Estrogen Protects the Inner Retina from Apoptosis and Ischemia-Induced Loss of Vesl-1L/Homer 1c Immunoreactive Synaptic Connections

Abstract

purpose. Protective effects of estrogen on nerve cells including retinal neurons have been described previously. However, subcellular effects on synaptic connectivity in mild ischemia more closely resembling ischemic conditions found in diabetic or sickle cell retinopathy and stenosis of the carotid artery have not been identified. The present study quantitatively analyzed effects of estrogen administration on synaptic connections of neurons in the ganglion cell layer (GCL) of the retina.

methods. Staining of Vesl-1L/Homer 1c (V-1L) immunoreactivity and TUNEL cytochemistry were used to quantify neuroprotective effects at the synaptic level in a model of mild retinal ischemia induced by temporary middle cerebral artery occlusion in the adult rat.

results. V-1L immunoreactivity was found in both synaptic layers, postsynaptic to glutamatergic ribbon synapses. Mild retinal ischemia led to a significantly higher percentage reduction in the number of V-1L-positive synapses in the inner plexiform layer (IPL) compared with the percentage of TUNEL-positive apoptotic neurons in the GCL. Estrogen prevented ischemia-induced loss of V-1L-immunoreactive synapses in the IPL and apoptosis of cells in the GCL.

conclusions. Immunoreactivity for V-1L can be used as a synaptic marker for early changes before more severe neurodegenerative events. The present results suggest that estrogen protects neurons in the GCL including RGCs from both apoptosis and early changes in synaptic connections associated with ischemia and potentially preceding apoptosis.

Retinal ganglion cells (RGCs), because of high sensitivity to cellular damage and neurotoxicity, offer a unique and effective model to study the mechanism of neurodegenerative disease progression. ^{1 2 3} Pathologically, these neurons are the primary targets of processes leading to the loss of vision in neurodegenerative diseases such as glaucoma, but they are also directly affected in acute diseases characterized by retinal ischemia, such as diabetic retinopathy, sickle cell retinopathy, and stenosis of the carotid artery. ^{4 5 6 7} These neurons are also known targets of steroid hormones. ^{8 9 10 11 12} As such, RGCs serve as an ideal model to evaluate the neuroprotective effects of steroid hormones, and at the same time, can offer important insights for development of novel treatments for retinal degeneration and for other neurodegenerative diseases, as well. 

Vesl/Homer proteins are cytosolic scaffold proteins that have been implicated in the clustering of neurotransmitter receptors and neuronal development and plasticity. The Vesl/Homer protein Vesl-1L/Homer 1c (V-1L) binds other proteins, such as metabotropic glutamate receptors and cation channels, and localizes them to glutamatergic synapses. ^{13 14 15 16 17 18 19} V-1L was used as a marker for the assessment of changes in synaptic connectivity preceding apoptosis and during early stages of apoptosis of RGCs, because it has several unique properties. As a postsynaptic clustering molecule, it has the property of linking neurotransmitter receptors, plasma membrane ion channels, intracellular calcium channels, and the cytoskeleton. ^{13 14 15 16 17 18 19 20 21} Therefore, V-1L serves as a unique marker to detect synaptic changes mediated or affected by extracellular signaling, intracellular signaling or cytoskeletal processes. V-1L is the ubiquitously expressed isoform of Vesl-1 with expression levels that are not affected by changing physiological conditions, as described for the conditionally expressed isoform of Vesl-1, Vesl-1S. ^{15 16 22} Therefore, V-1L can be used as a synaptic marker that is sensitive to overall changes in synaptic connectivity—that is, neurotoxicity-induced loss of synapses, but with expression levels that are not influenced by changed physiological conditions. ^{13 14 15 16 17 18 19 20 22} This specific function in cross-linking proteins at glutamatergic synapses in the central nervous system (CNS) was used in the present study to identify the distribution of ischemia-induced changes of postsynaptic elements of RGCs. 

The neuroprotective effects of estrogens have been extensively assessed in animal models of cerebral ischemia. After the initial report of neuroprotection with estrogens in a model of transient cerebral ischemia, ²³ different estrogens have been demonstrated to protect the brain from different forms of ischemic damage. ^{24 25 26 27 28 29 30 31 32 33 34 35 36} Collectively, the ability of estrogens to preserve neuronal function in various experimental models of neurodegeneration suggests that they exert equally efficacious neuroprotective effects in retinal degeneration. 

The absence of estrogens, such as occurs after menopause, can have pathologic effects on vision. Age-related risks for development of ocular diseases can be reduced with estrogen treatment. ³⁷ Furthermore, just as postmenopausal women have been documented as having a higher prevalence of and risk of developing Alzheimer’s disease, ³⁸ they are also at a higher risk for development of macular degeneration—a risk that seems to be reduced with estrogen replacement. ^{39 40} Shorter duration of estrogen exposure—that is, a shorter duration between menarche and the onset of the menopause, also increases a woman’s risk of age-related macular degeneration, ⁴¹ further supporting the hypothesis that estrogen treatment may help stave off ocular tissue degeneration. Direct investigation of estrogen’s protective effects against ischemia-reperfusion-induced retinal damage has demonstrated that 17β-estradiol reduces leukocyte accumulation and consequent retinal damage, particularly in the inner retina. ⁴² Mechanistically, excitotoxic cell death in the retina (such as that which occurs after ischemia) may be, at least in part, a consequence of changes in GABAergic signaling. ^{43 44} Alternatively, estrogen may also exert its protective effects by influencing the vasculature or blood flow to the eye. To this end, the beneficial effect of estrogens in reducing the risk of developing age-related eye diseases, such as glaucoma and macular degeneration, has been linked to its positive effects on vascular hemodynamics. ⁴⁵  

The present study tested the hypothesis that estrogen (17 β-estradiol) exerts protective effects on early changes in the synaptic connections between cone bipolar cells and RGCs associated with mild retinal ischemia and subsequent apoptotic events. Because of the profound and widespread effects that strong neurotoxic insults have on the retina, it is often difficult to differentiate between causal and secondary factors of neurotoxicity and necrotic or apoptotic events in neurons. Therefore, in the present study, we used an animal model, characterized by mild retinal ischemia, that is suitable for use in detecting early causal events leading to neurodegeneration. The use of model systems of mild ischemia rather than strong neurotoxic insults allowed us to evaluate early changes potentially relevant to the prevention of neurodegeneration and acute retinal diseases. 

Methods

All experiments were approved by the University of North Texas Health Science Center (UNTHSC) Institutional Animal Care and Use Committee and were performed in compliance with the guidelines for the welfare of experimental animals issued by the NIH and in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

In Vivo Model of Mild Retinal Ischemia

Retinal ischemia was induced by transient occlusion of the middle cerebral artery (MCA) in adult female Sprague-Dawley rats, as described previously. ^{23 26 27} One-hour endovascular MCA occlusion caused a partial infarction of the ipsilateral ophthalmic artery and induced mild retinal ischemia affecting approximately 4% ± 1% of neurons in the ganglion cell layer (GCL), but not other retinal neurons, excluding secondary effects induced by presynaptic neurons of RGCs (see Fig. 3 ). After the 1-hour MCA occlusion, animals were reperfused for 24 hours. Blood samples were obtained at the time of MCA occlusion and after death, to determine plasma concentrations of estrogen. Rats were ovariectomized (OVX) two weeks before temporary MCA occlusion surgery. The OVX rats were divided into two groups that received either placebo replacement or estrogen replacement. All animals received placebo or hormone replacement by a single subcutaneous injection 2 hours before MCA occlusion surgery, as described previously. ^{23 26 27 46} Estrogen was dissolved in inert corn oil (also used for placebo controls) and a dose of 100 μg/kg body weight was chosen to produce high physiological and pharmacological estrogen concentrations (50–500 pg/mL serum, with a peak concentration at 2 hours after injection ⁴⁶ ). 

Tissue Preparation

After ischemia treatment and 24 hours of reperfusion, rats were anesthetized deeply with halothane and decapitated. A detailed description of the preparation of the retinal tissue for light and electron microscopic immunocytochemistry is given in Brandstätter et al. ⁴⁷ Briefly, the eyes were opened and immersion fixed in 4% (wt/vol) paraformaldehyde (PA) in phosphate buffer (PB; 0.1 M, pH 7.4) for 30 minutes. After fixation, the retinas were dissected from the eyecup, cryoprotected in a graded series of sucrose in PB (10%, 20%, 30%) and sectioned vertically at 12-μm thickness on a cryostat. For electron microscopy, the fixation was in 0.05% glutaraldehyde and 4% (wt/vol) PA in PB for 10 minutes, followed by incubation in 4% (wt/vol) PA in PB for 40 minutes. After fixation for electron microscopy, the retinas were washed in PB and cryoprotected as described for light microscopy. 

Tissues from three individual animals for each condition were used in the study, and sections with the same eccentricity (0.5 to 2 mm lateral to the optic nerve as a reference point) were compared. Orientation of the eye and the retina tissue during processing was maintained by labeling the eye with a permanent marker and by making asymmetric incision into the retina. 

Antiserum against V-1L

An affinity-purified polyclonal antiserum against V-1L raised in rabbit against a GST-V-1L fusion protein was used in the present study. The specificity of this antiserum has been characterized. ²²  

Light Microscopic Immunocytochemistry

Immunocytochemical labeling was performed by the indirect fluorescence method, as described previously. ⁴⁸ The V-1L antiserum was used at a dilution of 1:1000, and sections were incubated in the primary antiserum overnight at 4°C. The binding sites of the primary antiserum were revealed by a secondary antiserum, fluorescence-conjugated goat anti-rabbit IgG (Alexa 594; Molecular Probes, Eugene, OR) diluted 1:500. In control experiments, either the primary or secondary antiserum was omitted, or the primary antiserum was incubated with the antigen (GST-V-1L fusion protein) against which the antiserum had been raised (10-fold excess of the antigen, wt/wt) for 1 hour before application to the sections resulting in only background label in each case. 

Fluorescence TUNEL Cytochemistry

Cells that had entered advanced stages of apoptosis were detected with a TUNEL assay performed with a fluorometric TUNEL system kit (DeadEnd; Promega, Madison, WI; fluorescein label, green fluorescence) according to the manufacturer’s instructions. The fluorescently labeled sections were counterstained with 4′,6′-diamino-2-phenylindole (DAPI, blue fluorescence; Molecular Probes) to visualize the total number of cells. 

Light Microscopic Analysis

Fluorescent images were collected digitally on a microscope set up for epifluorescence (Microphot FXA; Nikon, Melville, NY). Excitation wavelengths were selected by a computer-controlled filter wheel (Ludl Electronic Products Ltd., Hawthorne, NY). A single-emission filter was used that allows the passage of DAPI (nuclear staining), fluorescent emission wavelengths (TUNEL staining, Alexa 488; Molecular Probes) and (immunocytochemistry, Alexa 594; Molecular Probes), depending on the excitation filter selected, which allowed multiple labels to be captured and the images overlaid without any spatial shifting of the image data. Images were captured with a cooled charge-coupled device (CCD) camera (Photometrics SenSys; Roper Scientific, Tucson, AZ) as 10-bit, 1024 × 1024-pixel, gray-scale images. The camera and microscope automation were computer controlled (IP Laboratory Spectrum; Scanalytics, Fairfax, VA). Images were deconvolved (Microtome plug-in or IP Laboratory Spectrum; Vaytek, Fairfield, IA). Deconvolved images were merged in the computer system to determine relative label distribution, count individual V-1L immunoreactive synapses in defined volumes of retina tissue, and visualize colocalization of labels (TUNEL staining). Images were reconstructed from 12 deconvolved optical sections through a 12-mm-thick vertical cryostat section and were merged in the system software. Tissues from three individual animals for each condition were used in the study, and sections with the same eccentricity (0.5–2 mm lateral to the optic nerve as a reference point) were compared. Orientation of the eye and the retina tissue during processing was maintained by labeling the sclera with a permanent marker and by making asymmetric incision into the retina. Sections from each animal (both the ipsilateral experimental tissue that had been subjected to ischemia and the contralateral control tissue were used from each animal) were analyzed after deconvolution and three-dimensional reconstruction (as just described); six Z-stacked images for each condition were quantified. Immunoreactive profiles were counted by using automated image acquisition and intensity analysis software (SimplePCI; Compix Inc., Cranberry Township, PA). Software parameters that were used to identify immunoreactive profiles (fluorescence intensity, area, volume) were kept constant for all analyses of different experimental conditions. Results are expressed in number of immunoreactive profiles per unit volume. Statistical analysis of immunoreactive labels and of the TUNEL assays was performed on computer by paired t-test (SPSS; SPSS Science Inc., Chicago, IL). 

Pre-embedding Immunoelectron Microscopy

The labeling for pre-embedding immunoelectron microscopy was performed as described in detail previously. ⁴⁷ Briefly, after dissection and cryoprotection, retinas were frozen and thawed repeatedly to enhance tissue penetration by the antiserum. Small pieces of retina were embedded in agar, and vertical sections (50 μm thick) were cut with a vibratome. The primary antiserum was used at the same concentration and diluted in the same medium, but without Triton X-100, as used for light microscopy. Tissue sections were incubated in primary antiserum for 4 days at 4°C. Binding sites of the primary antiserum were visualized with a biotinylated goat anti-rabbit IgG secondary antiserum diluted 1:100 (Vector Laboratories, Burlingame, CA) and a peroxidase-based enzymatic detection system (Vectastain Elite ABC kit; Vector Laboratories). The reaction product was silver intensified and gold toned. Control experiments were performed as described for light microscopic immunocytochemistry. The analysis was performed with an electron microscope (EM910; Carl Zeiss Meditech, Thornwood, NY). Cell types were identified by using well-established anatomic criteria, such as the position of processes in defined sublayers of the inner plexiform layer, the size and morphology of neuronal processes, as well as the presence of electron-dense material and the presence and quantity of synaptic vesicles. Cone bipolar cell terminals were distinguished from rod bipolar cell terminals by their difference in size, the presence of multiple ribbon synapses and the absence of postsynaptic ganglion cell processes at rod bipolar cell terminals. 

Results

V-1L Expression in the Retina

V-1L immunoreactivity was present in both plexiform layers of the rat retina (Fig. 1) . The immunoreactivity appeared predominantly as punctate staining indicative of clustering of V-1L at synapses. The presence of immunoreactive profiles of different sizes in the outer plexiform layer (OPL) indicates the expression of V-1L at both rod and cone photoreceptor terminals (Fig. 1) , as has been shown previously for other synaptically localized proteins. ^{49 50 51}  

We further analyzed the distribution of the V-1L immunoreactivity to identify the subcellular localization at the ultrastructural level. To detect V-1L immunoreactivity at the subcellular level, a very sensitive immunocytochemical method combining peroxidase staining with silver intensification and gold toning of the label was used. Because of the diffusion of 3,3′-diaminobenzidine (DAB) reaction product, the spatial resolution was lower than when using gold-coupled secondary antibodies. However, the current method enabled us to localize low concentrations of antigen with an antiserum that exhibits a high sensitivity to alterations in its antigen due to the fixation and embedding procedures. 

We detected V-1L immunoreactivity specifically localized to the postsynaptic elements at bipolar cell dyads in the inner plexiform layer (IPL) of rat retina. Ganglion cells expressed V-1L at their contacts to cone bipolar cells both in the ON- and in the OFF-pathway (Figs. 2A 2B , respectively). This specific immunoreactivity pattern is consistent with the clustering function of V-1L at synapses and with similar distribution patterns of V-1L-associated proteins, such as group I metabotropic glutamate receptors. ⁴⁸ Typically, both postsynaptic processes at the cone bipolar cell ribbon synapses were labeled for V-1L (Fig. 2) . At cone bipolar cell synapses, the two postsynaptic elements are typically a process of an amacrine cell and a dendrite of an RGC or two RGC dendrites. ⁵² At rod bipolar cell synapses, both postsynaptic elements belong to amacrine cells. ^{53 54} Because of poor tissue preservation, we were not able to identify the ganglion cell processes in each case by the absence or low number of synaptic vesicles and similar profiles in the dendritic subcompartment. However, in each case in which an identification of RGC dendrites by anatomic criteria (absence or low number of synaptic vesicles, identification of an axon in serial sections) was possible, they were always V-1L positive. Since we found V-1L immunoreactivity in all positively identified ganglion cell dendrites being postsynaptic elements at bipolar cell dyads in the IPL, all ganglion cell subtypes appear to be V-1L positive. Therefore, V-1L-immunoreactive synapses in the IPL can be used to assess processes affecting all ganglion cells. The specific label by V-1L immunoreactivity can be used to identify synaptic elements contributed by ganglion cell dendrites (Fig. 2) . We used this morphologic finding in subsequent experiments to identify changes in the synaptic connectivity of the retina and especially in the IPL where RGCs form contacts with their presynaptic partners. 

Effect of MCA Occlusion on Neurons in the GCL

In our experiments, we used a model for cerebral ischemia, temporary endovascular MCA occlusion, ^{23 26 27} which causes a partial infarction of the ipsilateral ophthalmic artery and leads to mild retinal ischemia. To further characterize the effect of the mild retinal ischemia model caused by temporary MCA occlusion we investigated the neurotoxic effects on retinal neurons using the TUNEL staining technique. Apoptotic cells were identified in vertical sections of control (eye contralateral to the ischemic insult), ischemic and estrogen-treated ischemic rat retinas using quantitative stereotaxis methods to determine stained profile counts. Figure 3 shows no apoptotic cells in the ganglion cell layer under control conditions (Fig. 3A) , whereas MCA occlusion for 1 hour and 24 hours of subsequent reperfusion (Fig. 3B) led to a highly reproducible number of apoptotic neurons in the GCL (Fig. 3B , arrows; 3.7% ± 1.1%). In estrogen-treated ischemia conditions (Fig. 3C) , no apoptotic neurons in the GCL were observed, indicating a complete rescue of cells including RGCs by estrogen treatment. Cells were counterstained with DAPI to visualize all cells (right panels in Fig. 3 ). Based on the previously published data identifying RGCs as the retinal cell types most susceptible to neurotoxicity, ^{1 2 3} it is reasonable to assume that RGCs in the GCL are the first cells that are affected by a mild neurotoxic insult. Additional indirect evidence that RGCs are the first targets of neurotoxicity in our model system and can be identified using the TUNEL staining technique are our ultrastructural results that all RGC subtypes appeared to be V-1L positive (Fig. 2) and the profound effect of ischemia on the number of V-1L-positive synapses in the IPL where RGCs stratify. However, it cannot be completely excluded that displaced amacrine cells are among the TUNEL-positive cells in the GCL. The number of apoptotic photoreceptor cells was not significantly different at all conditions (arrowheads in Fig. 3A 3B 3C indicate apoptotic photoreceptor cells; <0.2%). Apoptosis was presumably due to light-induced photoreceptor damage in albino rats and elevated sensitivity of their photoreceptors to light damage. ⁵⁵ The highly reproducible number of apoptotic photoreceptor cells typical in albino rats was present at the same frequency in all conditions. This served also as an internal control for the functionality of the cytochemical TUNEL assay in the same sections. 

Effect of Mild Retinal Ischemia on the Number of Vesl-1L/Homer 1c Immunoreactive Synapses

The distribution of Homer was quantitatively analyzed in the IPLs of retinas of OVX rats that had experienced retinal ischemia and were treated with a prolonged regimen of 17β-estradiol. The number of synapses in a given volume of all sublayers of the IPL was compared between animals with retinal ischemia, with or without estrogen treatment and controls. The number of V-1L-positive synapses was significantly decreased by approximately 23% in animals with retinal ischemia (Fig. 4B) compared with control animals without ischemia (contralateral to the ischemic insult: Fig. 4A ; contralateral to the ischemic insult after estrogen treatment: Fig. 4D ) and this effect was reversed by estrogen treatment (Fig. 4C) . Additional control experiments included animals that had undergone sham ischemia with and without estrogen treatment and the contralateral control eyes for all treatment conditions. The quantitative assessment of the effects of retinal ischemia with or without estrogen treatment on density of V-1L-positive synapse in defined volumes of rat retina IPL is summarized in Figure 4E . The loss of V-1L-immunoreactive synapses was uniform across the different sublaminas of the IPL, indicating that both ON- and OFF-pathways are affected similarly. Only under ischemic conditions was a significantly different number of V-1L-positive synapses detected (P < 0.01), whereas all control conditions showed no significant difference to one another and to the estrogen-treated group. The low number of apoptotic neurons in the GCL (Fig. 3) was accompanied by a strong decrease in glutamatergic, V-1L-immunoreactive synapses in all sublayers of the IPL (Fig. 4) , indicating that changes in the synaptic connectivity of RGCs are not only larger in number than the percentage of terminally apoptotic cells in the GCL, but that they also may precede more severe neuronal damage. 

Discussion

Endogenous steroid hormones as well as their metabolic products and chemical derivatives have been shown to mediate protection against cellular damage in a variety of organs and cell types, preventing the consequences of acute insults and degenerative diseases that ultimately lead to cell death. ^{24 25 26 27 28 29 30 31 32 33 34 35 36} Several mechanisms potentially mediating the protective effects of estrogen are being discussed currently, including effects on neurotransmitter signaling, the vasculature, immune responses, and cellular energy metabolism. ^{56 57} A number of clinical studies and surveys indicate beneficial effects of estrogen replacement therapy on the age-related risks for development of ocular diseases or discuss a correlation between reduced estrogen levels or exposure time and an increased risk for development of macular degeneration. ^{37 39 40 41} This indicates a potential role for estrogens and related compounds as neuroprotective agents for retinal neurons. A recent study investigated possible roles of estrogen ⁴² in an ischemia-reperfusion model leading to substantial damage and neurodegeneration of the retina that is characterized by a strong immune response and leukocyte accumulation. ^{42 58 59} Nonaka et al. ⁴² found that both the morphologic and physiological changes associated with a strong ischemic insult were partially attenuated by an acutely high pharmacologic dose of estrogen (0.1 mg/kg, 30 minutes before ischemia ⁴³ ). 

In the present study, we addressed two questions that are highly relevant for a potential role of estrogens in the treatment of retinal neurodegenerative diseases and acute neurotoxic insults to the retina: Which cells are primarily affected by ischemic insults and is the synaptic connectivity of retinal neurons affected before terminal apoptosis can be identified? To investigate these questions, we chose a model for mild retinal ischemia that affects only a small percentage of cells. Whereas previous studies were able to identify the inner retina as the target of strong retinal ischemia ^{42 58 59 60 61 62} and partial protection by estrogen, ⁴² we were able to detect neurons in the GCL, including RGCs, as the primary targets of a mild ischemic insult (Fig. 3) , which has potential relevance for similar pathophysiological processes with slow progression or mild onset characteristics (glaucoma, retinal ischemia during diabetic retinopathy, sickle cell retinopathy or stenosis of the carotid artery ^{4 5 6 7} ). Based on the previously published data identifying RGCs as the retinal cell types most susceptible to neurotoxicity, ^{1 2 3} it is reasonable to assume that RGCs are the first cells that are affected by a mild neurotoxic insult. 

Additional indirect evidence that RGCs are the first targets of neurotoxicity in our model system and can be identified with the TUNEL staining technique are our ultrastructural results that all RGC subtypes appear to be V-1L positive (Fig. 2) and the profound effect of ischemia on the number of V-1L-positive synapses in the IPL where RGCs stratify. However, it cannot be completely excluded that displaced amacrine cells are among the TUNEL-positive cells in the GCL. The combination of this finding (Fig. 3) with our ultrastructural localization of V-1L as a marker of postsynaptic elements of glutamatergic ribbon synapses in the retina (Fig. 2) allowed us to correlate ischemia and neuroprotective effects directly with the synaptic connectivity of RGCs (Fig. 4) . Vesl-1L as a postsynaptic clustering molecule has the unique property of linking neurotransmitter receptors, plasma membrane ion channels, intracellular calcium channels, and the cytoskeleton. ^{13 14 15 16 17 18 19 64 65} Therefore, Vesl-1L serves as a unique marker to detect synaptic changes mediated or affected by extracellular signaling, intracellular signaling or cytoskeletal processes. Vesl-1L is expressed by all retinal ganglion cells in dendrites postsynaptic to bipolar cell ribbon synapses and therefore is not restricted to subclasses of RGCs (Fig. 2) . Also, Vesl-1L is the ubiquitously expressed isoform of Vesl-1 with expression levels that are not affected by changing physiological conditions, as described for the conditionally expressed isoform of Vesl-1, Vesl-1S. ^{15 16 22} Therefore Vesl-1L can be used as a synaptic marker that is sensitive to overall changes in synaptic connectivity (i.e., neurotoxicity-induced loss of synapses), but with expression levels that are not influenced by changed physiological conditions. The distribution of V-1L coincided well with the synaptic expression sites of the best-characterized binding partner of V-1L, mGluR 1 and 5, in the retina. ⁴⁸  

The low statistical variance and the significance level of the present results indicate that Vesl-1L/Homer-1c immunoreactive synapses can function as an early indicator of neurodegeneration and neuroprotection and that the protective effects of estrogen on neurons in the GCL, especially V-1L-positive contacts of RGCs to cone bipolar cells in the IPL, are highly specific and reproducible. This could provide the basis for the development of novel treatments for neurodegenerative diseases and acute neurotoxic insults to RGCs. 

Whereas in a previous study estrogen had only modest protective effects on the ganglion cell layer and RGCs after transient retinal ischemia induced by temporary ligation of the optic sheath and nerve ⁴² the present study reports a complete rescue of neurons in the GCL from apoptosis and changes in synaptic connections of RGCs. This difference in the efficacy of estrogen may be attributable to two factors: Ligating the optic nerve can lead to permanent or transient damage of the optic nerve and/or the axons of RGCs, as well as to the impairment of axonal transport of RGCs, which has been shown to play a role in RGC function, viability, and survival. ^{63 64 65} This severe damage, which has been found to be associated with irreversible morphologic changes, ^{63 64 65} is potentially unresponsive to estrogen’s neuroprotective effects. ^{66 67}  

In the present study, cell death was solely induced by ischemia and the subsequent signaling events associated with decreased blood supply to the retina. Estrogen has been shown to be a potent neuroprotectant under these conditions of low oxygenation and supply of nutrients (e.g., leading to oxidative stress). ^{24 25 26 27 28 29 30 31 32 33 34 35 36} A second factor may be related to the fact that approximately 4% of neurons in the GCL, approximately 23% of V-1L-positive synapses in the IPL, and presumably no interneurons were affected by the mild retinal ischemia, which may indicate that the model mimics a slow-onset neurodegeneration similar to glaucoma and retinal ischemia during diabetic retinopathy, sickle cell retinopathy, or stenosis of the carotid artery, ^{4 5 6 7} that is responsive to neuroprotection by estrogen. 

²  Present address: Leiden University Medical Centre, Department of Neurology and Neurophysiology, Leiden, The Netherlands.

 Supported by a grant from the University of North Texas Health Science Center at Fort Worth Intramural Research Program (PK) and a Young Investigator Award from The National Alliance for Research on Schizophrenia and Depression (PK).

 Submitted for publication November 25, 2002; revised February 7 and March 18, 2003; accepted March 21, 2003.

 Disclosure: S. Kaja, None; S.-H. Yang, None; J. Wei, None; K. Fujitani, None; R. Liu, None; A.-M. Brun-Zinkernagel, None; J.W. Simpkins (P); K. Inokuchi, None; P. Koulen, 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.

 

Micrograph of a vertical cryostat section through rat retina immunostained with the antiserum against V-1L. Immunofluorescence was found in both synaptic layers, the IPL and the OPL. Arrows: larger immunoreactive profiles in the OPL, presumably cone photoreceptor synapses; arrowheads: smaller V-1L-immunoreactive puncta in the OPL, presumably rod photoreceptor synapses. Strong punctate immunoreactivity was found throughout the IPL and was excluded from the nuclear layers. The image was reconstructed from 12 deconvolved optical sections through a 12-μm-thick vertical cryostat section that had been merged in imaging software. This Z-stack image allowed V-1L immunoreactivity to be visualized in a three-dimensional space of a defined volume of retina tissue. Scale bar, 25 μm.

Figure 2.

 

Electron micrographs show the postsynaptic localization of V-1L immunoreactivity at glutamatergic cone bipolar cell ribbon synapses in the IPL of rat retina. Arrowhead: presynaptic ribbon in the bipolar cell axon terminal of (A) ON- and (B) OFF-cone bipolar cells. (✶) The two postsynaptic elements at the bipolar cell dyad. V-1L immunoreactivity, represented by the electron-dense precipitates close to the active zone of the ribbon synapse, is always confined to the postsynaptic elements. V-1L was expressed at both postsynaptic elements, indicating that dendrites of ganglion cells, constituting one or both elements of the bipolar cell axon terminal dyad, are labeled. Scale bar, 200 nm.

Figure 3.

 

TUNEL staining of vertical cryostat sections through rat retina after induction of mild retinal ischemia by a 1-hour endovascular MCA occlusion followed by reperfusion for 24 hours, leading to a partial infarction of the ipsilateral ophthalmic artery. (A, C) Apoptotic cells fluorescently stained by TUNEL cytochemistry (green); (B, D) the same sections labeled with DAPI (blue) to counterstain all cell nuclei and thereby identify the total cell number. Three different experimental conditions are shown, retinas contralateral (A) or ipsilateral (B) to the ischemic lesion from OVX rats that had received placebo replacement therapy or retinas ipsilateral to the ischemic lesion from OVX rats that had received estrogen replacement therapy (C; single subcutaneous injection 2 hours before MCA occlusion surgery, 100 μg/kg body weight). Plasma estrogen concentrations were below detection limits in sections in (A) and (B) and at 50 to 500 pg/mL in the section in (C), with a peak concentration at 2 hours after injection. ⁶³ TUNEL-positive neurons in the GCL (arrows) were observed only in ischemic retinas obtained from the ipsilateral eye of placebo treated (B) or control animals. In the GCL, 3.7% ± 1.1% of neurons were labeled as apoptotic. Apoptotic neurons in the GCL were absent in contralateral retinas of placebo-treated (A) or control animals and in ipsilateral (C) and contralateral retinas of estrogen-treated animals. Typical for albino rats—presumably induced by slow photodamage—a small number (<0.2%) of apoptotic photoreceptor cells was seen at the same frequency under all conditions (arrowheads: TUNEL-positive apoptotic photoreceptor cells). This finding served also as an internal control for the functionality of the cytochemical assay. Scale bar, 100 μm.

Figure 4.

 

Effects of mild retinal ischemia and estrogen treatment on the number of V-1L-immunoreactive synapses in the IPL of the rat retina. (A–C) Vertical cryostat sections through rat retinas show the distribution of V-1L immunoreactivity under different experimental conditions after induction of mild retinal ischemia. All four different experimental conditions are shown: retinas contralateral (A) or ipsilateral (B) to the ischemic lesion from OVX rats that had received placebo replacement therapy or retinas ipsilateral (C) or contralateral (D) to the ischemic lesion from OVX rats that had received estrogen replacement therapy. The strong punctate immunoreactivity throughout the IPL was reduced in the ischemic condition (B) when compared with control (A) or estrogen treated animals (C, D). The general distribution of immunoreactivity was not changed in any of the experimental conditions. No reduction in the number of either large or small immunoreactive profiles in the OPL (presumably cone and rod photoreceptor synapses, respectively) were observed. The images were reconstructed from 12 deconvolved optical sections through a 12-μm-thick vertical cryostat section and were merged in the imaging system software. These Z-stack images allowed visualization of V-1L immunoreactivity in a three-dimensional space of a defined volume of retina tissue. (E) The effects of mild retinal ischemia and estrogen treatment on the number of V-1L-immunoreactive synapses in the IPL of the rat retina are graphically summarized after quantification of six Z-stack images for each condition. Four different experimental conditions are shown: retinas contralateral (control) or ipsilateral (ischemia) to the ischemic lesion from OVX rats that had received placebo replacement therapy or retinas ipsilateral (ischemia + estrogen) or contralateral (control + estrogen) to the ischemic lesion from OVX rats that had received estrogen replacement therapy (dose, 100 μg/kg body weight). Data are presented as mean ± SEM for each condition and were compared with control data for statistical analysis by paired t-test. A significant change in the number of V-1L-immunoreactive synapses in the IPL was observed only in ischemic retinas (P < 0.01). Estrogen replacement therapy prevented this ischemia-induced loss of V-1L-positive synapses. Scale bar, (A–D) 25 μm.

The authors thank Margaret and Sara Koulen for excellent support.