Hyperglycemia Enhances Constriction of Retinal Venules via Activation of the Reverse-Mode Sodium-Calcium Exchanger

Retinopathy is one of the most common microvascular complications of diabetes and a major cause of vision impairment and blindness in working-age adults (1). Hyperglycemia, a hallmark of diabetes, is closely linked to the retinal pathology, including progression from early disruption of retinal perfusion for proper oxygen supply to the development of edema and fluid accumulation during the late stages of the disease (2). Although alteration of microcirculatory function in the retina is considered a critical event contributing to altered blood flow and tissue edema with diabetes, the precise vasomotor regulatory mechanisms remain unclear. It is well recognized that venular resistance influences hydrostatic pressure and fluid homeostasis at the level of the capillaries (3). Constriction of retinal venules is expected to increase retinal venous resistance and upstream capillary pressure, which could consequently reduce retinal perfusion and promote edema (4), especially during the conditions when the compensatory or defense mechanisms are exhausted under disease states such as diabetes.

In accord with this idea, clinical evidence has shown increased retinal venous pressure and decreased retinal blood flow in patients with retinopathy at late stages of diabetes (5,6), including those with diabetic macular edema (7). However, whether diabetes or hyperglycemia affects responses of retinal venules to endogenous vasoconstrictors is unclear. The potent vasoconstrictor endothelin-1 (ET-1) has been implicated in the pathogenesis of diabetic retinopathy partly based on elevated levels of this peptide in the retinal tissue (8) of diabetic rats and vitreous fluid (9) of patients with diabetic retinopathy. Elevated plasma/serum or retinal tissue levels of other vasoconstrictors, such as thromboxane A₂ (TXA₂) (10), norepinephrine (11), and angiotensin II (12), have also been detected in experimental diabetes. Therefore, it is important to investigate whether diabetes/hyperglycemia alters venular reactivity to these endogenous vasoconstrictors in the retina.

It is generally accepted that calcium entry into the vascular smooth muscle cells triggers vasoconstriction (13). In some blood vessels, this process is mediated by activation of L-type voltage-operated calcium channels (L-VOCCs) (13). However, our recent study showed that L-VOCCs play no role in retinal venular constriction to ET-1 despite the fact that extracellular calcium entry is required (14). An alternative mechanism for calcium entry is activation of the sodium-calcium exchanger (NCX) (15). This antiporter membrane protein can operate in both forward and reverse modes. The forward-mode NCX prevents calcium overload in cells and uses the energy that is stored in the electrochemical gradient of sodium by allowing three sodium ions to flow down their gradient across membranes in exchange for export of one calcium ion (16). However, when intracellular levels of sodium ions rise beyond a critical point, especially under pathological conditions, NCX begins importing calcium ions in the reverse mode (16). Interestingly, previous studies demonstrated that ET-1 can exert both inotropic and hypertrophic effects on cardiomyocytes by promoting extracellular calcium influx through the reverse-mode NCX (17,18).

The activation of reverse-mode NCX also has been shown to cause increased calcium entry in cultured cardiomyocytes exposed to high levels of glucose (19). However, the contribution of reverse-mode NCX to agonist-induced constriction of retinal venules under normal and diabetic/hyperglycemic conditions has not been investigated. We therefore addressed these issues by examining the direct response of isolated retinal venules to ET-1 and other vasoconstrictor agonists (U46619, norepinephrine, phenylephrine, and angiotensin II) and investigating whether in vitro and in vivo hyperglycemia influence these vasoconstrictions via reverse-mode NCX activation. We also used biochemical and molecular tools to assess the impact of hyperglycemia on vitreous humor levels of vasoconstrictor substances and mRNA expression of the target receptors for vasoconstriction in retinal venules.

All animal procedures were performed in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Baylor Scott & White Health Institutional Animal Care and Use Committee. Domestic (Yorkshire) male pigs (age range, 6–10 weeks; weight range, 9–20 kg) were purchased from Real Farms (San Antonio, TX). Diabetes (i.e., in vivo hyperglycemia) was induced by selective ablation of pancreatic β-cells with intravenous injection of streptozocin (Zanosar) (200 mg/kg in saline) via an ear vein (48 pigs), as described in detail in our previous studies (20,21). The control euglycemic group was intravenously injected with saline (59 pigs).

Fasting blood glucose levels were obtained every other day using a Bayer Contour glucometer (Bayer Corporation, Pittsburgh, PA). Pigs were treated with insulin (Humulin 70/30, 2–8 units) (Lilly, Indianapolis, IN) if blood glucose was sustained >600 mg/dL to keep the level between 250 and 600 mg/dL. After 2 weeks, pigs were sedated with Telazol (4–8 mg/kg, intramuscularly), anesthetized with 2–5% isoflurane, and intubated. The eyes were harvested as described previously (22).

Single second-order retinal venules (1–1.5 mm in length without side branches) were dissected from surrounding neural/connective tissues and then cannulated on each end with glass micropipettes containing physiological saline solution (PSS) with 5 mmol/L d-glucose and 1% albumin (14). In another cohort, 25 mmol/L d-glucose in the PSS was used. Vessels were pressurized to 5 cmH₂O intraluminal pressure without flow by two independent pressure reservoirs, and their inner diameter was recorded using videomicroscopic techniques throughout the experiments, as described previously (14).

Cannulated, pressurized retinal venules were bathed in PSS-albumin at 36–37°C to allow development of basal tone (stable within 60–90 min). To evaluate the effect of in vivo hyperglycemia on vasomotor function, diameter changes to cumulative administration of ET-1 (1 pmol/L to 10 nmol/L) (Bachem, Torrance, CA), TXA₂ receptor agonist U46619 (0.1 nmol/L to 1 μmol/L) (Cayman Chemical, Ann Arbor, MI), angiotensin II (0.1 nmol/L to 10 μmol/L) (MilliporeSigma, St. Louis, MO), nonselective α-adrenergic receptor (AR) agonist norepinephrine (1 nmol/L to 10 μmol/L) (Cayman Chemical), or α₁-AR agonist phenylephrine (1 nmol/L to 10 μmol/L) (Cayman Chemical) were recorded and compared in vessels isolated from diabetic and control pigs. Retinal venules were exposed to each concentration of agonist for 10 min until a stable diameter was attained. Only one agonist was evaluated in an individual vessel. The contribution of ET_A receptors (ET_ARs) in vasoconstriction to ET-1 was evaluated after incubation with respective antagonist BQ123 (1 μmol/L) (14). The contribution of TXA₂ receptors (TXA₂Rs) in vasoconstriction to U46619 was examined in the presence of respective antagonist SQ29548 (10 μmol/L) (23). The relative roles of α₁-ARs and α₂-ARs in vasoconstriction to norepinephrine or phenylephrine were evaluated in the presence of the selective α₁-AR antagonist prazosin hydrochloride (10 μmol/L) (24) or the α₂-AR antagonist atipamezole (10 μmol/L) (25). The role of reverse-mode NCX in agonist-induced constrictions was examined in the presence of its selective inhibitors KB-R7943 (10 μmol/L) (26) or SEA0400 (10 μmol/L) (27). In some vessels, the role of L-VOCCs in the vasoconstriction to ET-1 was evaluated in the presence of L-type calcium channel inhibitor nifedipine (1 μmol/L) (14). All vessels were pretreated with antagonists or inhibitors extraluminally for at least 30 min.

The direct effect of high glucose on vasoconstrictions to ET-1, U46619, and norepinephrine was evaluated in vitro after intraluminal incubation of vessels from 49 nondiabetic pigs with 25 mmol/L d-glucose for 2 h (20), and the results were compared with the treatment with normal 5 mmol/L d-glucose as the control. The contribution of NCX in agonist-induced vasoconstrictions was evaluated after coincubation of 5 mmol/L or 25 mmol/L glucose-PSS with KB-R7943 (10 μmol/L) for 2 h. In some vessels, vasoconstriction to ET-1 (0.1 nmol/L) was assessed in the presence of 20 mmol/L l-glucose (MilliporeSigma) plus 5 mmol/L d-glucose to determine the stereospecific impact of glucose on this response. Osmolarity in all of the 25 mmol/L glucose solutions was balanced to 290 mOsm by reducing the NaCl concentration to avoid a potential hyperosmolarity effect.

Drugs were obtained from Tocris Cookson (Ellisville, MO) except as specifically stated otherwise. ET-1 and angiotensin II were dissolved in water, BQ123 and nifedipine were dissolved in ethanol, and phenylephrine, norepinephrine, U46619, SQ29548, prazosin hydrochloride, atipamezole, KB-R7943, and SEA0400 were dissolved in DMSO. Subsequent concentrations of these drugs were diluted in PSS. The final concentration of DMSO or ethanol in the vessel bath did not exceed 0.1% by volume. Vehicle control studies indicated that this final concentration for these two solvents had no effect on vessel viability, vasoconstrictor responses, or maintenance of basal tone (data not shown).

Total RNA was isolated from retinal venules (sample pooled from both eyes) and neural retina tissue via RNeasy mini kit (QIAGEN, Crawley, U.K.), as described previously (14,28). To perform real-time PCR experiments, specific primer sets presented in Table 1 were used. The same amount of total RNA for each sample was used to synthesize cDNA, followed by real-time PCR with the Applied Biosystems QuantStudio 6 Flex Real-Time PCR System (Life Technologies Corporation, Carlsbad, CA), as described previously (14).

Vitreous humor was obtained after harvesting the eyes from control and diabetic pigs. Undiluted vitreous samples were centrifuged at 12,000 rpm for 10 min at 4°C, and the supernatants were stored at −80°C for subsequent analysis within 3 months. Levels of ET-1 protein and TXB₂, a stable TXA₂ lipid metabolite, in the vitreous were measured using an ET-1 ELISA kit (ADI-900-020A; Enzo Life Sciences, Farmingdale, NY) and a TXB₂ ELISA kit (ADI-900–002; Enzo Life Sciences), respectively, according to the manufacturer’s instructions.

At the end of each functional experiment, the vessel was relaxed with 0.1 mmol/L sodium nitroprusside in EDTA (1 mmol/L)-Ca²⁺–free PSS to obtain its maximum diameter at 5 cmH₂O intraluminal pressure (14). Diameter changes in response to agonists were normalized to the resting diameter and expressed as percentage changes in diameter (14). Data are reported as mean ± SEM, and n represents number of vessels (one per pig for each agonist and treatment group for functional studies) or pigs (for molecular/biochemical studies). The Student t test or repeated-measures two-way ANOVA, followed by the Bonferroni multiple-range test, was used to determine the significance of experimental interventions, as appropriate. GraphPad Prism 6.0 software (GraphPad Software, La Jolla, CA) was used for the analysis. P < 0.05 was considered significant.

Porcine retinal venules were isolated and pressurized at 5 cmH₂O with an average maximum diameter of 130 ± 1 µm (total 100 vessels) (Fig. 1). All vessels in this study developed similar levels of stable basal tone by constricting to ∼93% of the maximum diameter. Administration of ET-1, U46619, or norepinephrine caused concentration-dependent constriction of retinal venules from nondiabetic pigs (Fig. 1). Threshold concentrations of ET-1 and U46619 for venular constrictions were ∼1 pmol/L and 0.1 nmol/L, respectively. Maximum vasoconstrictions were ∼50% of resting diameters at the highest concentrations of 10 nmol/L ET-1 and 1 μmol/L U46619 (Fig. 1). The threshold concentration of norepinephrine for venular constriction was ∼0.1 μmol/L, and the maximum constriction was ∼75% of resting diameter at the highest concentration of 10 μmol/L (Fig. 1). However, administration of angiotensin II or phenylephrine caused minor but significant constriction of venules to ∼90% of resting diameter at the highest concentration of 10 μmol/L (Fig. 1).

Retinal venules isolated from the 2-week diabetic pigs (449 ± 19 mg/dL) and age-matched control pigs (97 ± 3 mg/dL) developed a comparable level of resting diameter (control: 121 ± 2 µm; diabetes: 120 ± 2 µm) and basal tone by constricting to ∼93% of the maximum diameter. However, retinal venules from diabetic pigs exhibited enhanced constrictions to ET-1, U46619, and norepinephrine (Fig. 2A). Maximum constrictions to the highest concentrations of these three agonists was ∼25–30% greater in diabetic than in control venules. In contrast, angiotensin II and phenylephrine did not alter basal tone of retinal venules except for modest constriction (∼10%) observed at 10 μmol/L, and there was no enhanced vasoconstriction with diabetes (data not shown).

Retinal venules isolated from nondiabetic pigs developed a comparable level of basal tone after intraluminal exposure to the normal (5 mmol/L) or high (25 mmol/L) level of d-glucose for 2 h. Vasoconstrictions to ET-1, U46619, and norepinephrine (Fig. 2B) were significantly increased by high d-glucose with increased maximum constriction ∼15–25% compared with the vessels exposed to normal d-glucose. The high l-glucose solution (20 mmol/L l-glucose plus 5 mmol/L d-glucose) did not enhance constriction of retinal venules to 0.1 nmol/L ET-1 (Supplementary Fig. 1).

The ET-1–induced constriction of retinal venules exposed to euglycemia or hyperglycemia (in vivo and in vitro) was abolished in the presence of the ET_AR antagonist BQ123 (Fig. 3A and B). A greater amount of ET_AR mRNA was expressed in retinal venules than in neural retina under control and diabetic conditions, but the expression in both tissues was not influenced by in vivo hyperglycemia (Fig. 3C). The ET-1 protein level was greater in the vitreous isolated from diabetic (2.9 ± 0.4 pg/mL ≈ 1.2 ± 0.2 pmol/L) compared with control (1.4 ± 0.2 pg/mL ≈ 0.6 ± 0.1 pmol/L) pigs (Fig. 3D).

Constriction of retinal venules from control pigs to U46619 was completely blocked in the presence of TXA₂R antagonist SQ29548 (Fig. 4A). The TXA₂R mRNA expression in retinal venules was greater than that in neural retina but was not altered in both tissues by diabetes (Fig. 4B). Concentrations of TXB₂, the stable TXA₂ metabolite, were comparable in the vitreous isolated from control (24 ± 4 pg/mL ≈ 65 ± 9 pmol/L) and diabetic (17 ± 3 pg/mL ≈ 47 ± 9 pmol/L) pigs (Fig. 4C).

The α₂-AR antagonist atipamezole abolished norepinephrine-induced constriction of retinal venules from control pigs, whereas α₁-AR antagonist prazosin did not alter this vasoconstriction (Fig. 5A). Expression of α₁-AR mRNA in retinal venules and neural retina tissue was unaffected by diabetes but was less in neural retina than in venules only after diabetes (Fig. 5B). The α_2A-AR mRNA in venules was greater than that in neural retina and was unaffected by diabetes in both tissues (Fig. 5C).

In the presence of the reverse-mode NCX inhibitor KB-R7943, retinal venules exposed to euglycemia or hyperglycemia (in vivo and in vitro) lost ∼50% of basal tone (Fig. 6). The KB-R7943 treatment also prevented the enhanced constrictions of retinal venules to ET-1 (Fig. 6A and B), U46619 (Fig. 6A and B), and norepinephrine (Fig. 6A and B) under hyperglycemia (in vivo and in vitro) without altering normal vasoconstrictor responses in euglycemia. An additional reverse-mode NCX inhibitor SEA0400 also reduced basal tone nearly 44% and blocked the enhanced venular constriction to 0.1 nmol/L ET-1 under in vivo hyperglycemia but did not affect ET-1–induced vasoconstriction under euglycemia (Supplementary Fig. 2). The L-VOCC inhibitor nifedipine reduced basal tone of venules exposed to euglycemia or hyperglycemia (in vitro and in vivo) to the same level induced by KB-R7943 but did not affect the enhanced vasoconstrictions to ET-1 by in vitro or in vivo hyperglycemia (Supplementary Fig. 3). The amount of NCX1–3 isoform mRNA expressions in retinal venules (Supplementary Fig. 4A) was greater than that in neural retina tissue (Supplementary Fig. 4B) but was unaltered by in vivo hyperglycemia.

Hyperglycemia during diabetes has been shown to alter microvascular function and perfusion in the retina (21,29). However, there is little information regarding vasomotor reactivity of retinal venules under normal and hyperglycemic states, especially the ability to constrict to endogenous substances. The main findings of the current study are that porcine retinal venules constrict in response to ET-1, U46619, and norepinephrine but not to angiotensin II and phenylephrine; vasoconstrictions to ET-1, U46619, and norepinephrine are augmented during in vitro and in vivo hyperglycemia; and activation of the reverse-mode NCX appears to mediate these enhanced vasoconstrictor responses.

Direct evidence for vasoconstrictor function of retinal veins is limited to two recent studies showing that ET-1 elicits constriction of these vessels in pigs (14,30). To further our understanding of vasomotor regulation of retinal venules, we evaluated herein the responsiveness of these microvessels to ET-1 and other endogenous vasoconstrictors (Fig. 1). As shown previously, ET-1 elicited potent constriction of isolated retinal venules with a threshold concentration of ∼1 pmol/L (14). We extended these findings to show for the first time the ability of retinal venules to constrict in response to activation by TXA₂R (U46619) and α-AR (norepinephrine) agonists. We also documented for the first time that hyperglycemia (both in vitro and in vivo) enhances venular responses to these vasoconstrictors (Fig. 2). In contrast to other peripheral vascular segments (31,32), retinal venules appear to lack sufficient signaling for vasoconstriction via angiotensin II receptors and α₁-ARs, because their specific agonists (angiotensin II and phenylephrine) evoked nominal changes in venular tone.

We recently showed that ET-1 elicits constriction of retinal venules by activating ET_ARs (14). The current study corroborates and extends this finding by demonstrating that pharmacological blockade of ET_ARs prevents constriction of retinal venules to ET-1 during hyperglycemia (Fig. 3A). In addition, both in vitro and in vivo hyperglycemia enhanced ET-1–induced constriction of retinal venules in a similar manner without altering the ET_AR mRNA expression level (Fig. 3A–C). It seems that hyperglycemia, while increasing vascular reactivity to ET-1, does not alter ET_AR expression, at least at the transcription level (Fig. 3C). Furthermore, the ability of in vitro exposure to high d-glucose but not high l-glucose to enhance ET-1–induced vasoconstriction indicates that the venular reactivity was specifically influenced by the functional hyperglycemia (Supplementary Fig. 1). The combination of increased vasomotor activity of retinal venules to ET-1 and elevated ET-1 concentration in the vitreous of diabetic pigs in our study (Fig. 3D) suggests the potential contribution of this vasoactive peptide to the retinal venous pathology. If the enhanced ET-1–induced constriction of retinal venules is sustained, this could potentially increase upstream transcapillary pressure and promote both edema and reduction in retinal perfusion under conditions when the compensatory or defense mechanisms cannot offset the disturbance (33), such as during the late stages of diabetes (4,5). Elevated levels of ET-1 in the vitreous humor during diabetes may exert a local impact on the diameter of arterioles, pericyte-containing capillaries, and venules in the retinal microcirculation. Previous evidence has shown that in vitro exposure of retinal pericytes to high glucose reduces the ability of these cells to contract in response to ET-1 (34). Conversely, 2 to 12 weeks of hyperglycemia in pigs with streptozocin-induced diabetes did not alter the constriction of isolated retinal arterioles to ET-1 (21). Taken together, these earlier findings and our current results with retinal venules suggest a differential impact of hyperglycemia/diabetes on the vasoconstrictor function at different segments of vessels in the retinal microcirculation. It appears reasonable to speculate that elevated levels of ET-1 in the vitreous humor during diabetes may have a more pronounced impact on diminishing retinal perfusion by increasing constriction of retinal venules. This contention is supported by previous reports of increased levels of ET-1 in the vitreous (9) and reduced retinal blood flow (35) in patients with diabetes. Furthermore, diminished blood flow in retinal veins of diabetic mice can be improved by treatment with an ET_AR antagonist (29). Collectively, these previous observations, along with our current findings, suggest that enhanced retinal venular constriction to ET-1, via ET_AR, may contribute to the early retinal complications of diabetes. Future in vivo studies will be necessary to characterize the reduction of retinal blood flow in association with enhanced retinal venular constriction in diabetic pigs.

TXA₂ is a lipid product of arachidonic metabolism synthesized from TXA₂ synthase by activated platelets and vascular cells (36,37). After cellular release, TXA₂ can activate vascular smooth muscle TXA₂Rs, leading to vasoconstriction. Because TXA₂ is quickly and spontaneously converted to the stable inactive TXB₂ in aqueous solution, experimental studies commonly evaluate TXA₂ function with the TXA₂ analog U46619. This stable TXA₂R agonist causes constriction of retinal arterioles (38), but its vasomotor influence on retinal venules has not yet been determined. In the current study, U46619 elicited constriction of retinal venules with a threshold concentration of ∼0.1 nmol/L under euglycemic conditions. Pharmacological blockade of TXA₂Rs abolished U46619-induced vasoconstriction, indicating a functional vasomotor role for these receptors in porcine retinal venules (Fig. 4A). Moreover, in vitro and in vivo hyperglycemia had no effect on the threshold concentration of U46619 to constrict retinal venules, but they both enhanced the maximum vasoconstriction (Fig. 2). In vivo hyperglycemia caused this augmented response without altering transcription of TXA₂R mRNA (Fig. 4B). To our knowledge, this is the first evidence of TXA₂R expression in the retinal microcirculation, and it extends previous reports detecting this receptor in human (39) and rodent (40) neural retina tissue.

It also appears that local production of TXA₂ (estimated from the levels of stable TXA₂ metabolite TXB₂) within the vitreous is not altered within 2 weeks of diabetes (Fig. 4C). However, we are unable to rule out the potential contribution from the vascular or serum level of TXA₂ with hyperglycemia in view that elevated production of TXB₂ from cultured bovine retinal endothelial cells and rabbit aorta after in vitro exposure to high glucose (36) and in vivo hyperglycemia (41), respectively, has been reported. Experimental type 1 diabetes in rats also significantly increases the platelet-generated TXB₂ levels in the serum within 2 weeks of hyperglycemia (10). Moreover, intravenous administration of a TXA₂ synthase inhibitor or a TXA₂R antagonist can reverse the diminished retinal blood flow after 3–4 weeks of type 1 diabetes in rodents (42,43). Taken together, our present findings and previous evidence from other investigators support a potential role for TXA₂R activation in retinal vascular disturbances during diabetes.

Although the retinal microcirculation apparently lacks autonomic innervation (44), the existence of α-ARs in retinal microvessels has been suggested based on pharmacological studies. The α₁-AR agonist phenylephrine has been shown to constrict retinal arterioles in mice, cows, and rabbits (24,45,46). However, our findings showed that phenylephrine only elicited minor constriction of porcine retinal venules (Fig. 1), indicating that these vessels contain few or no functional α₁-ARs. Brimonidine, an α₂-AR agonist, evokes constriction of isolated second-order arterioles from pigs (28). Comparably, we found constriction of second-order retinal venules to the nonselective α-AR agonist norepinephrine in a manner sensitive to the blockade of α₂-ARs but not of α₁-ARs (Fig. 5A). These results indicate that α₂-ARs are responsible for constrictor action of norepinephrine in retinal venules. At the molecular level, we identified for the first time the expression of α₁- and α_2A-AR mRNAs in retinal venules (Fig. 5B and C). The α_2A-AR expression in these vessels is consistent with the dominant expression of this α-AR subtype in porcine retinal arterioles (28) and human retinal tissue (47). Similar to its effects on ET-1 and U46619, in vivo hyperglycemia augmented norepinephrine-induced constriction of retinal venules without modifying expression of α-AR (α₁ and α_2A) mRNA levels (Fig. 5B and C). This appears to be the first report of the direct impact of hyperglycemia on constriction of venules to norepinephrine during diabetes, and it is consistent with the increased norepinephrine-induced contraction of the large portal vein isolated from diabetic rats (48).

A common mechanism for constriction of blood vessels involves entry of extracellular calcium into the vascular smooth muscle (13). Two transmembrane proteins that have been shown to contribute to regulation of calcium entry into cells are reverse-mode NCXs (15) and L-VOCCs (13). To determine whether increased activity of reverse-mode NCX or L-VOCC contributes to the enhanced agonist-induced constriction of retinal venules during hyperglycemia, we treated vessels with their cognate inhibitors. Pharmacological blockade of reverse-mode NCX (Fig. 6) prevented the enhanced constriction of retinal venules to ET-1, U46619, and norepinephrine after exposure to in vitro and in vivo hyperglycemia, suggesting a possible role of NCX activation during diabetes. This conclusion is supported by the ability of two structurally different selective inhibitors of reverse-mode NCX, KB-R7943 (26) (Fig. 6) and SEA0400 (27) (Supplementary Fig. 2), to block the hyperglycemia-enhanced constriction of retinal venules to ET-1. In contrast, L-VOCC blockade did not prevent the enhanced vasoconstriction, indicating that these channels were not involved in the altered vascular reactivity by hyperglycemia (Supplementary Fig. 3). NCX does not appear to contribute to agonist-induced venular constrictions during euglycemia, because NCX blockade had no effect on these responses. Interestingly, our data suggest that L-VOCC and NCX both contribute to the maintenance of basal tone of retinal venules, because their respective inhibitors reduced the resting diameter to the same degree. At the molecular level, mRNAs for all three NCX isoforms (NCX1–3) were detected in porcine retinal venules, but in vivo hyperglycemia did not affect their expression (Supplementary Fig. 4). Future functional studies are warranted to evaluate whether a specific NCX isoform is activated in retinal venules by hyperglycemia.

A possible mechanism triggered by hyperglycemia upstream of NCX is activation of the sodium-hydrogen exchanger (NHE). Hyperglycemia can lead to intracellular acidosis and elevation of hydrogen ions, which could increase activity of the NHE to remove these ions from the cell in exchange for entry of sodium ions (49). As a result, the intracellular sodium ion concentration may rise sufficiently to activate the reverse-mode NCX. The possible link between NHE and reverse-mode NCX will be a reasonable next step to investigate. It is worth noting that reverse-mode NCX blockade can protect cardiac and vascular complications of hyperglycemia by preventing excess calcium entry in cultured cardiomyocytes (19) and endothelial cells (50) exposed to high glucose (25 mmol/L). On the basis of our current findings that pharmacological blockade of reverse-mode NCX prevented the increased constriction of retinal venules to multiple endogenous receptor pathways, therapeutic targeting of this specific protein may help improve retinal perfusion during the early stages of diabetes.

Acknowledgments. The authors are grateful to Angie Hitt and the animal facility staff in the Department of Comparative Medicine at Baylor Scott & White Health for their technical assistance with animal care.

Funding. This work was supported by the Baylor Scott & White-Central Texas Foundation, the Ophthalmic Vascular Research Program of Baylor Scott & White Health (L.K.), the Kruse Chair Endowment (L.K.), the Retina Research Foundation (L.K. and T.W.H.), and National Institutes of Health National Eye Institute R01-EY-023335 and R01-EY-024624 (T.W.H.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.-L.C. designed and conducted experiments, analyzed data, and wrote the manuscript. W.X. conducted experiments. R.H.R. and L.K. reviewed and edited the manuscript. T.W.H. designed experiments and reviewed and edited the manuscript. T.W.H. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Data and Resource Availability. All data generated or analyzed during this study are included in the published article and its online Supplementary Data.

Prior Presentation. Parts of this study were presented in poster form at the Association for Research in Vision and Ophthalmology 2017 Annual Meeting, Baltimore, MD, 7–11 May 2017, and at the Microcirculatory Society Annual Meeting at Experimental Biology 2018, San Diego, CA, 21–25 April 2018.