Elevated Glucose Levels Increase Retinal Glycolysis and Sorbitol Pathway Metabolism

Transcription

1 Elevated Glucose Levels Increase Retnal Glycolyss and Sorbtol Pathway Metabolsm Implcatons for Dabetc Retnopathy Mara K. Van den Enden,* Jens R. Nyengaard,^ Eva Ostrow,% Judth H. Burgan,X and Joseph R. Wllamson% Purpose. To assess effects of elevated glucose levels on retnal glycolyss and sorbtol pathway metabolsm. Methods. Freshly solated retnas from normal male Sprague-Dawley rats were ncubated for 2 hours at 37 C, ph 7.45, n Krebs bcarbonate-hepes buffer contanng 5, 10, 20, or 30 mm glucose. Glycolytc metaboltes, sorbtol, and fructose were measured n extracts of retna and medum. Results. Elevated glucose levels ncreased retnal levels of sorbtol and trose phosphates, decreased.sra-glycerol-3-phosphate levels, ncreased lactate and fructose producton, and ncreased the retnal lactate-pyruvate rato (ndcatve of an ncreased cytosolc rato of free NADH-NAD + lke that nduced by hypoxa). An nhbtor of aldose reductase (AL 4114) normalzed sorbtol, fructose, trose phosphates, and the lactate-pyruvate rato wthout affectng lactate producton or sra-glycerol 3-phosphate levels. Conclusons. Elevaton of retnal glucose levels causes a hypoxa-lke redox mbalance "pseudohypoxa" that results from ncreased oxdaton of sorbtol to fructose n the second step of the sorbtol pathway. Ths redox mbalance provdes a plausble explanaton for mpared regulaton of retnal blood flow (n the- absence of vascular structural changes) n humans wth dabetes and n nondabetc acutely hyperglycemc anmals. These fndngs, together wth other observatons, suggest that ths redox mbalance precedes, and may contrbute to, hypoxc and schemc retnopathy assocated wth dabetes. Invest Ophthalmol Vs Sc. 1995; 36: Although hemodynamc changes are consdered to play an mportant role n the pathogeness of dabetc retnopathy, the natural hstory of retnal blood flow changes, the factors that medate them, and ther role n the development of endstage (prolferatve) retnopathy are unclear. 1 " 4 Focal areas of capllary closure (acellular capllares) and nonperfused capllares appear to develop relatvely early after the onset of dabetes, and t s postulated that hypoxc and schemc From the * Department of Internal Medcne, AZ. Mddelhem Hosptal, Antwerp, Belgum; the ^Stereologcal Research Laboratory, Aarhus Unversty, Aarhus Q Denmark; and the %Pathobgy Department, Washngton Unversty School of Medcne, St. Lous, Mssour. Supported by Natonal Eye Insttute grant (EY06600), the Megevdenskabens Fremme and Novo Nordc Foundaton, and by the Klo Dabetes and Vascular Research Foundaton. Submtted for publcaton Aprl 20, 1994; revsed February 13, 1995; February 23, Propretary nterest category: N. Reprnt requests: Joseph R. Wllamson, Washngton Unversty School of Medcne, Department of Pathology, Box 8118, 660 S. Eucld Avenue, St. Lous, MO retna adjacent to such areas releases vasoactve agents and growth factors that ncrease blood flow and ntate vasoprolferatve changes, ncludng the formaton of shunts. The mechansm (s) responsble for capllary closure and nonperfuson, and whether these changes are preceded and caused by blood flow changes, are unknown. Dscordant reports regardng whether retnal blood flow s ntally ncreased or decreased after the onset of poorly controlled dabetes n humans and anmals may relate to methodologc dfferences and problems n assessng blood flow, dfferences n duraton and severty of dabetes, and speces examned. 2 " 4 Several nvestgatons have demonstrated ncreased retnal blood flow, such as that n kdney and scatc nerve, 5 " 8 n rats wth poorly controlled dabetes of a few weeks duraton, 5 ' 9 " 12 although not all reports are n agreement Recent studes ndcate that retnal and scatc nerve blood flow are ncreased n nonketotc rats but are unchanged or decreased n ketotc Investgatve Ophthalmology & Vsual Scence, July 1995, Vol. 36, No. 8 Copyrght Assocaton for Research n Vson and Ophthalmology 1675

2 1676 Investgatve Ophthalmology & Vsual Scence, July 1995, Vol. 36, No. 8 dabetc rats despte comparable blood glucose levels and vascular conductance n both groups (unpublshed observatons, 1995). These fndngs are consstent wth evdence that renal blood flow s decreased n severely dabetc rats In contrast to dscordant reports regardng whether retnal blood flow s ncreased or decreased by dabetes, there s general agreement that hyperglycerha of only a few hours duraton (produced by ntravenous glucose nfuson) ncreases retnal and renal blood flow n nondabetc humans and anmals. 17 " 21 Because blood flow n retna and other tssues s ncreased nstantly by hypoxa, 22 ' 23 the fndng that ncreased retnal blood flow n nondabetc, acutely hyperglycemc rats s prevented by confuson of pyruvate and by nhbton of the sorbtol pathway, 21 coupled wth evdence that pyruvate also attenuates hypoxc and schemc myocardal njury, 23 " 25 suggests that ncreased retnal blood flow nduced by acute hyperglycema, dabetes, and hypoxa may be medated by a redox mbalance common to hypoxa and to ncreased sorbtol pathway metabolsm,.e., an ncreased rato of ncotnamde adenne dnucleotde (reduced form)-ncotnamde adenne dnucleotde phosphate (oxdzed form) (NADH-NAD + ). The plausblty of ths hypothess s supported by evdence that: (1) the retnal lactate-pyruvate rato s ncreased (ndcatve of an ncreased cytosolc rato of free NADH-NAD + ) n dabetc rats 826 ' 27 ; (2) exposure to ncreased glucose levels and ncreased flux of glucose through the sorbtol pathway are assocated wth an ncreased lactate-pyruvate rato n several cells and tssues 8 ' 23 ' 28 " 37 ; (3) ncreased blood flow n ocular tssues, scatc nerve, and kdney n dabetc rats s prevented by nhbton of the sorbtol pathway 5 ' 8 ; and (4) exposure of normal human red cells and retna to elevated sorbtol levels (at physologcal glucose levels) also ncreases the lactate-pyruvate rato 38 (and unpublshed observatons, 1993), as well as blood flow and vascular albumn permeaton n skn chamber granulaton tssue vessels. Ths vascular dysfuncton s prevented by coadmnstraton of pyruvate and by an nhbtor of sorbtol dehydrogenase 39 (and unpublshed observatons, 1994). In the frst step of the sorbtol pathway, glucose s reduced to sorbtol (coupled to oxdaton of NADPH to NADP + ) by aldose reductase. In the second step of the pathway, sorbtol s oxdzed to fructose (coupled to reducton of NAD + to NADH) by sorbtol dehydrogenase. Because the enzymes of the sorbtol pathway are located n the cytoplasm, an ncreased rate of oxdaton of sorbtol to fructose wll favor an ncreased cytosolc rato of free NADH/NAD +. Ths same redox mbalance develops n hypoxc and schemc tssues because of mpared reoxdaton of NADH to NAD + by the mtochondral electron transport chan. Because the "hypoxa-lke" effects of elevated glucose levels on NADH-NAD + ratos and vascular functon appear to be ndependent of tssue oxygenaton, 23 ' 36 ' 37 we have used the term pseudohypoxa to descrbe them. 23 An ncreased NADH-NAD + rato s assocated wth ncreased blood flow regardless of the cause,.e., true hypoxa, hyperglycema, galactosema, cyande posonng, carbon monoxde posonng, or elevaton of extracellular lactate levels suffcently to ncrease the ntracellular lactate-pyruvate (NADH-NAD + ) rato. 23 ' 40 The current experments were undertaken to assess the effects of elevated glucose levels on retnal metabolsm n vtro, n partcular on the retnal lactate-pyruvate rato, a surrogate parameter of the cytosolc rato of free NADH-NAD +. Exposure of retnas from normal rats to 20 mm to 30 mm glucose for 2 hours n vtro nduces hypoxa-lke ncreases n lactate-pyruvate ratos that are attenuated by an nhbtor of aldose reductase and n trose phosphates that are prevented by pyruvate as well as by nhbtors of aldose reductase. METHODS Anmals and Materals Retnas were obtaned from eyes of normal male Sprague-Dawley rats weghng ~250 g. One group of rats was pretreated wth an nhbtor of aldose reductase, AL 4114 (2,7-dfluoro-4,5-dmethoxyspro[9Hfluorene-9,4'-mdazoldne]-2',5'-done), 41 a gft of Alcon Laboratores Inc., mxed n the chow n an amount to provde ~0.04 mmol/kg per day for 7 days before removng the retnas. All rats used n these experments were housed and cared for n accordance wth the gudelnes of the Unversty Commttee for the Humane Care of Laboratory Anmals, NIH gudelnes on laboratory anmal welfare, and the ARVO Statement for the Use of Anmals n Ophthalmc and Vson Research. Rats were housed ndvdually, allowed food and water ad lbtum, and were on a 12- hour lght-12-hour dark cycle. Durng the lght cycle, ambent room lght s ~65 foot candles and cage lght s 1 to 2 foot candles. Expermental Protocols Forty-two rats were anesthetzed wth sodum pentobarbtal (45 mg/kg body weght) 15 mnutes before surgery. Both eyes were enucleated and placed n cecold Krebs bcarbonate-hepes (KH) buffer, ph 7.45, after whch the anmals were exsangunated. Several corneal ncsons were made, the lens was removed, and the retna was gently separated from the chorod and placed n ce-cold KH buffer untl the start of the ncubaton. The composton of the KH buffer (n mmol/l) was: 110 NaCL, 4.5 KC1; 0.5 MgCl 2, 1.0 CaCl 2-2H 2 O,

3 Retnal Glycolyss and Sorbtol Pathway Metabolsm Na 2 HPO 4, 30 Hepes, 15 NaHCO 3, and 0.1 L-argnne plus 0.05% bovne serum albumn. The buffer was gassed wth humdfed 95% O 2-5% CO 2 (Po 2 ~ 500 torr) at 37 C for 2 hours before use. Although average retnal oxygen levels are n the range of 3%, 42 retnas n these studes were ncubated n buffer equlbrated wth 95% O 2 because of evdence that ths hgh Po 2 s requred for mantenance of electrophysologcal functon of ncubated retnas. 43 The need for ths hgh Po 2 to mantan electrophysologcal functon n vtro reflects the mportance of oxdatve metabolsm n mantanng retnal functon and the dependence of ntraretnal Po 2 on dffuson of oxygen from the retnal surfaces n the absence of oxygen delvery by the vasculature. The retnas were ncubated for 2 hours at 37 C n KH buffer contanng 5, 10, 20, or 30 mm D-glucose ± 21 //M AL 4114 or 3 mm pyruvate. For the assessment of glycolytc ntermedates, three retnas were ncubated n 30 ml buffer n 50 ml Erlenmeyer flasks. For quantfcaton of sorbtol, wyo-nostol and fructose three retnas were ncubated n 3 ml buffer n test tubes. After the addton of retnas to the buffer, the ndvdual flasks and tubes were agan gassed wth 95% O 2-5% CO 2 for the frst 10 mnutes of the ncubaton. No change n Po 2 (measured wth a Cornng 170 ph/blood Gas Analyzer) was detectable n medum sampled at the begnnng and end of ncubatons. Incubatons were termnated by rapdly removng retnas from the medum. For the assessment of glycolytc ntermedates, retnas and medum were mmedately extracted wth perchlorc acd as descrbed below under the secton Metabolc and Bochemcal Assays. For quantfcaton of polyols and fructose, retnas and medum were frozen separately n lqud ntrogen untl analyss. Metabolc and Bochemcal Assays Glycolytc Intermedates. After separatng retnas and medum, the combned retnas were extracted by addng 50 fa of ce-cold 3 N perchlorc acd. After 20 mnutes (at 4 C), 100 fa of dstlled H 2 O was added, and the samples were centrfuged at 14,000g for 5 mnutes at 4 C. The supernatant was removed and assayed for determnaton of metaboltes (pyruvate, lactate, swrglycerol 3-phosphate, fructose 1,6-bsphosphate, dhydroxyacetone phosphate, and glyceraldehyde 3-phosphate) by standard enzymatc methods. 44 ' 45 Fructose 1,6-bsphosphate was converted to dhydroxyacetone phosphate and glyceraldehyde 3- phosphate by aldolase and measured together wth endogenous dhydroxyacetone phosphate and glyceraldehyde 3-phosphate as trose phosphates. 29 The tssue pellet was saved for measurement of DNA, 46 to whch all metaboltes were normalzed. Alquots of ncubaton medum were extracted by addng 55 //I of ce-cold 12 N perchlorc acd to 1 ml of medum. After centrfugaton at 14,000g for 5 mnutes at 4 C, the supernatant was taken for determnaton of lactate and pyruvate. The lactate-pyruvate rato was used as a surrogate estmate of the cytosolc rato of free NADH-NAD + because t provdes a more relable measure of ths parameter than measurement of NADH and NAD + n tssue extracts (t s not possble to dstngush between mtochondral and cytosolc pools, and the reduced and oxdzed nucleotdes are not extracted wth equal effcency because the reduced form s bound to enzymes more tghtly than the oxdzed form). 47 Sorbtol and myo-inostol. The frozen retnas were resuspended n 1 ml of 0.02% sodum azde n water, 200 fa of nternal standard (unformly labeled 13 C- sorbtol and 2 H 6 -w)o-nostol) was added, and the samples were heated for 15 mnutes at 100 C. The extracts were deprotenzed wth 150 fa of 0.3 N barum hydroxde and 150 fa of 0.3 N znc sulfate. After centrfugaton, sorbtol and wyo-nostol n the supernatant were quantfed as ther butylboronate dervatves by gas chromatography-mass spectrometry. 48 The pellet was saved for measurement of DNA content. 46 Fructose. Unextracted ncubaton medum was assayed for fructose content by a fluorometrc enzymatc method usng Resazurn (7-hydroxyphenoxazn-3-one 10 oxde) (Sgma, St. Lous, MO) Metaboltes measured n the retna (lactate, pyruvate, sorbtol, myonostol, total trose phosphates, and sra-g3p) are referred to as retnal content or levels, whereas metaboltes measured n the medum alone (fructose) or n combnaton wth the retna (lactate and pyruvate) are referred to as retnal producton or medum content/ level as approprate. Statstcal Analyses All data are expressed as the mean ± standard devaton. Analyss of varance of all parameters was performed (on natural logarthm transformed data) wth the SAS general lnear models procedure. 51 Overall dfferences among expermental groups for each parameter were frst assessed by the Van der Waerden test 52 ; ndvdual par-wse comparsons were evaluated by least square means analyss only f the Van der Waerden test was sgnfcant at P < 0.05 for a gven parameter. A nonparametrc Blom transformaton 53 of all data was performed before assessment of ndvdual par-wse group dfferences. Uncorrected P values based on two-taled tests of sgnfcance are shown for the relevant comparsons. RESULTS Effect of Elevated Glucose Levels on Retnal Glycolyss Lactate producton (medum 4- retna) by retnas n 5 mm glucose was 26 ± 5 nmol//g DNA (retnal lactate

4 1678 Investgatve Ophthalmology & Vsual Scence, July 1995, Vol. 36, No. 8 ncrease n lactate content versus only an approxmately 30% ncrease n pyruvate. AL 4114 ex vvo decreased lactate-pyruvate ratos 27% for retnas n 5 mm glucose (P = 0.002) and 16% for retnas n 30 mm glucose (P = 0.019). AL 4114 gven n vvo and 40 A. Lactate < mm Glucose FIGURE 1. Relatonshp between retnal lactate-pyruvate ratos (retnal content of lactate and pyruvate) and lactate producton (lactate n medum + retna) for retnas ncubated n 5,10, 20, and 30 mm glucose. Sgnfcandy dfferent from 5 mm glucose: *P < ; fp = ; %P = Sgnfcandy dfferent from 10 mm glucose: P = ; #P = accounted for ~1.5% of total lactate producton). Lactate producton ncreased lnearly between 5 and 20 mm glucose; t was ncreased maxmally (~31%) at 20 mm glucose, P = , versus 5 mm glucose (Fg. 1). Lactate producton by retnas n 30 mm glucose was slghtly less than n 20 mm glucose, but the dfference was not statstcally sgnfcant. AL 4114 ex vvo had no effect on lactate producton by retnas n 5 or 30 mm glucose (Fg. 2A). Lactate producton by retnas n 30 mm glucose plus AL 4114 n vvo and ex vvo was slghtly lower (11%) than for retnas n 30 mm glucose alone (P = 0.06); 3 mm pyruvate had no effect on lactate producton by retnas n 30 mm glucose. Pyruvate producton (medum -I- retna) by retnas n 5 mm glucose was 2.13 ± 0.55 nmol//xg DNA (retnal pyruvate accounted for ~0.8% of total pyruvate producton). Pyruvate producton dd not dffer for retnas n 5, 10, 20, or 30 mm glucose (pyruvate producton by retnas n 5 and 30 mm glucose s shown n Fg. 2B). AL 4114 ex vvo ncreased pyruvate producton 32% by retnas n 5 mm glucose (P = 0.001) and 13% by retnas n 30 mm glucose (P= 0.024) (Fg. 2B). AL 4114 n vvo and ex vvo ncreased pyruvate producton ~ 47% by retnas n 30 mm glucose (P < ). Retnal lactate-pyruvate ratos (based on retnal content of lactate and pyruvate) were 24.0 ± 4.7 for retnas n 5 mm glucose. The ratos ncreased lnearly wth ncreasng glucose levels (Fg. 1) between 5 and 30 mm glucose (n contrast to lactate producton, whch plateaued at 20 mm glucose as shown n Fg. 1) and were elevated ~60% at 30 mm glucose (P < versus 5 mm glucose) (Fgs. 1, 2C). Increased retnal lactate-pyruvate ratos n 30 versus 5 mm glucose were accounted for by an approxmately twofold o c 30 E 25 < Z Q o I Q I B. Pyruvate *t F * C. Lactate/Pyruvate rato ; t 5 + AL4114 ex vvo. **4 +. t 1 «-- : t# 30 l ; 4-*-%. 30+ AL4114 ex vvo 30 + AL4114 n ex vvo 30 + PYR FIGURE 2. Effects of AL 4114 and 3 mm pyruvate on total accumulaton (retna + medum) of lactate (A) and pyruvate (B) and on the retnal lactate-pyruvate rato (lactate and pyruvate n retna only (C) after ncubaton n 5 or 30 mm glucose for 2 hours at 37 C. AL 4114 n vvo was admnstered for 7 days at a dose of 0.04 mmol/kg per day. AL 4114 ex vvo was added to the ncubaton medum at a concentraton of 21 (JM. See Methods for further detals. Data for ndvdual rats are shown, as are the mean ± SD on the left sde. (A) Sgnfcandy dfferent from 5 mm glucose: *P = ; fp = ; %P= (B) Sgnfcandy dfferent from 5 mm glucose: *P = Sgnfcantly dfferent from 30 mm glucose: fp < ; P = Sgnfcandy dfferent from 30 mm glucose + AL 4114 ex vvo: %P = (C) Sgnfcantly dfferent from 5 mm glucose: *P < ; -\P = Sgnfcantly dfferent from 30 mm glucose: %P< ; P= Sgnfcandy dfferent from 30 mm glucose + AL 4114 ex vvo: #P =

5 Retnal Glycolyss and Sorbtol Pathway Metabolsm 1679 < Q O Q o E 4 T " t t -- I 5 30 A. Trose Phosphates * * :: 30 + AL4114 n + ex vvo T B. sn - G3P * # 30 + PYR FIGURE 3. Effects of AL 4114 and 3 mm pyruvate on retnal trose phosphate (A) and sn-g3p (B) levels after ncubaton n 5 or 30 mm glucose for 2 hours at 37 C. See legend to Fgure 2 for detals of ncubaton condtons. (A) Sgnfcantly dfferent from 5 mm glucose: *P < Sgnfcantly dfferent from 30 mm glucose: fp < ; %P = (B) Sgnfcantly dfferent from 5 mm glucose: *P < ; tp = ; %P = Sgnfcantly dfferent from 30 mm glucose: P = Sgnfcantly dfferent from 30 mm glucose + AL 4114 n vvo and ex vvo: #P = ex vvo completely normalzed lactate-pyruvate ratos for retnas n 30 mm glucose. Normalzaton of retnal lactate-pyruvate ratos n 30 mm glucose by AL 4114 n vvo and ex vvo was accounted for by a 17% decrease n retnal lactate and a 29% ncrease n pyruvate; for retnas ncubated n 5 mm glucose, the reducton n lactate-pyruvate ratos by AL 4114 ex vvo was accounted for by a 20% decrease n lactate content and a 12% ncrease n pyruvate (data not shown). The correlaton between lactate producton and lactate-pyruvate rato at all glucose concentratons (usng natural logarthm transformed data) was not statstcally sgnfcant (r = 0.33, P= 0.064) because of the further ncrease n lactatepyruvate rato at 30 versus 20 mm glucose despte no further ncrease n lactate producton. Trose phosphate levels were 16.5 ± 3.4 pmol//ltg DNA for retnas n 5 mm glucose (Fg. 3A); they were ncreased 35% by 30 mm glucose (P < 0.01). Ths ncrease was completely prevented by AL 4114 n vvo and ex vvo, whch decreased trose phosphates to levels 39% (P= ) and 18% (P= 0.37) below those n 30 and 5 mm glucose, respectvely. Trose phosphates for retnas n 30 mm glucose + 3 mm pyruvate were decreased stll further,.e., ~40% lower than for retnas n 5 mm glucose (P < 0.01). srtc3? levels were 8.4 ±1.3 pmol///g DNA for retnas n 5 mm glucose (Fg. 3B); they were decreased -21% by 30 mm glucose (P = ). AL 4114 n vvo and ex vvo slghtly attenuated ths effect of 30 mm glucose on sn-gw levels, whch remaned 13% lower than for retnas n 5 mm glucose (P = 0.041) and dd not dffer sgnfcantly from retnas n 30 mm glucose (P = 0.27); 3 mm pyruvate caused 5W-G3P levels to fall another 21% (P = versus 30 mm glucose). Effect of Elevated Glucose Levels on Sorbtol Pathway Metaboltes and rayo-inostol Levels Sorbtol levels were 2.9 ± 0.7 pmol///g DNA for retnas n 5 mm glucose; they were ncreased 5.7-fold by 30 mm glucose (P < ) (Fg. 4A). AL 4114 ex vvo decreased retnal sorbtol levels 27% n 5 mm glucose (P = 0.015) and 84% n 30 mm glucose (P < ) to levels 9% lower than those n 5 mm glucose alone (P = 0.16); 3 mm pyruvate decreased retnal sorbtol levels 16% n 30 mm glucose (P = versus 30 and P = versus 5 mm glucose). Fructose producton was 7.4 ± 0.9 pmol//ltg DNA for retnas n 5 mm glucose; t was ncreased threefold by 30 mm glucose (P < ) (Fg. 4B). AL 4114 ex vvo decreased retnal fructose producton 54% n 5 mm glucose (P < ) and 72% n 30 mm glucose (P< ). Thus, for retnas n 30 mm glucose + AL 4114 ex vvo, fructose producton was only 13% hgher than by retnas n 5 mm glucose alone (P = 0.79); 3 mm pyruvate had no effect on fructose producton by retnas n 30 mm glucose. Retnal rayo-nostol levels were 33 ± 2 pmol///g DNA after ncubaton n 5 mm glucose; they were unaffected by 30 mm glucose ± AL 4114 ex vvo but were decreased ~15% by AL 4114 ex vvo n 5 mm glucose (P = , data not shown). DISCUSSION These experments, together wth evdence dscussed n the ntroducton, support the hypothess that sorbtol pathway-lnked vascular dysfuncton n dabetc rats s largely the consequence of an ncreased cytosolc rato of free NADH-NAD + resultng from ncreased oxdaton of sorbtol to fructose (coupled to reducton of NAD + to NADH) by sorbtol dehydrogenase (Fg. 5). Ths hypoxa-lke redox mbalance has potentally mportant mplcatons for the pathogeness of retnal vascular dysfuncton early after the onset of poorly controlled dabetes, as well as for later vascular struc-

6 1680 Investgatve Ophthalmology & Vsual Scence, July 1995, Vol. 36, No. 8 <20 Z Q a 10 o a A A. Sorbtol * t# ## ex vvo ex vvo FIGURE 4. Effects of AL 4114 and 3 mm pyruvate on retnal sorbtol levels (A) and on fructose accumulaton n the medum (B) after ncubaton n 5 or 30 mm glucose for 2 hours at 37 C. See legend to Fgure 2 for detals of ncubaton condtons. (A) Sgnfcantly dfferent from 5 mm glucose: *P < ; fp = 0.002; %P = Sgnfcantly dfferent from 30 mm glucose: F< #P = Sgnfcantly dfferent from 30 mm glucose + AL 4114 ex vvo: \P < (B) Sgnfcandy dfferent from 5 mm glucose: *P < Sgnfcandy dfferent from 30 mm glucose: fp < Sgnfcantly dfferent from 30 mm glucose + AL 4114 ex vvo: %P < attenuates lactate-pyruvate ratos n the retna and scatc nerve n dabetc rats 8 and n the scatc nerve of nondabetc rats. 54 The lkelhood that the ncreased lactate-pyruvate (NADH-NAD + ) rato n retnas ncubated n 30 mm glucose s not medated by Pasteur and Crabtree effects 55 ' 56 s supported by observatons that t s ndependent of lactate producton (n the current experments and n those of Nyengaard et al 36 ), prevented completely by AL 4114, and observed n human erythrocytes 29 ' 35 n whch oxdatve phosphorylaton s absent. The fndng that AL 4114 decreased retnal lactate-pyruvate ratos wthout affectng lactate producton (and also decreased fructose producton) by retnas ncubated n 5 mm glucose as well as n 30 mm glucose ndcates that even at physologcal glucose levels, glucose flux through the sorbtol pathway s suffcent to modulate the cytosolc rato of free NADH-NAD +. Increased retnal lactate producton at elevated glucose levels s consstent wth observatons n glomerul, endoneurum, granulaton tssue, aorta, and cultured retnal vascular cells exposed to elevated glucose levels and wth elevated lactate levels n retnas from dabetc rats. 623 ' 2857 " 60 The observatons that lactate producton was ncreased maxmally by 20 mm glucose (31% versus 5 mm glucose) and that most of the ncrease occurred between 10 and 20 mm glucose agree wth those of Keen and t Glucose NADPH NADP+ t Sorbtol tural changes assocated wth clncally sgnfcant dabetc retnopathy. Effects of Elevated Glucose Levels on Retnal Lactate-Pyruvate Ratos and on Glycolyss Increased lactate producton (lactate n retna and medum) and lactate-pyruvate ratos for retnas n 20 to 30 mm glucose versus 5 mm glucose ndcate that elevated glucose levels mmc the effects of hypoxa on glycolyss and on cytosolc free NADH/ NAD + (lactate-pyruvate). 23 The lkelhood that the ncreased lactate-pyruvate rato s medated by ncreased metabolsm of glucose through the sorbtol pathway s supported by the fndng that t was prevented completely byal 4114, whch also prevented ncreased retnal sorbtol levels and fructose producton. Ths nterpretaton s supported by evdence that the nhbton of sorbtol dehydrogenase FIGURE 5. Interrelatonshps between glycolyss, de sorbtol patfway, and lpd metabolsm: effects of elevated glucose levels. PI

7 Retnal Glycolyss and Sorbtol Pathway Metabolsm 1681 Chlouveraks. 61 Although Wnkler commented that lactate producton by normal rat retnas was maxmal at 5 mm glucose, 62 data shown n hs Fgure 4 suggest that t was ncreased ~35% n 20 versus 5 mm glucose. The hgher lactate-pyruvate ratos n retnas at 30 mm glucose versus 5 mm glucose are consstent wth evdence that the retnal NADH/NAD + rato s ncreased n dabetc rats 826 ' 27 and that lactate - pyruvate ratos are ncreased n many tssues (heart, lens, endoneurum, glomerul, aorta, and erythrocytes) exposed to elevated glucose levels n vvo, n vtro, or both. 8 ' 2328 " 37 (and unpublshed observatons, 1994). Elevated retnal lactate-pyruvate ratos n dabetc rats mght be attrbutable to several mechansms other than elevated glucose levels per se,.e., decreased blood flow resultng n hypoxa and/or other metabolc and hormonal mbalances assocated wth the dabetc mleu. On the other hand, ncreased lactate-pyruvate ratos n retnas (from normal rats) ncubated at 30 mm glucose versus 5 mm glucose at dentcal O 2 tensons n vtro s attrbutable to effects of elevated glucose levels per se. These observatons support the concepts of hyperglycemc pseudohypoxa 23 and "spurous anoxa." 6I Keen and Chlouveraks 61 suggested that many functonal and structural retnal changes nduced by dabetes mght "be nterpreted as responses to the spurous 'anoxc' stmulus of hgh lactate concentraton, spurous n that they result not from anoxa but from ncreased glycolyss.'' The current observatons (together wth other evdence dscussed n ths artcle) support the mportance of the ncreased lactate-pyruvate rato rather than ncreased glycolyss n medatng hyperglycema and dabetes-nduced retnal vascular dysfuncton; however, the fundamental concept that hyperglycema and the dabetc mleu mmc effects of hypoxa on retnal metabolsm and vascular functon s supported by both nvestgatons as well as by observatons of Nyengaard et al. 36 In contrast to the ncreased retnal lactate-pyruvate ratos nduced by elevated glucose levels n the current experments and n dabetc rats, 8 ' 26 ' 27 Thomas et al 63 reported that lactate-pyruvate ratos n cultured human retnal pgment epthelal cells were the same n 5 and 20 mm glucose (134 and 114, respectvely). These very hgh ratos at 5 mm glucose (compared to those n whole retnas, ether freshly solated or ncubated) 8 ' 26 ' 36 suggest that ther paradgm s unsutable for assessng glucose effects on the redox state of pyrdne nucleotdes. As evdence for n vvo relevance of ther observatons, Thomas et al 29 noted ther RPE cell ratos were comparable to those n ncubated human erythrocytes; those ratos also are hghly unphysologcal. 64 ' 65 Effects of Elevated Glucose Levels on Redox Senstve Metaboltes: Trose Phosphates and sra-g3p Increased trose phosphate levels nduced by 30 mm glucose are attrbutable to the ncreased lactate-pyruvate (NADH/NAD + ) ratos that favor nhbton of glyceraldehyde-3 phosphate dehydrogenase (Fg. 5). As a consequence, fructose 1,6-bsphosphate, dhydroxyacetone phosphate (DHAP), and glyceraldehyde 3-phosphate accumulate. Ths nterpretaton s supported by the fndngs that AL 4114 completely prevented the ncreased lactate-pyruvate rato and that both AL 4114 and pyruvate prevented the ncrease n trose phosphates nduced by 30 mm glucose (Fg. 3A). An ncreased cytosolc rato of free NADH-NAD + also favors reducton of DHAP to 5n-G3P (Fg. 5), the frst step n one pathway for de novo synthess of 1,2- dacyl-.m-glycerol. 23 ' 66 Whether sn-gsp levels ncrease depends on the rate at whch 57^G3P s metabolzed further (Fg. 5). The fndng that snrgsp levels were decreased by 30 mm glucose (and not normalzed by AL 4114) s consstent wth observatons n granulaton tssue and suggests that ncreased lpd synthess (ndependent of sorbtol pathway actvty) nduced by 30 mm glucose favored ncorporaton of S7&-G3P nto lpds at a faster rate than t was syntheszed. Ths explanaton s supported by evdence for ncreased de novo synthess and tssue levels of DAG and other lpds n cells and tssues exposed to elevated glucose levels. 23 ' 59 ' 66 ' 67 It also s consstent wth evdence that snrgsp dehydrogenase actvty n some tssues s not hgh enough to establsh equlbrum between ts substrates (sra-g3p and DHAP) and the free nucleotde cofactors NADH and NAD These explanatons are consstent wth elevated sn-g3y* levels n the lens of dabetc rats 68 and n human erythrocytes exposed to hgh glucose levels, 35 ' 38 tssues n whch lpd synthess s lkely to be low. The fndng that retnal snrg3p levels were decreased further n 30 mm glucose + pyruvate versus 30 mm glucose (Fg. 3B) s consstent wth observatons n kdney, granulaton tssue, and human erythrocytes 6,38,39,60 Ths effect of pyruvate can be explaned by ncreased oxdaton of NADH to NAD + (coupled to reducton of pyruvate to lactate) by lactate dehydrogenase (Fg. 5) and the mpact of the resultant decrease n NADH-NAD + on oxdaton of glyceraldehyde 3-phosphate to 1,3-bsphosphoglycerate and the reducton of DHAP to sra-g3p. Dscordant Effects of AL 4114 and Pyruvate on Sorbtol Pathway Metabolsm Increases n retnal sorbtol levels and medum fructose producton by 30 mm glucose, and ther preventon by AL 4114, are consstent wth numerous obser-

8 1682 Investgatve Ophthalmology & Vsual Scence, July 1995, Vol. 36, No. 8 vatons n dabetc anmals. 523 ' 5758 It s of nterest that fructose producton at 5 mm glucose was ~2.6-fold hgher than sorbtol levels and that AL 4114 reduced sorbtol levels by only 27% whle decreasng fructose by ~54% (Fg. 4). These fndngs ndcate that the fructose was derved prmarly from sorbtol and that retnal sorbtol levels account for only ~33% of total sorbtol syntheszed; the other 67% was oxdzed to fructose. The markedly dscordant effects of AL 4114 and pyruvate on retnal sorbtol levels and fructose producton (Fg. 4), together wth evdence that both agents prevent sorbtol pathway-lnked vascular dysfuncton, 21 ' 66 ' 69 support the concluson that sorbtol pathway-lnked ncreases n retnal blood flow n dabetc rats result largely from the redox mbalance caused by ncreased sorbtol oxdaton to fructose rather than from putatve osmotc effects of elevated sorbtol and/ or fructose levels 31 ' 70 or metabolc and redox changes lnked to reducton of glucose to sorbtol (Fg. 5). Aldose reductase nhbtors prevent the redox mbalance by decreasng sorbtol levels and reducton of NAD + to NADH by sorbtol dehydrogenase. In contrast, pyruvate prevents the redox mbalance by ncreasng the rate of reoxdaton of NADH to NAD + wthout substantally alterng sorbtol levels or fructose accumulaton (Fg. 5). These observatons suggest that sorbtol pathway-medated abnormaltes develop when cells are unable to oxdze NADH to NAD + fast enough to keep pace wth ncreased reducton of NAD + to NADH by sorbtol dehydrogenase when sorbtol levels are elevated. Although pyruvate also s a free radcal scavenger, 71 scavengng of free radcals would not account for the effects of pyruvate on trose phosphate and sw-g3p levels n the current experments. Implcatons for Dabetc Retnopathy The ncreased lactate-pyruvate (NADH-NAD + ) rato caused by ncreased oxdaton of sorbtol to fructose s ndcatve of reductve stress,.e., producton and accumulaton of excess reducng equvalents (n the form of NADH). 72 ' 73 Accumulaton of reducng equvalents favors formaton of reactve oxygen speces (resultng n oxdatve stress) 23 ' 7273 ; t also nfluences the actvty of numerous enzymes that use NADH and NAD + as cofactors for coupled oxdaton-reducton reactons. 23 The lkelhood that retnal vascular dysfuncton n acutely dabetc rats s medated largely by (sorbtol pathway-nduced) reductve and oxdatve stress s supported by evdence that free radcal scavengers such as probucol and superoxde dsmutase, as well as nhbtors of ntrc oxde synthase, attenuate vascular dysfuncton assocated wth acute hyperglycema n nondabetc anmals, dabetes, and hypoxa, and schema-reperfuson njury. 9 ' 12 ' 20 ' 23 ' 74 The hypoxa-lke redox mbalance caused by elevated glucose levels provdes a plausble explanaton for mpared regulaton of retnal blood flow n subjects wth dabetes and n acutely hyperglycemc nondabetc anmals n response to varous nterventons (vasoactve agents, changes n blood pressure, autonomc stmulaton, and changes n Po 2 of nspred gas mxtures). 318 ' 75 " 78 The NADH-NAD+ rato s a senstve parameter of the balance between rates of NAD + reducton and of NADH oxdaton; we suggest that ths rato modulates a cascade of events that medate blood flow changes n response to changes n retnal oxygenaton. An ncreased rato sgnals the need for more oxygen that s met by an ncrease n blood flowmedated by vasodlator prostaglandns and ntrc oxde. 23 ' 79 Ths sgnalng cascade for ncreasng blood flow s actvated when the redox mbalance s nduced by hyperglycema as well as by hypoxa. However, because the redox mbalance nduced by hyperglycema s ndependent of tssue oxygenaton, blood flow remans elevated despte ncreased oxygen delvery. Because the retnal vasculature remans dlated and blood flow s ncreased on account of the hgh NADH-NAD + rato, responses to vasoactve stmul also wll be blunted. Subsequent vascular structural changes (.e., scleross wth decreased complance) would further lmt responses to vasoactve agents. Ths explanaton for mpared retnal blood flow regulaton s consstent wth evdence that nterventons (nhbtors of aldose reductase, nhbtors of sorbtol dehydrogenase, and pyruvate) that prevent the redox change also prevent ncreased blood flow n acutely hyperglycemc and dabetc rats. 5 ' 8 ' 21 ' 23 ' 39 ' 66 ' 69 The fndng that a hypoxa-lke redox mbalance develops after exposure to elevated glucose levels for no more than 2 hours n vtro (coupled wth evdence that retnal blood flow s ncreased by n vvo hyperglycema of comparable duraton) 20 ' 21 ndcates that ths redox mbalance precedes, and may contrbute to, development of hypoxc and schemc dabetc retnopathy. Because the mechansms by whch hyperglycema and hypoxa ncrease NADH-NAD + ratos are ndependent and addtve, 23 ' 36 less severe hypoxa or schema would be requred to cause hypoxc dysfuncton and njury n retnas reductvely stressed by the dabetc mleu. Thus, ncreased producton of vascular endothelal growth factor, whch has been mplcated n the pathogeness of prolferatve dabetc retnopathy 80 ' 81 and whch s stmulated by hypoxa, would occur at hgher oxygen tensons n retnas already reductvely stressed by the dabetc mleu. The lnear ncrease n retnal lactate-pyruvate (NADH-NAD + ) ratos wth ncreasng glucose levels between 5 and 30 mm glucose s consstent wth evdence that the onset and progresson of dabetc retnopathy and other complcatons of dabetes are correlated wth the severty of dabetes.

9 Retnal Glycolyss and Sorbtol Pathway Metabolsm 1683 The dsappontng effects of aldose reductase nhbtors n clncal trals n humans wth dabetes may seem paradoxcal n vew of ther marked effcacy n preventng dabetes-nduced mcrovascular and neural abnormaltes n anmal models of dabetes. 23 ' 82 Several factors may account for these dsappontng results: the dose of aldose reductase nhbtors used n the clncal trals was nadequate to block the sorbtol pathway; clncal trals have been largely reversal rather than preventve studes (most of the anmal studes have been preventatve); the duraton of the trals has been much too short n vew of the duraton requred to demonstrate the benefcal effects of mproved glucose control on dabetc complcatons n the recently completed Dabetes Control and Complcatons Tral 83 ; and other metabolc mbalances assocated wth the dabetc mleu (.e., ncreased oxdaton of fatty acds) 23 may play an mportant role n the pathogeness of retnopathy n humans wth dabetes. In concluson, exposure of retnas from normal rats to elevated glucose levels mmcs the effects of hypoxa n causng an ncreased rato of lactate-pyruvate (NADH-NAD + ) whch, under the condtons of these experments, s lnked to ncreased flux of glucose through the sorbtol pathway. These fndngs suggest that reductve stress nduced by the dabetc mleu precedes, and may contrbute to the development of, hypoxc and schemc dabetc retnopathy. Key Words aldose reductase, dabetc retnopathy, hypoxa, redox, sorbtol Acknowledgments The authors thank Alcon Laboratores Inc. for the gft of AL They also thank A. M. Faller for techncal assstance and M. Hmmelmann for secretaral assstance n the preparaton of ths manuscrpt. References 1. Stefansson E, Landers MB, Wolbarsht ML. Oxygenaton and vasodlataton n relaton to dabetc and other prolferatve retnopathes. Ophthalmc Surg. 1983; 14: Grunwald JE, Rva CE, Bane J, Brucker AJ. Total retnal volumetrc blood flow rate n dabetc patents wth poor glycemc control. Invest Ophthalmol Vs Sc. 1992;33: Patel V, Rassam S, Newsom R, Wek J, Kohner E. Retnal blood flow n dabetc retnopathy. Br Med J. 1992:305: Feke GT, Buzney SM, Ogasawara H, et al. Retnal crculatory abnormaltes n type 1 dabetes. Invest Opthalmol Vs Sc. 1994;35: Tlton RG, Chang K, Puglese G, et al. Preventon of hemodynamc and vascular albumn fltraton changes n dabetc rats by aldose reductase nhbtors. Dabetes. 1989:38: Tlton RG, Baer LD, HarlowJE, Smth SR, Ostrow E, Wllamson JR. Dabetes-nduced glomerular dysfuncton: Lnks to a more reduced cytosolc rato of NADH/NAD +. Kdney Int. 1992;41: Ido Y, McHowat J, Chang K, et al. Neural dysfuncton and metabolc mbalances n dabetc rats: Preventon by acetyl-l-carntne. Dabetes. 1994;43: Tlton RG, Chang K, Nyengaard JR, Van den Enden M, Ido Y, Wllamson JR. Inhbton of sorbtol dehydrogenase: Effects on vascular and neural dysfuncton n streptozocn-dabetc rats. Dabetes. 1995; 44: Corbett JA, Tlton RG, Chang K, et al. Amnoguandne, a novel nhbtor of ntrc oxde formaton, prevents dabetc vascular dysfuncton. Dabetes. 1992; 41: Sutera SP, Chang K, Marvel J, Wllamson JR. Concurrent ncreases n regonal hematocrt and blood flow n dabetc rats: Preventon by sorbnl. AmJPhysol. 1992;263:H945-H Crngle SJ, Yu DY, Alder VA, Su EN. Retnal blood flow by hydrogen clearance polarography n the streptozotocn-nduced dabetc rat. Invest Ophthalmol Vs Sc. 1993;34: Tlton RG, Chang K, Hasan KS, et al. Preventon of dabetc vascular dysfuncton by guandnes: Inhbton of ntrc oxde synthase versus advanced glycaton end-product formaton. Dabetes. 1993;42: Puglese G, Tlton RG, Speedy A, et al. Effects of very mld versus overt dabetes on vascular haemodynamcs and barrer functon n rats. Dabetologa. 1989; 32: Bursell SE, Clermont AC, Shba T, Kng GL. Evaluatng retnal crculaton usng vdeo fluorescen angography n control and dabetc rats. Curr Eye Res. 1992; 11: Hll MA, Larkns RG. Alteratons n dstrbuton of cardac output n expermental dabetes n rats. Am] Physol. 1989;257:H57l-H Hostetter TH, Troy JL, Brenner BM. Glomerular hemodynamcs n expermental dabetes melltus. Kdney Int. 1981; 19: Atherton A, Hll DW, Keen H, Young S, Edwards EJ. The effect of acute hyperglycaema on the retnal crculaton of the normal cat. Dabetologa. 1980; 18: Ernest JT, Goldstck TK, Engerman RL. Hyperglycema mpars retnal oxygen autoregulaton n normal and dabetc dogs. Invest Ophthalmol Vs Sc. 1983; 24: Clermont A, Bursell SE, Wolpert H, et al. The response of the retnal crculaton to acute changes n blood glucose. ARVO Abstracts. Invest Ophthalmol Vs Sc. 1993; 34: Hasan KS, Chang K, Allson W, et al. Glucose-nduced ncreases n ocular blood flow are prevented by amnoguandne and L-NMMA, nhbtors of ntrc oxde synthase. ARVO Abstracts. Invest Ophthalmol Vs Sc. 1993; 34: Hasan KS, Santago JV, Wllamson JR. Acute hyperglycema-nduced ncreases n regonal blood flow are

10 1684 Investgatve Ophthalmology 8c Vsual Scence, July 1995, Vol. 36, No. 8 prevented by Pyruvate and by Tolrestat. ADA Abstract. Dabetes. 1993; 42 (suppl 1):189A. 22. Fallon TJ, Maxwell D, Kohner EM, et al. Retnal vascular autoregulaton n condtons of hyperoxa and hypoxa usng the blue feld entoptc phenomenon Ophthalmology. 1985; 92: Wllamson JR, Chang K, Frangos M, et al. Hyperglycemc 'pseudohypoxa' and dabetc complcatons. Dabetes. 1993; 42: Bunger R, Mallet RT, Hartman DA. Pyruvate-enhanced phosphorylaton potental and notropsm n normoxc and postschemc solated workng heart: Near-complete preventon of reperfuson contractle falure. EurJ Bochem. 1989; 180: Cavalln L, Valente M, Rgobello MP. The protectve acton of pyruvate on recovery of schemc rat heart: Comparson wth other oxdzable substrates. / Mol Cell Cardol. 1990;22: Heath H, Rutter AC, Beck TC. Reduced and oxdzed pyrdne nucleotdes n the retnae from alloxan-dabetc rats. Vson Res. 1962;2: Graymore CN. Bochemstry of the retna. In: Graymore CN, ed. Bochemstry oftheeye. London: Academc Press; 1970; Morrson AD, Clements RS Jr, Wnegrad Al. Effects of elevated glucose concentratons on the metabolsm of the aortc wall. / Cln Invest. 1972;51: Travs SF, Morrson AD, Clements RS Jr, Wnegrad Al, Osk FA. Metabolc alteratons n the human erythrocyte produced by ncreases n glucose concentraton: The role of the polyol pathway. / Cln Invest. 197l;50: Varma SD, Knoshta JH. Sorbtol pathway n dabetc and galactosemc rat lens. Bochm Bophys Ada. 1974; 338: Knoshta JH, Fukush S, Kador P, Merola LO. Aldose reductase n dabetc complcatons of the eye. Metabolsm. 1979;28: Cheng H-M, Gonzalez RG, von Saltza I, Chylack LT Jr, Hutson NJ. Glucose flux and the redox state of pyrdne dnucleotdes n the rat lens. Exp Eye Res. 1988; 46: Lou MF, Dckerson JRJr, Garad R, York BM Jr. Glutathone depleton n the lens of galactosemc and dabetc rats. Exp Eye Res. 1988;46:5l Tsubota K, Krauss JM, Kenyon KR, Lang RA, Mglor S, Cheng H-M. Lens redox fluorometry: Pyrdne nucleotde fluorescence and analyss of dabetc lens. Exp Eye Res. 1989;49: Frangos M, Santago JV, Wllamson JR. In human erydrocytes, galactose nduces greater mbalances n glycolyss than glucose. ADA Abstract. Dabetes. 1991; 40 (suppl 1):254A. 36. Nyengaard JR, Ostrow E, Burgan JR, Smth SR, Wllamson JR. Interactons between hyperglycema- and hypoxa-nduced redox changes n rat retnas. ARVO Abstracts. Invest Ophthalmol Vs Sc. 1994; 35: Wllamson JR, Ostrow E, Burgan J, Smth S, Nyengaard J. Hyperglycema- and hypoxa-nduced redox changes n rat (scatc nerve) endoneurum. ADA Abstract. Dabetes. 1994; (suppl 1):17A. 38. Frangos M, Smth S, Santago J, Klo C. Sorbtol-nduced mbalances n glycolyss n human erythrocytes are reduced by pyruvate. ADA Abstracts. Dabetes. 1990; 39 (suppl 1):274A. 39. Wllamson JR, Chang K, Ostrow E, Allson W, Harlow J, Klo C. Sorbtol-nduced ncreases n vascular albumn clearance (VAC) are prevented by pyruvate but not wyo-nostol. ADA Abstracts. Dabetes. 1989;38:94A. 40. Guyton AC, RossJM, Carrer O Jr, Walker JR. Evdence for tssue oxygen demand as the major factor causng autoregulaton. Crc Res. 1964;14,15(suppl l):i-60-i Reddy VN, Ln L-R, Gbln FJ, Lou M, Kador PF, Knoshta JH. The effcacy of aldose reductase nhbtors on polyol accumulaton n human lens and retnal pgment epthelum n tssue culture. / Ocul Pharm. 1992;8: Boulton M, Patel B, Khalq A, Morarty P, Jarvs-Evans J, McLeod D. Modulators and mleu n preretnal neovascularsaton. Eye. 1992; 6: Wnkler BS. The electroretnogram of the solated rat retna. Vson Res. 1972; 12: Bergmeyer H-U. Methods of Enzymatc Analyss. Wenhem: Verlag Cheme; Lowry OH, Passoneau JV. A Flexble System of Enzymatc Analyss. Orlando: Academc Press; Burton K. A study of the condtons and mechansm of the dphenylamne reacton for the colormetrc estmaton of deoxyrbonuclec acd. Bochem J. 1956;62: Wllamson DH, Lund P, Krebs HA. The redox state of free ncotnamde-adenne dnucleotde n the cytoplasm and mtochondra of rat lver. Bochem J. 1967; 103: Eades DM, Wllamson JR, Sherman WR. Rapd analyss of sorbtol, galacttol, manntol, and myo-nostol mxtures from bologcal sources. J Chromatogr. 1989;490:l Ameyama M. Enzymc mcrodetermnaton of D-glucose, D-fructose, D-gluconate, 2-keto-D-gluconate, aldehyde, and alcohol wth membrane-bound dehydrogenases. In: Wood WA, ed. Methods n Enzymology: Carbohydrate Metabolsm. New York: Academc Press; 1982;89: Soyama K, Etsuko O. Improved procedure for determnng serum D-arabntol by resazurn-coupled method. Cln Chm Ada. 1987; 168: Anderson TW. An Introducton to Multvarate Analyss. New York: Wley & Sons; Lehmann EL. Nonparametrcs: Statstcal Methods Based on Ranks. San Francsco: Holden-Day; Blom G. Statstcal Estmates and Transformed Beta Varables. New York: Wley & Sons; Gesen K, Utz R, Grotsch H, Lang HJ, Nmmesgern, H. Sorbtol-accumulatng pyrmdne dervatves. Arznem-Forsch/DrugRes. 1994;44 (II): Koobs DH. Phosphate medaton of the Crabtree and Pasteur effects: Does a change n energy metabolsm enhance the potental for malgnancy? Scence. 1972; 178: Rmmer T, Lnsenmeer RA. Resstance of dabetc rat electroretnogram to hypoxema. Invest Ophthalmol Vs Sc. 1993; 34:

11 Retnal Glycolyss and Sorbtol Pathway Metabolsm Poulsom R, Mrrlees DJ, Earl DCN, Heath H. The effects of an aldose reductase nhbtor upon the sorbtol pathway, fructose-1-phosphate and lactate n the retna and nerve of streptozotocn-dabetc rats. Exp Eye Res. 1983; 36: Heath H, Kang SS, Phlppou D. Glucose, glucose-6- phosphate, lactate and pyruvate content of the retna, blood and lver of streptozotocn-dabetc rats fed sucrose- or starch-rch dets. Dabetologa. 1975; 11: Lee T-S, Saltsman KA, Ohash H, Kng GL. Actvaton of proten knase C by elevaton of glucose concentraton: Proposal for a mechansm n the development of dabetc vascular complcatons. Proc Natl Acad Sd USA. 1989;86: Kawamura T, Smth S, Wllamson JR. Glucose-nduced metabolc changes n tssue chamber granulaton tssue. ADA Abstracts. Dabetes. 1990;39:192A. 61. Keen H, Chlouveraks C. Metabolc factors n dabetc retnopathy. Internatonal Symposum on the Bochemstry of the Retna: Bochemstry of the Retna. London: Academc Press; 1965: Wnkler BS. Glycolytc and oxdatve metabolsm n relaton to retnal functon. JGen Physol. 1981; 77: Thomas TP, Porcellat F, Kato K, Stevens MJ, Sherman WR, Greene DA. Effects of glucose on sorbtol pathway actvaton, cellular redox, and metabolsm of myo- Inostol, phosphonostde, and dacylglycerol n cultured human retnal pgment epthelal cells. / Cln Invest. 1994;93:27l Huckabee WE. Relatonshps of pyruvate and lactate durng anaerobc metabolsm: I: Effects of nfuson of pyruvate or glucose and of hyperventlaton. J Cln Invest. 1958; 37: Tlton WM, Seaman C, Carrero D, Pomell S. Regulaton of glycolyss n the erythrocyte: Role of the lactate/pyruvate and NAD/NADH ratos. J Lab Cln Med. 1991;118: Wolf BA, Wllamson JR, Easom RA, Chang K, Sherman WR, Turk J: Dacylglycerol accumulaton and mcrovascular abnormaltes nduced by elevated glucose levels, f Cln Invest. 1990;87: Craven PA, Davdson CM, DeRuberts FR. Increase n dacylglycerol mass n solated glomerul by glucose from de novo synthess of glycerolpds. Dabetes. 1990; 39: Gonzalez A-M, Sochor M, McLean P. Effect of expermental dabetes on glycolytc ntermedates and regulaton of phosphofructoknase n rat lens. Bochem BophysRes Commun. 1980;95: Wllamson JR, Ostrow E, Eades DM, et al. Glucosenduced mcrovascular functonal changes n nondabetc rats are stereospecfc and are prevented by an aldose reductase nhbtor. J Cln Invest. 1990;85: Thurston JH, McDougal DB, Jr, Hauhart RE, Schulz DW. Effects of acute, subacute, and chronc dabetes on carbohydrate and energy metabolsm n rat scatc nerve: Relaton to mechansms of perpheral neuropathy. Dabetes. 1994; 44: O'Donnell-Tormey J, Nathan CF, Lanks K, DeBoer CJ, de la Harpe J. Secreton of pyruvate: An antoxdant defense of mammalan cells. / Exp Med. 1987; 165: Jaeschke H, Klenwaechter C, Wendel A. NADH-dependent reductve stress and ferrtn-bound ron n allyl alcohol-nduced lpd peroxdaton n vvo: The protectve effect of vtamn E. Chem-Bol Interactons. 1992;81: Dawson TL, Gores GJ, Nemnen A-L, Herman B, LemastersJJ. Mtochondra as a source of reactve oxygen speces durng reductve stress n rat hepatocytes. Amf Physol. 1993;C961-C Tlton RG, Chang K, Faller AM. Probucol prevents vascular proten leakage nduced by dabetes, glucose, and glycated protens. ADA Abstract Dabetes. 1993; (suppl 1):189A. 75. Snclar SH, Grunwald JE, Rva CE, Braunsten SN, Nchols CW, Schwartz SS. Retnal vascular autoregulaton n dabetes melltus. Ophthalmology. 1982; 89: Grunwald JE, Rva CE, Brucker AJ, Snclar SH, Petrg BL. Altered retnal vascular response to 100% oxygen breathng n dabetes melltus. Ophthalmology. 1984; 91: Langan LP, Clark CV, Allaw J, Hll DW, Keen H. Responses of the retnal crculaton to systemc autonomc stmulaton n dabetes melltus. Eye. 1989; 3: Patel V, Rassam SMB, Chen HC, Kohner EM. Oxygen reactvty n dabetes melltus: Effect of hypertenson and hyperglycaema. Cln Sd. 1994; 86: Wllamson JR, Kawamura T, Klo C, et al. Current concepts of the role of polyol metabolsm and related metabolc abnormaltes n dabetc complcatons. In: Sakamoto N, Albert KGMM, Hotta N, eds. Pathogeness and Treatment of NIDDM and ts Related Problems. Amsterdam: Elsever Scence BV; 1994: D'Amore PA. Mechansms of retnal and chorodal neovascularzaton. Invest Ophthalmol Vs Sd. 1994; 35: Aello LP, Avery RL, Arrgg PG, et al. Vascular endothelal growth factor n ocular flud of patents wth dabetc retnopathy and oder renal dsorders. NEngl JMed. 1994;331: Frank RN. The aldose reductase controversy. Dabetes. 1994;43: Dabetes Control and Complcatons Tral (DCCT) Research Group. The effect of ntensve treatment of dabetes on the development and progresson of longterm complcatons n nsuln-dependent dabetes melltus. NEngl JMed. 1993; 329: