Neuroprotection of retinal ganglion cell function and their central nervous system targets

7,8-Dihydroxiflavone Maintains Retinal Functionality and Protects Various Types of RGCs in Adult Rats with Optic Nerve Transection

To analyze the neuroprotective effects of 7,8-Dihydroxyflavone (DHF) in vivo and ex vivo, adult albino Sprague-Dawley rats were given a left intraorbital optic nerve transection (IONT) and were divided in two groups: One was treated daily with intraperitoneal (ip) DHF (5 mg/kg) (n = 24) and the other (n = 18) received ip vehicle (1% DMSO in 0.9% NaCl) from one day before IONT until processing. At 5, 7, 10, 12, 14, and 21 days (d) after IONT, full field electroretinograms (ERG) were recorded from both experimental and one additional naïve-control group (n = 6). Treated rats were analyzed 7 (n = 14), 14 (n = 14) or 21 d (n = 14) after IONT, and the retinas immune stained against Brn3a, Osteopontin (OPN) and the T-box transcription factor T-brain 2 (Tbr2) to identify surviving retinal ganglion cells (RGCs) (Brn3a⁺), α-like (OPN⁺), α-OFF like (OPN⁺Brn3a⁺) or M4-like/α-ON sustained RGCs (OPN⁺Tbr⁺). Naïve and right treated retinas showed normal ERG recordings. Left vehicle-treated retinas showed decreased amplitudes of the scotopic threshold response (pSTR) (as early as 5 d), the rod b-wave, the mixed response and the cone response (as early as 10 d), which did not recover with time. In these retinas, by day 7 the total numbers of Brn3a⁺RGCs, OPN⁺RGCs and OPN⁺Tbr2⁺RGCs decreased to less than one half and OPN⁺Brn3a⁺RGCs decreased to approximately 0.5%, and Brn3a⁺RGCs showed a progressive loss with time, while OPN⁺RGCs and OPN⁺Tbr2⁺RGCs did not diminish after seven days. Compared to vehicle-treated, the left DHF-treated retinas showed significantly greater amplitudes of the pSTR, normal b-wave values and significantly greater numbers of OPN⁺RGCs and OPN⁺Tbr2⁺RGCs for up to 14 d and of Brn3a⁺RGCs for up to 21 days. DHF affords significant rescue of Brn3a⁺RGCs, OPN⁺RGCs and OPN⁺Tbr2⁺RGCs, but not OPN⁺Brn3a⁺RGCs, and preserves functional ERG responses after IONT.

Animal Models of LED-Induced Phototoxicity. Short- and Long-Term In Vivo and Ex Vivo Retinal Alterations

Phototoxicity animal models have been largely studied due to their degenerative communalities with human pathologies, e.g., age-related macular degeneration (AMD). Studies have documented not only the effects of white light exposure, but also other wavelengths using LEDs, such as blue or green light. Recently, a blue LED-induced phototoxicity (LIP) model has been developed that causes focal damage in the outer layers of the superior-temporal region of the retina in rodents. In vivo studies described a progressive reduction in retinal thickness that affected the most extensively the photoreceptor layer. Functionally, a transient reduction in a- and b-wave amplitude of the ERG response was observed. Ex vivo studies showed a progressive reduction of cones and an involvement of retinal pigment epithelium cells in the area of the lesion and, in parallel, an activation of microglial cells that perfectly circumscribe the damage in the outer retinal layer. The use of neuroprotective strategies such as intravitreal administration of trophic factors, e.g., basic fibroblast growth factor (bFGF), brain-derived neurotrophic factor (BDNF), ciliary neurotrophic factor (CNTF) or pigment epithelium-derived factor (PEDF) and topical administration of the selective alpha-2 agonist (Brimonidine) have demonstrated to increase the survival of the cone population after LIP.

Neuroprotective Effect of Brimonidine against Facial Nerve Crush Injury in Rats via Suppressing GFAP/PAF Activation and Neuroinflammation

OBJECTIVE

This study aimed to investigate the potential neuroprotective action of brimonidine against facial nerve crush injury in rats and the possible underlying mechanisms.

METHODS

Sixty Wistar adult rats were randomly and equally divided into 3 groups: 40 rats underwent unilateral facial nerve crush injury and were administered with either saline (intraperitoneal, n = 20) or brimonidine 1 mg/kg/day (intraperitoneal, n = 20) for 5 consecutive days. Functional and electromyographic recovery was recorded postoperatively. The facial nucleus of 5 mice in each group was analyzed for mRNA expression levels of GFAP, PAF, NT-4, P75NTR, NF-κB, TNF-α, IL-6, and α2-ARs by qRT-PCR.

RESULTS

Brimonidine promoted the recovery of vibrissae movement, eyelid closure, and electrophysiological function in a rat model of nerve crush injury. Hematoxylin and eosin staining and electron microscopy showed significant recovery of Schwann cells and axons in the brimonidine group. Brimonidine attenuated the crush-induced upregulation in GFAP and PAF mRNA (p < 0.05), as well as enhanced the mRNA levels of NT-4 and P75NTR (p < 0.05), while decreased the expression of NF-κB, TNF-α and IL-6 (p < 0.05).

CONCLUSIONS

Brimonidine could promote the recovery of facial nerve crush injury in rats via suppressing of GFAP/PAF activation and neuroinflammation and increasing neurotrophic factors. Brimonidine may be apromising candidate agent for the treatment of facial nerve injury.

Aerobic exercise delays retinal ganglion cell death after optic nerve injury

Aerobic exercise has been shown to play a crucial role in preventing neurological diseases and improving cognitive function. In the present study, we investigated the effect of treadmill training on retinal ganglion cells (RGCs) following optic nerve transection in adult rats. We exercised the rats on a treadmill for 5 d/week (30 min/d at a rate of 9 m/min) or placed control rats on static treadmills. After 3 weeks of exercise, the left optic nerve of each rat was transected. After the surgery, the rat was exercised for another week. The percentages of surviving RGCs in the axotomized eyes of inactive rats were 67% and 39% at 5 and 7 days postaxotomy, respectively. However, exercised rats had significant more RGCs at 5 (74% survival) and 7 days (48% survival) after axotomy. Moreover, retinal brain-derived neurotrophic factor (BDNF) protein levels were significantly upregulated in response to exercise compared with those in the axotomized eyes of inactive rats. Blocking BNDF signaling during exercise by intraperitoneal injections of ANA-12, a BDNF tropomyosin receptor kinase (TrkB) receptor antagonist, reduced the number of RGCs in exercised rats to the level of RGCs in the inactive rats, effectively abolishing the protection of RGCs afforded by exercise. The results suggest that treadmill training effectively rescues RGCs from neurodegeneration following optic nerve transection by upregulating the expression of BDNF.

Functional and morphological alterations in a glaucoma model of acute ocular hypertension

To study short and long-term effects of acute ocular hypertension (AOHT) on inner and outer retinal layers, in adult Sprague-Dawley rats AOHT (87mmHg) was induced for 90min and the retinas were examined longitudinally in vivo with electroretinogram (ERG) recordings and optical coherent tomography (OCT) from 1 to 90 days (d). Ex vivo, the retinas were analyzed for rod (RBC) and cone (CBC) bipolar cells, with antibodies against protein kinase Cα and recoverin, respectively in cross sections, and for cones, horizontal (HZ) and ganglion (RGC) cells with antibodies against arrestin, calbindin and Brn3a, respectively in wholemounts. The inner retina thinned progressively up to 7d with no further changes, while the external retina had a normal thickness until 30d, with a 20% thinning between 30 and 90d. Functionally, the a-wave showed an initial reduction by 24h and a further reduction from 30 to 90d. All other main ERG waves were significantly reduced by 1d without significant recovery by 90d. Radial sections showed a normal population of RBCs but their terminals were reduced. The CBCs showed a progressive decrease with a loss of 56% by 30d. In wholemount retinas, RGCs diminished to 40% by 3d and to 16% by 30d without further loss. Cones diminished to 58% and 35% by 3 and 7d, respectively and further decreased between 30 and 90d. HZs showed normal values throughout the study. In conclusion, AOHT affects both the inner and outer retina, with a more pronounced degeneration of the cone than the rod pathway.

Topical Brimonidine or Intravitreal BDNF, CNTF, or bFGF Protect Cones Against Phototoxicity

Purpose

To develop a focal photoreceptor degeneration model by blue light-emitting diode (LED)-induced phototoxicity (LIP) and investigate the protective effects of topical brimonidine (BMD) or intravitreal brain-derived neurotrophic factor (BDNF), ciliary neurotrophic factor (CNTF), or basic fibroblast growth factor (bFGF).

Methods

In anesthetized, dark-adapted, adult female Swiss mice, the left eye was dilated and exposed to blue light (10 seconds, 200 lux). After LIP, full-field electroretinograms (ERG) and spectral-domain optical coherence tomography (SD-OCT) were obtained longitudinally, and reactive-Iba-1⁺monocytic cells, TUNEL⁺ cells and S-opsin⁺ cone outer segments were examined up to 7 days. Left eyes were treated topically with BMD (1%) or vehicle, before or right after LIP, or intravitreally with BDNF (2.5 μg), CNTF (0.2 μg), bFGF (0.5 μg), or corresponding vehicle right after LIP. At 7 days, S-opsin⁺ cone outer segments were counted within predetermined fixed-size areas (PFA) centered on the lesion in both flattened retinas.

Results

SD-OCT showed a circular region in the superior-temporal left retina with progressive thinning (207.9 ± 5.6 μm to 160.7 ± 6.8 μm [7 days], n = 8), increasing TUNEL⁺ cells (peak at 3 days), decreasing S-opsin⁺ cone outer segments, and strong microglia activation. ERGs were normal by 3 days. Total S-opsin⁺ cones in the PFA for LIP-treated and fellow-retinas were 2330 ± 262 and 5601 ± 583 (n = 8), respectively. All neuroprotectants (n = 7–11), including topical BMD pre- or post-LIP, or intravitreal BDNF, CNTF, and bFGF, showed significantly greater S-opsin⁺ cone survival than their corresponding vehicle-treated groups.

Conclusions

LIP is a reliable, quantifiable focal photoreceptor degeneration model. Topical BMD or intravitreal BDNF, CNTF, or bFGF protect against LIP-induced cone-photoreceptor loss.

Translational Relevance

Topical BMD or intravitreal BDNF, CNTF, or bFGF protect cones against phototoxicity.

Shared and Differential Retinal Responses against Optic Nerve Injury and Ocular Hypertension

Glaucoma, one of the leading causes of blindness worldwide, affects primarily retinal ganglion cells (RGCs) and their axons. The pathophysiology of glaucoma is not fully understood, but it is currently believed that damage to RGC axons at the optic nerve head plays a major role. Rodent models to study glaucoma include those that mimic either ocular hypertension or optic nerve injury. Here we review the anatomical loss of the general population of RGCs (that express Brn3a; Brn3a⁺RGCs) and of the intrinsically photosensitive RGCs (that express melanopsin; m⁺RGCs) after chronic (LP-OHT) or acute (A-OHT) ocular hypertension and after complete intraorbital optic nerve transection (ONT) or crush (ONC). Our studies show that all of these insults trigger RGC death. Compared to Brn3a⁺RGCs, m⁺RGCs are more resilient to ONT, ONC, and A-OHT but not to LP-OHT. There are differences in the course of RGC loss both between these RGC types and among injuries. An important difference between the damage caused by ocular hypertension or optic nerve injury appears in the outer retina. Both axotomy and LP-OHT induce selective loss of RGCs but LP-OHT also induces a protracted loss of cone photoreceptors. This review outlines our current understanding of the anatomical changes occurring in rodent models of glaucoma and discusses the advantages of each one and their translational value.

Retinal neurodegeneration in experimental glaucoma

In rats and mice, limbar tissues of the left eye were laser-photocoagulated (LP) and ocular hypertension (OHT) effects were investigated 1 week to 6 months later. To investigate the innermost layers, retinas were examined in wholemounts using tracing from the superior colliculi to identify retinal ganglion cells (RGCs) with intact retrograde axonal transport, melanopsin immunodetection to identify intrinsically photosensitive RGCs (m(+)RGC), Brn3a immunodetection to identify most RGCs but not m(+)RGCs, RECA1 immunodetection to examine the inner retinal vessels, and DAPI staining to detect all nuclei in the GC layer. The outer retinal layers (ORLs) were examined in cross sections analyzed morphometrically or in wholemounts to study S- and L-cones. Innervation of the superior colliculi was examined 10 days to 14 weeks after LP with orthogradely transported cholera toxin subunit B. By 2 weeks, OHT resulted in pie-shaped sectors devoid of FG(+)RGCs or Brn3a(+)RGCs but with large numbers of DAPI(+)nuclei. Brn3a(+)RGCs were significantly greater than FG(+)RGCs, indicating the survival of large numbers of RGCs with their axonal transport impaired. The inner retinal vasculature showed no abnormalities that could account for the sectorial loss of RGCs. m(+)RGCs decreased to approximately 50-51% in a diffuse loss across the retina. Cross sections showed focal areas of degeneration in the ORLs. RGC loss at 1m diminished to 20-25% and did not progress further with time, whereas the S- and L-cone populations diminished progressively up to 6m. The retinotectal projection was reduced by 10 days and did not progress further. LP-induced OHT results in retrograde degeneration of RGCs and m(+)RGCs, severe damage to the ORL, and loss of retinotectal terminals.

Effects of ocular hypertension in the visual system of pigmented mice

To study the effects of ocular hypertension (OHT) on the visual system of C57BL/6 pigmented mice, the limbal and episcleral veins of the left eye were laser photocoagulated (LP). LP increased the intraocular pressure during the first five days (d), reaching basal values at 7d. To investigate the effect of OHT on the retinal ganglion cell (RGC) retrograde axonal transport, hydroxistilbamidine methanesulfonate (OHSt) was applied to both superior colliculi (SCi) and the retinas were dissected 2 or 4 weeks after LP. To determine RGC survival, these same retinas were immunoreacted against Brn3a (general RGC population) and melanopsin (intrinsically photosensitive RGCs, m⁺RGCs). To study whether OHT affected non-RGC neurons in the ganglion cell layer (GCL), RGCs were immunodetected with Brn3a and all GCL nuclei counterstained with DAPI in a group of animals examined 4 weeks post-LP. Innervation of the SCi was examined at 10 days, 8 or 14 weeks after LP with the orthogradely transported cholera toxin subunit-B. OHT resulted in diffuse and sectorial loss of OHSt+RGCs (50% at 2 weeks and 62% at 4 weeks) and in a comparable loss of Brn3a+RGCs at the same time intervals. m+RGCs decreased to 59% at 2 weeks and to 46% at 4 weeks, such loss was diffuse, did not parallel the sectorial loss of the general RGC population and was more severe in the superior-temporal retina. In the GCL, cell loss is selective for RGCs and does not affect other non-RGC neurons. The retinotectal innervation appeared significantly reduced at 10 days (55.7%) and did not progress further up to 14 weeks (46.6%). Thus, LP-induced OHT results in retrograde degeneration of RGCs and m+RGCs, as well as in the loss of CTB-labelled retinotectal terminals.

Laser-induced ocular hypertension in adult rats does not affect non-RGC neurons in the ganglion cell layer but results in protracted severe loss of cone-photoreceptors

To investigate the long-term effects of laser-photocoagulation (LP)-induced ocular hypertension (OHT) in the innermost and outermost (outer-nuclear and outer segment)-retinal layers (ORL). OHT was induced in the left eye of adult rats. To investigate the ganglion cell layer (GCL) wholemounts were examined at 1, 3 or 6 months using Brn3a-immunodetection to identify retinal ganglion cells (RGCs) and DAPI-staining to detect all nuclei in this layer. To study the effects of LP on the ORL up to 6 months, retinas were: i) fresh extracted to quantify the levels of rod-, S- and L-opsin; ii) cut in cross-sections for morphometric analysis, or; iii) prepared as wholemounts to quantify and study retinal distributions of entire populations of RGCs (retrogradely labeled with fluorogold, FG), S- and L-cones (immunolabeled). OHT resulted in wedge-like sectors with their apex on the optic disc devoid of Brn3a(+)RGCs but with large numbers of DAPI(+)nuclei. The levels of all opsins diminished by 2 weeks and further decreased to 20% of basal-levels by 3 months. Cross-sections revealed focal areas of ORL degeneration. RGC survival at 15 days represented approximately 28% and did not change with time, whereas the S- and L-cone populations diminished to 65% and 80%, or to 20 and 35% at 1 or 6 months, respectively. In conclusion, LP induces in the GCL selective RGCs loss that does not progress after 1 month, and S- and L-cone loss that progresses for up to 6 months. Thus, OHT results in severe damage to both the innermost and the ORL.

A novel in vivo model of focal light emitting diode-induced cone-photoreceptor phototoxicity: neuroprotection afforded by brimonidine, BDNF, PEDF or bFGF

We have investigated the effects of light-emitting diode (LED)-induced phototoxicity (LIP) on cone-photoreceptors and their protection with brimonidine (BMD), brain-derived neurotrophic factor (BDNF), pigment epithelium-derived factor (PEDF), ciliary neurotrophic factor (CNTF) or basic fibroblast growth factor (bFGF). In anesthetized, dark adapted, adult albino rats a blue (400 nm) LED was placed perpendicular to the cornea (10 sec, 200 lux) and the effects were investigated using Spectral Domain Optical Coherence Tomography (SD-OCT) and/or analysing the retina in oriented cross-sections or wholemounts immune-labelled for L- and S-opsin and counterstained with the nuclear stain DAPI. The effects of topical BMD (1%) or, intravitreally injected BDNF (5 µg), PEDF (2 µg), CNTF (0.4 µg) or bFGF (1 µg) after LIP were examined on wholemounts at 7 days. SD-OCT showed damage in a circular region of the superotemporal retina, whose diameter varied from 1,842.4±84.5 µm (at 24 hours) to 1,407.7±52.8 µm (at 7 days). This region had a progressive thickness diminution from 183.4±5 µm (at 12 h) to 114.6±6 µm (at 7 d). Oriented cross-sections showed within the light-damaged region of the retina massive loss of rods and cone-photoreceptors. Wholemounts documented a circular region containing lower numbers of L- and S-cones. Within a circular area (1 mm or 1.3 mm radius, respectively) in the left and in its corresponding region of the contralateral-fellow-retina, total L- or S-cones were 7,118±842 or 661±125 for the LED exposed retinas (n = 7) and 14,040±1,860 or 2,255±193 for the fellow retinas (n = 7), respectively. BMD, BDNF, PEDF and bFGF but not CNTF showed significant neuroprotective effects on L- or S-cones. We conclude that LIP results in rod and cone-photoreceptor loss, and is a reliable, quantifiable model to study cone-photoreceptor degeneration. Intravitreal BDNF, PEDF or bFGF, or topical BMD afford significant cone neuroprotection in this model.

Changes in the inner and outer retinal layers after acute increase of the intraocular pressure in adult albino Swiss mice

In adult albino mice the effects of increased intraocular pressure on the outer retina and its circuitry was investigated at intervals ranging 3-14 weeks. Ocular hypertension (OHT) was induced by cauterizing the vessels draining the anterior part of the mice eye, as recently reported (Salinas-Navarro et al., 2009a). Electroretinographic (ERG) responses were recorded simultaneously from both eyes and compared each other prior to and at different survival intervals of 2, 8 or 12 weeks after lasering. Animals were processed at 3, 9 or 14 weeks after lasering, and radial sections were obtained in the cryostat and further processed for immunocytochemistry using antibodies against recoverin, gamma-transducin, Protein Kinase C-alpha (PKC-alpha), calbindin or synaptophysin. The synaptic ribbons were identified using an antibody against the protein bassoon, which labels photoreceptor ribbons and nuclei were identified using TO-PRO. Laser photocoagulation of the perilimbar and episcleral veins of the left eye resulted in an increase in mean intraocular pressure to approximately over twice its baseline by 24 h that was maintained for approximately five days reaching basal levels by 1 week. ERG recordings from the different groups of mice showed their a-, b-wave and scotopic threshold response (STR) amplitudes, when compared to their contralateral fellow eye, reduced to 62%, 52% and 23% at 12 weeks after lasering. Three weeks after lasering, immunostaining with recoverin and transducin antibodies could not document any changes in the outer nuclear layer (ONL) but both ON-rod bipolar and horizontal cells had lost their dendritic processes in the outer plexiform layer (OPL). Sprouting of horizontal and bipolar cell processes were observed into the ONL. Fourteen weeks after lasering, protein kinase-C antibodies showed morphologic changes of ON-rod bipolar cells and calbindin staining showed abnormal horizontal cells and a loss of their relationship with their presynaptic input. Moreover, at this time, quantitative studies indicate significant diminutions in the number of photoreceptor synaptic ribbons/100 microm, and in the thickness of the outer nuclear and plexiform layer, when compared to their fellow eyes. Increased intraocular pressure in Swiss mice results in permanent alterations of their full field ERG responses and in changes of the inner and outer retinal circuitries.

Ocular hypertension impairs optic nerve axonal transport leading to progressive retinal ganglion cell degeneration

Ocular hypertension (OHT) is the main risk factor of glaucoma, a neuropathy leading to blindness. Here we have investigated the effects of laser photocoagulation (LP)-induced OHT, on the survival and retrograde axonal transport (RAT) of adult rat retinal ganglion cells (RGC) from 1 to 12 wks. Active RAT was examined with fluorogold (FG) applied to both superior colliculi (SCi) 1 wk before processing and passive axonal diffusion with dextran tetramethylrhodamine (DTMR) applied to the optic nerve (ON) 2 d prior to sacrifice. Surviving RGCs were identified with FG applied 1 wk pre-LP or by Brn3a immunodetection. The ON and retinal nerve fiber layer were examined by RT97-neurofibrillar staining. RGCs were counted automatically and color-coded density maps were generated. OHT retinas showed absence of FG+ or DTMR+RGCs in focal, pie-shaped and diffuse regions of the retina which, by two weeks, amounted to, approximately, an 80% of RGC loss without further increase. At this time, there was a discrepancy between the total number of surviving FG-prelabelled RGCs and of DMTR+RGCs, suggesting that a large proportion of RGCs had their RAT impaired. This was further confirmed identifying surviving RGCs by their Brn3a expression. From 3 weeks onwards, there was a close correspondence of DTMR+RGCs and FG+RGCs in the same retinal regions, suggesting axonal constriction at the ON head. Neurofibrillar staining revealed, in ONs, focal degeneration of axonal bundles and, in the retinal areas lacking backlabeled RGCs, aberrant staining of RT97 characteristic of axotomy. LP-induced OHT results in a crush-like injury to ON axons leading to the anterograde and protracted retrograde degeneration of the intraocular axons and RGCs.

Retinal ganglion cell population in adult albino and pigmented mice: a computerized analysis of the entire population and its spatial distribution

In adult Swiss albino and C57 pigmented mice, RGCs were identified with a retrogradely transported neuronal tracer applied to both optic nerves (ON) or superior colliculi (SCi). After histological processing, the retinas were prepared as whole-mounts, examined and photographed under a fluorescence microscope equipped with a motorized stage controlled by a commercial computer image analysis system: Image-Pro Plus((R)) (IPP). Retinas were imaged as a stack of 24-bit color images (140 frames per retina) using IPP with the Scope-Pro plug-in 5.0 and the images montaged to create a high-resolution composite of the retinal whole-mount when required. Single images were also processed by specific macros written in IPP that apply a sequence of filters and transformations in order to separate individual cells for automatic counting. Cell counts were later transferred to a spreadsheet for statistical analysis and used to generate a RGC density map for each retina.

RESULTS

The mean total numbers of RGCs labeled from the ON, in Swiss (49,493+/-3936; n=18) or C57 mice (42,658+/-1540; n=10) were slightly higher than the mean numbers of RGCs labeled from the SCi, in Swiss (48,733+/-3954; n=43) or C57 mice (41,192+/-2821; n=42), respectively. RGCs were distributed throughout the retina and density maps revealed a horizontal region in the superior retina near the optic disk with highest RGC densities. In conclusion, the population of mice RGCs may be counted automatically with a level of confidence comparable to manual counts. The distribution of RGCs adopts a form of regional specialization that resembles a horizontal visual streak.

A computerized analysis of the entire retinal ganglion cell population and its spatial distribution in adult rats

In adult albino (SD) and pigmented (PVG) rats the entire population of retinal ganglion cells (RGCs) was quantified and their spatial distribution analyzed using a computerized technique. RGCs were back-labelled from the optic nerves (ON) or the superior colliculi (SCi) with Fluorogold (FG). Numbers of RGCs labelled from the ON [SD: 82,818+/-3,949, n=27; PVG: 89,241+/-3,576, n=6) were comparable to those labelled from the SCi [SD: 81,486+/-4,340, n=37; PVG: 87,229+/-3,199; n=59]. Detailed methodology to provide cell density information at small scales demonstrated the presence of a horizontal region in the dorsal retina with highest densities, resembling a visual streak.