2009; 15:10-25 <http://www.molvis.org/molvis/v15/a2>
Received 27 September 2008 | Accepted 31 December 2008 | Published 9 January 2009
1Neuroscience Research Institute, University of California, Santa Barbara, CA; 2Department of Molecular, Cellular, and Developmental Biology, University of California, Santa Barbara, CA
Correspondence to: Steven K. Fisher, Ph.D., Neuroscience Research Institute, University of California Santa Barbara, Santa Barbara, CA, 93106-5060; Phone: (805) 893-3637; FAX: (805) 893-2005; email: firstname.lastname@example.org.
Purpose: To describe changes induced by retinal detachment in the ultrastructure and organization of rod terminals and their connections with B-type horizontal cell (HC) axon terminals and rod bipolar cell (RB) dendrites.
Methods: Sections from control, 3 day, 7 day, and 28 day detached feline retinas were prepared for confocal immunofluorescence, light microscopy, and electron microscopy (EM). In addition, 100 μm-thick vibratome sections were immunolabeled with markers for photoreceptor terminals, HCs, and RBs. More than 40 rod spherules were studied in 90 nm-thick serial sections by transmission EM to greater detail changes in their ultrastructure and innervation.
Results: Following retinal detachment, many rod terminals retracted varying distances toward their respective cell bodies in the outer nuclear layer (ONL). In retinas detached for 1 to 4 weeks, an altered synaptic vesicle population and associated ribbons were found in all retracting terminals. Many rod somata in the distal ONL seemed to lack synaptic terminal structures altogether. In a retina detached for 1 week, EM showed that less than half of the retracted terminals remain in contact with RB dendrites. In contrast, almost every surviving spherule was contacted by neurite outgrowths from the axon terminals of the B-type HC. Although retracted spherules had several presynaptic structures similar to those in normal retina, numerous changes occurred in their overall synaptic architecture. The spherule’s invagination was shallower, contained fewer postsynaptic processes, and often had “opened,” allowing swollen HC processes apposing the synaptic ribbon to directly contact other processes of the outer plexiform layer (OPL) neuropil. Whereas in normal cat retina each HC “lobe” comes from a different axon terminal system, after detachment, the opposing lateral elements can stem from the same terminal. The innervating RB dendrites that branched off stout RB dendritic trunks that extended up into the ONL were thinner than normal, unbranched, often electron dense, and lacked organelles. When present, most merely lay adjacent to retracting spherules rather than enter any synaptic invagination that might still occur.
Conclusions: Immunocytochemistry enabled RB and HC neurites to appear postsynaptic to retracted rod terminals. However, at the ultrastructural level, HCs seemed to more consistently retain connection with the retracted spherules than the RBs. The highly conserved architecture of the rod spherule was lost as the invagination opened and postsynaptic contacts became fewer. It would seem that the lack of RB central elements as well as the drastic alterations in the architecture of most retracted terminals would necessarily alter the physiology of this complex synapse.
Most central nervous system (CNS) neurons have multiple, often hundreds of presynaptic active zones. The mammalian rod photoreceptor is unique in its reliance on a single, complex presynaptic structure with a large active zone [1,2]. In feline retina where the rods number 400,000/mm2 over much of the retina , the postsynaptic targets of their terminals are the axon telodendria of the B-type horizontal cell (HC) and rod bipolar cell (RB) dendrites [4,5]. Several investigations have shown that most of the pre- and postsynaptic components of this spherule complex are essentially the same across mammalian species. For example, the height and width of the arcuate synaptic ribbon, the surface area of the active zone itself, and the volume of extracellular space within the deep invagination are virtually the same from cat to rat to human; moreover, unlike the output terminals of most retinal neurons, rod spherules change little in size with increasing distance from visual centers, such as a fovea or area centralis [1,6]. This highly conserved microstructure argues for a successful solution to the physiologic and energetic demands of a synapse where glutamate release is almost continuously varying with light intensity . The cytoarchitecture of this ribbon synaptic complex provides a mechanism for the high rates of sustained transmitter release by vesicle exocytosis [1,2,8-11]. The structural organization of the spherule underlies the sensitivity of the rod photoreceptor and its ability to rapidly accommodate changes in illumination . It would seem that changes in any component of this synaptic complex have the potential to alter the rod’s communication with second order neurons. Our earlier studies suggested that retinal detachment produces a widespread disorganization of rod terminals in the outer plexiform layer (OPL) with many retracting deep into the outer nuclear layer (ONL) [13,14]. These observations indicate that these synapses may be physically altered in ways that could prevent normal synaptic transmission.
There is a large body of existing literature on retinal circuitry in the feline retina [4,5,15-20]. Detailed information about the organization of its rod synapse [1,2,6] makes the cat retina a good choice for studying changes in rod connectivity as a result of detachment.
It has been shown previously that between 1 and 3 days after detachment, rod terminals begin retracting  and both HCs and RBs sprout neurites that grow into the ONL . While the stout RB outgrowths tend to be single, short, and unbranching, the processes from B-type HC axon terminals can occur singly or in clusters and can achieve great lengths [14,21], often growing out of the neural retina into the subretinal space. A similar pattern of terminal retraction and neurite outgrowth from HCs and RBs has also been reported in the late stages of the RCS retinal degeneration , retinal degenerations associated with the nob2 mutant mouse [23,24]; with the knockouts of bassoon [25,26], CaBP2 , and retinoschisin ; with the mid-stages of the double rod–cone knockout ; as well as in retinopathy induced by vigabatrin —all suggesting that they are common responses to photoreceptor degeneration. Moreover, the retraction of photoreceptor axons and terminals has also been reported in cultured porcine retinal explants prepared for transplantation studies , a phenomenon that can be prevented by treating the explants with a cAMP analog or the adenylyl cyclase stimulant, forskolin . Prior to this there has been no detailed high resolution studies of the transformations that rod synapses undergo as spherules pull back from the OPL or of the RB and HC axon terminal neurites (HCat) neurites that appear to grow toward them in what may be an attempt to maintain physiologic connection.
Experimental retinal detachments were made in adult cat eyes (Felis domesticus) as previously described . Briefly, a solution of sodium hyaluronate, consisting of 0.25% Healon in balanced salt solution (Pharmacia, Piscataway, NJ), was infused via a fine glass micropipette between the retina and the retinal pigment epithelium (RPE). The Healon is needed to maintain the detachment.
The experimental use of the animals was conducted in compliance with both the guidelines of the UCSB IACUC and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Cats were maintained on a 12:12 light dark cycle under ordinary room illumination. The specimens examined in this study stemmed from earlier investigations.
The samples consisted of 100 μm agarose-embedded sections of normal cat retina and retinas that had been detached for 3, 7, or 28 days. Tissue was fixed and held in 4% paraformaldehyde in sodium cacodylate buffer until they were sectioned without dehydration as described previously [14,34]. Sections were blocked for 2 h to overnight at 4 °C in 1:20 normal donkey serum (Jackson ImmunoResearch Laboratories, West Grove, PA) in PBS containing 0.5% BSA, 0.1% Triton X-1009, and 0.1% sodium azide (PBTA). Sections were then incubated overnight in various combinations of the following primary antibodies all diluted with PBTA: 1:500 calretinin (rabbit polyclonal; Chemicon, Temecula, CA), 1:100 α-protein kinase C (rabbit polyclonal; BioMol Research Labs, Plymouth Meeting, PA), 1:100 biotinylated neurofilament (mouse monoclonal 70 and 200 kDa subunits; Biomeda, Foster City, CA; biotinylation done by Vector Labs, Burlingame, CA), and 1:500 VAMP2 (synaptobrevin, mouse monoclonal; Synaptic Systems GmbH Göttingen, Germany). These primary antibodies were visualized using secondary antibodies conjugated with various fluorochromes: donkey anti-rabbit or anti-mouse immunoglobulin G conjugated to either 1:200 Cy3 or Cy2 (Jackson ImmunoResearch Laboratories), and streptavidin conjugated to 1:100 Cy5 (Jackson ImmunoResearch Laboratories). Samples were rinsed, mounted in 5% n-propyl gallate in glycerol, and examined with a BioRad 1024 confocal microscope (BioRad, Hercules, CA). Optical slices of 0.5 μm were viewed with a 40X/1.3 N.A. objective. Some images consisted of multiple planes of focus projected as a z-series; the number of these planes is included in the accompanying figure captions.
Retinal samples of normal cat retina (n=2) and cat retinas detached for 3 days (n=2), 7 days (n=4), and 28 days (n=3) were processed for light (LM) and electron microscopy (EM) as detailed in Fisher et al. . Sampled regions were excised from central superior retina. Sections were taken on an LKB V ultramicrotome (LKB-Produktor AB, Bromma, Sweden). For LM, 1-μm-thick sections were stained with saturated aqueous paraphenylenediamine and photographed using an Olympus BX60 Research microscope equipped with a Microfire digitial camera (Optronics, Goleta, CA). For EM, serial sections were collected and mounted on formvar-coated slot grids, stained with uranyl acetate followed by lead citrate, and examined at 80 kV on a JEOL JEM-1230 transmission electron microscope (JEOL USA Inc., Peabody, MA) equipped with an AMT digital camera system (Advanced Microscopy Techniques Corp., Danvers, MA). More than 40 rod spherules were examined using serial thin sections.
All transmitted and confocal images were captured digitally and required no additional image processing once correct exposures or laser settings were determined. Transmission EM images were taken using standard EM sheet film negatives that were scanned on a Microtek ScanMaker i900 (Microtek Lab, Inc., Carson, CA). The images were then inverted, the contrast was adjusted, the images assembled, and labels were applied using Adobe Photoshop CS2 software (Adobe Systems, Inc., San Jose, CA).
The most relevant morphological effects of detachment in the first month were evident in the outer retina (compare Figure 1A to Figure 1B,C; also see ). As photoreceptors lose their outer segments and inner segments and some rods die by apoptosis , rod terminals retracted from the OPL toward their cell bodies. In the OPL of the normal cat retina, rod terminals were clustered tightly together just beneath the inner edge of the ONL and above and around the cone terminals (Figure 1D). The retraction of rod spherules was underway by day 3 when there was a loosening of their clustering and evidence of a few terminals moving up into the lower ONL. The rod terminals themselves had yet to shrink by day 3 (compare Figure 1D,E). Horizontal cell hypertrophy was already evident 3 days after detachment with swollen perinuclear cytoplasm and obvious vacuoles in the OPL (Figure 1E). Rod spherule retraction was common at 7 days when they became thin, acquired a distinctive teardrop shape, and stained dark (arrows, Figure 1B,F). The retraction of so many rod terminals at this time point interrupted the tight configuration of the OPL neuropil, making it difficult to recognize in some regions (Figure 1B,F). However, these distinctive retracting terminals so obvious at 7 days were hard to find at 28 days when remaining rod nuclei fronted directly on a narrowed OPL neuropil containing swollen HC processes (Figure 1G).
As rod terminals retracted after detachment, immunocytochemical labeling revealed distinctive neurite outgrowths from both HCs and RBs. Based on labeling patterns, it has been established that most HC outgrowths (Figure 2D) arise from the axon terminal system of B-type cells . Similar outgrowths from RBs (Figure 2B) were easily identified because of their specificity for labeling with anti-protein kinase C-α . Neurites of both types often terminated near retracted rod terminals (arrows, Figure 2B,D). It is likely that there was both outgrowth and pruning of dendrites from RBs. In normal cat retina (Figure 2A), fine RB dendrites contacted 15–50 rod spherules  see Figure 1A in . After detachment, these numerous fine dendrites rapidly decreased in number so that by 7 days (Figure 2B) only a few stout dendrites arose from each RB cell body. There appeared to be a similar process of simultaneous neurite outgrowth and pruning in HC axon terminals (compare Figure 2C,D). These morphological changes after detachment all suggested significant changes at the ultrastructural level in rod synapses.
In normal cat retina the photoreceptor terminals and their postsynaptic processes are tightly packed together. The spaces between them are occupied by thin Müller cell processes. Typically, rod spherules (rs; Figure 3A) are arrayed together in the outer OPL between the lowest tier of rod cell nuclei and the OPL neuropil. Cone pedicles (cp; Figure 3A) tend to lie just at the outer edge of the neuropil.
The OPL of a cat retina detached for 7 days (Figure 3B) can be highly disorganized compared to that in normal attached cat retina (Figure 3A). No photoreceptor terminals were evident along its distal border in this region (top 1/3 of Figure 3B). However, other regions of the OPL in detached retina at both 7 and 28 days still had clusters of rod spherules in their usual position in the distal OPL. In some cases these surrounded surviving cone pedicles (Figure 3C). Particularly conspicuous at 7 days after detachment were the many electron-dense rod axons and spherules that appeared to be retracting into the ONL (arrows, Figure 3D,E), corresponding to the dark-stained structures observed by LM (Figure 1B,F). These teardrop-shaped terminals were more elongated and thinner (approximately 1–2.5 μm in width) than those in normal retina (3–5 μm) . As spherules withdrew from the OPL, they lost connections with cone terminal telodendria that normally contacted a subpopulation of spherules . The cone pedicles themselves underwent significant morphological changes but remained anchored in position along the border of the OPL [37,38].
After detachment, HCs appeared to undergo a type of hypertrophy, resulting in swollen perinuclear cytoplasm, enlarged dendrites and axon telodendria (Figure 3B,D,E). The perinuclear cytoplasm of the B-type HC is normally crowded with organelles [4,15,17], but these are dramatically reduced by 7 days following detachment (Figure 1F). One HC, measuring 17×14.5 μm had an 8 μm-diameter nucleus lying in a largely empty perinuclear cytoplasm (data not shown). Similar cells were seen on LM to the right in Figure 1F and by EM in Figure 3B, where two enlarged dendritic trunks from one HC course off at right angles to each other.
Many detailed ultrastructural studies on normal cat retina have demonstrated that every spherule contains 1 or 2 synaptic ribbons, each associated with a “synaptic unit” comprised of 2 invaginating HC axon terminal (HCat) processes and 1 or more RB dendrites [1,4,6,15]. These innervating processes enter the base of the spherule via a single “opening” or “hilus” only 0.24 μm in diameter, through which they all must pass [1,2,6,15,39,40] (see paired arrowheads in Figure 4A,B,D). Beneath the synaptic ribbon was an electron-dense arciform density whose extent was more easily appreciated in radial aspect when the arched ribbon was viewed en face (asterisks, Figure 4B).
Retracting terminals still possessed several of the typical components of presynaptic machinery including mitochondria, synaptic ribbons, arciform densities, and synaptic vesicles. Cat spherules usually have more than one mitochondrion that can lie close to the synaptic site, but also can extend distally into the rod axons (Figure 4L and Figure 5A,B). Retracting terminals also contained 1 or 2 ribbons. These ribbons have a normal thickness (30–35 nm) and height (0.2–0.25 μm), but are shorter than the 2.3±0.3 μm reported for ribbons in normal cat spherules . Overall, the rod spherule synaptic ribbons showed greater abnormalities in substructure with increasing displacement from the OPL, but they were never found floating free in the cytoplasm away from the membrane. The fate of the arciform density in retracting terminals was unclear since they underlay some ribbons (Figure 4C-E,H,K) but were not evident with many others (Figure 4F,J and Figure 5C). Curiously, basal membrane densifications were sometimes seen unaccompanied by any other typical presynaptic structures (Figure 4L). Normal spherules were filled with synaptic vesicles, both surrounding the synaptic ribbons and scattered throughout the presynaptic cytoplasm (Figure 4A,B). The vesicles were spherical and a uniform 35–50 nm in diameter in control as well as detached retinas (Figure 4C-H). Partially retracted terminals showed a range of vesicle populations from dense (Figure 4C,D) to sparse (Figure 4L). Fully retracted terminals, common in the 28-day detachments (Figure 4I), usually had many presynaptic features lying in their perinuclear cytoplasm, but relatively few vesicles (Figure 4F,I,J), although some had many (Figure 4E). In contrast to the 7-day detached retina, however, profiles of actively retracting terminals were rare in the 28-day detached retinas. Additionally, many rod nuclei in the outer ONL of the detached retinas apparently had no axon and lacked any of the typical presynaptic structures.
After detachment (Figure 4C-L and Figure 5), the pattern of innervation of the spherules was altered. Even when spherules were in their usual location (Figure 4C), they were often turned relative to one another. The invaginations were reduced in depth, complexity, and integrity. Invagination width and depth are features that vary little from species to species. Rao-Mirotznik and coworkers  reported a consistent invagination depth and width of about 1 μm. Some retracting spherules in 7 day detachments had similar dimensions, especially those that have retracted the least. Overall, most had depths (and widths) ranging from 0.5 to 0.75 μm (see examples in Figure 4C,D,F-I and Figure 5C), yet many had no invaginations at all (Figure 4D,E,J). The less complex synaptic invaginations of retracting spherules in the 7 day detachments contained fewer postsynaptic processes than usual—often limited to a pair of HC lobes and no central elements from the RB dendrites (Figure 4D,E and Figure 5C). Most rod ribbons still opposed two HC processes, but many appeared to be lobes from branches of the same, rather than different, HC axon terminals as occurred in normal cat retina . We also noted a lack of “fingers” of rod terminal cytoplasm projecting into the invaginations as has been reported in normal cat and human rods [1,6,39,40]. The overall integrity of the spherule invagination was altered because invariably rod terminal cytoplasm fully encapsulated the postsynaptic processes and isolated them from other spherules as well as from the fine glial processes of the encompassing Müller cells. Normally the only opening to this invagination is at its base via the narrow “hilus” . Although many retracting terminals maintained a hilus, not all were at the base and, more importantly, the opening itself was frequently wider (Figure 4F,G and Figure 5C,D) or lacking altogether. In such cases HC lateral elements contacted the spherules directly in open configurations where they also opposed other swollen HCat telodendria and Müller cell processes (see * in Figure 4C-E,H and Figure 5C,D). Often the rod terminal cytoplasm encapsulating the invagination pulled back asymmetrically such that one HC lateral element was invaginated while its partner is not (Figure 4D,H; spherule 3 in Figure 5D). These open configurations can result in the swollen HC lateral elements of adjacent spherules directly apposing one another.
The dilation of the hilus after detachment and the resulting opening up of the invagination revealed another feature of rod spherule architecture: the synaptic ridge. Literally a shallow evagination that projects as a ridge between the HC lateral elements, the synaptic ridge underlay the arciform density  and was flanked by the 2 linear active zones along the ribbon base and immediately above the bifurcation ridge , where released glutamate diffused between apposed HC lobes with their AMPA receptors and spread down to the RB dendrites with their mGluR6 receptors. The trough-like arciform density has long been proposed to offer some sort of rigidity to the ribbon complex. This synaptic ridge was revealed in this study when the synaptic invagination opened and the rod terminal membrane pulled back and flattened (Figure 4C-E,H and Figure 5C,D). It projected a short distance between apposing HC lobes. This configuration may be short-lived during the retraction phase since fully retracted terminals can contain ribbons that directly appose postsynaptic processes without arciform densities (Figure 4J).
The density of cytoplasmic organelles appeared sparse in the B-type HC after detachment by comparison to that in normal feline retina where their distinctively dark axon terminal processes occupied much of the distal OPL neuropil and contained numerous microtubules, neurofilaments, and scattered mitochondria in a granular cytoplasm . Such processes were not evident in the detached cat retina, but instead the outer OPL contained a population of circular processes (Figure 1B,C,E-G) containing minimal ground substance or cytoskeleton (Figure 3B,D,E), occasional mitochondria with few cristae (Figure 5A,B), and, notably, no ribosomes (Figure 5A-D). Axon terminal processes can be differentiated from Müller cell processes because the former lack ribosomes, while the latter have many . These profiles, typically ranging from 1 to 2 μm in diameter (Figure 3D,E and Figure 5A-D), often were adjacent to one another in the outer OPL, no longer separated from each other by intervening Müller cell processes. Serial sections through a small cluster of retracted and partially retracted rod spherules demonstrated that processes arising from them became the lateral elements postsynaptic to the rod ribbons (Figure 5A-D). Thus the axon terminal underwent similar ultrastructural changes to those observed in its cell body. In normal cat retina, the HC axon terminal branches inside the invagination contained a population of light-cored vesicles . Similar vesicles occurred in the swollen terminal branches after detachment (Figure 4C-E,H and Figure 5B-D); they were of less uniform size than the synaptic vesicles of the rod spherules, ranging from 45 to 110 nm in diameter with most of the smaller vesicles being circular and the larger ones more elongated. The swollen processes contacting the spherules varied in size from 0.4 up to 0.7 μm in diameter, much larger than typical HC lateral elements (approximately 0.25 μm in diameter) . The processes of the OPL neuropil in detached retinas were surrounded by expanded extracellular space, a feature never associated with the tightly packed OPL of normal retina.
The RB dendritic trunks that grew into the ONL after detachment (Figure 3F and Figure 4E) were quite stout, 0.3 to 0.5 μm in caliber. Key to their identification was the presence of the “helical organelle,” a system of spiraling and flattened tubules of smooth endoplasmic reticulum that immediately underlay the cell’s membrane [39,40]. Small, dark, vermiform terminal processes only 0.08-μm thick abruptly branched off from these RB dendritic trunks. These branchlets lacked organelles and coursed along Müller cell processes until they contacted their target spherules. Most, however, ultimately terminated close to a spherule but did not enter its invagination (Figure 4C,G).
In response to retinal detachment, Müller cell nuclei were seen in the distal retina (Figure 3D,E). At 7 days after detachment, the nuclei of migrating Müller cells were ovoid, averaging 7×5 μm with faint nucleoplasm devoid of heterochromatin (Figure 1F and Figure 3B,D,E) and differed in appearance from normal Müller cell nuclei, which were compact, angular, electron-dense and restricted to the inner portions of the inner nuclear layer (INL). While Müller cell processes in the normal ONL were thin and inconspicuous, after detachment their hypertrophy created dense radial columns of Müller cell cytoplasm full of intermediate filaments, microtubules, ribosomes, scattered mitochondria, and large arrays of rough endoplasmic reticulum (Figure 3B,C,E,F and Figure 4E,J). As the packing density of cells in the ONL declined and rod terminals retracted from the OPL, remaining retinal processes appeared juxtaposed to, rather than ensheathed by, these thickened Müller cell processes (Figure 3D,E, Figure 4D-F,J,L, and Figure 5A).
The first few days after detachment represent a critical period of outer segment degeneration, RPE apical surface remodeling, the beginning of Müller cell reactivity, and a wave of apoptotic cell death that can kill up to 20% of the rod population [13,36]. After this, apoptosis proceeds but subsides, Müller cell hypertrophy continues until large glial scars form intra- and extraretinally. Seemingly of critical importance to information processing by the retina, photoreceptor synaptic terminals undergo a series of changes resulting in significant alteration to this critical primary synapse. By day 3 of this critical period, many rod terminals have begun the process of retraction to their respective cell bodies. The end product is a large population of rods with their terminals far distal to the OPL and showing a highly modified structural arrangement. Their deep synaptic invaginations are lost and synaptic ribbons are shortened with, in some cases, a greatly reduced population of vesicles. By a month after detachment fewer terminals appear in the active state of retraction as if a new steady-state has been attained . Rod terminal retraction has been described in many retinal degenerations as well as in sheets of porcine photoreceptors prepared for retinal transplantation , suggesting that it might be a common response to photoreceptor injury. The underlying mechanisms leading to retraction are yet to be fully understood, although recent studies have suggested that the RhoA-ROCK pathway as well as cAMP may be involved [32,41].
Terminal retraction is accompanied by dramatic changes in second order neurons, easily identified by specific confocal immunolabeling studies in several species and several retinal degenerations. The combination of rod terminal retraction and neurite sprouting by the postsynaptic targets, the RBs and B-type HCats [14,21], suggests that rod synaptic transmission has been compromised.
The rod spherule in the feline retina has been well studied and its characteristics measured in detail [1,2,6]. The release of 100 vesicles/second/spherule is estimated to be the minimum rate of glutamate exocytosis necessary to account for the rod’s broad range of sensitivity [2,42]. Unique among CNS neurons, the rod photoreceptor relies on a single active zone to transmit its signal. The synaptic zone is folded, not planar as in most other CNS synapses, and its cleft volume is 20 times larger than that of most conventional synapses [for review see 10]. This synaptic interface is contained within a single deep invagination of the rod axon terminal. The volume of the extracellular space within the invagination averages 0.21 μm3 and varies little from spherule to spherule, and the length of the linear active zone is remarkably similar across species at approximately 2.5 μm [1,6]. This allows glutamate concentrations to be achieved within the cleft sufficient to simultaneously stimulate glutamate receptors on the invaginating processes of both RBs and HCats [1,2]. It is thought that glutamate receptors inside the synaptic invagination as well as those clustered outside prevent “spillover” of glutamate from one spherule to another that could degrade signal fidelity [see 2,11]. While the exact mechanism by which synaptic ribbons regulate tonic glutamate release remains controversial , the ribbon’s surface area is essentially invariant . A ribbon will typically have around 130 docking sites for vesicles primed for release (twice the number found at the largest conventional synapses known), along with 5 times that number of vesicles tethered to the ribbon above the linear active zone. This type of data points to some central role of the ribbon’s architecture in achieving the rate of transmitter exoctyosis necessary for rod function. Our results showed that much of the critically important organization of the rod spherule is drastically altered as spherules retract following retinal detachment.
It is well documented that photoreceptor outer segments undergo severe degeneration within a week of detachment. Our data showed that rod spherules also undergo structural “degeneration” that seems likely to affect the physiology of synaptic transmission. Like the outer segment, many rod synapses “deconstruct” their highly structured three-dimensional organization. Significantly, the elaborate fingers of rod cytoplasm that project into the invagination and may effect local glutamate concentration [1,2] appear highly vulnerable to the effects of injury. We found little evidence of such fingers in the retracting terminals. The opening up or virtual loss of the rod hilus also reflects an important step in the restructuring of the invagination. In general, the amount of structural disruption to a spherule’s invagination correlates with the degree of its displacement from the OPL, but even then the severity of this deconstruction can vary between spherules—some have invaginations of almost normal depth (spherule 1, Figure 5), others are shallower (spherule 2, Figure 5), some have partial invaginations where only one lobe of a pair of HCat lateral elements is surrounded by rod cytoplasm (Figure 4G,H, spherule 3, and Figure 5), and others have no invagination whatsoever (* in Figure 4D,E). In the latter case, it is particularly striking to see images of paired HCat lobes lined up on either side of a ribbon, both in direct contact with the spherule but also lying exposed to other neuronal and glial cell processes in the altered OPL neuropil. Regulating cleft volume and preventing glutamate spillover would seem impossible in these highly altered synapses.
Another serious challenge to normal spherule function following detachment is the loss of synaptic contacts with other cells— particularly with RBs. Fewer than half of the spherules we examined in serial sections had any recognizable RB contacts. That RBs also experience dendritic pruning, leaving only 1 to 4 neurite outgrowths per cell recognizable by immunofluorescence microscopy [14,21], would also suggest that many spherules lose their RB innervation, thus compromising the rod pathway. Dendritic pruning of RBs has also been reported in other retinal degenerations [43-46]. There is no specific evidence that any of the contacts made by the modified spherules are functional. While spherules in normal cat retina receive contact from two different RBs , we never saw evidence of more than one RB dendrite ending close enough to a retracted spherule to warrant classifying it as a “contact.” These RB dendritic outgrowths probably retained their mGluR6 receptors, because they were found on ectopic RB dendritic outgrowths in the nob2 mutant mouse [23,24], in the bassoon knockout mouse , and in the RCS rat , so that if contact was reestablished after retinal detachment normal synaptic transmission could resume. Even though our study revealed that processes from HCats were much more successful in remaining closely apposed with retracting spherules, rarely did we see spherules contacted by more than one axon terminal system, a distinct difference from the situation in normal cat retina . In general, there were fewer postsynaptic HC processes in the remaining invaginations, and they appeared undifferentiated and structurally abnormal.
Another consequence of retinal detachment in feline retina is the extreme hypertrophy that the B-type HCs and Müller cells undergo. The swollen, pale B-type HC cell bodies and dendritic trunks in detached samples were immediately apparent by both LM and EM. Our data indicate that this swelling also affects the terminals that contact rod spherules. Thus B-type HCs are highly reactive to detachment with abnormal swelling, robust neurite sprouting, and an upregulation of neurofilaments [21,47,48]. HC hypertrophy is reminiscent of reactions seen in axotomized neurons elsewhere in the CNS where, despite an increase in neurofibrillar components, the cells otherwise appear swollen and sparse in organelles. These HCs however are not axotomized by detachment but may experience a reduction in synaptic input to their axon terminals as rods undergo degeneration.
The more familiar example of cell hypertrophy after retinal detachment is that seen in the early reaction of Müller cells [13,49,50]. According to our data, Müller cell processes fill in areas of the OPL where rod spherules and other neuronal processes are lost. HC and Müller cell hypertrophy results in a rapid change in the immediate cellular environment for the surviving rod terminals and cone terminals. Large expanses of Müller cell cytoplasm now populate the OPL. Although there is no physiologic data, the extensive hypertrophy of the B-type cells may well alter their electrical properties. Thus retinal detachment leads to many changes in the photoreceptor synaptic milieu; as photoreceptors degenerate, terminals and postsynaptic processes are lost, and there is a distinct loosening of the densely packed OPL. Processes generated by the hypertrophy of Müller cells can result in large columns of fibrillar Müller cell cytoplasm interrupting the regular array of neurons in the ONL as well as processes in the OPL.
Limiting the loss of photoreceptor outer segments has always been thought to be a key element in lessening damage caused by retinal detachment. However, preventing the remodeling of photoreceptor synaptic terminals and subsequent loss of synaptic connectivity may be just as important to visual recovery. This study represents a start at understanding the structural reactions of photoreceptor synapses to detachment. There is still much to be learned. Serial section studies of this kind are time-consuming, tedious, and limited in resolution in the z axis. A promising approach may be to study photoreceptor synapses in relatively thick sections by electron tomography with intermediate voltage EM . Similar studies of ribbon synapses in the ear have uncovered striking levels of detail not available in studies of the kind done here .
We are grateful for the technical expertise of Mr. Kellen Betts. Portions of these data were published as an abstract in Linberg et al., ARVO Annual Meeting; 2006 April 30-May 4; Fort Lauderdale, FL. This study was supported by NIH grant EY000888 (to S.K.F.), NCRR SIG grant 1S10RR017753–01, and NSF ITR grant 0331697.