Endothelial function can be modulated by acute hyperglycemia

Abstract

Endothelial dysfunction is a known consequence of chronic hyperglycemia and a major pathogenic factor for microvascular diseases such as diabetic retinopathy. The effect and exposure period to acute hyperglycemia on the vascular endothelium, and its ability to recover from such exposure is less well understood. Here, we used an isolated perfused eye preparation to study the effect of acute hyperglycemia on the ocular microvascular endothelium of normoglycemic and 1-, 2-, 3-, and 4-week streptozotocin (STZ)-induced diabetic rats. The acetylcholine (Ach)-induced vasodilatory response was measured during sequential exposure to 6 mM (normoglycemic), 12 mM (mild hyperglycemic), 24 mM (hyperglycemic) and then again to 6 mM (normoglycemic) perfusates. Eyes were then processed histologically for examination of capillary density, capillary diameter, pericyte distribution, endothelial nitric oxide synthase (eNOS) distribution and accumulation of advanced glycated end products (AGEs). Ach responses were significantly enhanced in normoglycemic, 1-,2-, 3- and 4-week diabetic eyes after less than 2-h of high glucose (24 mM) exposure. Upon return to normoglycemia, Ach-induced responses in 1-, 2- and 3-week diabetic eyes were comparable with normoglycemic eyes. In the 4-week diabetic eyes, Ach-induced vasodilatory responses remained significantly enhanced despite restoration of normoglycemia. Capillary density and capillary diameter did not change significantly after 1-, 2-, 3- and 4-weeks of STZ-induced diabetes. Pericyte distribution significantly increased in all vascular layers of the 3- and 4-week diabetic eyes. eNOS immunoreactivity significantly increased in 1-week diabetic eyes. Significant AGEs immunolabelling was detected in the vascular basement membrane and intracellularly in 1-week STZ rats. These findings suggest that whilst endothelial function can recover after short term hyperglycemia, prolonged exposure (≥ 4 weeks) results in incomplete restoration, indicating a potential transition towards irreversible dysfunction.

Introduction

Microvasculature and cellular metabolism are closely linked such that the relationship between hyperglycemia and microvascular dysfunction is proposed to be bidirectional and constitutes a vicious cycle¹. Microvascular damage resulting from hyperglycemia may contribute to the pathogenesis of diabetes and its complications such as diabetic retinopathy and diabetic kidney diseases^2–5. Hyperglycemia is known to alter cardiovascular function as demonstrated by enhanced endothelia-mediated dilation⁶. However, the exact effects of acute hyperglycemia on the vasculature, particularly microvasculature, are remarkably varied depending on the type of vessel, duration and methods of induction of hyperglycemia and the response to endothelia-mediated agents^2,7. The reversibility of altered endothelial function upon correction of acute hyperglycemia in diabetes is also poorly understood⁷.

Acute increase in blood glucose can occur after consumption of energy drinks⁸, after a meal—postprandial, or induced by stress^9,10. Stepwise increase/concentration-dependent induction of endothelial dysfunction by acute hyperglycemia has been demonstrated in normal and diabetic subjects using flow mediated dilatation technique on the brachial artery¹¹. It is unclear whether a similar level of endothelial dysfunction also takes place at microvascular level in response to acute hyperglycemia. It is of interest to know to what extent the microvascular function is affected by acute hyperglycemia as the consequences of advanced microvascular dysfunction such as retinopathy, nephropathy and neuropathy can be dire. It is therefore also of interest to know whether such endothelial dysfunction, if present, is recoverable upon returning to normoglycemia.

The retinal endothelium plays a key role in retinal homeostasis and endothelial dysfunction is a major pathogenic factor in the development of diabetic retinopathy^3,12. Some of the functions of the retinal endothelia include local sensing of hemodynamic forces, local production of vasoactive factors, the mediation of responses to circulating vasoactive factors and direct control of local blood flow via vasomotor tone control¹³. Nitric oxide (NO), produced by endothelial cells through the endothelial nitric oxide synthase (eNOS), plays a major role for many of these endothelial functions; however, endothelial dysfunction can result in disruptions of eNOS expression, reduced NO production and bioavailability¹⁴.

Formation of advanced glycation end products (AGEs) is commonly associated with the onset and progression of diabetic retinopathy¹⁵. AGEs are the permanent product of nonenzymatic glycation of proteins and lipids; forming in the vascular basement membrane and then accumulating intracellularly¹⁶. They are responsible for thickening of capillary basement membrane, quenching the bioactivity of NO and its impairing vasodilatory function, breakdown of the blood-retinal-barrier and formation of acellular and occluded capillaries^15–17. AGE’s have previously been demonstrated to accumulate in a time-dependent fashion in the retina of diabetic rats, with progressive increases observed at 2, 4 and 8 months of diabetes duration¹⁸. It is unclear whether AGEs are already starting to accumulate from the onset of diabetes.

The vasodilatory response of endothelial cells is a reliable surrogate marker of endothelial function/dysfunction^19,20. In this study, we used an arterially perfused isolated eye preparation to examine the effect of hyperglycemia on Ach-induced endothelial-mediated vasodilatation. The isolated perfused eye technique is advantageous in being tissue organ specific and free from systemic influences²¹. In this study we examined the extent of Ach-dependent endothelial vasodilatory function affected by different levels and exposure periods of hyperglycemia, and whether such effects can be restored upon return to normoglycemia. Histological analysis of the retinal vasculature was performed to assess capillary density, capillary diameter, pericyte distribution, eNOS distribution and accumulation of AGEs. We provide new information regarding the acute vulnerability of retinal endothelial cells to hyperglycemia-induced functional changes, the ability of these cells to recover function following reversal of acute hyperglycemia and the cellular responses to prolonged hyperglycemia- including pericyte changes that precede overt vascular remodeling. Specifically, we provide valuable information regarding the importance of normoglycemia for maintaining endothelia function within the ocular vasculature. This study aids the understanding of diabetic retina disease, which remains one of the most common causes of blindness worldwide²².

Results

Body weight and glucose

Mean body weight and non-fasting blood glucose levels for each study group are summarized in Table 1.

Table 1.

Summary of weight and unfasted blood glucose of control and STZ-induced diabetic rats.

	n	Body weight (g)	Blood glucose (mM)
Combined control	11	388.36 ± 95.41	8.40 ± 1.24
9-week-old	6	305.83 ± 31.42	8.05 ± 1.48
12-week-old	5	413.42 ± 29.29	8.88 ± 0.67 
1-week STZ	9	311.14 ± 11.01	31.66 ± 7.02*
2-week STZ	5	311.47 ± 9.97	34.46 ± 6.29*
3-week STZ	5	330.92 ± 12.41	30.45 ± 6.29*
4-week STZ	5	365.88 ± 13.72	30.53 ± 7.25*

Significant values are in bold.

*P < 0.05 between control and specific time-point of STZ.

Endothelial dependent vasodilatory responses during and after exposure to acute hyperglycemia in control eyes

To determine the comparability of vasoactive responses between normoglycemic rats of two age groups, we measured and compared the acetylcholine-induced dose-dependent responses of 10 eyes, from 5 normoglycemic rats at 9 weeks of age with those of 12 eyes, from 6 normoglycemic rats at 12 weeks of age. Two-way ANOVA of the dose-dependent response curves showed no statistically significant difference at any glucose concentration-baseline 6 mM (P = 0.545), 12 mM (P = 0.586), 24 mM (P = 0.932) or final 6 mM (P = 0.864) glucose perfusate (Supplementary Fig. S1A–D). Therefore, normoglycemic rats from both age groups were pooled to form the control group used for all comparative analyses.

Compared to the Potassium Ringers precontracted basal pressure, Ach induced a vasodilatory response at all concentrations of Ach boluses tested in a dose-dependent manner, irrespective of glucose concentration in the perfusate. Two-way ANOVA of the dose–response curves showed a significantly enhanced response to Ach during exposure to 12 mM and 24 mM glucose perfusate, compared to baseline 6 mM glucose concentration (P = 0.002 and P < 0.001, respectively; Fig. 1A). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant increase in Ach-induced vasodilatory response at all concentrations of Ach (P < 0.05). The greatest enhancement in vasodilatory response was between 6 and 24 mM glucose at 10⁻⁴ M Ach with the respective percentage of perfusion pressure averaging 83.12 ± 4.1% and 72.78 ± 5.4% of the pre-contracted level, respectively (P < 0.001; Fig. 1A). Re-exposure to 6 mM glucose after acute exposure to 12 mM and 24 mM glucose found comparable, endothelium-dependent vasodilatory responses to 6 mM glucose at the beginning of the test series (P = 0.324; Fig. 1A).

Fig. 1.

Acetylcholine-induced vasodilatation in the ocular vasculature of control and 1-, 2- 3- and 4-week STZ-induced diabetic rats during acute exposure to at different glucose concentrations. Ocular vascular responses to Ach at increasing log Molar concentrations were recorded during the initial perfusion with 6 mM glucose (baseline), then during 12 mM glucose perfusate, then 24 mM glucose and then back to 6 mM glucose perfusate (final). Two-way repeated measures ANOVA considers the dose-responses of all Ach concentrations tested. Statistical significance between dose–response curve of different groups is denoted by an asterisk [*] and a bracket indicating the groups compared. (A) Significantly larger Ach-induced vasodilatory responses were observed during exposure to 12 mM and 24 mM glucose, compared to baseline 6 mM glucose. Upon re-exposure to 6 mM glucose, the Ach-induced dose responses returned to baseline level. (B) In the 1-week STZ rats, significant increase in Ach-induced vasodilatory responses were observed only during 24 mM glucose exposure when compared to baseline values. Ach-induced vasodilatory responses returned to baseline levels, following normalization of hyperglycemia. (C) In the 2-week STZ rats, there was a statistically significant difference between baseline 6 mM and 12 mM glucose; and baseline 6 mM and 24 mM glucose. (D) In the 3-week STZ rats, there was a statistically significant difference between baseline 6 mM and 24 mM glucose; and baseline and final 6 mM glucose. (E) In the 4-week STZ rats, there was a statistically significant difference between baseline 6 mM and 12 mM glucose; baseline 6 mM and 24 mM glucose; and baseline and final 6 mM glucose. Data presented as the normalized percentage change in perfusion pressure ± SEM.

Endothelial dependent vasodilatory responses during and after exposure to acute hyperglycemia in 1-, 2-, 3- and 4-week diabetic eyes

We examined the impact of 1-, 2-, 3- and 4-week STZ-induced hyperglycemia exposure on the Ach-induced vasodilatory response. In 1-week STZ rats, perfusion with 12 mM glucose Potassium Ringers perfusate did not induce a significant change in the overall dose-dependent Ach induced vasodilatory response compared to baseline 6 mM glucose exposure (P = 0.219; Fig. 1B). Two-way RM ANOVA of the dose responses curves showed a significantly enhanced response to Ach during exposure to 24 mM glucose perfusate, compared to baseline 6 mM glucose concentration (P < 0.001; Fig. 1B). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant increase in Ach-induced vasodilatory response at all concentrations of Ach (P < 0.05). Re-exposure to 6 mM glucose after acute exposure to 12 mM and 24 mM glucose found comparable, endothelium-dependent vasodilatory responses to 6 mM glucose at the beginning of the test series (P = 0.331; Fig. 1B).

In the 2-week STZ rats, acute exposure to 12 mM and 24 mM glucose significantly enhanced the overall dose-dependent Ach induced vasodilatory response, compared to baseline 6 mM glucose exposure (P = 0.002 and P < 0.001, respectively; Fig. 1C). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant increase in Ach-induced vasodilatory responses at all concentrations of Ach (P < 0.05). The greatest enhancement in vasodilatory response was between 6 and 24 mM at 10⁻¹⁰ M Ach with the respective percentage of perfusion pressure averaging 82.26 ± 2.83 and 74.06 ± 2.4% of the pre-contracted level, respectively (P < 0.001; Fig. 1C). Re-exposure to 6 mM glucose after acute exposure to 12 mM and 24 mM glucose found comparable, endothelium-dependent vasodilatory responses to 6 mM glucose at the beginning of the test series (P = 0.413; Fig. 1C).

In the 3-week STZ rats, acute exposure to 24 mM glucose significantly enhanced the overall dose-dependent Ach induced vasodilatory responses compared to baseline 6 mM glucose exposure (P < 0.001; Fig. 1D). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant increase in Ach-induced vasodilatory responses at all concentrations of Ach (P < 0.05). Re-exposure to 6 mM glucose after acute exposure to 12 mM and 24 mM glucose found significantly attenuated endothelium-dependent vasodilatory responses compared to baseline pre high glucose exposure (P < 0.001; Fig. 1D). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant decrease in Ach-induced vasodilatory responses at all concentrations of Ach (P < 0.05).

In the 4-week STZ group, acute exposure to 12 mM and 24 mM glucose significantly enhanced the Ach induced vasodilatory responses, compared to baseline 6 mM glucose exposure (all P < 0.001; Fig. 1E). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant increase in Ach-induced vasodilatory responses at all concentrations of Ach (P < 0.05). Re-exposure to 6 mM glucose, found significantly attenuated Ach induced vasodilatory responses compared to baseline pre high glucose exposure (P < 0.001). Bonferroni’s post-hoc analysis of individual Ach concentrations showed a statistically significant decrease in Ach-induced vasodilatory responses at all concentrations of Ach (P < 0.05).

Comparison of acute hyperglycemia-induced changes in acetylcholine-mediated vasodilatory responses between control and diabetic rats

We compared the Ach-induced vasodilatory responses of the 1-, 2-, 3- and 4-week STZ-induced diabetic rat eyes to control eyes for each respective perfusate glucose level. Dose dependent Ach-induced vasodilation was observed in diabetic eyes as in control eyes. In the 1-week STZ group, comparison of the overall vasodilatory responses during exposure to 6 mM and 12 mM glucose found comparable dose responses in the diabetic and control rats (P = 0.167, 0.074, respectively; Fig. 2A,B). Exposure to 24 mM glucose perfusate found significantly larger Ach-induced vasodilatory dose responses in the 1-week diabetic eyes (P = 0.015, Fig. 2C). Bonferroni’s post-hoc analysis showed this effect to be true for all Ach concentrations at 24 mM glucose (P < 0.05). Re-exposure to 6 mM glucose after acute hyperglycemic exposure found comparable, endothelium-dependent vasodilatory responses between the control and 1-week STZ eyes (P = 0.606; Fig. 2D).

Fig. 2.

Comparison of Ach-induced perfusion pressure changes between control and 1-, 2-, 3 or 4-week STZ-induced diabetic rat eyes during perfusion with (A) 6 mM glucose (baseline) (B) 12 mM glucose (C) 24 mM glucose and (D) 6 mM glucose (final) in the perfusate. Statistical significance between dose–response curve of different groups is denoted by an asterisk [*] and a bracket indicating the groups compared. Two-way ANOVA considers the dose-responses of all Ach concentrations and found a statistically significant difference between the control and 1-week STZ rats at 24 mM glucose only; there was a significant difference between the control and 2-week STZ rats at 12 mM glucose only. There was a statistically significant difference between the control and 3-week STZ rats at baseline 6 mM, 24 mM and final 6 mM glucose. There was a statistically significant difference between the control and 4-week STZ rats at baseline 6 mM, 12 mM, 24 mM and final 6 mM glucose. Data presented as the normalized percentage change in perfusion pressure ± SEM.

Comparison of the overall vasodilatory responses during acute exposure to baseline 6 mM, 12 mM, 24 mM and final 6 mM glucose perfusate found comparable Ach-induced dose responses between the control and 2-week STZ eyes (P = 0.428, 0.065, 0.216 and 0.635 respectively; Fig. 2A–D).

Ach induced vasodilatory responses in the 3-week STZ group were significantly greater during exposure to baseline 6 mM glucose perfusate, compared to the control group (P = 0.005; Fig. 2A). Bonferroni’s post-hoc analysis showed this effect to be true for all Ach concentrations (P < 0.05). Comparison of the overall vasodilatory responses during exposure to 12 mM glucose found comparable dose responses between the control and 3-week STZ rats (P = 0.523; Fig. 2B). Exposure to 24 mM glucose found significantly larger Ach-induced vasodilatory response in the 3-week STZ group (P < 0.001; Fig. 2C). Post-hoc analysis showed this effect to be true for all Ach concentrations (P < 0.05). Re-exposure to 6 mM glucose after acute exposure to 12 mM and 24 mM glucose found comparable, endothelium-dependent vasodilatory responses between the control and 3-week STZ rats (P = 0.642; Fig. 2D).

In the 4-week STZ group, Ach-induced vasodilatory responses were significantly greater than the control group at baseline 6 mM, 12 mM and 24 mM glucose concentrations (all P < 0.001; Fig. 2A–C). Post-hoc analysis showed this to be true for all Ach concentrations (P < 0.05). Re-exposure to 6 mM glucose after acute exposure to 12 mM and 24 mM glucose found significantly enhanced, endothelium-dependent vasodilatory responses in the 4-week STZ group compared to controls (P < 0.001; Fig. 2D). Bonferroni’s post-hoc analysis showed this effect to be true for all Ach concentrations (P < 0.05).

Histological analysis

Capillary density

When measurements across all three retinal plexuses were analyzed irrespective of disease state, capillary density was greatest in the DCP (control: 23.7 ± 1.19%, 1-week STZ: 22.78 ± 0.15%; 2-week STZ: 20.50 ± 0.50%; 3-week STZ: 21.75 ± 0.43%; 4-week STZ: 22 ± 0.71%), followed by the SVP (control: 20.25 ± 0.66%, 1-week STZ: 20.55 ± 0.15%; 2-week STZ: 20 ± 0.71%; 3-week STZ: 20.5 ± 0.50%; 4-week STZ: 20.75 ± 0.83%) and ICP (control: 11.12 ± 1.05%, 1-week STZ: 11 ± 0.11%; 2-week STZ: 10.75 ± 0.43%; 3-week STZ: 10.25 ± 0.43%; 4-week STZ: 10.25 ± 0.43%). There was no significant difference in capillary density between control and 1-, 2-, 3- or 4-week STZ rats in any plexus (all P > 0.05; Fig. 3A–G).

Fig. 3.

Morphometric comparisons of capillary density, capillary diameter and pericyte distribution between control and 1-, 2-, 3- or 4-week STZ-induced diabetic rats. Representative confocal images of the (A, D) superficial, (B, E) intermediate and (C, F) deep vascular layers labelled using lectin (green), anti-αSMA (red) and Hoechst (blue). (G) Capillary density and (H) capillary diameter did not change significantly after 1-, 2-, 3- or 4- weeks of STZ-induced diabetes. (I) Pericyte count significantly increased in the SVP, ICP, and DCP after 3- and 4-weeks of STZ induced diabetes. Asterisk (*) indicates statistical significance (Unpaired t-test, P < 0.05). Data presented as the mean ± SD. Scale bar: 50 µm.

Capillary diameter

Mean capillary diameter was largest in the SVP (control: 7.3 ± 0.14 µm; 1-week STZ: 7.24 ± 0.17 µm; 2-week STZ: 7.12 ± 0.06 µm: 3-week STZ: 7.21 ± 0.02 µm; 4-week STZ: 7.22 ± 0.04 µm) followed by the DCP (control: 7.32 ± 0.07 µm, 1-week STZ: 7.34 ± 0.11 µm; 2-week STZ: 7.19 ± 0.05 µm; 3-week STZ: 7.24 ± 0.04 µm; 4-week STZ: 7.25 ± 0.05 µm) and ICP (control: 7.11 ± 0.09 µm, 1-week STZ: 7.19 ± 0.14 µm; 2-week STZ: 7.01 ± 0.02 µm; 3-week STZ: 7.03 ± 0.09 µm; 4-week STZ: 7.04 ± 0.12 µm). There was no significant difference in capillary diameter between control and 1-, 2-, 3- or 4-week STZ rats in any plexus (P > 0.05; Fig. 3A–H).

Comparison of pericyte distribution between control and diabetic rats

Irrespective of disease state, pericytes were present in all retinal vascular plexuses (Fig. 3A–F). The SVP (control: 38.62 ± 1.86 per 318 µm², 1-week STZ: 38.12 ± 1.05 per 318 µm²; 2-week STZ: 42.25 ± 0.83 per 318 µm²; 3-week STZ: 46.12 ± 0.31 per 318 µm²; 4-week STZ: 54.75 ± 1.85 per 318 µm²) and DCP (control: 38.25 ± 2.48 per 318 µm², 1-week STZ: 38.63 ± 0.67 per 318 µm²; 2-week STZ: 42.75 ± 1.29 per 318 µm² 3-week STZ: 48.64 ± 0.75 per 318 µm²; 4-week STZ: 59.25 ± 0.81 per 318 µm²) had similar pericyte counts. The ICP had the least number of pericytes (control: 20.25 ± 0.83 per 318 µm², 1-week STZ: 19.63 ± 1.40 per 318 µm²; 2-week STZ: 21.75 ± 1.08 per 318 µm²; 3-week STZ: 29.84 ± 1.02 per 318 µm²; 4-week STZ: 31.25 ± 1.45 per 318 µm²). There was no significant difference in pericyte count between control and 1- or 2-week STZ rats in any plexus (all P > 0.05; Fig. 3I). After 3 weeks of STZ-induced diabetes, pericyte count significantly increased in all vascular plexuses, with the greatest increase observed in the DCP (all P < 0.001; Fig. 3I). Pericyte count was also significantly greater in the SVP, ICP and DCP of the 4-week STZ rats, compared to controls, with the greatest increase observed in the DCP (all P < 0.001; Fig. 3I).

Fig. 5.

Immunodetection of AGEs in control and 1-week STZ-induced diabetic rats. Representative confocal images of the (A, B) SVP, (C, D) ICP, and (E, F) DCP, perfusion labelled for lectin (green) shown in the left panels and AGEs (yellow) shown in the right panels. (G) Immunolabelling of AGE was only detected in the 1-week STZ rats, in all vascular layers and vessel orders; localized to the vascular basement membrane and intracellularly. Asterisk (*) indicates statistical significance (Unpaired t-test, P < 0.05). Data presented as the percentage of AGE positive area ± SD. Scale bar: 50 µm.

Comparison of eNOS distribution between control and diabetic rats

Immunolabelling of eNOS was present in all layers and vessel orders of the retinal vasculature (Fig. 4A–F). Irrespective of disease state, eNOS immunoreactivity was greatest in the SVP (control: 12.6 ± 0.45%; 1-week STZ: 17 ± 1.09%), followed by the DCP (control: 12 ± 0.52%, 1-week STZ: 16.8 ± 1.18%) and ICP (control: 9.3 ± 0.26%, 1-week STZ: 12 ± 0.63%). After 1-week of STZ induced diabetes, eNOS immunolabelling significantly increased in all vascular plexuses and across all vessel orders, compared to controls (all P < 0.001; Fig. 4G). The greatest increase in eNOS immunolabelling was observed in 5th order arterioles, increasing from 19.5 ± 1.61 to 32 ± 1.16% (P < 0.001; Fig. 4G).

Fig. 4.

Comparison of eNOS immunolabeling in the (A, B) SVP, (C, D) ICP and (E, F) DCP of the control and 1-week STZ rats. The left panels show images with lectin (green) and Hoechst (blue) staining, and the right panels show the same region with eNOS (yellow) labelling only. (G) eNOS labelling significantly increased in all vascular layers and vessels orders after 1-week of STZ-induced diabetes. Asterisk (*) indicates statistical significance (Unpaired t-test, P < 0.05). Data presented as the percentage of eNOS positive area ± SD. Scale bar: 50 µm.

Immunodetection of AGEs accumulation in control and diabetic rats

AGEs immunoreactivity was not detected in any vascular plexus or vessel order of the control group (Fig. 5A,C,E). Significant AGEs immunolabelling was detected in the vascular basement membrane and intracellular confines of all vascular layers and vessel orders of the 1-week STZ eyes (All P < 0.001; Fig. 5B,D,F). Immunolabelling of AGEs was greatest in the SVP (12.5 ± 1.61%) followed by the DCP (5.6 ± 0.77%) and ICP (3.7 ± 0.86%) (Fig. 5G). The greatest increase in AGE immunolabelling was observed in first-order arterioles (A1), followed by a stepwise decrease across A2 through A6 (P < 0.05; Fig. 5G).

Discussion

This study investigated the endothelial dependent vasodilatory responses of ocular endothelial cells to varying levels and durations (1–4 weeks) of hyperglycemia and after the restoration of normoglycemia. The major findings of this study are as follows: (1) Endothelial-dependent vasodilatory responses were significantly enhanced in the ocular microcirculation of control, 1-, 2-, 3- and 4-week STZ-induced diabetic rats after less than 2-h of high glucose (24 mM) exposure. (2) Endothelial-dependent vasodilatory responses at baseline 6 mM glucose exposure were significantly enhanced in the 3- and 4-week STZ rats, compared to the control group. (3) Endothelial-dependent vasodilatory responses during high glucose exposure were significantly more enhanced in the 1-, 3- and 4-week STZ rats, compared to the controls. (4) Upon re-exposure to normoglycemic conditions the Ach-dependent vasodilatory responses of the control, 1- and 2-week STZ rats returned to pre-hyperglycemic level, with the responses in the 1- and 2-week STZ rats also comparable to normoglycemic eyes. In the 3-week STZ rats, whilst Ach-induced vasodilatory responses were significantly attenuated compared to the group’s own baseline pre high glucose exposure, these responses were comparable to normoglycemic eyes. (5) Upon re-exposure to normoglycemic conditions, the 4-week STZ group exhibited only partial recovery: although vasodilatory responses were significantly attenuated compared to the group’s own baseline 6 mM glucose condition, they remained elevated relative to normoglycemic controls. (6) Capillary density and capillary diameter did not change significantly in any retinal vascular plexus after 1-, 2-, 3- and 4-weeks of STZ-induced diabetes. (7) Pericyte count significantly increased in the SVP, ICP and DCP of the 3- and 4-week STZ rats. (8) eNOS immunoreactivity was significantly greater in all retinal vascular layers and vessel orders of the 1-week STZ rats, compared to age-matched controls. (9) Significant accumulation of AGEs was present in all vascular layers and vessel orders after 1-week of STZ induced diabetes.

Microvascular endothelial cells are particularly vulnerable to hyperglycemic injury as they cannot downregulate their glucose transport rate effectively when glucose concentration is elevated, leading to intracellular hyperglycemia^23,24. Ach-induced dilatation is an accepted method to assess endothelial-dependent function as it is mediated by endothelial nitric oxide released by the vascular endothelium^25–27.

We showed that in the control eyes, Ach, induced dose-dependent vasodilatory responses within the ocular microvasculature, irrespective of glucose concentrations. The response of the ocular microvasculature in our study is comparable to the macrovascular response in healthy young adults demonstrated in a recent study by Horton et al.⁶ In that study, hyperglycemia induced using dextrose infusion resulted in a significant increase in brachial artery flow-mediated dilation (ratio of geometric mean of 1.34) compared to the normoglycemic group⁶. In the same study, Horton and colleagues evaluated microvascular perfusion in forearm skeletal muscle using ultrasound and reported a significant increase in microvascular flow velocity and microvascular blood flow following induced hyperglycemia (ratio of geometric mean 1.39 and 2.34, respectively). These findings suggest that in a non-diseased state, there are sufficient mechanisms to compensate for the decreased nitric oxide (NO) bioavailability due to hyperglycemia through compounds such as an endothelium-derived hyperpolarizing factor (EDHF)²⁸. The augmented vasodilatory response reported in small resistance vessels may be linked to EDHF having a greater contribution than NO to vasodilatation in these vessels^29–31. The enhanced vasodilatation may be due to the putative nature of EDHF; up-regulation of pathways that increase the conductivity of myoendothelial junctions which support the hyperpolarization of endothelial cells; and/or production of epoxyeicosatrienoic acids that upregulate the production or reactivity to EDHF^32,33. Supporting this, Nacci and colleagues demonstrated that L-N^G-nitro-L-arginine methyl ester (L-NAME) nearly abolished vasodilation to Ach in control mice. In contrast, vasodilation in 1-week STZ-treated mice, was only partially impaired by L-NAME, suggesting the involvement of alternative pathways- such as EDHF³⁴. Moreover, studies using eNOS knockout mice have shown preserved Ach-induced vasodilatory responses despite the absence of NO³⁵. Taken together, these findings support the plausibility of our interpretation and suggest that early in diabetes, the vasculature may undergo functional adaptation involving both NO and EDHF pathways.

Following 1-week of STZ-induced diabetes, eNOS expression in the retinal vasculature was significantly greater than the age-matched controls. These changes may underlie the NO mediated, enhanced Ach-induced vasodilatory response to 24 mM glucose observed in the 1-week diabetic rats and suggest a functionally coupled and responsive eNOS system during early hyperglycemia. Our finding is similar to increased eNOS mRNA expression, and NO production reported in human aortic endothelial cells³⁶. In that study by Cosentino and colleagues, human aortic endothelial cells exposed to high glucose (22.2 mmol/L) for 5 days, expressed a twofold increase in eNOS gene expression, 40% increase in NO production as measured by nitrite levels but more than a threefold increase in superoxide anions. In 2-week STZ mice, retinal endothelial cells maintained in high glucose had significantly increased eNOS expression and activity, as well as formation of superoxide and nitrotyrosine³⁷. There are also reports of augmented NO-dependent vasodilation in the early stages of diabetes, prior to the onset of oxidative stress mediated eNOS dysfunction^34,38.

Prolonged hyperglycemia may exacerbate redox balance, resulting in eNOS producing superoxide instead of NO-the mechanism termed “eNOS uncoupling”³⁹. This mechanism may in part explain the impaired endothelial-dependent relaxation more consistently reported in the later stages of diabetes, often associated with sustained oxidative stress and depletion of tetrahydrobiopterin^39,40. Supporting this, Pieper⁴¹ reported a triphasic response of increased (24 h post STZ-injection), unaltered (1–2 weeks post STZ-injection) and impaired (8 weeks post STZ-injection) endothelial-dependent relaxation within the same diabetic animal model, suggesting a progressive decline in NO production with increasing duration of hyperglycemia. The incomplete restoration of endothelial function due to prolonged hyperglycemia (≥ 4 weeks) in our study, indicates the beginning of functional decline, potentially due to redox imbalance and early uncoupling.

Significant intracellular AGEs formation has been reported in cultured endothelial cells exposed to a hyperglycemic environment for 1-week⁴². We showed that following 1-week of experimental diabetes, significant accumulation of AGEs occurred in the SVP, ICP and DCP, confirming in vivo accumulation after 1-week exposure to hyperglycemia. AGEs form in two stages, namely initial and final (advanced). The initial stage is a reversible reaction occurring within a few hours and involves reducing sugars reacting non-enzymatically with the free amino group of proteins, lipids and/or nucleic acids to form an unstable Schiff base^43,44. In the next stage, lasting several weeks, the Schiff base is subject to intramolecular rearrangement (the Amadori reaction), which produces a more stable and partly reversible, early glycation product. In the presence of oxygen, Amadori reaction products become phenotypically brownish yellow in color and referred to as glycoxidation⁴⁴. Further Maillard reactions such as polymerization, oxidation, dehydration and cyclization events lead to the formation of final AGEs⁴⁵. Whilst AGEs form naturally during aging, the rate of AGE synthesis is accelerated by hyperglycemia and oxidative stress¹⁵. As such accumulation of AGEs is considered a major factor in the development and progression of diabetes related vascular disease. In our study, the enhanced vasodilation after acute glucose elevation in STZ-induced rats occurred alongside increased eNOS immunoreactivity, despite mild AGE accumulation. This suggests that at this early stage (1-week post-STZ induction), AGEs have not yet reached levels sufficient to impair endothelial function. Instead, we propose that mild AGE accumulation or early glycation signaling may stimulate low-level oxidative or inflammatory responses, triggering a compensatory upregulation of eNOS and NO production. This adaptive mechanism has been described in other models of early diabetes models, where endothelial cells initially respond to metabolic stress by transiently enhancing eNOS expression and NO availability^41,46.

Following the formation of advanced, irreversible AGEs, they are known to interact with cellular receptors such as RAGE (Receptor for Advanced Glycation End-products), triggering oxidative stress, apoptosis, inflammation, and initiate the development of “hyperglycemic memory”^15,47. These events perpetuate the continued presence of hyperglycemic stress, despite restoration of normoglycemia. However, during the early stages of diabetes, RAGE expression may not consistently co-localize with AGE accumulation. For instance, Kase et al.⁴⁸ reported accumulation of AGEs in the central retinal artery, vein and across multiple retinal layers of diabetic donor eyes, yet RAGE expression was absent at these sites and throughout the retina. Similarly, other experimental models have shown AGE receptor-1 and -2 expression-primarily expressed on pericytes and smooth muscle cells did not change significantly with either induction of diabetes or exogeneous AGE infusion in control animals¹⁸. Thus, it is likely that upregulation of RAGE expression patterns may be more tightly associated with the progression of diabetic retinopathy rather than an early pathogenic event in diabetic retinopathy.

Prolonged hyperglycemia is a major risk factor for endothelia disease⁴⁹. Increased oxidative stress from the enhanced activation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase⁵⁰, excessive reactive oxygen species (ROS) production³⁶ and the depletion of superoxide dismutase activity⁵¹ are some mechanisms that can reduce bioavailability of NO and contribute to the impaired endothelial-dependent relaxation. The duration of time after which hyperglycemia-induced endothelial alteration can recover is unclear. It is known that a relatively prolonged period of endothelial alteration measured by way of retinal perfusion precedes the development of permanent retinal structural injury as in the form of microaneurysms^12,52. Current clinical management of diabetic retinopathy (DR) remains largely reactive, with interventions typically initiated only after the appearance of structural lesions such as microaneurysms, hemorrhages, or neovascularization—hallmarks of established retinal injury. Standard therapies, including intravitreal anti-VEGF injections, corticosteroids, and laser photocoagulation, are effective in stabilizing or improving vision but primarily address late-stage manifestations of the disease^53,54. The concept of a defined ‘tolerance period’—a phase in early diabetes during which the retinal endothelium retains functional responsiveness or is transiently augmented—has important translational implications. Defining the therapeutic window during which reversal of endothelial alteration through the treatment of hyperglycemia can therefore improve the management of diabetic retinal disease and shift the clinical paradigm from reactive to proactive care. The concept of a tolerance period has precedent in other vascular beds. For instance, studies of large vessels (e.g., aorta and mesenteric arteries) have shown that short-term hyperglycemia may transiently enhance or preserve endothelial responses, which are later blunted or lost with prolonged exposure. Pieper⁴¹ described a triphasic pattern in the aorta of STZ-induced diabetic rats where endothelial-dependent relaxation was enhanced at 24 h, unchanged at 1–2 weeks, and impaired by 8 weeks. Similarly, Nacci and colleagues demonstrated that mesenteric arteries from 1-week diabetic mice retained partial endothelium-dependent relaxation, while responses were significantly impaired at later time-points (≥ 8-weeks), highlighting a reversible phase that precedes irreversible dysfunction³⁴. Tight glycemic control has been demonstrated to transiently improve endothelial function, even after hyperglycemic injury, indicating a window during which the endothelium remains functionally plastic^55,56. This concept is further supported by large-scale clinical trials such as the Diabetes Control and Complications Trial (DCCT) and UK Prospective Diabetes Study (UKPDS), which have shown that early and sustained glucose control significantly reduces the risk of DR progression^57,58.

Following 1-, 3- and 4-weeks of experimental hyperglycemia we found that at 24 mM glucose concentration, the diabetic eyes demonstrated a significantly enhanced vasodilatory response than controls. This suggests some dysregulation of the modulatory function of the endothelium after 1-, 3- and 4- weeks of hyperglycemia. Under physiological conditions, the endothelium modulates vascular tone through a tightly regulated balance of vasodilators—including nitric oxide (NO), prostacyclin, and EDHFs—and vasoconstrictors, such as endothelin-1 and thromboxane A₂^59–61. In our study, the marked enhancement of Ach-induced vasodilation after 1-, 3- and 4-weeks of STZ-induced diabetes—particularly in the presence of elevated glucose—suggests that this regulatory balance may be temporarily shifted in favor of vasodilation. This may be associated with upregulation of eNOS expression, increased NO bioavailability or increased sensitivity to acetylcholine. Although this enhanced vasodilatory response may appear beneficial superficially, it more likely reflects an adaptive or compensatory effect, indicating underlying dysregulation of endothelial signaling pathways. Importantly, the loss of full reversibility of endothelial function at later time point of STZ (≥ 4-weeks), despite restoration of normoglycemia support the concept of a limited “tolerance period” in which the endothelium retains its responsiveness to stimuli but becomes increasingly susceptible to irreversible damage. Our results therefore support three important concepts regarding the pathophysiology of diabetic retina disease: (i) Acute hyperglycemia can directly alter endothelia function in the retinal circulation in diabetes. (ii) Normoglycemia is critical for maintaining normal endothelial function in the retina in early diabetes and (iii) There exists a time-sensitive window of endothelial reversibility, after which functional recovery—even with glycemic control—is compromised. Our results also provide the biologic basis for understanding the findings of landmark clinical studies such as the UKPDS that have exemplified the importance of normoglycemia in preventing the progression of diabetic retinopathy⁵⁷.

An important future direction of this research will be to refine the duration of hyperglycemia after which recovery of endothelial alterations is no longer possible. Our results demonstrate that whilst altered endothelial function can recover after short-term hyperglycemia (≤ 3-weeks), prolonged exposure (≥ 4-weeks) results in incomplete restoration, indicating a potential transition toward irreversible dysfunction. It is likely that significant recovery of endothelia function will not be possible after a prolonged duration of chronic hyperglycemia due to diabetes. Studies in diabetic rats have shown that six months of strict glucose control following two months of poor glycemic control significantly reduced the progression of retinopathy⁶². However, in the same rodent study it was found that there was no observable benefit when poor glucose control surpassed six months, suggesting that hyperglycemia needs to be addressed within a defined therapeutic time frame before irreversible retinal injury manifests. Strategies to recover retinal endothelial alterations prior to the onset of structural manifestations that denote irreversible retinal injury (e.g., microaneurysms) have the potential to significantly alter the paradigm of retinal disease management. Using the vasoactive properties of endothelia to gauge retinal injury and test possible treatment strategies may facilitate earlier intervention and lead to better outcomes.

Materials & methods

General

The study was approved by The University of Western Australia Animal Ethics Committee. All animal procedures conformed to the Australian Code for the care and use of animals for scientific purposes. The authors complied with the ARRIVE guidelines⁶³.

81 eyes from 31 young adult male Sprague–Dawley rats from 6 study groups were used for vasoactivity and histology studies. As part of the vasoactivity study, the control group comprised of 22 eyes from 11 normoglycemic rats at 9 and 12 weeks of age. The 1-week STZ group comprised of 12 eyes from 7 rats at 9 weeks of age having been successfully STZ induced to hyperglycemia for 1 week. The 2-, 3- and 4-week STZ groups comprised of 9 eyes from 5 rats having been successfully induced to hyperglycemia for the corresponding time points. 4 eyes from 2, 9-week-old normoglycemic rats were used for the “osmolarity control” experiments. 4 eyes from two, 1-week STZ rats were used as part of the “prolonged exposure control” experiment. 3 additional 9-week old control, and three, 1-week STZ rats were included specifically for histological analysis without prior vasoactivity study (n = 9 eyes from 9 normoglycemic rats or 9 eyes from 9, 1-week STZ rats). 4 eyes were excluded due to a failed viability test or not achieving a pre-contraction pressure not exceeding 45 mmHg.

Six-week-old rats were acclimatized for two weeks on standard rat chow and tap water ad libitum. Diabetes was induced by a single intraperitoneal injection of streptozotocin (S0130; Sigma-Aldrich, St. Louis, MO 70 mg/kg) dissolved in sodium citrate buffer (0.05 M, pH 4.5). The control group was treated with an injection of an equal volume of the vehicle. Diabetes was confirmed by the presence of an unfasted blood glucose value of greater than 250 mg/dl and evidence of polyuria 72 h post injection. Experiments were conducted 1-, 2- 3- or 4-week post STZ-injection.

Anesthesia

Rats were deeply anesthetized with an intraperitoneal injection of Ketamine Hydrochloride (Ceva, Glenorie, NSW, Australia 75 mg/Kg) and Medetomidine Hydrochloride (Ilium, Glendenning, NSW, Australia 0.5 mg/Kg), their blood glucose measured (AlphaTRAK2) via tail vein prick and then euthanized with Sodium Pentobarbitone (Virbac, Milperra, NSW, Australia 160 mg/Kg). Both eyes were enucleated, ensuring the presence of a long optic nerve and ophthalmic artery (5–7 mm), and stored in ice-cold, carbogen-bubbled (95% O₂, 5% CO₂) Sodium Ringer’s solution prior to cannulation.

Isolated perfused eye preparation

Figure 6A shows the isolated arterially perfused eye setup, as described in our previous papers^21,64,65. An advantage of the isolated perfused rat eye preparation is that whilst the entire ocular circulation is perfused, the total vascular resistance to perfusion is dominated by the microvasculature. Two sets of equipment were used so that both eyes from the same animal could be experimented on simultaneously. Under microscopic visualization, tissue was dissected, and the ophthalmic artery cannulated and secured onto a glass pipette (tip diameter 100–120 μm), using an 8.0 suture. The eye was seated in a temperature-controlled bath (37 °C, confirmed before the start of each experiment) containing Sodium Ringers solution and perfused with carbogen-bubbled Sodium Ringers or Potassium Ringers solution (95% O₂, 5% CO₂, pH 7.4) at a constant flow rate of 5 μl/min using a computer-controlled syringe pump. This flow rate was determined in previous study using the same technique⁶⁴ The perfusion pressure was measured by a Cobe pressure transducer. The pipette, syringe, pressure transducer, high pressure liquid chromatography (HPLC) valve, and the syringe pump formed the fluid delivery pathway. The HPLC valve enabled a 20 μl drug bolus to be delivered to the perfusate stream without pressure artefacts or air bubbles. Figure 6B shows a typical chart recording of perfusion pressure change in response to increasing log concentrations (Ach 10⁻¹⁰, 10⁻⁸, 10⁻⁶ and 10⁻⁴ M) of Ach in a control rat eye. Null response from a dummy injection prior to Ach administration confirms the integrity of the system.

Fig. 6.

Experimental setup. (A) Photograph of isolated perfused rat eye and schematic (Inset) are provided. Both panels demonstrate the temperature controlled organ bath, constant perfusion flow (5 μl/min) by the computer-controlled syringe pump, the transducer that measures the input pressure and the HPLC valve where bolus doses of the vasoactive agents are added to the perfusion line, leading into the cannula and the eye in an environmentally controlled organ bath⁶⁴. (B) Raw data showing perfusion pressure changes in response to bolus acetylcholine of increasing [log M] (Ach 10⁻¹⁰, 10⁻⁸, 10⁻⁶ and 10⁻⁴ M) as a function of time. Arrows mark time of bolus delivery and horizontal bar shows time scale along the x-axis. Note that a dummy injection was used before Ach administration.

Solutions and drugs

Sodium Ringers comprises of NaCl 115 mM, CaCl₂ 1.3 mM, NaHCO₃ 25 mM, MgSO₄·7H₂O 1.2 mM, K₂HPO₄ 2.499 mM, MOPS 2 mM and glucose 6 mM. Potassium Ringer’s solution was made by replacing NaCl with equimolar KCl (115 mM). Potassium Ringers at differing concentrations of glucose (12 mM & 24 mM glucose, pH 7.4) were prepared by adding appropriate amount of glucose. Acetylcholine chloride (A6625; Sigma-Aldrich, St. Louis, MO) was dissolved in saline to make stock solution at 1 M and aliquoted for storage at − 20 °C. Dilutions were made fresh daily in Potassium Ringers solution and the corresponding glucose concentration.

Experimental regime

Figure 7 presents a schematic of the experimental timeline. Once perfusion commenced, 3 × 2 IU of heparin boluses were delivered to the ocular vasculature to prevent coagulation in the microvasculature and the system was left to stabilize until the basal pressure was stable at 10–13 mmHg, usually 30-min. Following a successful viability test as shown by a sharp increase in perfusion pressure (> 17 mmHg) in response to a 20 µl bolus of 115 mM Potassium Ringers solution, the ocular vasculature was induced to continuously contract using equal molar Potassium Ringers solution (115 mM with 6 mM glucose) as the perfusate and given a 30-min stabilization period for the increased perfusion pressure to be stabilized (45–55 mmHg). A 20 µl sham injection of 6 mM glucose in 115 mM Potassium Ringers was given through the valve to confirm an artifact-free null response. Four sets of dose-dependent responses were recorded using increasing log Molar concentrations of Acetylcholine (Ach). Ach dose-dependent response series were induced and recorded during perfusion of 115 mM Potassium Ringers containing either 6 mM, 12 mM or 24 mM glucose. The sequence of the four Ach dose-dependent response series was in the order of 6 mM, 12 mM, 24 mM followed by 6 mM glucose in 115 mM Potassium Ringers. The Ach boluses were loaded sequentially at increasing concentrations at time intervals of approximately 8-min into the perfusate line. At the end of each Ach dose-dependent response series, residual solution within the perfusion system and incubation bath were completely flushed out and replaced with the perfusate at a specified glucose concentration ready for the next Ach dose–response series. Irrespective of experimental group, all eyes underwent the same experimental regime of Ach dose-dependent response in 6 mM glucose, then 12 mM, then 24 mM and finally back to 6 mM. The use of 12 mM and 24 mM glucose concentrations are representative of moderate and severe hyperglycemia, based on clinical and experimental relevance. In individuals with type 2 diabetes, fasting glucose ranges between 11 and 16 mM, with postprandial levels often exceeding 20 mM^6,26,66. Similarly, STZ-induced diabetic rats exhibit blood glucose levels above 20 mM, whilst normoglycemic levels range between 5 and 8 mM^67,68. In vitro models commonly use glucose concentrations of 11–30 mM to span a physiologically and pathologically relevant range for investigating hyperglycemia-induced alterations in endothelial function^23,56.

Fig. 7.

Schematic of experiment regime used on all rat eyes. Steps 1 to 6 indicate the sequence of events. Each isolated rat eye was perfused initially using Sodium Ringers during which there was a (1) 30-min stabilization period. This was followed by (2) sham injection and viability test using bolus delivery of 115 mM Potassium Ringers. The isolated rat eye was then continuously perfused using Potassium Ringers at a specified glucose concentration (6 mM, 12 mM, 24 mM and back to 6 mM). During exposure of the isolated rat eye to each glucose concentration, increasing log_M concentrations of Ach were loaded at 8-min intervals for measurement of responses as change in perfusion pressure. All eyes, irrespective of experimental group, underwent all 6 steps of the experimental regime in the same sequence.

Control experiments

Two types of control experiments were performed: the first addressed the potential confounding effect of increased osmolarity at higher glucose concentrations. The second addressed whether the observed changes in vasodilatory response was confounded by the prolonged exposure time of higher glucose concentration with the consecutive 12 mM and 24 mM glucose perfusion.

Osmolarity control: The “experimental regime” was repeated in 4 eyes from 2 normoglycemic rats using the 6 mM glucose perfusate as the base and adding equimolar concentration of mannitol (12 mM perfusate = 6 mM glucose + 6 mM mannitol; 24 mM perfusate = 6 mM glucose + 18 mM mannitol) as an inert osmotic agent. Perfusion with the equimolar 12 mM and 24 mM perfusate significantly attenuated the overall dose-dependent Ach-induced vasodilatory response, compared to baseline 6 mM glucose exposure (all P < 0.001; Supplementary Fig. S2). Re-exposure to baseline 6 mM found comparable, endothelium-dependent vasodilatory responses to 6 mM glucose at the beginning of the test series (P = 0.374; Supplementary Fig. S2).

Prolonged exposure control: The “experimental regime” was repeated in 4 eyes from two 1-week STZ rats without the use of 12 mM glucose perfusate. Increase in glucose concentration from 6 to 24 mM significantly enhanced the Ach induced vasodilatory response (P = 0.007; Supplementary Fig. S3). Re-exposure to 6 mM glucose, after acute exposure to 24 mM glucose found compared, endothelium-dependent vasodilatory responses to 6 mM glucose at the beginning of the test series (P = 0.632; Supplementary Fig. S3).

Isolated ocular perfusion labelling

At the end of the experiments, isolated ocular perfusion labelling was performed to examine capillary density, capillary diameter, distribution of pericytes, distribution of eNOS and accumulation of AGEs. To validate histological analysis, 12 additional eyes from three control rats and three 1-week STZ rats were perfusion labelled after being freshly enucleated without prior vasoactivity study.

Eyes were prepared using our previously described techniques^69–71. In brief, the ophthalmic artery was cannulated with a glass micropipette (∼120 um) and perfused with the following solutions in sequence, at a rate of 80 µl per minute (unless otherwise stated): washed with 0.1 M phosphate buffer (PB) (10 min), fixed with − 20 °C 100% methanol (20 min) and permeabilized with 1% Triton-X-100 in PB (6 min). The following mixture of primary antibodies and fluorescently tagged labels were then manually pushed (3 boluses of 160 µl, at 20-min intervals): Lectin- fluorescein isothiocyanate (FITC, L4895; Sigma-Aldrich; 60 µg), mouse anti-α-smooth muscle actin (α-SMA) (A2547 Sigma-Aldrich; 1:25 dilution), rabbit anti- endothelial Nitric Oxide Synthase (eNOS) (AB5589 Abcam; 1:25 dilution), goat serum (10%), Hoechst 33,342 (94,403 Merck; 1:1000) and 0.1% Triton-X-100 in PB. To co-label for AGEs, the protocol was modified only to replace the primary antibody for eNOS with rabbit anti-AGEs (AB23722 Abcam; 1:25 dilution). After washing with 0.1 M PB (10-min), secondary antibodies with fluorescent markers were manually pushed (3 boluses of 100 µl, at 15-min intervals): goat anti-rabbit conjugated to Alex Fluor 633 (A21071 ThermoFisher Scientific; 1:200) and 0.1 M PB. Post-perfusion the eyes were decannulated, the retina dissected from the eye cup and floated in secondary antibodies for α-SMA overnight: goat anti-mouse conjugated to Alexa Fluor 555 (Abcam AB150114; 1:200) and 0.1 M PB. Retinas were subsequently washed in 0.1 M PB for a day then flat mounted in RapidClear^®1.47 (Sunjin Lab Co., Taiwan) for confocal microscopy.

Image acquisition

Labelled retinal wholemounts were imaged on a Nikon C1 plus confocal system (Nikon Corporation, Tokyo, Japan) equipped with four solid-state lasers at wavelengths 405 nm, 488 nm, 551 nm and 635 nm. Three low magnification images per retinal wholemount were acquired within two optic disc distances using the Plan Apo 20 × objective lens (NA = 0.75; WD = 1.0 mm), with a step size of 1 µm and a field of view of 0.635 mm × 0.635 mm to measure capillary density, AGEs accumulation and e-NOS distribution. Three high magnification images were acquired per specimen using a Plan Apo 40 × oil lens (NA = 0.75; WD = 0.16 mm) for quantification of vessel diameter and pericyte distribution.

Image analysis

1. Capillary density—The density of blood vessels was measured for each of the three vascular layers in the rat retina of control, 1-, 2-, 3- and 4-week STZ rats. Segmentation of each vascular layer was performed using ImageJ (National Institutes of Health, Bethesda, MD, USA) software guided by the co-localization of nuclei staining in each retinal layer. The superior vascular plexus (SVP) extends from the inner limiting membrane to the outer border of the ganglion cell layer (GCL). The intermediate capillary plexus (ICP) extends from the border of the GCL to the middle one-half of the inner nuclear layer (INL). The deep capillary plexus (DCP) extends from the middle one-half on the INL to the outer border of the outer plexiform layer (OPL)⁷². Z-projected image of each vascular plexus was binarized using the ImageJ “threshold tool”, stratifying the image into features of interest (i.e. the vessels) versus background. The ImageJ “area fraction measurement tool” was then used to quantify the vessel density, measured as the percentage of lectin positive area.

2. Capillary diameter—A blinded observer was instructed to select 10 capillaries evenly and randomly from each vascular layer, in each image (30 vessels measured per layer, per retinal wholemount) of the control, 1-, 2-, 3- and 4-week STZ rats. The chosen vessels were measured at midpoint along their course, away from bifurcations and bulges, which indicate sites of endothelia nuclei. A line was drawn perpendicular to the capillary between the endothelium walls to measure the luminal diameter of the capillary. Individual measurements were combined to calculate mean capillary diameter and standard deviation. Measurements were obtained for each of the three vascular plexuses.

3. Pericyte distribution—Pericytes are abluminal cells with dome shaped nuclei that partially ensheathe capillaries⁶⁹. Our previous studies found the combination of nuclei and α-SMA labelling were sufficient for accurate identification of pericytes^69–71. The total number of pericytes was manually counted for the control, 1-, 2-, 3- and 4-week STZ group using NIS-Elements AR (version 5.30.05). This was achieved by scrolling through the lectin, α-SMA and nuclei co-labelled z-stacks for each vascular plexus as specified in the capillary density section and supplemented with a generated three-dimensional volume render to allow analysis at varying angles of rotation.

4. eNOS distribution—Segmentation of each vascular layer was performed as described in the “capillary density” section. The ImageJ “threshold tool” was used to binarize each z-projected image, stratifying the image into features of interest (i.e. eNOS immunolabelling) versus background. The “area fraction tool” was then used to quantify the distribution of eNOS, measured as the percentage of eNOS positive area within the specified retinal vascular plexus or vessel order for the control and 1-week STZ group.

5. Immunodetection of AGEs—Quantification of AGEs in the control and 1-week STZ group was performed as described in the “eNOS distribution” section.

Data analysis

The stabilized baseline perfusion pressure of the ocular vasculature during the initial 6 mM glucose in Sodium Ringers perfusion was subtracted from all pressure responses of the Potassium Ringers and subsequent dilation responses to Ach boluses. Ach-induced dilation responses were then expressed as a normalized percentage change from the Potassium Ringers induced pre-contraction pressure at the beginning of the corresponding dose–response series (Figs. 1 and 2).

All statistical analysis was performed using the statistics program, SigmaPlot 12.5. All data underwent Shapiro–Wilk normality testing and homogeneity of variance testing prior to analysis. Two-way repeated measures ANOVA were used to compare dose response curves (intra-group), with the perfusate glucose concentration as the first factor, the Ach concentration as the second factor and P < 0.05 as the acceptance level. Standard two-way ANOVA was used to compare dose response curves of the same glucose concentration between the control and STZ groups. Bonferroni’s post-hoc test was used for group comparisons at each concentration of Ach with an acceptance level of P < 0.05. Unpaired t-test was used to compare capillary density, capillary diameter, pericyte distribution, eNOS distribution or AGEs accumulation between the control and STZ groups at each vascular plexus (SVP, ICP or DCP). Results are presented as mean ± standard deviation (SD) unless otherwise stated.

Author contributions

DY and PY conceptualized the study. HQ performed all experiments, data collection, data analysis and manuscript drafting. DY and PY provided technical support for experimental setup preparation and assisted with manuscript drafting. CB contributed to manuscript drafting and revising it critically for important intellectual content. JD assisted in experimental setup preparation and animal handling. DY was responsible for supervision, project administration and final approval of submitted version. All authors contributed to final manuscript editing and submission. DY is the guarantor of this work, and as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Funding

National Health and Medical Research Council of Australia Investigator Grant (APP1173403) and the Stan Perron Charitable Foundation, Perth, Western Australia.

Data availability

The datasets used during the study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-12612-4.