Response of Capillary Cell Death to Aminoguanidine Predicts the Development of Retinopathy: Comparison of Diabetes and Galactosemia

Abstract

purpose. To examine the relationship between early retinal capillary cell apoptosis and late histologic lesions of diabetic retinopathy and to compare the effects of aminoguanidine (AMG) on the retinopathies caused by diabetes and galactose feeding.

methods. Rats with alloxan-induced diabetes and rats fed a 30% galactose diet (known to induce diabetic-like retinopathy) were assigned randomly to receive diet with (2.5 g/kg diet) or without AMG. After 6 to 8 months of diabetes or galactosemia, retinal trypsin digests were prepared, and capillary cell apoptosis was quantitated using the Tdt-mediated dUTP nick-end labeling (TUNEL) reaction in association with morphologic evidence of nuclear fragmentation. At 18 months duration, pericyte ghosts and acellular capillaries were quantitated in the isolated vasculature. Several advanced glycation end products (AGEs) were measured at 4 months of study and at 18 months of study by established methods to assess biochemical effects of AMG.

results. As expected, both diabetic and galactosemic rats showed increased frequency of TUNEL-positive capillary cells at 6 to 8 months and vascular lesions characteristic of retinopathy at 18 months. AMG inhibited both the early apoptosis and late histopathology in the diabetic rats, but neither of these abnormalities in the galactosemic rats. In contrast to its preventative effect on retinopathy in the diabetic rats, AMG showed no inhibitory effect on levels of hemoglobin AGE, or tail collagen pentosidine, fluorescence, and thermal breaking time. Diabetes of 4 months’ duration did not cause a detectable increase in retinal levels of several AGEs.

conclusions. The frequency of early apoptosis in retinal microvascular cells predicted the development of the histologic lesions of retinopathy in diabetes as well as in galactosemia. The beneficial effect of AMG on retinal lesions in diabetes is exerted on pathways that are either not operative or are less important in galactosemia and that may not relate to the accumulation of AGEs.

The background stages of diabetic retinopathy are believed to lead to sight-threatening ischemic and proliferative retinopathy when a critical number of retinal capillaries become nonperfused. Histologic analysis of retinal areas that were nonperfused in vivo has indicated that nonperfused vessels are acellular. ¹ How the capillary cells die is unclear, but we have found that both retinal pericytes and endothelial cells undergo accelerated death by a process consistent with apoptosis in humans and animals with diabetes and in nondiabetic animals fed galactose. ²  

The relationship of microvascular cell death to the histologic lesions of diabetic retinopathy is not yet firmly established, but accelerated death of the retinal capillary cells precedes development of the retinopathy. ² We reasoned that if the capillary cell death plays a critical role in the development of the retinopathy, the two processes should have a concordant response to pharmacologic therapy. In particular, because the increase in rate of apoptosis precedes the appearance of microangiopathy, an early effect of therapy on the rate of capillary cell death should predict the late histologic outcome of the retinopathy. 

We tested this proposition using diabetic and galactose-fed rats treated with aminoguanidine (AMG). In rats made diabetic or experimentally galactosemic, a retinal microangiopathy develops that is consistent with the early stages of the retinopathy in patients with diabetes, including the presence of excessive numbers of acellular capillaries. ³ Hammes et al. ^{4 5 6} reported that AMG inhibits the development of retinal lesions in diabetic rats. The beneficial effect of AMG on experimental diabetic retinopathy was attributed to interference with accumulation of advanced glycation end products (AGEs), but the evidence rested solely on semiquantitative in situ measurement of tissue fluorescence, ⁴ which is a nonspecific sequela of multiple oxidative modifications of proteins and lipids. ⁷ We thus measured several specific AGEs in the rats treated with AMG in an effort to correlate the biochemical action of the drug with its effects on apoptosis and retinopathy. After initiation of our study, a study of galactosemic rats reported that AMG had no beneficial effect on galactose-induced retinopathy, ⁸ but assessment of retinopathy in that study was not quantitative, and no data on drug levels or AGEs were presented. 

Materials and Methods

Animals

Sprague–Dawley male rats weighing approximately 200 g were assigned randomly to a diabetic, galactosemic, or control group. Diabetes was induced with alloxan (45 mg/kg body weight, intravenously), and insulin was given as needed (0–2 U of neutral protamine Hagedorn [NPH] insulin subcutaneously, two times per week) to achieve slow weight gain without preventing hyperglycemia and glycosuria. Thus, diabetic rats were insulin-deficient but not grossly catabolic. Experimental galactosemia was induced in nondiabetic rats by feeding rat chow containing 30% galactose. Diabetic rats and galactose-fed rats were assigned randomly to receive control diet or diet supplemented with AMG (2.5 g/kg food). This concentration of AMG was calculated to provide approximately the same amount of drug as that resulting from 0.5 g AMG per liter of drinking water used by other investigators. ⁴ Treatment of animals conformed to the ARVO Statement on Treatment of Animals in Ophthalmic and Vision Research, as well as to specific institutional guidelines. Rats were studied for retinal microvascular cell death at a time when no statistically significant increase in the frequency of vascular lesions is yet detectable ² (8 months of diabetes and 6 months of galactose feeding), and for histopathology at 18 months. An additional cohort of rats was studied after 4 months of diabetes or galactose-feeding. AGEs were measured in tissues of these diabetic animals, and effects of a larger dose of AMG (2 g/l in the drinking water) on retinal microvascular cell death was assessed in these galactose-fed animals. 

Metabolic Indices

Glycohemoglobin (GHb; an estimate of the average level of hyperglycemia over the previous 2–3 months) was measured at least two to three times per year in each animal by affinity chromatography (Glyc-Affin; Pierce, Rockford, IL) after an overnight fast. Twenty-four hour urine volume also was used as a parameter of the severity of glycemia, and was measured over 2 to 3 consecutive days on at least three occasions per year in each animal. Body weight and average daily food consumption were measured weekly. Plasma AMG levels in blood (nonfasted) were measured by Alteon (Ramsey, NJ) by reversed-phase high-performance liquid chromatography (HPLC). 

In Situ Cell Death and Histopathology

The retinal vasculature was isolated by trypsin digestion ³ of the retinas. Cell death was examined in a subset of animals using the Tdt-mediated dUTP nick-end labeling (TUNEL) method (In Situ Cell Death Detection kit; Boheringer–Mannheim, Indianapolis, IN) coupled with morphologic detection of chromatin fragmentation and/or apoptotic bodies. The TUNEL procedure and identification of the fluorescent signals, as well as cellular attribution and counts of TUNEL-positive nuclei, were performed as previously described. ² By matching the fluorescein-specific signals with the corresponding nuclear images viewed in light microscopy after the preparations had been stained with periodic acid Schiff and hematoxylin, TUNEL-positive nuclei were mostly attributed to pericytes and endothelial cells. Twenty percent of the images of TUNEL-positive chromatin were classified as undetermined, because their morphology did not permit confident cellular attribution. The number of TUNEL-positive nuclei is expressed per whole retina. 

The retinal trypsin digests of the animals killed at 18 months were evaluated for vascular histopathology. Acellular capillaries and pericyte ghosts, the empty pockets in the basement membranes at the sites from which pericytes have disappeared, were counted as previously described. ³ The number of acellular capillaries was expressed per square millimeter of retina and the number of pericyte ghosts was reported per 1000 capillary cells. All counts of apoptotic cells and histologic lesions were performed in a masked fashion. 

Measurements of AGEs

Hemoglobin-AGE.

Hemoglobin (Hb)-AGE was measured as an index of intracellular AGE accumulation in rats with 8 to 15 months duration of diabetes or galactosemia. ⁹ Measurements were performed by Alteon by immunoassay. 

AGEs in Tail Tendon Collagen.

Chemically characterized AGEs (pentosidine ¹⁰ and the newly identified products of α-dicarbonyl reaction with proteins methylglyoxal-lysine dimer [MOLD] and glyoxal-lysine dimer[ GOLD] ¹¹ ) and protein-bound fluorescence were measured in tail collagen at the 18-month time point. Minced tail tendons were homogenized and digested with 1% collagenase (Clostridium histolyticum, type VII) for 16 hours. Fluorescence was measured in the digested material at excitation–emission wavelengths of 370/440 nm. Collagenase-digested material from tail tendon was subjected to acid hydrolysis in 6 N HCl at 110°C for 16 hours and was used for pentosidine measurement by a two-step HPLC method. ¹² MOLD and GOLD were measured by the HPLC assay described by Chellan and Nagaraj. ¹¹ The limit of detection in the assay for MOLD and GOLD was 0.5 picomoles. 

Tail tendon breaking time was measured as an index of cross-linking in rats with diabetes of 8 or 18 months’ duration. The measurement was performed on individual fibers of tail collagen using a modification of the method described by Monnier et al. ¹³ Briefly, five individual fibers of tail collagen were immersed in 7 M urea (40°C) with weight pulling the fibers down. The number of minutes until individual fibers broke was measured, and the middle three values were averaged (highest and lowest values were excluded). The weight had to be increased to 5.3 g for the 8-month old animals for the fibers to break in a reasonable amount of time. Collagen fibers from animals in the 18-month experiment did not break reproducibly by this method, even after increasing the weight hanging on the fibers, the temperature of the urea, or the incubation duration; therefore, no results will be reported for this group. 

AGEs in Retina.

Methylglyoxal-derived modification (MG-AGEs) of retinal proteins was measured by enzyme-linked immunosorbent assay (ELISA) ¹⁴ in retinal homogenates of rats with 4 months’ duration of diabetes (n ≥ 5 per group). The antibody generated against methylglyoxal-modified RNase A recognizes MOLD and argpyrimidine in addition to other uncharacterized AGEs, but not Nε-carboxymethyllysine. ¹⁴ One unit of MG-AGEs was defined as the amount responsible for 1% inhibition of antibody binding to the ELISA well. 

Statistical Analysis

The data are summarized and expressed as means ± SD. The five experimental groups were compared with the nonparametric Kruskal–Wallis test followed by Mann–Whitney tests. Analysis of variance (ANOVA) followed by Fisher’s multiple comparison tests yielded similar conclusions. 

Results

Metabolic Characteristics and AMG Levels

GHb and urine volumes were significantly greater in the diabetic and galactosemic rats, whether treated or not treated with AMG, than in control rats (Table 1 ; all P < 0.001). Similarly, blood glucose concentration was elevated in all diabetic groups (blood galactose was not measured in galactose-fed animals) and was not affected by AMG. The lower GHb and urine volumes in the galactosemic rats, when compared with the diabetic rats, were due most likely to blood hexose level being elevated only after meals in galactose-fed rats, in contrast to the sustained blood glucose elevation in the diabetic rats. Diabetic rats were hyperphagic, and their food consumption was greater than control or galactose-fed rats. Plasma levels of AMG in treated animals remained elevated throughout the day and were approximately similar in the diabetic and galactosemic animals (in diabetic rats, 8 AM and 6 PM concentrations were 0.54 ± 0.28 μg/ml and 0.65 ± 0.26μ g/ml, respectively; in galactose-fed rats, 8 AM and 6 PM concentrations were 0.50 ± 0.35 μg/ml and 0.30 ± 0.21μ g/ml, respectively). 

In an additional experiment, galactose-fed rats were treated with 2 g/l AMG in the drinking water for 4 months. GHb and body weight when rats were killed were, respectively, 3.7% ± 0.6% and 614 ± 67 g in control rats, 6.3% ± 0.5% and 508 ± 32 g in the galactose-fed rats, and 6.0% ± 1.0% and 440 ± 29 g in the galactose-fed rats receiving AMG. Plasma levels of AMG in these animals (8 AM; 2.1 ± 0.9 μg/ml) were four times higher than those observed with the AMG-supplemented diet reported above. 

Effects of AMG on Retinal Vascular Apoptosis and Histopathology

The retinal vessels of rats with 8 months’ duration of diabetes showed, as reported previously, ² a significantly greater number of TUNEL-positive cells than that observed in nondiabetic control rats (P < 0.01; Fig. 1 ). AMG administration prevented the diabetes-induced increase in TUNEL-positive pericyte nuclei (0.6 ± 0.9 versus 4.2 ± 3.1 per whole retina in the untreated diabetic group, P = 0.04), endothelial cell nuclei (0 versus 2.4 ± 2.3 in the untreated group, P = 0.003), and total nuclei (inclusive of pericyte, endothelial cell, and nuclei with undetermined cellular attribution; 0.8 ± 1.3 versus 8.8 ± 6.5 in the untreated group, P < 0.01). Diabetes of 18 months’ duration caused an increased number of acellular capillaries (19.6 ± 4.5/mm² retina versus 4.4 ± 3.0 in controls, P = 0.0001; Fig. 1 ), the development of which was completely prevented by AMG (6.3 ± 4.2/mm² retina; P = 0.0001 versus untreated diabetic rats). AMG also reduced the number of pericyte ghosts (5.6 ± 2.6/1000 capillary cells versus 12.8 ± 3.5/1000 cells in untreated diabetic rats; P = 0.0001), but the number remained larger than in control rats (1.4 ± 1.4; P < 0.01). 

The retinal vessels of rats with 6 months’ duration of experimental galactosemia showed more TUNEL-positive cells than the vessels of control rats (P < 0.05), and the number was not decreased in the AMG-treated group (Fig. 2) . Identical results were obtained in the additional cohort of galactosemic rats to which AMG was administered at a higher dose in drinking water. In these rats, 4 months of galactose feeding were sufficient to increase the number of TUNEL-positive retinal microvascular cells (5.8 ± 2.6 versus 0.7 ± 0.9 per whole retina in control rats, P < 0.02), and AMG did not prevent the increase (8.8 ± 7.2 per whole retina, P < 0.05 versus control rats) despite blood levels four times higher than those achieved with AMG added to the diet. AMG was ineffective also in protecting galactosemic rats from retinopathy. In the rats fed the galactose-rich diet for 18 months, AMG did not reduce the excess number of acellular capillaries (Fig. 2) and the large increase in pericyte ghosts (12.3 ± 4.7 versus 14.4 ± 5.0 in untreated galactosemic rats). 

Effects of AMG on AGE Levels

Hb-AGE, as well as tail collagen pentosidine, fluorescence, and thermal breaking time were all significantly greater than normal in diabetic rats (all P < 0.001; Fig. 3 ), and the level of AMG achieved in our diabetic rats did not reduce these parameters. Levels of MOLD and GOLD in tail collagen were below the lower limit of detection of the assay. Among the AGEs measured in the galactosemic rats, only tail collagen fluorescence showed a modest elevation (not significant), whereas Hb-AGE and pentosidine levels tended to be lower than normal, although not significantly so in this small sample (Fig. 3) . 

In the retina, diabetes of 4-months’ duration did not increase the levels of pentosidine (0.10 ± 0.08 picomole/micromole leucine versus 0.16 ± 0.14 in control rats) or methylglyoxal-modified protein (0.14 ± 0.04 versus 0.11 ± 0.07 arbitrary units/micromole leucine in control rats), and MOLD and GOLD were not detected. 

Discussion

In this study, AMG inhibited the accelerated death of retinal capillary cells and development of retinopathy in diabetic rats. This relationship between accelerated microvascular cell death and retinopathy is further strengthened by the finding that these two abnormalities also showed a concordant response to AMG in galactose-fed animals. In galactose-fed animals, however, neither cell death nor retinopathy was affected by AMG. Thus, in diabetes as well as in galactosemia, the rate of microvascular cell death predicted whether retinopathy eventually developed. We postulate that capillary cell death plays an important role in the development of the retinopathy. The mechanism by which AMG exerted its beneficial effects in diabetes is not clear, because the effect on retinopathy was not associated with changes in several well-characterized indices of advanced glycation. A similar inhibition of retinopathy by AMG has been observed recently by us in studies of dogs given the drug for 5 years. ¹⁵  

As observed previously in the retinal vessels of both diabetic and galactosemic rats, ² the increased number of cells positive in the TUNEL reaction showed chromatin fragmentation or formation of apoptotic bodies, consistent with an apoptotic mode of death. Furthering the concept that diabetes causes a proapoptotic environment in the retina, Li et al. ¹⁶ found overexpression of a member of the interleukin-1β–converting enzyme family in human diabetic pericytes, and we recently observed increased levels of the proapoptotic protein Bax in the retina and dying pericytes of human eye donors with diabetes. ¹⁷ Although only the use of interventions that selectively inhibit events unique to the apoptotic process will ultimately establish the role of apoptosis in retinal capillary obliteration, the growing body of evidence favors such a role. Insofar as pericytes show minimal if any replicative capabilities in the adult retina, ¹⁸ their accelerated apoptosis can readily account for the pericyte dropout and formation of ghosts in diabetic retinopathy. The proinflammatory ¹⁹ and procoagulant ²⁰ characteristics of apoptotic endothelial cells may be a trigger for occlusive events eventually leading to obliteration of capillaries. 

Apoptosis and the morphologic manifestations of retinopathy were present both in diabetes and galactosemia, but the divergent effects of AMG on the two retinopathies suggest dissimilar events upstream of apoptosis. In the retinopathy of diabetic rats, there is an AMG-sensitive step that apparently is absent in galactosemic rats. AGEs and their Amadori precursors tend to be formed in greater quantities in diabetes than in galactosemia, as exemplified in this study by the values of pentosidine, Hb-AGE, and GHb, and it therefore seemed reasonable to suspect that the AMG-sensitive step in diabetic rats might be the accumulation of AGEs. The accumulation of representative AGEs in diabetic rats, however, was not prevented by AMG as administered in our study. 

The apparent discrepancy with previously reported inhibitory effects of AMG on formation of Hb-AGE, ⁹ pentosidine, ²¹ collagen cross links, ²² and fluorescent compounds ^{4 23} may have several explanations. First, the blood levels of AMG achieved in our diabetic rats were on the lower side of the range reported in rodents treated with the drug, ^{4 24} and may not have been sufficient to inhibit AGE formation. Rigorous comparison of the relationship between blood concentration and tissue effects of AMG in different studies is made difficult by the fact that blood levels of the drug have been measured and/or reported very seldom, and that such levels show large variations even in rats treated with similar doses of the drug by the same investigators. ^{4 24} Second, some studies showing an effect of AMG on AGE levels may have overestimated the degree of inhibition because they were of short duration. Booth et al. ²⁵ have reported that in vitro AMG decreased the rate of AGE formation but not the final amount of AGE formed, most likely because the drug inhibits the late kinetic stages of glycation much less efficiently than the early stage. It is also of note that in several long-term studies showing beneficial effects of AMG on sequelae of diabetes, inhibition of AGE accumulation by AMG was either not observed for all AGE species or all tissues, ^{21 26} or was mimicked by compounds (such as methylguanidine ²⁷ ) not expected to inhibit advanced glycation. ²⁸ Consistent with our findings, Degenhardt et al. ²⁹ have reported recently that AMG inhibits albuminuria in diabetic rats without inhibiting the formation of AGEs in skin collagen. Therefore, beneficial effects of AMG on the complications of diabetes do not necessarily correlate with its inhibition of parameters of AGE accumulation. 

Two caveats in the interpretation of our results are that AGEs in the rats studied for the long-term were not measured in the retina, and that only some AGE species were tested. Hammes et al. ^{4 6} attempted to assess effects of AMG on retinal AGEs by quantitating in situ fluorescence of retinal arterioles at wavelengths characteristic of AGEs. These wavelengths, however, are not specific for AGEs, and likely include also a myriad of oxidation products. Because we and others have found AMG to inhibit oxidative stress and other biochemical processes, ^{28 30 31 32 33 34 35 36 37} ascribing the reduction of fluorescence in AMG-treated diabetic rats to inhibition of AGEs may not be justified. It remains possible that AMG may achieve in the retina concentrations higher than in other tissues and/or may be more effective on AGEs other than those tested to date. 

The finding that AMG was able to effectively prevent experimental diabetic retinopathy without inhibiting accumulation of Hb-AGEs (intracellular) and extracellular AGE has several implications. One is that the beneficial effect of AMG on diabetic retinopathy may be mediated by one or more of the many actions of the drug that are unrelated to inhibition of AGEs. ^{28 30 33 35 36 37} AMG treatment prevents biochemical manifestations of retinal oxidant stress (as measured by accumulation of thiobarbituric acid-reactive substances) and activation of protein kinase C, ³⁷ but these effects were similar in diabetic and galactose-fed rats, and therefore may not be related to the unique AMG-sensitive step in the development of diabetic retinopathy. A second intriguing implication that has possible clinical relevance pertains to the systemic drug levels found to inhibit the retinopathy; retinopathy was inhibited in diabetic rats (present study and Reference 4) at plasma levels substantially lower than those currently achieved in clinical studies. Use of the drug at low doses may inhibit diabetic retinal microangiopathy while lessening drug-induced side effects. 

 Supported by Public Health Service Grants EY00300 and EY09122, Juvenile Diabetes Foundation International Grant 196075, and the George and Frances Levin Endowment. MM was the recipient of a fellowship from the Manpei Suzuki Diabetes Foundation, Tokyo, Japan.

 Submitted for publication March 13, 2000; revised May 18, 2000; accepted May 26, 2000.

 Commercial relationships policy: N.

 

Table 1.

 

Characteristics of Experimental Groups throughout 18 Months of Diabetes or Galactosemia

	n	GHb (%)	Fasting Blood Glucose (mg/dl)	24-Hour Urine Volume (ml)	Final BW (g)	Daily Food Intake (g/day)
Normal	14	3.7 ± 0.2	62 ± 9	8 ± 4	576 ± 111	28 ± 4
Diabetes
Control	9	13.5 ± 1.4	275 ± 33	106 ± 22	292 ± 65	47 ± 9
+AMG	9	12.7 ± 2.0	312 ± 65	113 ± 31	298 ± 49	54 ± 10
Galactosemia
Control	13	5.4 ± 0.5	67 ± 7	33 ± 8	447 ± 41	28 ± 3
+AMG	15	5.5 ± 0.7	66 ± 9	32 ± 6	431 ± 23	31 ± 5

Figure 1.

 

Effect of AMG on retinal microvascular cell death (left) and acellular capillaries (right) in rats with diabetes of 8 and 18 months’ duration, respectively. Microvascular cell death was studied with the TUNEL reaction on retinal trypsin digests. N, normal rats; D, diabetic rats; D+A, diabetic rats receiving AMG. Each bar represents the mean ± SD of the number of TUNEL-positive cells or acellular capillaries counted in the number of rats indicated in parentheses. *P < 0.01 versus control rats;** P < 0.01 versus untreated diabetic rats.

Figure 2.

 

The absence of effect of AMG on retinal microvascular cell death (left) and acellular capillaries (right) in rats with galactosemia of 6 and 18 months’ duration, respectively. N, normal rats; G, galactose-fed rats; G+A, galactose-fed rats receiving AMG. *P < 0.05 versus control rats. The ineffectiveness of AMG in preventing microvascular cell death in galactose-fed rats was confirmed also in an additional cohort of rats.

Figure 3.

 

The absence of effect of AMG on AGEs (Hb-AGE, tail collagen pentosidine, fluorescence, and breaking time) in diabetic and galactose-fed rats. These parameters were measured at durations of diabetes or galactosemia of 8 to 15 months, 18 months, 18 months, and 8 months, respectively.

The authors thank Vincent Monnier for the method to measure collagen breaking time.