Remote Ischemic Preconditioning Protects Retinal Photoreceptors: Evidence From a Rat Model of Light-Induced Photoreceptor Degeneration

Abstract

Purpose: To test whether remote ischemic preconditioning (RIP) is protective to photoreceptors, in a light damage model, and to identify mechanisms involved.

Methods: A pressure cuff was used to induce ischemia (2 × 5 minutes) in one hind limb of 4- to 6-month-old albino Sprague-Dawley rats raised in dim, cyclic light (12 hours 5 lux, 12 hours dark). Immediately following the ischemia, rats were exposed to bright continuous light (1000 lux) for 24 hours. After 7-day survival in dim, cyclic light conditions, retinal function was assessed using the flash electroretinogram (ERG) and retinal structure was examined for photoreceptor survival and death, as well as for stress. Messenger RNA and protein expression of growth factors and brain-derived neurotrophic factor (BDNF) receptors was also assessed at 7-day survival.

Results: Bright light exposure reduced the amplitude of the a- and b-waves of the ERG, upregulated the expression of glial fibrillary acidic protein (GFAP) by Müller cells, increased the number of dying (TUNEL+) photoreceptors, and reduced the number of surviving photoreceptors. Remote ischemic preconditioning mitigated all of these bright light–induced effects. Remote ischemic preconditioning–induced protection was associated with increased retinal expression of BDNF and its low-affinity receptor NGFR.

Conclusions: The present study provides evidence, for the first time, that RIP protects photoreceptors against bright light–induced photoreceptor degeneration. This observation is consistent with previous reports of RIP-induced protection of the inner retina and of other vital organs. Brain-derived neurotrophic factor may play a role in mediating the RIP-induced neuroprotection through activation of NGFR.

The photoreceptors of the mammalian retina are vulnerable to a range of stressful stimuli, including metabolic toxins,¹ excessive or even physiological levels of light,^2,3 hypoxia or hyperoxia,^3–6 and more chronic stresses, such as aging.⁷ 

The “conditioning” of the retina to stabilize its photoreceptors in the face of environmental stress has been tested in a range of experiments. In several studies, the conditioning intervention that stabilized photoreceptors was known to be potentially damaging. Exposure to ischemia,⁸ hyperthermia,⁹ bright light^10,11 or physiological levels of light,⁶ nearby mechanical damage,¹² and gamma rays¹³ were all effective, in appropriately designed experiments, in making photoreceptors resistant to degeneration; all are also potentially damaging. 

The conditioning intervention tested in the present study is termed remote ischemia. The idea that putting a tourniquet on a leg or arm might stabilize the retina in the face of stress arose from evidence of ischemic conditioning of heart muscle¹⁴ (for a recent review see Ref. 15). These studies demonstrated that an episode of sublethal ischemia in a patch of heart muscle reduced the injury caused by reperfusion after a subsequent, more severe period of ischemia, and that the protection involved the activation by ischemia of endogenous self-protective mechanisms, reviewed in Brooks and Andrews,¹⁵ in the affected muscle. This fundamental observation has been expanded in four ways. 

First, it became clear that ischemia-induced protection is not specific to heart muscle; the same protective response to local ischemia has since been reported in the brain^16,17 and retina.^8,18 Many tissues, perhaps all vascularized tissues, can respond to sublethal ischemia by producing signals that induce a protective response. Second, it became clear that, although the conditioning ischemia may be localized (e.g., blockage of one coronary artery), the protection spreads beyond the ischemic site. A key observation in this field is that ischemia in a patch of heart muscle protects the whole heart.¹⁹ Third, it became clear that ischemia in one tissue could induce protection in remote tissues. Called remote ischemic preconditioning (RIP), this effect is now in increasing use clinically. In humans, the conditioning ischemia is usually induced in one arm using a conventional tourniquet. Limb ischemia has been shown to reduce the cerebral lesion and functional loss caused by a stroke,²⁰ to reduce retinal damage caused by local ischemia,²¹ to reduce acute renal injury in a range of clinical settings, and to reduce liver injury in ischemia (reviewed in Refs. 22–24). Fourth, it became clear that the protective response induced by ischemia (local or remote) is effective not only against ischemia–reperfusion injury, but also against a number of tissue-destructive stresses. For example, RIP protects ganglion cells from degeneration after injury to the optic nerve²⁵; protects the lung against injury induced by ischemia–reperfusion injury in muscle²⁶; and protects the brain from hypothermic circulatory arrest.²⁷ 

These observations suggest that the phenomenon of ischemia-induced protection is a general response, inducible in any tissue, effective against any stress, and that the factors that mediate the protection must spread spatially from the site of ischemia throughout the body. This general response may underlie, for example, the strong correlation between moderate exercise and lower morbidity.²⁸ The benefits of exercise, these observations suggest, arise because exercise induces ischemia of skeletal muscle, which induces body-wide upregulation of tissue-protective mechanisms. 

This concept needs testing in specific models, however, and its limits need definition. In an earlier study, we showed that remote ischemia enhances the function of the normal rat retina, increasing the response of photoreceptors to light.²⁹ In this study, we have tested whether remote ischemia can stabilize photoreceptors against damage caused by exposure to bright light, and whether the trophic factor brain-derived neurotrophic factor (BDNF), which is known to mediate photoreceptor stability in other models,^30,31 contributes to that stability. 

Methods

Animals

Sprague-Dawley rats were sourced from the Animal Resource Centre (Perth, WA, Australia). Animals were raised from birth in controlled scotopic conditions (12 hours at 5–8 lux, 12 hours dark, 22°C). Normal chow (WEHI, Barastoc, VIC, Australia) and water were available ad libitum. All experimental methods and animal care procedures were approved by the University of Sydney Animal Ethics Committee (K22/6-2011/1/5563), and were in compliance with the ARVO Statement for use of Animals in Ophthalmic and Vision Research. All experiments were performed on animals aged 4 to 6 months. 

Experimental Design

The experimental design is shown in Figure 1. The ERG was recorded and a week later one limb was made ischemic for 2 × 5 minutes. Fifteen to 30 minutes later the animal was exposed to bright continuous light (details below) for 24 hours. One week after exposure to bright light, the ERG was recorded again; the animal was euthanized, the eyes were taken for examination, and a blood sample was taken. 

Figure 1

 

Experimental groups. The blood pressure cuff symbol indicates ischemia of the lower limb for two 5-minute periods separated by a reperfusion period of 5 minutes. The lamp symbol shows when animals were exposed to bright light for 24 hours to induce degeneration of photoreceptors. The red arrows indicate when ERGs were performed. At 7 days after light exposure, the animals were euthanized, the retinas were collected, and a blood sample was taken.

There was one design variation within the series. Animals in an initial cohort were anesthetized (details below) during ischemia of the limb, while later cohorts were not anesthetized but were gently restrained by a skilled handler. This variation allowed us to assess whether the anesthetic contributed to the protective outcome. The experimental groups were controls; light damaged (LD) without conditioning; remote ischemia conditioned (anesthetized) and light damaged (RIP-LD); and remote ischemia conditioned (awake) and light damaged (RIP-LD awake). There were six animals in each group. 

Induction of Limb Ischemia

Where RIP was applied under anesthesia, each animal was injected with ketamine (60 mg/kg) and xylazine (7 mg/kg intraperitoneal; Parnell Manufacturing Pty Ltd, Alexandria, NSW, Australia). Where RIP was applied without anesthesia, animals were wrapped in a towel and restrained by a skilled operator. Sham RIP experiments involved the animal's being wrapped in a towel and restrained with the foot extended. 

A cuff developed for the measurement of blood pressure in human neonates (size 2; Cas Medical Systems, Branford, CT, USA) was wrapped around the upper hind limb of the rat as proximally as possible. The cuff was inflated manually (DS56; Welch Allyn, Inc., Skaneateles Falls, NY, USA), and was maintained at a pressure exceeding normal systolic blood pressure (160–200 mm Hg). Ischemia was monitored by three measures: the pressure in the cuff, shown by the sphygmomanometer; the color of the footpad skin, observed directly (it became blueish); and the temperature of the footpad, using a thermistor probe (Thermistor Pod; AD Instruments Pty Ltd, Bella Vista, NSW, Australia) taped against the skin. 

Pressure was applied for two 5-minute periods, separated by a 5-minute period of reperfusion. During this period of reperfusion and at the end of the second period of ischemia, both the change in skin color and the drop in skin temperature (typically ∼2°C) reversed promptly, as previously reported.²⁹ 

Exposure to Bright Light

Rats assigned to light damage groups were exposed to bright continuous light to induce retinal degeneration. The animals were separated into individual boxes with transparent Plexiglass covers. The light source was a cool white fluorescent tube (5500 K, 1000 Watts, XX657; Ericsson, Willoughby, Ohio, USA) and delivered 1000 lux, measured using a portable lux meter (LX1010B; Shanghai Handsun Electronic Co., Ltd, Shanghai, China) held 5 cm above the floor of the box. After 24-hour exposure, rats were returned to normal housing under scotopic conditions. This follows previous studies, including several from this laboratory.^32–34 

Electroretinogram

Preparation.

In keeping with our previous study,²⁹ rats were dark adapted overnight (12–15 hours) and recordings performed the following morning. Under dim red light illumination, rats were anesthetized by intraperitoneal injection of ketamine (60 mg/kg) and xylazine (7 mg/kg), a dose sufficient to maintain effective anesthesia for 45 to 60 minutes. Mydriasis was achieved with topical application of atropine sulfate (1.0 %; Bausch & Lomb Australia Pty Ltd, Macquarie Park, NSW, Australia). Proxymetacaine (0.5%; Alcon Laboratories Pty Ltd, Frenchs Forest, NSW, Australia) was applied topically for corneal anesthesia and Carbomer (2 mg/g polyacrylic acid; Novartis Pharmaceuticals, North Ryde, NSW, Australia) for corneal hydration. A thread was loosely drawn around the eyeball to minimize lid activity. The animal was supported on a platform warmed by internal circulating water at 40°C; this reliably maintained body temperature at 37°C to 38°C as monitored by a rectal temperature probe (Harvard Apparatus, Holliston, MA, USA). 

Full-Field ERG Recordings.

With the animal laid on the warmed platform, the head was positioned with the right eye exposed to a Ganzfeld integrating sphere (Photometric Solutions International, Huntingdale, VIC, Australia). Once the setup was complete and the dim red light removed, 10 minutes of dark adaptation was allowed before commencement of recording. 

The ERG was recorded between a custom-made 4-mm platinum positive electrode lightly touching the cornea and a 2-mm diameter Ag/AgCl pellet electrode (No. E206; SDR Clinical Technology, Middle Cove, NSW, Australia) placed in the mouth. Both were referenced to a stainless steel ground (23 g × 1.25 inch; Terumo Medical, Somerset, NJ, USA) inserted into the skin of the rump. Signals were recorded with band-pass setting of 0.3 to 1000 Hz, with a 2-kHz acquisition rate (using a Powerlab 4SP system; AD Instruments Pty Ltd, Bella Vista, NSW, Australia). 

The light stimuli used to elicit the ERG were brief light-emitting diode flashes. The flash was programmed to be of 1- to 2-ms duration and −4.4 to 2.0 log scot cd.s.m⁻² intensity. Luminous energy was calibrated (IL1700; International Light Research, Peabody, MA, USA) to give rodent (λ_max = 502 nm) scotopic (Z-CIE luminosity filter) luminous measures (cd.s.m⁻²). 

Eleven intensities of flash were used over the range stated above. At lower flash intensities (−4.4 to −0.3 log scot cd.s.m⁻²), responses were averaged from four flashes delivered at 1 Hz. At higher intensities (0.4–2.0 log scot cd.s.m⁻²), fewer responses were averaged and interstimulus intervals were increased, up to 90 seconds. The amplitudes of the a- and b-waves were measured as shown in Figure 1 of our previous study²⁹ and Figure 2A. 

Figure 2

 

Remote ischemic preconditioning mitigated loss of retinal function caused by LD. (A) The ERG evoked by a bright flash. The a-wave was measured from the baseline to the early negative peak (left red arrow); the b-wave was measured from the early negative peak to the subsequent positive peak (right red arrow). (B) Representative waveforms elicited by a flash of intensity 2.0 log scot cd.s.m⁻². Records are shown for the RIP group (gray dash), for the control group (black), for the LD group (red), for the RIP-LD group (blue), and for the RIP-LD awake group (blue dash). For the latter three groups, the ERG was recorded 7 days after exposure to damaging light. (C) a- and b-wave amplitudes for individual animals and group means from the five experimental groups. Post hoc (Tukey's) comparisons confirmed (P < 0.05, n = 6) that amplitudes are greater in the RIP-LD awake and RIP-LD groups than the LD group (blue versus red, asterisks). (D) Comparison of the mean amplitudes ± SEM of the a-waves recorded for the LD (red) and RIP-LD (blue) groups elicited by flashes of intensity −1.3 to 2.0 log scot cd.s.m⁻². Two-way ANOVA comparison with Tukey's post hoc test identified intensities at which the RIP-LD values are significantly greater (*P < 0.05, **P < 0.01, ***P < 0.001, n = 6). (E) Comparison of the mean amplitudes ± SEM of the b-waves recorded for the LD (red) and RIP-LD (blue) groups elicited by flashes of intensity −4.4 to 2.0 log scot cd.s.m⁻². Two-way ANOVA comparison with Tukey's post hoc test identified intensities at which the RIP-LD values are significantly greater (*P < 0.05, **P < 0.01, ***P < 0.001, n = 6).

Histologic Studies

Tissue Collection.

Immediately following completion of the posttreatment ERG, rats were euthanized (60 mg/kg Lethabarb intraperitoneal; Virbac, Regent Park, NSW, Australia). From each animal, one eye was allocated for histologic studies, while the other was dissected to remove the retina for molecular studies. For the eye destined for histology, a stitch was inserted in the superior portion of the conjunctiva prior to enucleation to aid later orientation, and the eyeball pierced with a 25-gauge needle to aid fixation. For the eye destined for molecular studies, an incision was made through the cornea and the vitreous humor was expelled. The retina was removed with sterilized forceps, placed in a nuclease-free tube, and snap frozen in liquid nitrogen. 

Histology.

Eyes were fixed in 4% paraformaldehyde in PBS at 4°C for 2 hours, rinsed in PBS, and cryoprotected overnight in 30% sucrose in PBS. Eyes were embedded in optimum cutting temperature compound (TissueTek; Sakura, Alphen aan den Rijn, The Netherlands) and frozen indirectly in isopentane cooled by liquid nitrogen. Retinas were sectioned sagittally at 20 μm using a cryostat (CM1850; Leica, North Ryde, NSW, Australia). At the appearance of the optic nerve head, sections were collected on gelatin- and poly-L-lysine–coated slides and stored at −20°C. 

Sections were stained with the TUNEL technique, to detect apoptotic cells using in situ cell death detection kit (Roche, Basel, Switzerland), and were immunolabeled with antibodies against glial fibrillary acidic protein (GFAP) and the BDNF receptors NGFR, NTRK2 and SORL1. 

Sections were prepared for TUNEL labeling with a 10-minute wash in terminal deoxytransferase (TdT) buffer (3 mM Trizma base, 14 mM sodium cacodylate, and 100 μM cobalt chloride), then incubated in a solution containing TdT (0.03 units/μL) and biotinylated dUTP (4 μM) for 1 hour at 37°C in TdT buffer. The reaction was terminated by a 15-minute wash in saline–sodium citrate buffer (300 mM sodium chloride, 30 mM sodium citrate). Sections were blocked with 10% goat or rabbit serum (Sigma-Aldrich Corp., St. Louis, MO, USA) in PBS for 30 minutes at room temperature. Sections were then incubated overnight at 4°C with either rabbit or mouse anti-GFAP (1:1000; Dako, Glostrup, Denmark), anti-NGFR (1:1000; Biosensis, Thebarton, SA, Australia), anti-NTRK2 (1:200; Santa Cruz Biotech, Dallas, Texas, USA), or anti-SORL1 (1:500; R&D Systems, Minneapolis, Minnesota, USA) in 1% goat or rabbit serum. The following day, sections were washed for 15 minutes in PBS prior to being incubated with a secondary antibody for 1 hour at room temperature, either goat anti-rabbit IgG Alexa 488 or rabbit anti-mouse IgG Alexa 594 (1:500; Life Technologies, Carlsbad, CA, USA) and for TUNEL labeling, Cy3-Streptavidin conjugate (1:1000, Life Technologies). After washing in PBS, sections were incubated for 2 minutes with the nuclear label bisbenzimide (1:10,000 wt/vol; Sigma-Aldrich Corp.). The slides were mounted with glycerol/gelatin (1:1; Sigma-Aldrich Corp.) and coverslipped. Negative controls were included with the substitution of primary antibody for 1% goat or rabbit serum. 

For systematic measurements of retinal parameters (thickness of the outer nuclear layer [ONL], numbers of TUNEL+ cells, extent of Müller GFAP labeling), each section studied was scanned from the superior edge to the optic disc to the inferior edge, in 400-μm steps; typically there were 10 to 12 such steps from the superior edge to the disc, and another 10 to 12 from the disc to the inferior edge. At each step an image was taken using either a Zeiss Axioplan 2 deconvolution microscope (Carl Zeiss, Gottingen, Germany) or Zeiss LSM 510 confocal microscope. The criteria for identifying TUNEL+ cells have been described previously.³⁵ In each field we measured the thickness of the ONL as a proportion of the thickness of the retina, from the inner to the outer limiting membrane. Where the section was labeled for GFAP, we also recorded the labeled length of the Müller cells as a proportion of retinal thickness. Counts and measurements were averaged over three sections per eye for four to six animals per group. 

Molecular Studies

RNA Isolation and Reverse Transcription.

Total RNA was isolated from individual retinas using TRIzol (Life Technologies) and purified using the RNAqueous kit (Life Technologies), according to manufacturer's instructions. RNA purity and concentration were determined by UV absorbance using a Nanodrop spectrophotometer (ND-1000; Thermo Scientific, Wilmington, DE, USA). RNA integrity was assessed using a 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). Retinal RNA (1 μg) was reversed transcribed into cDNA using the Transcriptor First Strand cDNA kit (Roche Applied Science, Penzberg, Germany) according to manufacturer's instructions. Resulting cDNA was diluted 1:20 in nuclease-free water. 

Real-Time PCR.

Primers for transcripts of interest were designed using either Primer Express Software (Applied Biosystems, Inc., Scorseby, Victoria, Australia) or Primer 3 (NCBI primer-blast). Primer sequences are given in the Table. Real-time PCR was performed on the Light Cycler 480 system (Roche Applied Science). Each 20-μL reaction contained SYBR PCR reaction mix (Roche Applied Science), 500 nM forward and reverse primers, and 5 μL diluted cDNA template. Each reaction was run in triplicate, with a negative control included for each primer pair. The thermal cycling conditions were 40 cycles of 95°C for 10 seconds, with differing annealing temperatures held for 10 seconds (Table), and extension and SYBR fluorescence measurements at 72°C for 15 seconds. The ΔΔCt relative quantification method was used to quantify transcript levels for genes of interest relative to the reference gene ribosomal protein lateral stalk subunit P0 (Rplp0). 

Table

 

Primers Used in q-PCR Measurements

 

Protein Extraction.

Individual retinas were placed in cell lysis RIPA buffer (Millipore, Bayswater, VIC, Australia) supplemented with protease inhibitor cocktail (Roche Applied Science) and homogenized using a tissue lyser (Qiagen, Venlo, Limburg, The Netherlands). Homogenate was centrifuged at 14,000g for 15 minutes at 4°C to obtain the protein supernatant fraction. Protein concentration was determined using an infrared spectrometer (Direct Detect, Millipore). 

Serum Extraction.

Blood was collected by a cardiac puncture and was left to stand at room temperature for 30 to 45 minutes. Following coagulation, blood was centrifuged at 12,000g for 10 minutes at 4°C. Serum was collected from the supernatant, aliquoted, and stored at −80°C until ELISA testing. 

Western Blots.

Retinal protein (40 μg) was separated by electrophoresis on 8% to 20% SDS-PAGE gradient gel (TGX; Bio-Rad, Hercules, California, USA) and transferred (Trans Turbo blot, Bio-Rad) to polyvinylidene fluoride membranes. After blocking in 5% (wt/vol) skim milk for 1 hour, the membranes were incubated with primary antibodies against BDNF (1:200, Santa Cruz Biotech), NGFR (1:200, Biosensis), NTRK2 (1:200, Santa Cruz Biotech), or β-actin (1:10,000: loading control; Santa Cruz Biotech) overnight at 4°C. After three 10-minute washes with Tris-buffer saline with 0.05% Tween-20, the membranes were incubated with horseradish peroxidase–conjugated goat anti-rabbit IgG or rabbit anti-mouse IgG (1:5,000, Millipore) for 1 hour at room temperature. Immunoreactive protein bands were visualized by chemiluminescence reagents (Luminata crescendo, Millipore) and recorded from a transilluminator (ChemiDoc, Bio-Rad). 

Enzyme-Linked Immunosorbent Assay.

A commercial kit was used to assess levels of BDNF in retinal and serum samples (BDNF Rapid, Biosensis). Brain-derived neurotrophic factor standards (100 μL of 4–500 pg/mL), serum samples (100 μL, 1:30 dilution), and individual retina protein extracts (100 μL, standardized to 1 mg/mL protein) were run in duplicate according to the manufacturer's instructions (BDNF Rapid, Biosensis). The colored reaction product was measured at 450 nm using a microplate reader (Clariostar; BMG Labtech, Ortenberg, Germany). Sample concentration was determined from standard concentration and absorbance measurements using four-parameter logistic nonlinear regression on MARS data analysis software (BMG Labtech). Results were expressed as concentration of BDNF (pg/mL for serum and pg/mg for retina). 

Statistical Analysis

Results from treatment groups were compared using a 1- or 2-way ANOVA with Tukey's post hoc analysis. For the data on ONL thickness and Müller cell labeling, inferior retina and superior retina were analyzed separately using a 1-way ANOVA with Tukey's test. All analyses were performed using statistical software (Graph Pad V5.01; La Jolla, CA, USA). 

Results

RIP Mitigated Loss of Retinal Function

The ERG was used to assess retinal function. The negative a-wave is generated by the photoreceptors, while the positive b-wave is considered to be generated by the Müller and bipolar cells of the inner retina.^36,37 The measurement of a- and b-waves is shown by the red arrows in Figure 2A; the left-hand arrow indicates how the amplitude of the a-wave was measured, and the right-hand arrow shows measurement of the amplitude of the b-wave. 

Representative ERG traces are shown in Figure 2B, elicited from animals in the five experimental groups in series 1; the responses were elicited with the same intensity of flash (2.0 log scot cd.s.m⁻²). Trends in amplitudes to this flash intensity are shown quantitatively in Figures 2C through 2E. In Figure 2C, amplitudes of the a- and b-waves are shown for individual animals; means for each group are also indicated. Overall, exposure to damaging light greatly reduced the amplitudes of the a- and b-waves, while RIP with or without anesthesia mitigated this reduction. Figures 2D and 2E show that the mitigation of ERG loss associated with RIP is evident over the intensity range used, for both a- and b-waves. 

RIP Mitigated Photoreceptor Degeneration

Our findings confirm previous reports of thinning of the ONL, caused by light damage,³⁸ including the preferential thinning of the ONL in retina 1 to 2 mm superior to the optic disc. Figures 3A through 3D and Figures 3E through 3H show representative images of superior and inferior light-damaged retina; the thinning of the ONL by LD is apparent when comparing Figure 3A with Figure 3B, and Figure 3E with Figure 3F. When the retina was conditioned by remote ischemia prior to exposure to the bright light, the thinning of the ONL was less, most clearly in inferior retina (Figs. 3F, 3G). The ONL is thick in the control (Figs. 3A, 3E) group, and its cell bodies are neatly stacked. The ONL is reduced in thickness by LD, especially in superior retina (Figs. 3B, 3E); this thinning is mitigated by RIP (Figs. 3C, 3G), although some disorganization of the ONL persists, particularly in superior retina (Fig. 3C). 

Figure 3

 

Remote ischemic preconditioning mitigated loss of photoreceptors caused by LD. (A–D) Sections of superior retina from control, LD, RIP-LD, and RIP groups labeled with bisbenzimide to show nuclei. Outer nuclear layer thickness was recorded as the thickness of the ONL (arrow at left labeled o), normalized to the thickness of the retina measured from the inner to the outer limiting membrane (arrow labeled r). (E–H) Corresponding sections of inferior retina. (A, B, E) The upper arrow indicates the inner limiting membrane, and the lower arrow indicates the outer limiting membrane. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. (I) Outer nuclear layer thickness was measured in 400-μm steps along the retina, from the superior to the inferior edge and normalized (ONL: OLM-ILM) in the retina, n = 5 (RIP and CON), n = 6 (RIP-LD and LD). The graph displays ONL: OLM-ILM means ± SEM for the four groups. Outer nuclear layer thickness differed between groups (2-way ANOVA, P < 0.0001). Intragroup comparisons show that RIP did not affect photoreceptor numbers in the retina not exposed to damaging light (RIP, Tukey's post hoc test, P > 0.05). Remote ischemic preconditioning mitigated the thinning of the ONL caused by light damage in the inferior retina (blue versus red, Tukey's post hoc test, **P < 0.01).

Figure 3I shows these trends quantitatively. Thinning of the ONL was apparent throughout the retina after exposure to damaging light (control versus LD). When the retina was preconditioned by ischemia, thinning was mitigated (RIP-LD versus LD), most clearly in inferior retina. The difference in thickness between the RIP-LD and LD groups in the inferior retina was statistically significant. 

RIP Reduced Müller Cell Stress

Light damage increases the expression in Müller cells of the cytoskeletal protein GFAP, a well-established marker of retinal stress.³⁹ In unstressed (control) retina, GFAP labeling was limited to the inner surface of the retina (i in Figs. 4A, 4E), where it is expressed by astrocytes.⁴⁰ After exposure to damaging light, the radially oriented processes of Müller cells also express GFAP, the labeling extending outward from the inner surface, reaching most of the distance across the retina, from its inner to its outer limiting membranes (Figs. 4B, 4F). This damage-induced expression of GFAP by Müller cell processes was mitigated by RIP, most markedly in inferior retina (Figs. 4C, 4G). The extent of GFAP labeling along Müller cells was quantified and normalized to the thickness of the retina (Fig. 4I); this normalization reduced the effect of obliquity of section on measures of retinal thickness (from the inner limiting membrane [ILM] to the outer limiting membrane [OLM]). The reduction in GFAP expression by RIP (Fig. 4I) was most marked in inferior retina, and reached statistical significance in post hoc comparisons between the LD and RIP-LD groups. That is, RIP appears to mitigate retinal stress caused by LD, although not to levels seen in control animals. 

Figure 4

 

Remote ischemic preconditioning mitigated retinal stress caused by light damage. (A–D, E–H) Sections of superior and inferior retina, respectively, labeled for the stress-inducible protein GFAP. In control retina (A, E) and in the RIP-treated group (D, H), the label is restricted to astrocytes at the inner limiting membrane (i). In LD retina, the labeling extended outward along the radial fibers of Müller cells (B, F) toward the outer limiting membrane (o). Remote ischemic preconditioning reduced the outward spread of labeling (C, G). (A, E) The upper arrow indicates the inner limiting membrane, and the lower arrow indicates the outer limiting membrane. (I) The length of GFAP labeling along Müller cells, normalized to the thickness of the retina (GFAP: OLM-ILM), was measured in 400-μm steps from the superior to the inferior edge of the retina, n = 5 (Con and RIP), n = 6 (RIP-LD and LD). The graph displays GFAP: OLM-ILM means ± SEM for four experimental groups. GFAP expression differed between groups (P < 0.0001, 2-way ANOVA). Remote ischemic preconditioning in the absence of photoreceptor damage did not affect GFAP expression of Müller cells (black versus gray); but RIP did significantly mitigate GFAP expression in Müller cells after light damage (blue versus red) in inferior retina (P < 0.05 [*] on a 2-way ANOVA test, using Tukey's post hoc test).

RIP Mitigated Photoreceptor Apoptosis

Light damage induces the apoptotic death of photoreceptors,⁴ as demonstrated with the TUNEL technique. Figures 5A and 5B show that TUNEL+ cells (red) are found predominantly in and just external to the ONL, as noted previously. Figures 5C and 5D show that TUNEL+ cells were rare in inferior retina (Figs. 5E, 5F) and in RIP and control groups. When the number of TUNEL+ cells in the ONL was counted for full retinal sections (scanned from the upper to the lower margin of the retina), the control and RIP groups showed very low numbers of TUNEL+ cells, as expected (Fig. 5G). The number of TUNEL+ cells was significantly higher in LD animals (Fig. 5G) than in the control retina. The number of TUNEL+ cells was lower in RIP-LD animals than in LD animals. 

Figure 5

 

Remote ischemic preconditioning reduced photoreceptor apoptosis caused by light damage. (A–F) Sections of retina from superior and inferior retina of the LD, RIP-LD, RIP, and control groups. The bright light caused apoptotic death of photoreceptor cells in the outer nuclear layer (ONL). Nuclei are labeled blue (bisbenzimide) and apoptotic cells are labeled red, using the TUNEL technique. TUNEL+ cells were evident in the LD and RIP-LID groups, predominantly in superior retina, where they were located in or subjacent to the ONL. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. In each part of the figure, the upper arrow indicates the inner limiting membrane, and the lower arrow indicates the outer limiting membrane. (G) Apoptotic cells in the ONL were counted in sections along the full length of the retina from the superior to the inferior edge (n = 5 all groups). The dot plot shows individual counts, means ± SEM, for RIP-LD (blue), LD (red), RIP (gray), and control (black) groups. LD caused an increase in TUNEL+ cell numbers above control levels (P < 0.0001, 1-way ANOVA). Remote ischemic preconditioning significantly mitigated this increase (blue versus red, *P < 0.05, Tukey's post hoc comparison).

RIP-Induced Retinal Protection Is Associated With Increased Retinal BDNF Expression

Exogenous application of growth factors, including BDNF, ciliary neurotrophic factor (CNTF), and basic fibroblast growth factor (bFGF), promotes the survival of photoreceptors following light damage.⁴¹ To test whether growth factors are involved in RIP-induced protection of photoreceptors, we assessed retinal transcript levels for these factors. 

Retinal levels of Bdnf mRNA transcripts were significantly and selectively raised in RIP-LD animals (Fig. 6A). Levels of transcripts for Cntf and Fgf2 (encoding bFGF) were raised in both LD and RIP-LD groups, suggesting that these genes are regulated by damage rather than by RIP (Figs. 6B, 6C). 

Figure 6

 

Remote ischemic preconditioning increases BDNF expression in light-damaged retina. Retinas were isolated 7 days following light damage and mRNA was quantified relative to a housekeeping gene (Rplp0). LD (red, n = 7), RIP-LD (blue, n = 8), and controls (black, n = 5). Using ELISA, protein expression for BDNF was determined from isolated retinas (n = 4 all groups). Graphs display means (bars) and SEM (error bars). (A) mRNA expression of Bdnf was higher in retinas conditioned by RIP and exposed to damaging light than in light-damaged or undamaged (CON) retinas (*P < 0.05, **P < 0.01, Tukey's post hoc comparison). (B) mRNA expression of Cntf was significantly (*P < 0.05, Tukey's post hoc test) raised above control levels in both LD and RIP-LD groups. Remote ischemic preconditioning of light-damaged retinas did not appear to alter CNTF expression. (C) mRNA expression of Fgf2 was significantly (*P < 0.05, Tukey's post hoc test) raised above control levels in both LD and RIP-LD groups. Remote ischemic preconditioning of light-damaged retinas did not appear to alter bFGF expression. (D) Retinal protein concentrations of BDNF in LD, RIP-LD, and control groups. Remote ischemic preconditioning was associated with a small but statistically significant (*P < 0.05, Tukey's post hoc test) increase of BDNF in light-damaged retina. (E) Remote ischemic preconditioning did not appear to alter BDNF protein levels in undamaged retina.

We also measured levels of the BDNF protein in retina, in the same experimental groups, using ELISA. A small but significant increase in retinal BDNF levels was noted in the RIP-LD group (Fig. 6D). When we tested (in a separate group of animals) whether RIP induces a change in retinal levels of BDNF protein in the absence of light damage (Fig. 6E), no difference was apparent. Hence, the rise in BDNF levels induced by RIP appears to be linked to the impact of light on the photoreceptors. 

BDNF Receptors: RIP-Induced Protection Is Associated With Increased p75NTR

Brain-derived neurotrophic factor can act through two receptors, the high-affinity receptor NTRK2 (also known as TrkB) and the low-affinity receptor NGFR (also known as p75NTR), which can form a complex with SORL1 (also known as sortilin). To explore whether expression of these receptors correlated with RIP-induced changes in BDNF levels, we measured the retinal expression of the genes Ntrk2, Ngfr and Sorl1, 7 days after light damage, normalizing expression to the housekeeping gene Rplp0. 

Expression of Ntrk2 mRNA was reduced in light-damaged animals, and this reduction was not mitigated by RIP (Fig. 7A). In contrast, mRNA transcripts for Ngfr were significantly increased in RIP-LD retinas relative to control and LD retinas (Fig. 7B). The expression of Sorl1 mRNA was not altered by RIP (Fig. 7C). 

Figure 7

 

Remote ischemic preconditioning increases expression transcripts for Ngfr, but not Ntrk2 or Ngf. Retinas were isolated 7 days following light damage and mRNA was quantified relative to a housekeeping gene (Rplp0). LD (red, n = 6–8), RIP-LD (blue, n = 8), and controls (black, n = 5). Protein was quantified relative to β-actin (n = 3 CON and LD, n = 4 RIP-LD, n = 5 RIP). Graphs display means (bars) and SEM (error bars). (A) mRNA expression Ntrk2, encoding the TrkB receptor, was not increased by RIP of the light-damaged retina. Unexpectedly, expression was lower in the RIP-LD group than in either LD or control groups; the difference was significant (*P < 0.05, Tukey's post hoc comparison) between RIP-LD and control groups. (B) Remote ischemic preconditioning in the light-damaged retina was associated with a statistically significant (*P < 0.05, Tukey's post hoc comparison) increase in transcripts for Ngfr, encoding p75NTR, compared to both LD and control groups. (C) mRNA expression of Sorl1, encoding sortilin, did not appear associated with RIP of the light-damaged retina. (D) mRNA expression of Ngf was suppressed in LD and RIP-LD groups compared to controls (***P < 0.001, Tukey's post hoc comparison).

When retinal protein levels of the two receptors were measured, normalized to β-actin expression, using Western blots (Fig. 8A), the values shown in Figures 8B and 8C were obtained. The protein levels of NTRK2 observed in LD, RIP-LD, control, and RIP groups were not significantly different (Fig. 8B). The protein levels of NGFR were higher compared to the control group. The RIP-LD group was significantly higher than the control group (Fig. 8C). Together, these findings suggest that the expression of NGFR correlates with RIP-induced retinal BDNF expression and with RIP-induced protection of photoreceptors. 

Figure 8

 

Remote ischemic preconditioning upregulates expression of the BDNF low-affinity receptor (NGFR) in the light-damaged retina. (A) Western blots for NGFR, NTRK2, and β-actin for retinas from control, LD, RIP-LD, and RIP groups. Upregulation is evident for the RIP-LD group. (B, C) The Western blots were assessed by densitometry, normalized to level of β-actin in each sample. Individual measurements are shown for each group; the error bars show mean relative density ratios ± SEM. Only the RIP-LD measures for NGFR were significantly different (**P < 0.01 Tukey's post hoc analysis, control versus RIP-LD).

In addition to being a low-affinity receptor for BDNF, NGFR is also the high-affinity receptor for nerve growth factor (NGF). To test whether NGF may have a role in RIP-induced neuroprotection, we measured Ngf mRNA transcripts in the retina. We observed a significant reduction in Ngf transcripts in both LD and RIP-LD retina relative to control retina (Fig. 7 D), suggesting that NGF is unlikely to mediate the protective effects of RIP in light-damaged rats. 

Increased BDNF Protein in the Retina Is Not Associated With Increased Serum BDNF

There are multiple locations in the body capable of synthesizing and releasing BDNF, including the brain and satellite muscle cells.^42,43 To understand whether the effect of remote ischemia could be mediated by an RIP-induced release of BDNF into the bloodstream, BDNF levels in the serum were measured by ELISA. At the 7-day time point used for the measures of photoreceptor death and retinal stress presented above, we did not detect a rise in serum levels of BDNF related to RIP, either in animals subjected to light damage (Fig. 9A) or in animals not exposed to damaging light (Fig. 9B). 

Figure 9

 

Serum BDNF levels do not vary with retinal levels. Serum was isolated at 7 days and concentrations of BDNF were detected using a sandwich ELISA. Graphs display means (bars) ± SEM (error bars). (A) Serum concentration of BDNF protein at 7 days did not vary significantly between LD (n = 4), RIP-LD (n = 3), and control groups (n = 3). (B) Serum concentration of BDNF protein at 7 days did not differ significantly between RIP (n = 10) and sham RIP groups (n = 9).

Discussion

Summary

This study provides the first demonstration that RIP protects photoreceptors against degeneration, preserving the amplitudes of the a- and b-waves of the ERG and the thickness of the ONL, reducing apoptosis in photoreceptors and retinal stress. Remote ischemic preconditioning also increased the mRNA and protein expression of BDNF and its low-affinity receptor NGFR in the damaged retina, suggesting a role for BDNF and this receptor in mediating RIP-induced protection. 

Protecting Photoreceptors by Stress Conditioning

Evidence has been gathered, at least since the 1980s, that the resistance of photoreceptors to damage, by bright light or by genetic mutations, can be regulated by conditioning the photoreceptors with a range of interventions, many of which are inherently stressful. Light-naïve photoreceptors in the retinas of dark-reared rats, for example, appear undamaged at the ultrastructural level but are destroyed by their first experience of daylight. Photoreceptors conditioned to daylight in a diurnal rhythm during rearing show shortened outer segments with damaged membranes, but are resilient when exposed to bright light.⁶ Physiological levels of light during rearing thus appear to be stressful, causing membrane damage to outer segments, and also to upregulate protective mechanisms in the photoreceptors or nearby cells. Correspondingly, conditioning the retina with potentially damaging interventions such heat,⁹ gamma rays,¹³ ischemia,⁴⁴ and nearby mechanical damage^45,46 increases the resistance of photoreceptors to damage by subsequent stress. The body-wide stress of exercise also appears in mice to stabilize photoreceptors, increasing their resilience to the stress of bright light,⁴⁷ and, in humans, to be associated with lower incidence of age-related macular degeneration.^48–50 Present results suggest that remote ischemia can be added to this list of stressful conditioning interventions that are capable of stabilizing photoreceptors. 

Our evidence that RIP can induce the protection of photoreceptors complements recent reports of RIP-induced protection of the inner layers of retina from ischemia–reperfusion injury and optic nerve transection.^21,25 It also complements our previous report that remote ischemia enhances the response of the retina to light, producing a supernormal ERG.²⁹ Remote ischemic preconditioning–induced enhancement of the function of an undamaged tissue had not been described previously; it was possible to detect in the retina because the ERG is a sensitive measure of retinal function, and remote ischemia can be delivered in a temporally precise way. The interaction between peripheral tissues such as muscle and skin, with a specialized sensory tissue such as the retina, may be multifaceted. It presumably involves the release of chemokines and trophic factors from ischemic tissue, as reviewed by Brooks and Andrews.¹⁵ 

Involvement of BDNF in RIP and Stress Conditioning

The remarkable range of protection-inducing interventions raises the question of whether each acts by a distinctive mechanism or whether all or some activate the same mechanism. Evidence on this point remains unsettled. On the one hand, studies of remote and direct ischemia-induced protection have reported the regulation of multiple pathways, including the suppression of proinflammatory genes, and activation of antioxidant and cell survival–promoting genes.^51–53 On the other hand, a recent study that tested the combination of two interventions (dietary saffron and photobiomodulation) reported no added benefit, suggesting that these two interventions, at least, activate common pathways.³³ Natoli et al.⁵⁴ examined the gene regulation induced by these two interventions, and identified several common pathways involving upregulation of antioxidant genes and suppression of inflammatory mediators. The idea that the several plant toxins known to induce neuroprotection at low doses might activate common pathways has already been suggested by Mattson and colleagues.^55,56 

This study reports evidence of a pathway involving the neurotrophic factor BDNF, a pathway not identified in studies of the retinal protection induced by photobiomodulation or dietary saffron. The upregulation of BDNF has been reported in prior studies of the effect of direct (nonremote) ischemia on the brain.^57,58 Here we extend this finding to the effect of remote ischemia on another part of the central nervous system, the retina. 

Exercise conditioning, which like RIP involves muscle ischemia, has been associated with BDNF-mediated photoreceptor protection in mice.³⁰ The increases in retinal BDNF reported in that study are similar to the increases reported here for the rat retina.³⁰ The present observation that photoreceptor resilience correlates with elevations in retinal BDNF complements previous evidence that BDNF can make photoreceptors resistant to light-induced damage or genetic mutation, whether through direct injection into the eye, or by transgene-induced upregulation or preimplantation of BDNF-producing cells.^{31,41,59–61} 

BDNF Signaling in the Retina

Brain-derived neurotrophic factor acts via the NTRK2 receptor to cause the simultaneous release of pro survival transcription factors and suppression of proapoptotic signals.⁶² Cone photoreceptors but not rod photoreceptors express TrkB, and it was a puzzling finding that elevations of retinal BDNF are able to protect both rod and cone photoreceptors.⁶³ One possible explanation was that BDNF acts via NTRK2 receptors on Müller cells to cause the release of FGF2, which has been shown to act directly on both photoreceptors.^64–67 In our study, however, we did not observe increases in retinal expression of the genes encoding TrkB or bFGF. 

An alternative BDNF signaling pathway involves BDNF acting via NGFR, a receptor that can stimulate either pro- or antiapoptotic signaling and is expressed by Müller cells, astrocytes, microglia, and endothelial cells.⁶⁸ Some studies report that elevation of NGFR expression in Müller cells contributes to photoreceptor degeneration.^64,69 However, knockdown of NGFR only partially reduces photoreceptor degeneration caused by damaging light and does not prevent photoreceptor degeneration from occurring in the rd mouse, suggesting that NGFR is not the sole mediator of photoreceptor degeneration.^70,71 Other studies have suggested that NGFR and SORL1 combine to form a receptor that mediates photoreceptor apoptosis.^72,73 In our study, however, there was no evidence of NGFR forming a coreceptor with SORL1, as Sorl1 mRNA did not change between treatment groups. 

NGFR comprises several extracellular domains and multiple intracellular “death domains,” and its activation can become pro survival when its conformation is altered, for example, in the presence of mature NGF.⁷⁴ Ischemic preconditioning in the brain has also been shown to increase NGFR expression, and this upregulation has been reported to be essential for neuroprotection.⁷⁵ In the retina there is also evidence that BDNF prolongs the survival of isolated bipolar cells via NGFR.⁷⁶ In our study, NGFR expression increased with light damage and increased further with RIP; its role in RIP-induced neuroprotection warrants further study. 

Clinical Implications and Future Directions

Remote ischemia has proved effective in increasing tissue resilience in periods of acute stress, for example, improving the outcome of pediatric surgery of the heart,⁷⁷ and in recovery from stroke.^20,78 The utility of remote ischemia for the treatment of long-term retinal degenerations, such as age-related macular degeneration or retinitis pigmentosa, will require exploration; less invasive interventions, like dietary saffron or photobiomodulation, seem more appropriate.^79,80 Further, it will be important to explore the variation of the protective effect with time after ischemia; the present observations were mostly made at 7 days after the exposure to damaging light. 

Understanding of the diversity of interventions that stabilize photoreceptors may open a range of therapeutic options appropriate for acute or chronic degenerations. For acute stress, however, remote ischemia may be particularly useful, because it can be generated quickly and precisely by application of a tourniquet to one limb. 

Acknowledgments

Supported by the Australian Research Council Centre of Excellence in Vision Science, and by the Sir Zelman Cowen Universities Fund. DMJ is supported by an Early Career Fellowship from the National Health and Medical Research Council (NHMRC) of Australia. 

Disclosure: A. Brandli, None; D.M. Johnstone, None; J. Stone, CSCM Pty Ltd (I S)