b-Wave amplitude 7 days aftera 60-minute period of retinal ischemia. The b-wave amplitude in rats withshort-term diabetes was significantly greater than that in nondiabetic controls.Similarly, rats with diabetes for 6 weeks also displayed significantly greaterb-wave amplitudes than controls. Asterisk indicates P<.001.Limit lines indicate SEM.
A, Normal electroretinogram (double-headedarrow indicates b-wave amplitude). B-F, Representative electroretinogramsrecorded 7 days after a 60-minute period of ischemia. B, Normoglycemic ratwith isotonic sodium chloride (saline) infusate. C, Hyperglycemic 5-day diabeticrat with saline infusate. D, Hyperglycemic 6-week diabetic rat with salineinfusate. E, Hyperglycemic glucose-injected rat with saline infusate (IP indicatesintraperitoneal). F, Normoglycemic rat with 5% glucose infusate.
Digitally processed images ofuntreated retina (A), retina of rats after 6 weeks of diabetes (B), and retinalsections from rats that received 60 minutes of ischemia and 7 days of reperfusion(C-H). C, Normoglycemic rat with isotonic sodium chloride solution (saline)infusate. D, Intraocular infusate of 5% 2-deoxyglucose. E, Five-day diabeticrat. F, Six-week diabetic rat. G, Glucose-induced preexisting hyperglycemia.H, Infusate of 5% glucose in a normoglycemic rat (for all parts, hematoxylin-eosin,original magnification ×250).
Thy-1 messenger RNA (mRNA) levelsin rats after 6 weeks of diabetes and control rats, 7 days after a 60-minuteperiod of ischemia. A, Representative polymerase chain reaction; top gel isfrom control eyes, and bottom gel, diabetic eyes (odd lanes, ischemic eyes;even lanes, untouched eyes). B, Ratio of Thy-1 to cyclophilin mRNA. The ratiowas significantly greater in diabetic rats than in controls. Asterisk indicates P<.001; limit lines, SEM.
b-Wave amplitudes after 45 minutes(A) and 60 minutes (B) of ischemia. A, Glucose-induced preexisting hyperglycemiapreserved the b-wave amplitude against 45 minutes of ischemia, but postischemiahyperglycemia did not. B, Preexisting hyperglycemia also significantly protectedthe b-wave amplitude when the duration of the ischemia was increased to 60minutes. Asterisk indicates P<.001; limit lines,SEM.
A, b-Wave amplitudes 7 days aftera 60-minute period of ischemia. The b-wave amplitude in isotonic sodium chloridesolution (saline)–infused eyes was significantly smaller than that inglucose-infused eyes but significantly greater than that in 2-deoxyglucose–infusedeyes and no different from that in lactate-infused eyes (P = .76). B, Ratio of Thy-1 to cyclophilin messenger RNA (mRNA). Thy-1mRNA was significantly preserved in the glucose-infused eyes. Asterisk indicates P<.001; dagger, P<.05; double dagger, P<.01;and limit lines, SEM.
Casson RJ, Chidlow G, Wood JPM, Osborne NN. The Effect of Hyperglycemia on Experimental Retinal Ischemia. Arch Ophthalmol. 2004;122(3):361-366. doi:10.1001/archopht.122.3.361
Copyright 2004 American Medical Association. All Rights Reserved.Applicable FARS/DFARS Restrictions Apply to Government Use.2004
To determine the effect of hyperglycemia and intraocular glucose deliveryon ischemic retinal injury.
Experimental diabetes was induced in age- and sex-matched Wistar ratsby an injection of streptozocin. The functional and structural retinal injuryin these rats after a period of pressure-induced retinal ischemia was comparedwith the injury in appropriate controls and with rats made hyperglycemic byan injection of systemic glucose. The effect of high intraocular pressure–inducedischemia with the use of several different isotonic substrates in the elevatedreservoir (isotonic sodium chloride solution, glucose, 2-deoxyglucose, andlactate) was also determined. Electroretinography, reverse transcriptase polymerasechain reaction, and histologic examination were used to assess the retinalinjury.
Streptozocin-induced diabetes, glucose injection–induced preexistinghyperglycemia, and intraocular glucose delivery during ischemia markedly reducedthe functional and structural ischemic retinal injury. Neither postischemichyperglycemia nor the intraocular delivery of lactate significantly affectedthe ischemic injury; however, the intraocular delivery of 2-deoxyglucose significantlyexacerbated the retinal injury.
Preexisting hyperglycemia and the intraocular delivery of glucose markedlyattenuate ischemic retinal injury.
These findings highlight fundamental differences in energy metabolismbetween brain and retina, have important implications for the pathophysiologyof diabetic retinopathy, and may lead to novel therapeutic strategies forischemic retinopathies.
PREEXISTING HYPERGLYCEMIA has a well-described deleterious effect onthe clinical outcome of cerebral ischemia1;however, information on the effect of hyperglycemia on retinal ischemia islacking. The many functional and structural similarities between retina andbrain and the causal association between long-term hyperglycemia and diabeticretinopathy2 suggest that hyperglycemia wouldalso exacerbate ischemic retinal injury; however, several factors may underminethis assumption. Unlike the brain, the isolated mammalian retina derives aconsiderable amount of adenosine triphosphate (ATP) from the conversion ofglucose to lactate, even in the presence of oxygen (aerobic glycolysis); and,importantly, it has the remarkable ability to maintain most of its ATP requirementin the absence of oxygen by anaerobic glycolysis.3- 6 Furthermore,an intraocular delivery of glucose before or during retinal ischemia has beenreported to attenuate ischemic histologic changes,7,8 andexperimental diabetes mitigates hypoxia-induced electroretinographic (ERG)changes.9 We therefore hypothesized that ifpreexisting hyperglycemia increased the availability of glucose to the ischemicretina, then it may in fact attenuate ischemic retinal injury, rather thanexacerbate it. Herein, we present evidence in support of this hypothesis anddiscuss the implications of the findings.
Procedures used in this study conformed to the Association for Researchin Vision and Ophthalmology Statement for the Use of Animals in Ophthalmicand Vision Research and were approved by the Home Office in England. MaleWistar rats (250-300 g) housed in a 12-hour light-dark cycle were used forall experiments; food and water were provided ad libitum. Anesthesia was achievedwith a combination of intramuscular fentanyl citrate (0.315 mg/mL) and fluanisone(10 mg/mL) (Hypnorm; Janssen Pharmaceutica, Beerse, Belgium), 0.4 ml/kg, anddiazepam, 0.4 mL/kg.
Baseline ERG recordings were taken 1 to 4 days before further intervention,as previously described.10 Diabetes was inducedby a single intraperitoneal injection of streptozocin (60 mg/kg), and controlrats received a single injection of the citrate buffer solvent. Four dayslater, the blood glucose level was measured from tail vein blood by meansof a blood glucose sensor (Abbott Laboratories, Medisense Products, AbbottPark, Ill). Diabetes was defined by a blood glucose level greater than 25mg/dL (13.8 mmol/L) at this measurement and at second measurement just beforeischemia. Diabetic rats were divided into 2 groups: group 1 underwent ischemia5 days after streptozocin injection, and group 2 rats underwent ischemia 6weeks later; streptozocin nonresponders were not used in further experiments.Rats received unilateral pressure-induced retinal ischemia for either 45 or60 minutes, as previously described.8 The ERGswere recorded 3 and 7 days later, after which the retinas were removed forreverse transcriptase polymerase chain reaction (RT-PCR) analysis or histopathologicalexamination.
Hyperglycemia was also induced by an intraperitoneal injection of glucose(2 g/kg) 20 minutes before ischemia, or at the end of the ischemic period.Control animals received isotonic sodium chloride solution (saline) (1 g/kg).The infusate used during the pressure-induced ischemia was routinely 0.9%saline, but for experimental purposes this was replaced with 5% glucose, aglycolysis inhibitor (5% 2-deoxyglucose), and 3.7% lactic acid.
Seven days after ischemia, retinal cyclophilin and Thy-1 messenger RNA(mRNA) levels were determined by means of RT-PCR as previously described.10,11 Briefly, total RNA was isolated fromwhole retinas, and first-strand complementary DNA synthesis was performedon 2 µg of DNase-treated RNA. The individual complementary DNA specieswere amplified in a 10-µL reaction, containing the 2-µL complementaryDNA aliquot, PCR buffer (10mM Tris hydrochloride, pH 8.3, 50mM potassium chloride),4mM magnesium chloride, 200µM of each deoxynucleotide triphosphate,4 ng/µL of both the sense and antisense primers, and Taq polymerase (2.5 U). Reactions were initiated by incubating at 94°Cfor 10 minutes, and PCRs (94°C, 15 seconds; 52°C, 55°C, or 56°C,30 seconds; 72°C, 30 seconds) were performed for a suitable number ofcycles followed by a final extension at 72°C for 3 minutes. Interexperimentalvariations were avoided by performing all amplifications in a single run.The PCR products of the primer pairs were separated on 1.5% agarose gels withthe use of ethidium bromide for visualization and yielded single bands correspondingto the expected molecular weights. The relative abundance of each PCR productwas determined by analysis of digital gel photographs software (Labworks;Ultra-Violet Products Ltd, Cambridge, England). For semiquantitative analysis,the ratio of the Thy-1 densitometric readings between the ischemic and controleyes was calculated and was normalized to the internal standard mRNA ratio(cyclophilin), which was assumed to be unaffected by the ischemia.
Terminally anesthetized rats were transcardially perfused with 50 mLof 10mM phosphate-buffered saline, followed by 4% paraformaldehyde. The eyeswere enucleated and immersion fixed for 1 hour in 4% paraformaldehyde, transferredto 10% neutral-buffered formalin overnight, and processed for routine paraffin-embeddedsections on an automated tissue processor (Shandon Pathcentre; Thermo Shandon,Inc, Pittsburgh, Pa). Eyes were embedded sagittally, and 5-µm serialsections including the optic nerve were cut with a rotary microtome (MicromHM 330; McBain Instruments, Chatsworth, Calif) and stained with hematoxylin-eosin.
The vitreous and retinas were removed from freshly killed rats, placedon ice, and weighed (the vitreous from both eyes was pooled, as were the 2retinas from each rat). After the retina samples were diluted with 1 mL ofdeionized water, the samples were sonicated, then centrifuged at 12 000g for 10 minutes. The supernatant was removed and the vitreoussamples were made up to 100 µL with distilled water. This volume wasthen added to a glucose hexokinase assay reagent (Sigma-Aldrich Corp, St Louis,Mo) for 15 minutes, and the spectroscopic absorbance was then read at 340nm. The final glucose concentration (milligrams per milliliter) in the vitreouswas determined by allowing for the dilution factor and comparing the absorbancewith a previously constructed calibration curve. The weight of glucose perweight of retina was determined in a similar manner.
To compare independent samples and normalize for slight day-to-day variationin the ERG, the ratio of the a- and b-wave amplitudes between paired eyes(one eye treated and the fellow untouched) was used as the unit of statisticalanalysis. Similarly, the paired-eye ratio of the RT-PCR densitometric readingswas used as the unit of statistical analysis. A 1-way analysis of variancewas used to compare means between 2 or more independent groups, and a Tukeyhonestly significantly difference test was used for post hoc comparisons.A Bonferroni correction was applied to the ERG measurements at 2 time points.All statistical determinations were performed with SPSS for Windows, version10 (SPSS Inc, Chicago, Ill), and all data are expressed as mean ± SEM;a P value of less than .05 was considered statisticallysignificant.
In the first set of experiments, we found a remarkable degree of protectionin the group 1 diabetic rats 3 and 7 days after a 45-minute period of ischemia:after 7 days, the mean b-wave amplitude (as a percentage of baseline) in thediabetic group (n = 9) was 89% ± 5%, but in the control group (n =9) it was only 24% ± 4% (P<.001); similarly,the reduction in Thy-1 mRNA (as a percentage of cyclophilin mRNA) was only5% ± 3% in the diabetic group (n = 6) compared with 22% ± 4%in the control group (n = 6; P = .02). When the durationof ischemia was increased to 60 minutes, the b-wave still exhibited betterpreservation in the diabetic group than the control group; the mean amplitudesare shown in Figure 1, and representativeERG tracings from an untreated, ischemic control and ischemic group 1 diabeticrat are shown in Figure 2A-C, respectively).The group 1 diabetic rats also displayed remarkable preservation of structuralintegrity (Figure 3E compared with Figure 3C).
The group 2 rats had no significant change in the ERG over time (datanot shown), but after a 60-minute period of ischemia, they displayed markedpreservation of the b-wave amplitude (Figure1 and Figure 2D). Similarly,the Thy-1 mRNA was remarkably preserved compared with the controls (Figure 4), as was the structural integrity(Figure 3F).
In the second group of experiments, an alternative method was used toinduce hyperglycemia: injecting rats with glucose (2 g/kg) 20 minutes beforeischemia. After 45 minutes of ischemia and 7 days of reperfusion, the meanb-wave amplitude (as a percentage of baseline) in the glucose-injected groupwas 95% ± 5%, but in the control group it was only 28% ± 4%(P<.001; Figure5A). Similarly, the reduction in Thy-1 mRNA (as a percentage ofbaseline) was only 5% ± 4% in the hyperglycemic group (n = 6) comparedwith 20% ± 3% in the control (P = .03; n =6). After 60 minutes of ischemia, the b-wave still exhibited better preservationin the hyperglycemic group than in the control group (Figure 5B). The hyperglycemic group also displayed marked preservationof structural integrity (Figure 3Gcompared with Figure 3C). However,when glucose (2 g/kg) was administered at the end of the period of ischemia,the mean b-wave amplitude (after 3 and 7 days of reperfusion) was not significantlydifferent from the control mean amplitude (Figure 5A).
Although the blood glucose level in the glucose-injected rats, at theonset of ischemia, was significantly lower (274 ± 25.2 mg/dL [15.2± 1.4 mmol/L]) than the blood glucose level in the streptozocin-induceddiabetic rats (420 ± 46.8 mg/dL [23.3 ± 2.6 mmol/L]; P<.001), there was no significant difference in the degree of protection.
The effect of using different energy substrates in the intraocular infusateduring the pressure-induced ischemia was also investigated. After ischemia,the glucose-infused eyes had significantly greater b-wave amplitudes thansaline-infused eyes (Figure 6A,and Figure 2F compared with Figure 2B), had significantly greater preservationof Thy-1 mRNA (Figure 6B), and showedbetter structural preservation than the saline-infused eyes (Figure 3H compared with Figure 3C). However, eyes infused with a glycolysis inhibitor (2-deoxyglucose)had significantly lower b-wave amplitudes than the saline-infused eyes (P = .01; Figure 6A)and had a corresponding exacerbation of structural injury (Figure 3D compared with Figure 3C). When isotonic lactate was used as the intraocular infusate,the mean b-wave amplitude was not significantly different from that of thesaline controls (P = .76; Figure 6A).
Table 1 compares the glucoseconcentration in the retina and vitreous of normoglycemic and hyperglycemicrats. The concentration of glucose in the vitreous was significantly greater(P<.001) in the hyperglycemic rats than in controls,in both the diabetic rats and the glucose-injected rats; however, the amountof free glucose in the retina was not significantly different between groups.
The b-wave amplitude of the ERG is a functional measure that is particularlysusceptible to ischemia12,13 andprovides a quantitative measurement of middle and inner retinal function,but it excludes information about retinal ganglion cells.14 Thy-1mRNA is a retinal ganglion cell marker, and its measurement by semiquantitativeRT-PCR provides a useful gauge of ischemic insult,11 whichpredominantly affects the inner retina; histopathological examination providesqualitative stuctural data about all retinal layers. Using these complementarytechniques, we found that preexisting hyperglycemia provided a remarkabledegree of protection against ischemic injury, a level of protection that,in our experience, is unrivaled by any other method of neuroprotection. Themost likely explanation for this phenomenon is that the elevated glucose levelsin the vitreous of the hyperglycemic rats caused relative preservation ofretinal ATP levels by anaerobic glycolysis during the period of ischemia.There are, however, several alternative explanations for this phenomenon thatneed to be considered.
Conceivably, the protective effect of experimental diabetes on acuteretinal ischemia could be nonmetabolic in nature, and perhaps be explainedby a diabetes-induced neuroprotective stress response. However, the findingthat a brief episode of preexisting hyperglycemia, with a concomitant increasein vitreous glucose concentration, also produced a profound degree of protectionmakes this explanation unlikely. In addition, the delivery of intraocularglucose during the period of ischemia protected against ischemic injury innormoglycemic rats and a glycolytic inhibitor (2-deoxyglucose) exacerbatedischemic injury, findings that further support the concept that the protectiveeffect is likely to be metabolic in nature.
Glucose has traditionally been considered the predominant energy substratefor cerebral neurons; however, evidence has recently accumulated suggestingthat, during neural activity, glial energy demands are met by anaerobic glycolysis,whereas neuronal energy demands are met by glial-derived lactate.15,16 Schurr et al17 reportedthat this glial-neuronal lactate shuttle is important for the recovery ofneuronal function after hypoxic injury. Hence, it is conceivable that theanaerobic buildup of lactate and its use in the immediate postischemic period,rather than the intraischemic use of glucose per se, was the protective factor.However, the finding that intraocular lactate was not protective, coupledwith the lack of protection afforded by hyperglycemia in the postischemicperiod, indicates that the anaerobic use of glucose during the ischemic periodrather than the use of accumulated lactate in the postischemic period waslikely to be the critical protective event.
Another protective mechanism using glucose is the possible increasein the activity of the pentose phosphate pathway with a corresponding increasein nicotinamide adenine dinucleotide phosphate and cellular reducing power,thereby decreasing free radical injury, particularly in the early reperfusionperiod. However, Winkler et al5 showed thatthe activity of the pentose phosphate pathway is not significantly increasedin the isolated retina when the concentration of glucose is raised from 5mMto 30mM.5 Furthermore, the complete lack ofneuronal protection by hyperglycemia in the early reperfusion period goesagainst the possibility that the pentose phosphate pathway is involved inthe neuronal protection.
The present results highlight fundamental differences between retinaland cerebral energy metabolism. Under aerobic conditions, brain tissue convertsonly 13% of its available glucose to lactate and obtains 94% of its ATP fromoxidative metabolism.18 Conversely, under aerobicconditions, the isolated rat retina converts 90% of its available glucoseto lactate, deriving 36% of its ATP from glycolysis.4 Hence,under anaerobic conditions the rat retina has "less distance to make up" and,by a 2-fold up-regulation of glycolysis (the Pasteur effect), can maintainATP levels at 50% to 70% of aerobic levels as long as glucose is abundant.4,5 Similarly, under aerobic conditions,human Müller cells convert 99% of available glucose to lactate and requireonly a 30% increase in glycolysis to maintain ATP production anaerobically.6
Several in vivo studies have also shown that vitreous glucose and retinalglycogen are used during ischemia.19- 22 Usinga pressure-induced ischemia model in rabbits, Weiss21 foundthat in the first 20 minutes of ischemia, retinal glycogen was used, but afterthis period the vitreous became an important source of glucose for anaerobicglycolysis. Weiss concluded that the vitreous glucose was the most importantenergy substrate for retinal ischemia lasting more than 20 minutes and thatthe exhaustion of anaerobic glycolysis was a crucial factor in determiningthe ischemic tolerance time of the rabbit retina.21 Inthe present study, diabetic rats and the rats that received a bolus of glucosedisplayed a similar degree of protection, despite the higher glucose levelsin the diabetics, which may reflect a saturation of the enzyme systems atmoderate glucose levels.
Given that rats store low amounts of retinal glycogen,23 andthe finding that the free retinal glucose levels were not significantly differentin the hyperglycemic and normoglycemic rats, it seems likely that vitreousglucose is an important energy source for the ischemic rat retina. The findingof elevated vitreous glucose levels in the diabetic rats was consistent withthe well-described correlation between blood glucose and vitreous glucoselevels.24 In addition, we found that vitreousglucose levels are rapidly elevated after a systemic bolus of glucose, a findingthat almost certainly accounts for the protection against ischemia that wasobserved in this group of rats. Hence, in summary, these experiments haveshown that the retina's relative lack of reliance on mitochondrial metabolismcan be exploited under anaerobic conditions in vivo by increasing glucoseavailability to the oxygen-starved retina.
The present study has implications for the pathophysiology of diabetes.Long-term hyperglycemia has a clear association with diabetic retinopathy,which commonly includes an ischemic element. Hence, although long-term hyperglycemiacan cause retinal microangiopathy2 and neuronalapoptosis,25 short-term hyperglycemia can paradoxicallypreserve neuronal structure and function after acute ischemia.
The finding that preexisting moderate hyperglycemia is protective againstretinal ischemia may also have clinical applications. Conceivably, retinalinjury caused by temporary ischemia in a variety of conditions, includingretinal detachment and venous occlusion, could be ameliorated by short-termmoderate hyperglycemia. In addition, the sustained administration of glucoseto the vitreous may represent a treatment strategy for conditions involvingan element of chronic retinal ischemia.
Corresponding author: Neville N. Osborne, DSc, Nuffield Laboratoryof Ophthalmology, Walton Street, Oxford OX2 6AW, England (e-mail: firstname.lastname@example.org).
Submitted for publication March 28, 2003; final revision received July15, 2003; accepted August 4, 2003.
This study was supported in part by grant QLK6-CT-2001-00385 from theEuropean Community PRO-AGE-RET (Protection Against Ageing in Retina) program,Brussels, Belgium.