ the colocalization of dystroglycan with dystrophin and β-dystrobrevin, ^{29 31} together with the presynaptic concentration of dystrophin and the nonoverlapping staining pattern of dystroglycan and synaptophysin ²⁹ are consistent with our results. However, this distribution is in contrast to other studies that report a concentration of dystroglycan in bipolar and horizontal cell processes, as well as a direct association of β-dystroglycan immunoreactivity with the photoreceptor ribbon synapse. ^{25 26 29} Because most of the studies used the 43DAG/8D5 monoclonal antibody and because this antibody is specific for β-dystroglycan in all species tested by using Western blotting and immunohistochemistry, and because the staining pattern was similar, using antibodies directed against other proteins of the dystrophin-associated protein complex, different results are unlikely to be caused by cross-reactivity of the antibody to unrelated proteins. Moreover, we used the sensitive method of silver intensification combined with gold toning to detect β-dystroglycan immunoreactivity, and we analyzed the immunoreactivity within a major part of the OPL covering four cone terminals, each of which contained several ribbon synapses. Thus, it is equally unlikely that the differences in the distribution and subcellular concentration of β-dystroglycan are due to the sensitivity of the staining method or due to the analysis of only single individual sections. Instead, we consider two possibilities to explain the different results regarding the precise localization of β-dystroglycan in the retinal OPL. One possibility is that dystroglycan has a different distribution in different species. Whereas in mouse, rat, and chick retinas, β-dystroglycan immunoreactivity is concentrated perisynaptically in photoreceptor processes, the immunoreactivity in rabbit, bovine, and human retinas could be concentrated in the processes of bipolar and horizontal cells. The second possibility is that the distribution is variable even within one species and may depend, for example, on the day/night cycle. In this respect, it is interesting to note that the processes extending from the photoreceptor terminals, which contain the vast majority of β-dystroglycan labeling, have been detected in a number of different species and have been shown to be highly dynamic and plastic structures, forming and disappearing reversibly during light/dark cycles. ^{41 42} Moreover, the localization of dystroglycan has been shown to be variable, even in retinas of a single species, i.e., the mouse. While β-dystroglycan had a perisynaptic localization in normal wild-type mice, the mdx ^3Cv mouse strain, which carries a mutation in the dystrophin gene and has an altered ERG, showed a strongly reduced immunoreactivity with antibodies against β-dystroglycan and dystrophin, as well as a redistribution of the immunoreactivity and a concentration at the active zone, ²³ suggesting that the localization of the DGC is flexible and is dependent on structures within the photoreceptor terminal. On the other hand, there is no apparent correlation between dystroglycan distribution and preponderance of a certain photoreceptor type, i.e., a similar distribution of the dystroglycan labeling was observed in the diurnal chick retinas, which are cone dominated, and in nocturnal rat and mouse, which have a rod-dominated retina. ²⁵  

Photoreceptor projections into the outer plexiform layer have been described in the literature, ^{41 43} but their functional significance remains unknown. It, therefore, remains to be determined what the function of the DGC within the photoreceptor terminals might be. One possibility is that the DGC in photoreceptor terminals serves a cytomechanical function, similar to its role in skeletal muscle tissue. ¹ Photoreceptor processes change their shape during light adaptation ⁴¹ and the DGC, through its association with the cytoskeleton, could play a role in either the dynamics of these changes or in the maintenance of the mechanical stability of the photoreceptor terminal. An alternative function of the DGC in photoreceptor terminals might be, in analogy to skeletal muscle fibers, that it serves as a scaffold to subcellularly concentrate proteins in a particular domain of the photoreceptors. ⁴  

How can mutations in proteins of a complex with a perisynaptic concentration result in an altered ERG, as observed in patients with muscular dystrophy and in several mutant mouse strains? While it is reasonable to assume that the altered ERG b-wave is due to a disruption of the synaptic communication between photoreceptors and their postsynaptic cells, the perisynaptic localization makes it unlikely that mutations in the DGC directly disturb synaptic vesicle exocytosis. Instead, the concentration of the complex in the extensions of photoreceptor terminals suggests an indirect effect of the mutations on synaptic transmission. An alternative possibility is suggested by the binding of numerous proteins to the DGC. Thus, in muscle fibers, one function of the DGC is to serve as a scaffold for the attachment of proteins, including nNOS and several kinases. ^{44 45 46} The concentration of these proteins at a particular subcellular localization appears to be important for their function, because loss of this subcellular concentration due to a lack of the DGC at the sarcolemmal plasma membrane interrupts their signaling pathway, leading to the generation of certain forms of muscular dystrophy. In photoreceptor terminals, the DGC could have a similar function, i.e., it could serve as a scaffold for the attachment of proteins. One potential candidate molecule that has a distribution within photoreceptor terminals that is remarkably similar to the distribution of dystroglycan is the NMDA receptor subunit NR1C2′. ⁴⁷ In addition, photoreceptor terminals also contain glutamate transporters, ⁴⁸ which might be anchored to the plasma membrane at a certain distance to the ribbon synapse to rapidly remove glutamate from the synaptic cleft. Finally, flat contacts between photoreceptors and bipolar cells are also found at the noninvaginated membrane portions of the terminal membrane in mammals. Mutations in proteins of the DGC might interfere with this subcellular concentration, which, as a consequence, might influence the ERG. 

In summary, our results suggest that the altered ERG of patients with muscular dystrophy and corresponding mouse models is caused by an altered function of photoreceptor terminals and that the defect does not directly involve bipolar or horizontal cells. The data also suggest that the perisynaptic area surrounding the invaginating dendrites of bipolar and horizontal cells at the photoreceptor terminal represents a molecularly specialized site where the synaptic transmission can be modulated. 

 This study was supported by the Deutsche Forschungsgemeinschaft (Kr 1039/7), the Forschungsfond of the University of Mainz (SK) and National Eye Institute Grant EY014227 (PK).

 Submitted for publication May 15, 2005; revised July 13, 2005; accepted November 1, 2005.

 Disclosure: H. Jastrow, None; P. Koulen, None; W.D. Altrock, None; S. Kröger, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Distribution of β-dystroglycan and Bassoon in the chick retina. Vertical (A, B) and horizontal (C) sections through an adult (A, B) and a postnatal day 1 chick retina (C) were double labeled with the monoclonal antibody 43DAG/8D5 against β-dystroglycan (red) and with a rabbit antiserum against Bassoon (green). Labeling of Bassoon was concentrated in both synaptic regions, the inner and the outer plexiform layer (IPL and OPL, respectively). In contrast, β-dystroglycan was almost exclusively found in the OPL. The retinal inner limiting membrane, a basal lamina that separates the retina from the vitreous humor and that contains β-dystroglycan immunoreactivity, ^{24 31} was removed during preparation of the section. (C) A horizontal section through the outer part of the P1 OPL. The focus was adjusted to a plane containing primarily cone photoreceptor terminals. Note the little overlap between dystroglycan and Bassoon immunoreactivity in the horizontal section, indicating a differential subcellular distribution of both proteins within individual photoreceptor terminals. Scale bars: (A): 20 μm; (B and C): 10 μm.

Figure 2.

 

Localization of β-dystroglycan immunoreactivity in the OPL of the chick retina. (A) A high-power electron micrograph that shows the ultrastructural localization of β-dystroglycan in a cone photoreceptor of a P1 chick retina. The immunoreactivity was exclusively present in the innermost parts of photoreceptor processes facing the OPL and the labeling was predominantly localized in the cytoplasm, because the antibody reacted with the intracellular C-terminal peptide of β-dystroglycan. Note the presence of immunolabeling in finger-like processes extending from the photoreceptor terminals into the OPL (A and B, arrows), and in isolated areas with no direct connection to any particular cell within the plane of sectioning (A and B, arrowheads). (B) The same section as (A) with an overlay of the segmented structures: the synaptic ribbons (green), horizontal cell processes (yellow), the nucleus (red), the immunoreactivity (orange), and the terminal of the cone that was further analyzed (blue), as well as a neighboring terminal (lavender). Scale bars, 0.5 μm.

Figure 3.

 

Reconstructed cone terminal viewed from the OPL. The entire set of 27 reconstructed serial sections of a cone terminal from a P1 retina viewed from the OPL is shown (A). The cell membrane and the cytoplasm of the terminal were marked in gray, immunoreactivity of the anti-β-dystroglycan antibody is shown in orange and ribbons are labeled in green. The transparency was adjusted for better visualization of the intracellular structures, including the synaptic ribbons and the dystroglycan immunoreactivity. Note the complex surface of the terminal within the OPL (A), which precluded a more detailed analysis. (B) The area boxed in (A) at higher magnification. Note the numerous processes extending from the terminal into the OPL. β-dystroglycan labeling was associated with most if not all processes. The troughs within the photoreceptor terminals contained the active zones of the ribbon synapses (marked in green) and were filled with the processes of horizontal cells or bipolar cell dendrites. Both cell types were not visualized for clearer demonstration of the distribution of the dystroglycan immunoreactivity.

Figure 4.

 

Comparison of the spatial distribution of ribbon synapses and β-dystroglycan immunoreactivity. A part of the photoreceptor terminal from a P1 chick retina that contained 5 individual ribbons was reconstructed from serial sections. The photoreceptor plasma membrane (gray), β-dystroglycan immunoreactivity (orange), as well as individual ribbons (green) are shown (A). To directly compare the spatial distribution of the dystroglycan immunoreactivity with the localization of the ribbons, the cell volume and the plasma membrane were digitally removed in (B) and only the ribbons and the β-dystroglycan immunoreactivity of the same volume of (A) is shown in (B). Note that the β-dystroglycan immunoreactivity was found at a distance to the synaptic ribbon and that the space immediately adjacent to the ribbons was devoid of label.

Figure 5.

 

Part of a reconstructed cone terminal shown at different viewing angles. Ten consecutive sections were isolated from the reconstructed terminal of the P1 retina and viewed at higher magnification to determine the distribution of β-dystroglycan immunoreactivity within the photoreceptor processes. The areas directly adjacent to the ribbon synapses were marked by asterisks in (B). No immunoreactivity was detected along the entire ribbon, demonstrating the complete absence of dystroglycan from the synaptic regions of photoreceptor synapses. Instead, labeling was associated with processes extending from the photoreceptors into the OPL. One process was outlined by a dashed line in (A). This process bifurcated with one branch extending further into the OPL and the other running parallel to the cone terminal surface. The immunoreactivity was concentrated at the tip and along the lateral wall of the process. The arrowhead indicates the tip of a process extending into the OPL which, in the consecutive section, was in direct continuity with the branched process outlined by the dashed line in (A).

The authors thank Alfred Maelicke (Department of Physiological Chemistry, University of Mainz, Germany), Heinz Wässle (MPI for Brain Research, Frankfurt, Germany), and Eckart Gundelfinger (Leibniz Institute for Neurobiology, Magdeburg, Germany) for support and encouragement, Christa Ziegler (MPI for Brain Research) and Anne Rohrbacher (University of Mainz) for expert technical assistance, and Lim Nam and Walter Hofer (MPI for Brain Research) for help with the electron microscopy.