Mouse Models of Retinal Ischemic Tolerance

Abstract

purpose. A brief period of noninjurious retinal ischemia, termed preconditioning, has been documented in rats to afford transient protection from retinal ischemic injury, a phenomenon known as ischemic tolerance. The present study was undertaken to develop and systematically characterize mouse models of ischemic tolerance.

methods. Retinal ischemic injury was caused by elevating intraocular pressure for 30, 45, or 60 minutes in chloral hydrate–anesthetized ND4 Swiss-Webster mice. Random animals were preconditioned 24 hours earlier with either 5 minutes of retinal ischemia or by exposing conscious animals to hypoxia (11% oxygen) for 2 hours. Flash electroretinograms were recorded 1 day and 1 week after ischemia. At 1 or 4 weeks after ischemia, eyes were perfusion fixed for microscopic examination and quantification of layer thickness and cell counts.

results. Retinal ischemia resulted in significant, duration-dependent reductions in inner retinal layer thickness and cell loss in the inner nuclear and ganglion cell layers. A duration-dependent attenuation in a- and b-wave amplitudes was concomitantly noted. The ischemic and hypoxic preconditioning treatments significantly attenuated the ischemia-induced changes in retinal morphology and function, even after 4 weeks of recovery. Tolerance was observed at 24 hours after ischemic preconditioning, but not at 72 hours.

conclusions. Two models of retinal ischemic tolerance are presented wherein ischemic or hypoxic preconditioning afforded morphologic and functional evidence of protection from retinal ischemic injury in mice. These two murine models should be useful for studies in mutant mice to elucidate endogenous genetic and molecular mechanisms of retinal protection that may then be used to design treatments for ischemic retinopathies.

Pharmacologic therapies for protecting the retina from ischemic injury remain elusive and less efficacious to date than expected, despite the recent growth in preclinical studies elucidating the underlying biochemical basis of cellular injury. A better understanding of endogenous neuroprotective and survival mechanisms may provide a novel therapeutic basis for treatment. Considerable evidence in brain ¹ has shown the existence of innate protective mechanisms that can be stimulated before ischemia and then afford a significant degree of protection against ischemic injury. Sublethal hypoxia or ischemia are two such preconditioning stimuli that can subsequently induce a state of ischemic tolerance. 

Recently, Roth et al. ² developed a rat model of retinal ischemic preconditioning wherein a 5-minute period of sublethal ischemia conferred nearly complete protection against a lethal ischemic insult administered 24 to 72 hours later. Similar in vitro models of retinal ganglion cell protection by preconditioning hypoxia-ischemia have been reported. ³ Both models offer advantages for identifying the mechanistic components underlying this protective response. 

Although studies of retinal ischemia in the mouse are scant, ^{4 5} we undertook the present study to develop versatile and reproducible mouse models of retinal ischemic tolerance. Our goal was to provide models suitable for using genetically engineered mice to help reveal in a nonpharmacologic manner the endogenous mechanistic basis that accounts for the typically robust protective effects of preconditioning. This report details morphologic and functional evidence of ischemic tolerance in the mouse retina, by using as the preconditioning stimuli either short-duration, sublethal retinal ischemia or a longer period of sublethal systemic hypoxia. 

Materials and Methods

All experiments were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Induction of Retinal Ischemia

All studies were conducted in adult (10–12 weeks of age) male Swiss-Webster ND4 mice (25–34 g; Harlan Sprague-Dawley, Indianapolis, IN). Animals were anesthetized intraperitoneally with chloral hydrate (350 mg/kg; Sigma Chemical Co., St. Louis, MO) and xylazine (4 mg/kg; Burns Veterinary Supply, Inc., Rockville Center, NY), which results in mean arterial blood pressure of 65 ± 5 mm Hg (n = 8) in the Swiss-Webster strain. Pupils were dilated with 1% tropicamide-2.5% phenylephrine hydrochloride (NutraMax Products, Inc., Gloucester, MA), corneal analgesia was achieved with 1 drop of 0.5% proparacaine hydrochloride (Bausch & Lomb Pharmaceuticals, Inc., Tampa, FL), and a contact lens was applied with saline. Retinal ischemia was induced for 30, 45, or 60 minutes by introducing into the anterior chamber (using a micromanipulator) a 32-gauge needle attached to a saline-filled reservoir (0.9% sodium chloride; Baxter, Deerfield, IL) raised above the animal to increase intraocular pressure (IOP) above systolic blood pressure (IOP increased to 90 mm Hg). Complete retinal ischemia was evident and verified in every animal under surgical microscopic examination by a whitening of the anterior segment of the eye and blanching of the retinal arteries. Core body temperature was maintained throughout the procedure at 37°C with a thermoregulated infrared heating lamp, while the eye was protected from light. At the end of the ischemic period, the needle was removed from the anterior chamber, and 1% atropine and 1% vetropolycin with hydrocortisone ointment (Fougera & Atlanta, Inc., Melville, NY) was applied to the conjunctival sac. The animals were allowed to recover in their home cages for 7 or 28 days. The fellow eye of each animal served as a nonischemic control, and results from these controls were pooled with additional data from eyes obtained from anesthetized but untreated nonischemic mice. Animals were randomized into nonischemic and ischemic treatment groups, and nonpreconditioned and preconditioned treatment groups. 

Ischemic Preconditioning Models

Two preconditioning procedures, temporary retinal ischemia and moderate systemic hypoxia, were tested for their ability to protect against retinal ischemia induced by elevated IOP. All animals were randomized for these preconditioning treatments as well, with experiments and histopathologic processing on preconditioned and nonpreconditioned animals run simultaneously at all opportunities. 

In the temporary retinal ischemia model of preconditioning, animals were anesthetized with chloral hydrate and xylazine and 2.5, 5, or 10 minutes of temporary retinal vessel occlusion was induced by occluding the retinal vascular trunk with a 5-mm straight aneurysm clip (mini-clip, model FE691K; Aesculap AG & CO. KG, Tuttlingen, Germany) that exhibited a 45- to 55-g closing pressure. No surgical preparation was necessary to apply the clip; the eye was partially proptosed for clip placement, and the clip was applied from the nasal side of the eye to clamp the entire optic nerve and vessel bundle. Blanching of both vessels and tissue was used to verify interruption of blood flow to the retina, and reperfusion of the retinal vasculature within 1 minute after release of the aneurysm clip, as reflected by filling of retinal vessels with blood and the overall blush of the retinal tissue, was confirmed through the surgical microscope, as well. A 30-, 45-, or 60-minute period of ischemia was induced 24, 72, or 168 hours later. Separate animals received the clip preconditioning without the subsequent period of ischemia (to serve as an ischemic preconditioning with sham-ischemia group) and were allowed to recover for periods equivalent to the ischemic group. 

In the moderate systemic hypoxia model of preconditioning, mice were placed into a closed 1.7-L plastic chamber through which air containing 8% or 11% oxygen (balance nitrogen) was continuously flushed at a rate of 1.5 L/min (quantified by an in-line flow-meter) for 2 hours. No more than three animals were placed in the chamber at any time. A 45- or 60-minute period of ischemia was induced 24 hours later in these hypoxia-preconditioned mice. Separate animals were exposed to systemic hypoxia, but their retinas were not rendered ischemic thereafter, to serve as a hypoxic preconditioning plus sham-ischemia group. Again, recovery periods for these animals matched those of the ischemic animals. 

Flash Electroretinography

Three scotopic flash electroretinograms (ERGs) were obtained in the experimental eye of each animal: a baseline ERG recorded 1 day before preconditioning or ischemia and two recovery ERGs recorded 1 and 7 days after preconditioning or ischemia. The recording procedure has been described in detail in an earlier publication. ⁶ In brief, animals were dark adapted overnight with free access to food and water. They were anesthetized as described earlier with chloral hydrate and xylazine and positioned dorsally in a customized headholder. Core temperature was maintained throughout the measurement period at 37.5 ± 1.0°C with a thermostatically controlled heating pad operated in conjunction with a rectal probe (model 73A; YSI, Yellow Springs, OH). Pupillary dilation (1.5–2.0 mm, measured at the end of each experiment) was maintained by superfusing the cornea with a pH-adjusted solution of 2% xylazine, ⁷ delivered to the eye through polyethylene (PE10) tubing terminating directly above the recording loop. 

A 2-mm-diameter stainless steel loop, which served as the recording electrode, was positioned gently on the corneal surface by a micromanipulator. The reference electrode (inactive) was a stainless-steel needle inserted under the skin behind the right ear, and a third electrode was placed under the skin on the animal’s back to serve as a ground. Recording leads were fed into a battery-powered differential amplifier (0.3–300 Hz, 1000×; model P55; Grass Instruments, Quincy, MA). Responses were digitized at 3 kHz (Micro 1401 digitizer; Cambridge Electronics, Somerville, MA) under the control of accompanying software (Signal; Cambridge Electronics) and custom averaging routines performed by computer (120-MHz Pentium; Intel, Mountain View, CA). Computer-triggered flashes were generated by a photostimulator set at I-16 (model PS 33-Plus; Grass Instruments). The 13-cm diameter xenon flash source was positioned in front of a white background 25 cm from the eye, and 30° anterior to the medial–lateral axis of the eyes. The 10-μs white light flash had an unattenuated intensity of 2.3 cd/sec · m², measured at the eye. Responses to unattenuated flashes were averaged over 5 to 10 presentations at a presentation rate of 0.1 Hz. At the end of the recording period, the conjunctiva was treated with antibiotic ophthalmic ointment. 

The amplitude of the a-wave was measured from the baseline to the trough of the a-wave, and b-wave measurements were referenced from the tip of the a-wave to the maximum positive value of the b-wave–oscillatory potential complex. ⁸ Measurements were performed by an observer blinded to the experimental condition of the animal. 

Histopathology and Injury Quantification

Anesthetized animals were transcardiac perfused (5 mL/min) with 50 mL 10-mM phosphate-buffered saline (Sigma ImmunoChemicals, St. Louis, MO), followed by 100 mL of 4% paraformaldehyde (Electron Microscopy Sciences, Fort Washington, PA). Eyes were enucleated and immersion fixed for 1 hour in 4% paraformaldehyde, transferred to 10% neutral-buffered formalin (Fisher Scientific, Fairlawn, NJ) overnight, and processed for routine paraffin-embedded sections on an automatic tissue processor (Histomatic 166A; Fisher Scientific). Eyes were embedded sagittally and 5-μm thick serial sections including the optic nerve were obtained by rotary microtome (model 2030; Reichert-Jung, Vienna, Austria). The slides were dried overnight at 37°C and stained with hematoxylin (Hematoxylin 7211; Richard-Allan Scientific, Kalamazoo, MI) and eosin (Eosin Y; Richard-Allan Scientific). 

Using image analysis software (Image-pro Plus; Media Cybernetics, Silver Spring, MD), the thickness of the different retinal layers were measured in triplicate in both superior and inferior quadrants at a fixed distance (750–1000 μm) from the edge of the optic disc. A minimum of three sections were used for each eye. The morphometric measurements we quantified included the distance from the outer limiting membrane to the inner limiting membrane (OLM-ILM), and the thickness of the inner nuclear layer (INL) and the inner plexiform layer (IPL). We also counted the total number of morphologically identified living cells in the INL and ganglion cell layer (GCL) in the same topographic region of the retina, to avoid the possibility of regional anatomic variation in the results. ⁹ These manual cell counts in the INL and GCL were performed over a length of 75 and 250 μm, respectively. All morphometric measures were performed in a masked fashion with regard to prior sham or experimental procedures. 

Statistical Analysis

Experimental differences in thickness and cell counts in the individual retinal layers and in the ERG a- and b-wave amplitudes were compared by the nonparametric ANOVA on ranks. The Kruskal-Wallis ANOVA on ranks method with the Dunn multiple range test was used for unpaired data comparisons made among the histopathologic data. The Friedman repeated-measures ANOVA on ranks method with the Dunnett multiple range test was used for the paired data comparisons made among the ERG data. For the graphic presentations in the figures, the data were normalized to the respective values measured in nonischemic fellow control eyes (histopathology) or during preischemic baseline conditions (ERGs). P < 0.05 was accepted as significant. 

Results

Morphometric analyses of fellow, nonischemic eyes from the different experimental groups revealed no significant differences among groups in layer thicknesses and cell counts, and the data from these eyes (n = 33) were therefore pooled to serve as a control group. These respective values were not different from those obtained in another groups of eyes that were punctured by the needle but not rendered ischemic. Based on the morphologic criteria we used for quantifying retinal injury, a graded, reproducible injury to the retina of Swiss-Webster ND4 mice resulted from 30, 45, and 60 minutes of retinal ischemia induced by elevated IOP. Similarly graded and parallel degrees of functional deficits were documented for 30- and 45-minute durations of ischemia by electroretinography. We used these end points to examine the endogenous ability of the retina to adapt to ischemic injury by preconditioning the retina a day before ischemia, either with brief, complete retinal ischemia lasting 5 minutes (ischemic preconditioning), or with systemic hypoxia (11% oxygen) for 2 hours (hypoxic preconditioning). Results in Figures 1 and 2A , and in Tables 1 and 2 , show that neither of these two preconditioning regimens was deleterious to the retina in and of themselves (when not followed by 30, 45, or 60 minutes of complete retinal ischemia by elevated IOP). Specifically, no significant changes in histopathologic morphometric measures or cell counts and no changes in ERG wave amplitudes were evidenced in animals subjected to the 5-minute ischemic preconditioning stimulus or the 11% hypoxia stimulus, except for a slight increase in a-wave amplitude after the ischemic preconditioning stimulus. Conversely, data in Table 1 show that morphologic signs of injury were evidenced when more severe stimuli were used (e.g., 10 minutes of ischemia or 2 hours of systemic hypoxia with 8% oxygen). In both cases, thinning was evidenced across the entire retina and in individual layers. With the 8% hypoxic stimulus, hyperplasia was noted in the INL and GCL. In contrast, we found that retinal ischemia for 2.5 minutes was insufficient to promote significant tolerance against a 45-minute period of retinal ischemia (n = 5; data not shown). Thus, as detailed in the next paragraph, the 5-minute ischemia and the 2-hour period of 11% oxygen hypoxia were adopted as two distinct noninjurious preconditioning stimuli that ultimately induced states of ischemic tolerance in the mouse retina. 

We used 30- and 45-minute elevations in IOP as moderately injurious ischemic insults in most of the experiments. Regarding functional impairment (Figs. 2A 2C) , 45 minutes of ischemia in these mice attenuated the a- and b-wave amplitudes severely. The a-wave amplitude was reduced by 61% and 60% 1 day and 1 week after ischemia, respectively, and the b-wave amplitude was attenuated by 76% and 79% at these same time points. The extent of a- and b-wave attenuation measured 1 day and 7 days after 30 minutes of ischemia (Fig. 2B) was less than that measured at corresponding times after 45 minutes of ischemia (Fig. 2C) , but not significantly so. For both the 30- and 45-minute ischemic insults, the ischemia-induced extent of functional impairment at 1 day and 1 week after ischemia was similar, with neither a worsening nor improvement in functional status over time. 

Histopathologically, we observed significant, duration-dependent reductions in retinal layer thickness and cell counts throughout the tissue at 1 week after ischemia (Fig. 3) . Specifically, in response to 45 minutes of ischemia, a 29%, 35%, and 44% reduction in the thickness of the OLM-ILM, the INL, and the IPL, respectively, was measured 1 week later. The number of cells in the INL was reduced 26%, and the number of cells in the GCL was reduced 53%. When the duration of the ischemic insult was reduced to 30 minutes, layer thinning and cell loss were correspondingly reduced. Specifically, we found an 18%, 16%, and 25% reduction in the thickness of the OLM-ILM, INL, and IPL, respectively; cell loss in the INL and GCL was 30% and 29%, respectively. 

The ability of the ischemic preconditioning regimen to afford electrophysiologic evidence of protection against retinal ischemia is summarized in Figure 2 . The extent of functional recovery induced by preconditioning varied inversely with the duration of the ischemic insult, so that recovery of ERG wave amplitudes was much greater in preconditioned eyes subjected to 30 minutes of retinal ischemia (Fig. 2B) relative to 45-minutes (Fig. 2C) . In the latter instance, ischemic preconditioning had no significant effect on a-wave amplitudes, but resulted in significant increases in b-wave amplitudes at both 1 day (14% recovery) and 7 days (14% recovery) after ischemia (Fig. 2C) . In animals subjected to 30 minutes of ischemia (Fig. 2B) , the a-wave recovered 102% 1 day later, and 93% 1 week later, to amplitudes that were not significantly different from baseline. The b-wave also recovered significantly, improving 59% and 46% at 1 day and 1 week after the insult, respectively. 

The ability of the ischemic and hypoxic preconditioning regimens to afford histopathologic evidence of protection against retinal ischemia is summarized in Figures 3 and 4 . When the retinas were preconditioned the day before the 30- or 45-minute ischemic insult with 5 minutes of ischemia, significant recovery in both layer thicknesses and cell counts were noted when measured at 1 week after ischemia. The extent of recovery was greater in eyes subjected to 30 minutes of ischemia (increases in layer thicknesses of 83% to 119% and increases in cell counts of 52% to 83%) relative to that measured in eyes subjected to 45 minutes (increases in layer thicknesses of 79%–89% and increases in cell counts of 35%–73%). 

To determine whether preconditioning by brief ischemia could afford significant protection against even more severe ischemic insults, we subjected mice to a 60-minute period of retinal ischemia. Histopathologic analyses of retinas obtained 1 week after ischemia revealed even greater injury, reflected in layer thinning and cell counts (Fig. 3) . Relative to the 45-minute period of ischemia, the thickness of the OLM-ILM was reduced an additional 15%. Additional 10% and 24% reductions in the thickness of the INL and IPL, respectively, were measured. In cell counts, further losses of 14% and 22% of the total cells in the INL and GCL, respectively, were evidenced. Relative to the nonischemic control, layer thicknesses decreased 43% to 68%, and cell counts were reduced 40% to 74% after ischemia of 60 minutes’ duration. Despite the greater severity of this injury, ischemic preconditioning the day before provided significant and robust protection, with significant improvements in the thickness of the OLM-ILM (77%) and IPL (53%), as well as an increase in cell viability in the GCL (49%). The preconditioning-induced improvement noted in INL thickness (76%) and number of viable INL cells (68%) did not reach statistical significance. 

When the animals were preconditioned with 2 hours of hypoxia (11% oxygen) the day before ischemia of 45 minutes, histopathologic evidence of protection was also observed at 1 week after ischemia (Fig. 4B) , but the protection was qualitatively distinct from, and not as robust as, that promoted when brief ischemia was used as the preconditioning stimulus (Fig. 4A) . We found increases of 45%, 60%, and 23% in OLM-ILM, INL, and IPL layer thickness, respectively, after hypoxic preconditioning, and increases in INL and GCL cell counts of 26% and 44%; however, these changes did not reach statistical significance when measured at this time (discussed later). No additional protective effect of hypoxic preconditioning was realized when the animals were preconditioned with two successive hypoxia challenges (11% oxygen) before a 45-minute ischemic insult; specifically, 3 days before the lethal ischemic insult, the mice were exposed for 2 hours to 11% oxygen, followed by exposure to 4 hours of 11% oxygen the day before lethal ischemia (n = 7; data not shown). 

To determine whether preconditioning induces long-lasting protection or simply delays the injury process, we subjected mice to 45 minutes of retinal ischemia and observed them for 4 weeks of recovery instead of 1 week. Both ischemic and hypoxic preconditioning regimens were studied in this experimental series. Our histopathologic analyses revealed that, in normal nonpreconditioned mice subjected to retinal ischemia, the thickness of the different retinal layers continued to decrease significantly over the additional 3 weeks after ischemia (Fig. 4) , with OLM-ILM thickness reduced an additional 21% and the thickness of the INL and IPL reduced an additional 21% and 30%. However, no additional decrease in cell numbers in the INL and GCL occurred over this 3-week period. In terms of preconditioning efficacy, animals preconditioned with systemic hypoxia (Fig. 4B) showed more robust long-term protection than those preconditioned with brief ischemia when analyzed 4 weeks after the lethal ischemic insult (Fig. 4A) . Specifically, the thickness of the OLM-ILM, INL, and IPL was increased significantly, by 40%, 38%, and 42%, respectively. Significant (89% and 50%) increases in INL and GCL cell counts were evidenced in these retinas, respectively. In the animals preconditioned with 5 minutes of ischemia, significant protection was still evidenced 4 weeks after the insult (Fig. 4A) , although the extent of protection was not as robust as that seen after 1 week and was not statistically significant across all end points. In particular, the thickness of the OLM-ILM and IPL both increased significantly (by 20%), and the INL cell number was significantly increased (by 46%). The trends for increases in INL thickness (14%) and GCL cell number (18%) were not statistically significant. 

Finally, to determine the time course over which the ischemia-tolerant state is sustained after preconditioning, animals were preconditioned with 5 minutes of ischemia and subjected to 45 minutes of retinal ischemia, either 3 or 7 days after preconditioning instead of 1 day later, as in all the foregoing studies. As shown in Figure 5 , no histopathologic protection was observed when the time window between this preconditioning stimulus and the lethal ischemia was extended to 3 or 7 days. 

Discussion

In the present study, we developed and characterized two mouse models of retinal ischemic tolerance. We used electroretinographic and histopathologic measures of retinal injury and protection to show that endogenous protective mechanisms serving to reduce ischemic injury can be induced in mouse retina by preconditioning a day before the lethal ischemic insult with noninjurious periods of either brief retinal ischemia or systemic hypoxia. As detailed in this section, the protection afforded by these stimuli was robust and long-lasting (to 4 weeks) and was reflected by significant preservation of the a- and b-waves of the ERG and by significant preservation of cellular and axonal retinal layer morphometry. 

Ischemic tolerance is now a well-described phenomenon in the central nervous system (CNS) and is under active investigation in brain. ¹ We reported a couple of years ago on the establishment of a rat model of retinal ischemic tolerance. ² In that study, significant protection resulted from using brief ischemia to precondition the retina against a subsequently applied ischemic insult of a duration to cause moderate-to-severe injury. Protection was evidenced both histologically and functionally. The present study extends these fundamental findings to the mouse, indicating that retinal ischemic tolerance, as expected, is inducible across species. We performed more extensive histopathologic analyses of the retinas than in the previous rat study, established two distinctly different, noninjurious preconditioning treatments to promote tolerance in the retina, and undertook long-term recovery studies to verify that the protection induced by preconditioning was indeed sustained. Although subsequent studies in the rat model have provided support for the participation of different signaling molecules in promoting ischemic tolerance, ^{10 11} pharmacologic approaches to elucidation of the mechanism can often be problematic. It was our intention to develop mouse models of retinal ischemic tolerance to allow for the future identification of the mechanistic basis of this phenomenon by using knockout and transgenic mice. 

The progressive injury induced in mouse retinas by 30, 45, and 60 minutes of complete retinal ischemia was extensively distributed throughout inner retinal structures, with duration-dependent reductions in layer thickness and cell counts, as well as ERG signal attenuation. Functional and histopathologic protection of the retina were demonstrated in response to two distinct preconditioning stimuli. Preconditioning with brief but direct retinal ischemia afforded significant protection even against the more severe (60-minute) ischemic insult. Both our ERG and histopathology results indicate that more robust protection of the retina can result from preconditioning against milder ischemic insults, as noted in the rat model of retinal ischemic tolerance. ¹² Pre-exposure to systemic hypoxia was also effective, particularly in providing sustained protection over a period of weeks. Given the growing evidence of distinct differences in ischemic injury mechanisms for gray and white matter, ¹³ the preservation of both cellular and axonal regions of the inner retina probably reflects the induction of a multifaceted mechanism of protection that increases the ischemic resistance of neurons and their axons, the surrounding oligodendrocytes, and perhaps Müller and vascular cells as well. Elucidation of the induction and expression mechanisms underlying this pancellular, panretinal protection remains the domain of future studies. 

Another important aspect of advancing new ischemic tolerance models is the demonstration that preconditioning stimuli, although promoting an adaptive response in the tissue, are noninjurious themselves. We verified in this study, using both functional and histopathologic approaches, that no injury resulted from the ischemic or the hypoxic challenges that served as effective preconditioning stimuli. Conversely, we showed that shorter periods of direct retinal ischemia were ineffective as a preconditioning stimulus, and that longer periods of ischemia resulted in significant histopathologic evidence of retinal injury. Similarly, more histopathologic injury was noted when a more severe hypoxic challenge of 8% oxygen for 2 hours was used as the preconditioning stimulus. Thus, the intensity and duration of hypoxia or ischemia must be titrated to trigger adaptive mechanisms in the retina without resulting in overt injury. 

To determine the extent to which the two preconditioning regimens we developed resulted in long-lasting protection against retinal ischemic injury, we undertook additional histopathologic examinations of retinas 4 weeks after ischemia. Our findings of a continued atrophy and thinning of both the cellular and axonal layers of the inner retina between 1 and 4 weeks after ischemia suggests that the common practice of assessing ischemic injury and the neuroprotective effects of drugs or even genetic manipulations within the first postischemic week may underestimate the ultimate extent of injury and/or protection. Ischemic preconditioning still afforded protection 1 month later, although not nearly as robust as that witnessed 1 week after ischemia. Conversely, the protection afforded by the hypoxic preconditioning stimulus resulted in about twice the level of protection at 4 weeks relative to the ischemic preconditioning stimulus. These differential short- and long-term outcomes resulting from the different preconditioning regimens indicate that each promoted somewhat unique states of ischemic tolerance. Direct ischemia induced protective mechanisms that were manifested more during the immediate postinsult period, but treatment efficacy faded somewhat over time. Systemic hypoxia promoted modest protection initially, but was much more effective at inducing longer-lasting adaptive changes in the retina. 

Oxygen sensing and adaptation to hypoxia are fundamental physiologic principles. For the most part, these adaptive responses are mediated at the cellular level by hypoxia-induced changes in gene expression. In retina, hypoxia is known to activate the genes for vascular endothelial growth factor, ¹⁴ insulin-like growth factor, ¹⁵ basic fibroblast growth factor, ¹⁶ transforming growth factor-β1, ¹⁶ lactate dehydrogenase, ¹⁷ and various heat-shock proteins, ³ some secondary to the activation of a transcription factor called hypoxia-inducible factor-1. ^{14 18} However, the relationship of the above changes to those specifically induced by the hypoxic preconditioning stimulus we used may not be directly translational, because the genomic response to a hypoxic stimulus can vary considerably with respect to the duration, intensity, and frequency of the imposed hypoxia, and the developmental state of the tissue so exposed. It is also important to point out that, because the hypoxia was systemic, adaptive responses may occur elsewhere in the CNS and in other tissues as well. Mechanistically, blood-borne mediators that trigger signal transduction pathways promoting cross-tissue tolerance may be operative under these conditions. Indeed, there is evidence of the direct ability of tumor necrosis factor-α and interleukin-1, cytokines produced and elaborated into the circulation in response to hypoxia, to promote ischemic tolerance in brain. ^{19 20} The involvement of particular mediators and the expression mechanisms underlying the development of injury resistance in these models of retinal ischemic tolerance still require positive identification. 

The length of time over which preconditioning promotes a state of ischemic tolerance is of considerable clinical interest. Based on results of our previous study in rats, ² we expected ischemic tolerance to persist for several days, but not as long as 1 week, after preconditioning. However, using histopathologic criteria, we found that protection induced by ischemic preconditioning was lost by 3 days. This shorter window of protection may result from the higher metabolic rate in CNS tissues of the mouse relative to rats, or other as-yet-unexplained species-, anesthesia-, or model-dependent differences. This is not to say that tolerance to ischemia is necessarily always transient, because this may also depend on the nature of the preconditioning stimulus and whether it is repeatedly presented. We did not determine the time window during which hypoxic preconditioning protects the mouse retina or the duration of tolerance after the presentation of multiple ischemic or hypoxic preconditioning treatments. We also did not attempt to determine how soon after either of these preconditioning stimuli that the retina exhibited any significant degree of tolerance, given the precedent that ischemic injury was actually exacerbated in rats when their retinas were rendered ischemic 1 hour after preconditioning. ² This observation, and the ability of protein synthesis inhibitors to block retinal ischemic tolerance in vivo ² and in vitro, ³ supports the concept that the development of a state of ischemic tolerance requires a finite period for changes in gene expression and the resultant effects of de novo protein synthesis to become manifest. 

In conclusion, we established two distinct models of retinal ischemic tolerance in mice that we characterized by functional and histopathologic assessments of injury protection. Both the ischemic and the hypoxic preconditioning stimuli were demonstrated to be noninjurious to the retina. In addition to using two different preconditioning stimuli to induce an ischemia-tolerant state, the models also differed in the nature and magnitude of regionally selective protection evidenced at 1 week and 1 month after ischemia. Electrophysiologic outcome measures appeared to be a more sensitive indicator of retinal injury than histopathologic ones, whereas histopathologic indices of preconditioning’s efficacy improved more robustly than the electroretinographic parameters. The longer-lasting protection resulting from the systemic hypoxia preconditioning stimulus may be more suitable for future studies of the underlying mechanistic basis of this response. It is our hope that elucidation of the molecular basis of ischemic tolerance will be aided considerably by using these models in various genetically engineered mice that are becoming widely available. With such mechanistic information, a better understanding of retinal ischemic injury and more rationally designed therapeutic strategies for limiting its deleterious effects may be realized. 

 Supported by National Eye Institute Grants EY-09678 (JMG) and EY-02687 (Department of Ophthalmology, Washington University), and the Edward Mallinckrodt, Jr Foundation (KKO).

 Submitted for publication October 9, 2001; revised January 23, 2002; accepted February 6, 2002.

 Commercial relationships policy: N.

 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.

 

Representative photomicrographs of hematoxylin and eosin stained 5-μm sections of Swiss-Webster ND4 mice retinas 1 week after the following treatments: (A) nonischemic control; (B) preconditioning with brief (5-minute) ischemia followed by sham lethal ischemia; (C) preconditioning with systemic hypoxia (2 hours, 11% oxygen) followed by sham lethal ischemia; (D) 45-minute (lethal) ischemia; (E) preconditioning with brief ischemia followed by 45-minute lethal ischemia; (F) preconditioning with systemic hypoxia (2 hours, 11% oxygen) followed by 45-minute lethal ischemia.

Figure 2.

 

Electroretinographic results in mice from the different treatment groups. (A) Representative pre- and posttreatment ERG traces (top to bottom) from: a nonpreconditioned, nonischemic control animal (no treatment); an animal subjected to ischemic preconditioning (PC) without ischemia; an animal subjected to ischemia (45 minutes) without preconditioning; and an animal subjected to ischemia (45 minutes) 1 day after ischemic preconditioning. (B) Amplitudes of the a- and b-waves 1 day and 1 week after 30-minute ischemia in untreated (n = 7) and ischemia-preconditioned (n = 8) animals. (C) Amplitudes of the a- and b-wave 1 day and 1 week after 45-minute ischemia in untreated (n = 7) and ischemia preconditioned (n = 5) animals. Results in (B) and (C) are normalized to the pretreatment baseline amplitudes in each eye. *P < 0.05 versus before treatment amplitude; #P < 0.05 versus respective ischemia group.

Table 1.

 

Normalized Histopathologic Measures for Different Preconditioning Treatments Followed by Sham Ischemia

Animal Group	n	Layer Thickness					Cell Counts
				OLM-ILM	INL	IPL	INL	GCL
Ischemic preconditioning (5 min of direct retinal ischemia)	4	103 ± 7	101 ± 7	107 ± 4	102 ± 3	107 ± 10
Ischemic preconditioning (10 min of direct retinal ischemia)	4	84 ± 4	87 ± 4	80 ± 6	94 ± 6	72 ± 7
Hypoxic preconditioning (2 hours of 11% oxygen)	4	105 ± 3	95 ± 4	102 ± 6	100 ± 8	114 ± 8
Hypoxic Preconditioning (2 hours of 8% oxygen)	4	85 ± 3*	81 ± 5*	79 ± 2*	112 ± 5*	115 ± 7

 Data are normalized to the nonischemic, nonpreconditioned fellow eyes in the ischemic preconditioning groups or to eyes from the nonischemic, nonhypoxic controls (n = 33) in the hypoxic preconditioning groups.

*   P < 0.05 versus the respective control by Friedman or Kruskal-Wallis ANOVA on ranks.

Table 2.

 

Normalized Electroretinographic Measures for Different Preconditioning Treatments Followed by Sham Ischemia.

Animal Group	n	a-Wave			b-Wave
				1 day	7 days	1 day	7 days
Nonpreconditioned, nonischemic controls	4	94 ± 11	145 ± 39	109 ± 18	123 ± 23
Ischemic preconditioning (5 min of direct retinal ischemia)	6	131 ± 8*	132 ± 16*	124 ± 7	111 ± 11
Hypoxic preconditioning (2 hours of 11% oxygen)	6	106 ± 5	122 ± 7	105 ± 3	115 ± 7

 Data shown are normalized to the respective baseline wave amplitudes before ischemic or hypoxic preconditioning.

*   P < 0.05 versus controls by Friedman ANOVA on ranks.

Figure 3.

 

Retinal ischemic tolerance induced by ischemic preconditioning (PC) against injury caused by 30-, 45-, or 60-minute ischemia. Histopathologic findings in mice subjected to 30- (n = 5), 45- (n = 7), or 60-minute (n = 5) retinal ischemia without pretreatment or after PC with brief (5-minute) retinal ischemia (n = 6, 6, and 4 for 30-, 45-, and 60-minute groups, respectively). Results are normalized to nonischemic, nonpreconditioned control eyes (n = 33). *P < 0.05 versus nonischemic control; #P < 0.05 versus respective ischemia group without preconditioning; +P < 0.05 versus 30-minute ischemia; @P < 0.05 versus 45-minute ischemia.

Figure 4.

 

Retinal protection by ischemic (A) and hypoxic (B) preconditioning (PC) was sustained over 4 weeks of recovery. Histopathologic findings in mice subjected to 45-minute retinal ischemia and recovered for 1 (n = 7) or 4 (n = 6) weeks without pretreatment relative to (A) animals preconditioned with brief (5-minute) retinal ischemia with recovery for 1 week (n = 6) or 4 (n = 6) weeks or (B) animals preconditioned with 2-hour systemic hypoxia (11% oxygen), with recovery for 1 week (n = 4) or 4 (n = 4) weeks. Results are normalized to nonischemic control eyes (n = 33). *P < 0.05 versus the nonischemic control; #P < 0.05 versus respective ischemia group (after 1- or 4-week recovery); +P < 0.05 versus ischemia after 1-week recovery.

Figure 5.

 

Time window for retinal protection by ischemic preconditioning (PC). Histopathologic findings in untreated mice subjected to 45-minute retinal ischemia (n = 7) and mice preconditioned with brief (5-minute) retinal ischemia followed by 45-minute retinal ischemia 1 (n = 6), 3 (n = 7), or 7 (n = 4) days later. Results are normalized to nonischemic control eyes (n = 33). *P < 0.05 versus nonischemic controls; #P < 0.05 versus ischemia.

The authors thank Amy H. Hwang for expert technical assistance and Mae O. Gordon and Brad Wilson for statistical consultations.