Exendin-4 promotes retinal ganglion cell survival and function by inhibiting calcium channels in experimental diabetes

Summary Progressive damage of retinal ganglion cells (RGCs) is observed in early diabetic retinopathy. Intracellular Ca2+ overload mediated by Ca2+ influx through voltage-gated Ca2+ channels (VGCCs) is involved in neurodegeneration, whereas glucagon-like peptide-1 (GLP-1) provides neuroprotection. However, whether GLP-1 plays a neuroprotective role in diabetic retinas by modulating VGCCs remains unknown. We found that eye drops of exendin-4, a long-acting GLP-1 receptor (GLP-1R) agonist, prevented the increase of L-type Ca2+ current (ILCa) densities of RGCs induced by 4-week hyperglycemia and promoted RGC survival by suppressing L-type VGCC (L-VGCC) activity in streptozotocin-induced diabetic rats. Moreover, exendin-4-induced suppression of ILCa in RGCs may be mediated by a GLP-1R/Gs/cAMP-PKA/ryanodine/Ca2+/calmodulin/calcineurin/PP1 signaling pathway. Furthermore, exendin-4 functionally improved the light-evoked spiking ability of diabetic RGCs. These results suggest that GLP-1R activation enhances cAMP to PP1 signaling and that PP1 inactivates L-VGCCs by dephosphorylating them, thereby reducing Ca2+ influx, which could protect RGCs against excitotoxic Ca2+ overload.


INTRODUCTION
Diabetic retinopathy (DR) is a major complication of diabetes mellitus (DM) and remains the leading cause of blindness globally. 1,2Due to the complexity of the disease and researchers' limited understanding of its pathogenic mechanisms, there are no effective therapeutic strategies for this disease.3][14][15][16] Ca 2+ overload is caused by a massive Ca 2+ influx through voltage-gated Ca 2+ channels (VGCCs) and ionotropic glutamate receptor channels.It is harmful to the cells because it constitutes the final common pathway for neuronal death under various pathological conditions, [17][18][19] including glaucoma patients and animal models, and rat models with retinal ischemia.[20][21][22][23] However, no studies have investigated the changes and roles of VGCCs in RGCs for DR development.L-type voltage-gated Ca 2+ channels (L-VGCCs) have been shown to be expressed in rat RGCs.24 L-VGCCs are high-voltage-activated ones that do not inactivate with time 25 ; therefore, their continuous activation may contribute significantly to the lethal Ca 2+ influx under pathological conditions. 22 Based othese research backgrounds, the first purpose of this work was to explore whether L-VGCCs in RGCs are altered in streptozotocin (STZ)-induced diabetic rats, a model of human type 1-like diabetes.
Therapeutic strategies based on neuroprotection in the early stages of DR could be effective in preventing visual impairment.Glucagonlike peptide-1 (GLP-1) is a metabolic hormone secreted from intestinal endocrine L-cells that stimulates glucose-dependent insulin secretion. 26Activation of GLP-1 receptor (GLP-1R) provides neuroprotection in a variety of experimental models of neurodegenerative disorders. 27,283][34][35][36] The presence of GLP-1R in the retina could serve as a potential new target for treating neurodegeneration through GLP-1R agonists.Actually, GLP-1 and GLP-1R agonists have been shown to prevent electroretinogram abnormalities and retinal neurodegeneration, protect the blood retinal barrier and also affect autophagy in diabetic animal models. 12,29,32,33However, whether GLP-1 and GLP-1R agonists exert neuroprotective effects by affecting the cellular calcium system in the retina is still unknown, although Gilman et al. reported that GLP-1 suppressed Ca 2+ currents and glutamate-induced currents in cultured rat hippocampal neurons. 37M upregulates I LCa densities and these effects were improved by Ex-4   We investigated whether the I LCa in RGCs in DM rats after 4 weeks of hyperglycemia was altered.Figure 2A shows typical I LCa traces that were recorded in a control RGC and in a diabetic RGC before and after Ex-4 application.Plotting the I-V curves of I LCa showed that hyperglycemia significantly and voltage-dependently enhanced I LCa densities (p < 0.0001 vs. control, two-way RM ANOVA, Figure 2B) and that perfusion of Ex-4 significantly suppressed the I LCa densities of diabetic RGCs (p < 0.05 vs. DM, Figure 2B).At À10 mV, I LCa density was increased to À53.40 G 21.96 pA/pF in DM group (p < 0.001, Figure 2C) from the control value of À36.29 G 19.12 pA/pF.However, this increase was significantly suppressed by additional Ex-4 (39.53 G 16.60 pA/pF, p < 0.01 vs. DM).
We further tested whether Ex-4 eye drops had an effect on diabetic RGCs by first demonstrating that Ex-4 administered via eye drops could reach the retina.We observed a significant increase in the amount of Ex-4 in the retinas 1 (0.082 G 0.009 ng) and 2 h (0.055 G 0.001 ng) after topical administration (Figure S2A).Ex-4 was then administered by eye drops (20 mg/kg) twice daily starting 2 weeks after the onset of DM and was prolonged for 2 more weeks (Figure 2D).Ex-4 did not reduce blood glucose levels or affect body weight in DM rats (Figures S2B and S2C), but did reduce I LCa in diabetic RGCs (Figure 2E).The I-V curve for the DM + Ex-4 group exhibited a significant downward shift compared to the DM + Saline (vehicle) group (two-way RM ANOVA, p < 0.01, Figure 2F).The Ex-4-induced suppression of I LCa density was observed at both À10 mV and 0 mV (p < 0.05, Figures 2F and 2G).

Topical administration of Ex-4 promotes diabetic RGC survival by suppressing L-VGCCs
To determine whether the effect of Ex-4 on L-VGCCs promotes RGC survival in DM retinas, we used the antibody for Brn3a, a specific marker for RGCs, 45 and counted the number of Brn3a-labeled (Brn3a+) RGCs in whole-mounted retinas.Saline, Ex-4, Ex-4+Ex-9-39, BayK-8644 or BayK-8644+Ex-4 was respectively administrated by eye drops twice daily for 2 weeks starting 2 weeks after the onset of DM (refer to Figure 2D); afterward, RGC survival was evaluated.Figure 3 shows the representative images of different areas of the retinas for different groups.The number of RGCs was assessed in 12 regions: 4 peripheral (Figure 4A), 4 middle (Figure 4B) and 4 central regions (Figure 4C) (for details see STAR Methods).As shown in Figures 4D and 4E, the effects of the 7 groups on the mean densities of surviving RGCs in the peripheral and middle regions were significant (one-way ANOVA, peripheral, p < 0.0001; middle, p < 0.0001).The mean densities of RGCs in the peripheral (1311 G 212.7 cells/mm 2 ) and middle regions (1873 G 218.6 cells/mm 2 ) in DM eyes were significantly lower than those in the control eyes (peripheral: 1639 G 189.1 cells/mm 2 , p < 0.0001, Figure 4D; middle: 2107 G 170.5 cells/mm 2 , p < 0.05, one-way ANOVA followed by Tukey's multiple comparisons test; Figure 4E).In DM eyes, Ex-4 markedly increased RGC survival because the RGC densities in the peripheral and middle regions were 1561 G 244.7 cells/mm 2 and 2081 G 295.5 cells/mm 2 in DM + Ex-4 eyes compared with 1292 G 184.8 cells/mm 2 (p < 0.001, Figure 4D) and 1877 G 212.0 cells/mm 2 (p < 0.05, Figure 4E) in DM + Saline eyes.These data indicate that topical administration of Ex-4 significantly enhances the survival of RGCs in DM eyes.
Different from the peripheral and middle regions of retinas, the effects of the 7 groups on the mean densities of surviving RGCs in the central region were not significant (p > 0.05, one-way ANOVA).Although the RGC density tended to decrease due to DM, no significant difference was observed between DM (2086 G 251.5 cells/mm 2 ) and control retinas (2304 G 309.9 cells/mm 2 , p > 0.05, Figure 4F).In addition, we examined overall RGC density of whole retina for the significance test and found that the mean RGC density in DM retinas (1757 G 192.9 cells/mm 2 ) was significantly lower than that in control retinas (2017 G 158.0 cells/mm 2 , p < 0.001, data not shown).Moreover, Ex-4 significantly enhances the survival of overall RGCs in DM eyes (1980 G 257.9 cells/mm 2 in DM + Ex-4 vs. 1782 G 164.5 cells/mm 2 , p < 0.01, in DM + Saline).
We also measured the area of the whole retina and found no significant difference between DM (57.20 G 3.3 mm 2 , n = 10) and control groups (56.95 G 3.3 mm 2 , n = 9) (p > 0.05, data not shown).All these results suggest that RGC reduction mainly occurs in the peripheral and middle retinas in 4-week DM rats and that topical administration of Ex-4 has a significant protective effect on RGCs in these regions, whereas RGCs in the central region were less affected by DM.

cAMP-PKA signaling pathway mediates Ex-4-induced suppression of I LCa
Following GLP-1R activation, the main intracellular signaling pathway stimulates Gs, which in turn activates cAMP-PKA in neurons and pancreatic beta cells. 46,47To investigate whether this pathway is involved, we added 3 mM GDP-b-S, a nonhydrolyzable G-protein inhibitor, into patch pipettes.When the I LCa amplitudes of RGCs reached a steady level (control), Ex-4 application did not change the I LCa (94.73 G 14.69% of control, p > 0.05, Figures 5A and 5B).When cell suspensions were pre-incubated with 10 mM NF-449, a Gsa protein inhibitor, for 20 min before recording, additional Ex-4 failed to change the I LCa in RGCs (98.14 G 12.85% of control, p > 0.05, Figures 5C and 5D).However, perfusion of 500 mM 8-Br-cAMP, a membrane-permeable cAMP analog, significantly suppressed the I LCa to 76.53 G 12.96% of control (p < 0.01, Figures 5E and 5F), thus mimicking the Ex-4 effect.Furthermore, applying 50 mM of RP-cAMP, a PKA inhibitor, did not change the I LCa (101.60 G 13.92% of control, p > 0.05, Figures 5G and 5H), and co-applying Ex-4 no longer suppressed the currents (96.22 G 12.04% of control, p > 0.05, Figures 5G and 5H).
Ca 2+ , calmodulin (CaM), calcineurin and protein phosphatase 1 (PP1) are involved in Ex-4-induced suppression of I LCa Since our recent work demonstrated that Ex-4 increased the intracellular Ca 2+ concentration ([Ca 2+ ] i ) in isolated rat RGCs via PKA activation, 48 we further investigated whether the Ex-4 effect on I LCa in RGCs depends on [Ca 2+ ] i .BAPTA (10 mM) was added to the recording pipette to chelate intracellular free Ca 2+ , additional Ex-4 did not suppress I LCa in RGCs (98.67 G 12.63% of control, p > 0.05, Figures 6A and 6B).No Ca 2+ was present in the extracellular solution for our experiments, indicating that changes in [Ca 2+ ] i are potentially due to an altered Ca 2+ release from intracellular calcium stores rather than changes in Ca 2+ influx via Ca 2+ channels.Inositol 1,4,5-trisphosphate (IP 3 )-and/or ryanodine-sensitive pathways mediate Ca 2+ release from intracellular calcium stores.Ex-4 application still suppressed I LCa in RGCs during intracellular infusion of 5 mg/mL of the IP 3 receptor antagonist heparin (64.91 G 19.02% of heparin alone, p < 0.01, Figures 6C and 6D).However, after internal infusion of 50 mM ryanodine, which depletes ryanodine-sensitive Ca 2+ sites, Ex-4 no longer suppressed the I LCa (98.57G 10.05% of ryanodine alone (p > 0.05, Figures 6E and 6F).
Previous studies showed that activating cAMP and/or PKA suppressed I LCa in neurons. 49,50However, these suppressing effects may be indirect because Ca 2+ has been shown to activate calcineurin by binding CaM, which relieves PP1 inhibition, and then PP1 inactivates L-VGCCs by dephosphorylating them. 51We tried to test whether this signaling pathway was also involved in the Ex-4 effect on I LCa .As shown in Figures 6G and 6H, after intracellular application of the CaM inhibitor W-7 (100 mM) for about 8 min, I LCa reached a steady level (control), and extracellular perfusion of Ex-4 failed to suppress the I LCa of RGCs (107.30G 13.40% of control).Similarly, during intracellular application of the calcineurin inhibitor FK-506 (50 mM) or the PP1 inhibitor okadaic acid (OA, 1 mM), adding Ex-4 no longer reduced the I LCa in RGCs (99.99 G 4.76% of control for FK-506, 101.30G 8.31% of control for OA, p > 0.05 for both, Figures 6I-6L).Moreover, applying the PP1 activator C2 Ceramide (500 nM) suppressed I LCa to 66.5 G 8.41% of control (p < 0.001, Figures 6M and 6N).These results suggest the involvement of the CaM/calcineurin/PP1 signaling pathway in the Ex-4 effect.
Although GLP-1R has also been shown to activate the downstream phospholipase C (PLC)/PKC signaling pathway, 52 our results did not support the involvement of this pathway in the Ex-4 effect on I LCa in RGCs.As shown in Figures S3A and S3B, during internal infusion of 10 mM U-73122, an inhibitor of PLC, perfusion of Ex-4 still suppressed the I LCa to 71.44 G 6.11% of control (n = 6, p < 0.0001).Moreover, in the presence of 10 mM Bis-IV, a PKC inhibitor, Ex-4 also significantly suppressed the currents (58.19 G 9.98% of control, n = 6, p < 0.001; Figures S3C and S3D).

Ex-4 ameliorates the functional changes in RGCs induced by DM
To examine whether DM causes functional abnormalities in RGCs, we performed high-throughput evaluations of light responses of RGCs using the multielectrode array (MEA) technique.Figure 7A shows the typical raster graphics of light-evoked spikes from the ON-, OFFand ON-OFF-RGCs at a light intensity of 5.86310 10 photons/cm 2 /s recorded by MEA in different groups.Light-evoked spike numbers increased as a function of light intensity in ON-RGCs in the control, DM, DM + Saline and DM + Ex-4 groups.In the DM groups, the total spike numbers of ON-, OFF-and ON-OFF-RGCs significantly decreased compared to those in corresponding control groups [two-way RM ANOVA, p < 0.0001 for ON-and ON-OFF-RGCs (for both on-and off-responses); p < 0.05 for OFF-RGCs, Figures 7B-7E], as evidenced by the significant downward scaling of the irradiance-response (I-R) curves, suggesting a decrease in photoresponse.However, the I-R curves of ON-and ON-OFF-RGCs in the DM + Ex-4 group significantly shifted upward compared to the DM group [p < 0.001 for ON-RGCs; p < 0.01 for ON-OFF-RGCs (for both on-and off-responses), Figures 7B, 7D, and 7E], suggesting an increased response gain due to Ex-4 administered by eye drops.In contrast, the I-R curve of OFF-RGCs in the DM + Ex-4 group was not significantly different from the DM group (p > 0.05, Figure 7C), indicating that Ex-4 does not work.No significant differences in total spike number were detected between the DM group and the DM + Saline group for the three types of RGCs (p > 0.05).

DISCUSSION
In this work, we first show that 4 weeks of hyperglycemia resulted in downregulation of rat retinal GLP-1 mRNA levels and upregulation of I LCa density in RGCs.Second, Ex-4 eye drops prevented upregulations of I LCa densities and retinal L-VGCC protein expression induced by hyperglycemia.Third, Ex-4 eye drops promoted RGC survival and its protective effects could be blocked by the competitive GLP-1R antagonist Ex-9-39 or the selective L-VGCC agonist BayK-8644.Fourth, Ex-4 significantly reduced I LCa in RGCs and the inhibitory role could be mediated by exciting the GLP-1R/Gs/cAMP-PKA/ryanodine/Ca 2+ /CaM/calcineurin/PP1 signaling pathway.Fifth, Ex-4 decreased phosphorylation of Cav1.2 in DM retinas.Sixth, Ex-4 improved the light-evoked spiking ability of ON-and ON-OFF-RGCs in DM rats.Our results established a mechanism by which activation of GLP-1R protects damaged RGCs via modulating L-VGCCs.A schematic diagram of possible mechanisms is shown in (Figure S4).These results suggest that L-VGCCs may be a therapeutic target for inhibiting early DR and topical administration of GLP-1R agonists could provide a noninvasive and effective approach in the treatment of the retinopathy independently of their hypoglycemic action.
1][22][23] This is the first work to show an upregulation of I LCa densities in diabetic RGCs, suggesting an increased Ca 2+ influx in RGCs at the early stage of DM.This is consistent with a previous study on dorsal horn neurons that showed an increase in activity for L-VGCCs in DM rats. 53Moreover, the expression levels of the Cav1.2 subunit in the DM retinas were significantly increased, which may be one of the reasons for the increase of I LCa in RGCs.GLP-1 and GLP-1R were expressed in the vertebrate retinas, [29][30][31][32][33][34][35][36] suggesting that the GLP-1 system plays an important role in the regulation of visual function.Actually, our recent study showed that GLP-1/Ex-4 significantly suppresses GABA receptor-mediated currents in rat RGCs through GLP-1R activation. 48Here we further found that perfusion of Ex-4 significantly suppressed I LCa in normal and diabetic RGCs and Ex-4 eye drops also reversed the upregulation of I LCa densities and expression levels of Cav1.2 induced by hyperglycemia.GLP-1/Ex-4 was consistently shown to suppress Ca 2+ currents in cultured hippocampal neurons and hypothalamic neurons. 37,54In contrast to these results, GLP-1 enhanced I LCa in canine cardiomyocytes. 55These data suggest that the GLP-1 effect on the Ca 2+ currents is cell-type-dependent.We demonstrated that RGC density decreased significantly in the peripheral and middle retinal regions in 4-week DM rats, but decreased more in the peripheral region.These results were consistent with previous studies showing RGC loss in STZ-induced 4-week diabetic rats. 5,56GC density in the central retinal region of DM rat also showed a decreasing trend.These results suggest that RGC death caused by hyperglycemia starts in the peripheral retina and gradually develops to the central retina.Similar damage also occurred in Ins2 Akita DM mice, with RGC loss being detected in the peripheral retina after 3 months of hyperglycemia but not in the central retina. 4Previous study showed that GLP-1 mRNA and protein were downregulated in retinas from donors with DM. 29 GLP-1 has been shown to cross the blood-brain barrier to exert effective effects in the central nervous system. 38,41However, whether GLP-1 can cross the blood-retinal barrier remains unknown.But since the endogenous GLP-1 has a very short half-life in plasma (only about 1-2 min), 26,57 due to degradation by the enzyme dipeptidyl peptidase-4, 26,44,[58][59][60][61] the amount of GLP-1 that can reach the retina through blood circulation may be limited.If GLP-1 does not cross the blood-retinal barrier, systemic GLP-1 could not activate its receptors within the retina.In addition, since GLP-1 is expressed by retinal neurons, it is reasonable to assume that GLP-1R in the retina respond to local release of GLP-1.We also identified a lower level of GLP-1 mRNA in DM rat retinas and Ex-4 eye drops reversed the RGC loss in DM rats without any effect on blood glucose levels.These results strongly suggest that Ex-4 has a direct neuroprotective effect that is independent of its ability to increase insulin secretion or lower blood glucose levels.Previous studies have also shown that systemic or local administration of GLP-1 or GLP-1R agonists reduces the loss of retinal neurons in diabetic animals. 12,29,33However, the mechanisms underlying the neuroprotection induced by GLP-1R agonists are not fully understood.We found that the protective effect of Ex-4 on diabetic RGCs was blocked by either the GLP-1R antagonist Ex-9-39 or the L-VGCC agonist BayK-8644, suggesting that the Ex-4 effect on RGCs is mediated by GLP-1R and L-VGCCs.In this regard, it has been shown that administering calcium channel blockers promotes RGC survival in patients and animal models with glaucoma 20 and in a retinal ischemic rat model. 23All of these data suggest that Ex-4 promotes RGC survival by reducing intracellular Ca 2+ overload by inhibiting L-VGCCs of RGCs in DM retinas.In addition, previous studies showed that the protective mechanisms of GLP-1R activation on DM retinal neurons included increased prosurvival signals and decreased proapoptotic signals, proinflammatory cytokines and extracellular glutamate, as well as antioxidant effects. 12,29,33,62hrough pharmacological dissections, we demonstrated that a distinct cAMP-PKA signaling pathway following GLP-1R activation may be responsible for the Ex-4 effect on I LCa in RGCs.This signaling pathway in GLP-1R-mediated effects has been demonstrated in both hippocampal neurons and pancreatic beta cells. 46,47However, the pathway by which PKA activation inhibited I LCa remains unknown.Our recent study showed that Ex-4-induced PKA activation causes an increase in [Ca 2+ ] i in rat RGCs. 48The elevation of [Ca 2+ ] i induced by Ex-4 was potentially the result of increased Ca 2+ release from ryanodine receptor-mediated intracellular stores.Activating cAMP and/or PKA has been shown to inhibit I LCa in tiger salamander rod and rat pinealocytes. 49,50PKA activation may indirectly suppress I LCa through the Ca 2+ /CaM/calcineurin/ PP1 signaling pathway, in which PP1 catalyzes the dephosphorylation of L-VGCCs, curtailing Ca 2+ entry. 51This pathway may also underlie the Ex-4 effect on I LCa because the Ex-4-induced inhibition of I LCa no longer occurred when CaM, calcineurin and PP1 were respectively blocked.
MEA recordings showed that the light-induced total spike number of ON-, OFF-and ON-OFF-RGCs were significantly decreased in 4-week DM rats, indicating that DM has a significant impact on overall RGC-mediated retinal outputs to higher centers.The ON and OFF signaling pathways and their interactions in the visual system were reported to be closely related to contrast detection 63 ; therefore, it is reasonable to speculate that declined visual contrast sensitivity, which has been considered an early sign of neural retinal dysfunction in diabetic patients and animals, [64][65][66] may be related to the diminished light responses of RGCs that we observed.Notably, Ex-4 eye drops significantly increased the light responses of ON-and ON-OFF-RGCs, but not OFF-RGCs.
8][69] Moreover, initial worsening of DR has been reported to be likely the consequence of rapid improvement of hyperglycemia (large reduction of HbA1c/rapid improvement of blood glucose levels). 34,70,71However, a recent paper shows that the early worsening of DR by rapid optimization of blood glucose levels is not a clinical problem in patient with mild or moderate DR. 72The development of new techniques for retinal imaging and functional assessment will help us to detect early changes and further to clarify whether therapies based on GLP-1R activation are good or bad for the human diabetic retina. 73

Limitations of the study
We did not further investigate the mechanism by which Ex-4 improved the light-evoked spiking ability of RGCs in DM rats.Previous studies demonstrated that GLP-1R immunostaining exists in the GCL and that there are weak positive signals in the inner nuclear layer and inner plexiform layer of retina, 29,30,[32][33][34][35][36] suggesting that Ex-4 may also play a neuroprotective role for the presynaptic cells of RGCs, such as amacrine cells and/or bipolar cells, whose somata are located in the inner nuclear layer and processes in the inner plexiform layer.However, the exact sites and mechanisms of these protective effects need to be explored further.5 weeks were purchased from SLAC Laboratory Animal Company (Shanghai, China).Since males are more susceptible to STZ than females, only male rats (which have a higher percentage of diabetes) were used to minimize the number of animals used.Rats were housed in a temperature-controlled room (25 G 1 C) under a 12:12-hour light/dark cycle, with food and water supplied ad libitum.All efforts were made to minimize animal suffering and to reduce the number of animals used.

Induction of diabetic rat model
Diabetes was induced in rats aged 4-5 weeks by a single intraperitoneal injection of STZ (70 mg/kg body weight, Sigma-Aldrich Corp., St. Louis, MO, USA) freshly dissolved in a sodium citrate buffer (0.1 M, pH 4.2) after fasting for 12 hours.Age-matched control rats received an equal volume of sodium citrate buffer.The blood glucose levels and body weight were measured before injection, 3 days after the STZ injection and once a week thereafter for 4 weeks (Figure S1).Rats with blood glucose levels exceeding 16.7 mM were considered to be diabetic. 74,75al-time quantitative reverse transcription PCR Rats were deeply anesthetized with an intraperitoneal injection of 25% urethane (10 mL/kg).Retinas were harvested, and total RNA (200 ng) was extracted and reverse-transcribed using commercially available kits (Takara Bio Inc., Shiga, Japan) according to the manufacturer's instructions.Quantitative RT-PCR was performed using TB Green Premix Ex Taq II (Takara Bio Inc.) on the Applied Biosystems QuantStudio 3 96-Well 0.2-mL Block.The primer sequences of GLP-1 and beta-2-microglobulin (B2M) were as follows: 5 0 -GGAGGGCCAGGCAG CAAAGG-3 0 (forward) and 5 0 -TCTGCGCCCAAGTTCCTCAGC-3 0 (reverse), 76 5 0 -ATCTTTCTGGTGCTTGTCTCT-3 0 (forward) and 5 0 -TGAGGT GGGTGGAACTGAGA-3 0 (reverse), 77,78 respectively.Quantitative PCR results were analyzed using the 2 -DDCt method 79,80 : Fold change = 2 -DDCt and DDCt = (Ct 1 -Ct 2 ) -(Ct 3 -Ct 4 ).Ct 1 and Ct 2 denoted the critical number of cycles for the target gene (GLP-1) and the housekeeping gene (B2M) respectively in the sample of a diabetic rat.Ct 3 and Ct 4 denoted the critical number of cycles for the target gene (GLP-1) and the housekeeping gene (B2M) in the sample of a control rat.

Topical ocular treatment
Referring to the previous study, 29 eye drops with drugs or vehicle (0.9% sodium chloride) were administered directly onto the superior corneal surface of both eyes using a micropipette after 2 weeks of hyperglycemia (Figure 2D).Some rats received eye drops (4-6 mL) of Ex-4 (40 mg/kg/ day), Ex-9-39 (60 mg/kg/day)+Ex-4, BayK-8644 (140 mg/kg/day) or BayK-8644+Ex-4.The treatment was repeated twice daily for 14 days.The age-matched control rats received eye drops of saline.At day 31 post-STZ, the rats were killed by intraperitoneal injection of 25% urethane (10 mL/kg) and the eyes were enucleated.The separated retinas were used for patch-clamp recording, western blotting and immunofluorescent labeling.

Enzyme linked immunosorbent assay (ELISA)
To estimate whether the topical administration of Ex-4 reaches the retina, Ex-4 concentration in the retinas of normal rat eyes 1 or 2 hours after administration of Ex-4 with eye drops (20 mg/kg) was evaluated using ELISA.The separated retinas were freshly dissected and sonicated in a lysis buffer containing 20 mM Tris (pH7.5), 150 mM NaCl, 1% Triton X-100, and a protease inhibitor cocktail (P8340, Sigma-Aldrich).The level of Ex-4 was determined by Exendin-4 (Heloderma suspectum) Enzyme Immunoassay Kit (Phoenix Pharmaceuticals, Inc., Burlingame, CA, USA) following the manufacturer's instructions.

Membrane protein extraction and western blot analysis
Western blot analysis was performed as described previously with minor modification. 75In brief, membrane protein was extracted from four retinas (as one sample) using a membrane protein extraction kit (K268, Biovision Inc., Milpitas, CA, USA).Membrane protein concentrations were determined using a standard bicinchoninic acid (BCA) assay kit (23235, Thermo Fisher Scientific, Waltham, MA, USA).Equivalent amounts of freshly extracted samples were loaded (for the detection of p-Cav1.2, the loading sample volume was 3 times that of the detection of Cav1.2, due to the low content of p-Cav1.2),subjected to 8% SDS-PAGE, and then transferred onto PVDF membranes (Immobilon-P, Millipore Co. Bedford, MA, USA).The blots were blocked for 2 hours at room temperature in a blocking solution consisting of 20 mM Tris-HCl (pH 7.4), 137 mM NaCl, 0.1% Tween-20 and 3% bovine serum albumin, and then in the same solution containing the antibody against Cav1.2 (1:500 dilution, ACC-003, Alomone, Jerusalem, Israel) or p-Cav1.2(1:200 dilution, 12674, Signalway Antibody, Greenbelt, MD, USA) overnight at 4 C. Na + -K + -ATPase (1:20000 dilution, ab76020, Abcam, Cambridge, UK) was used as a loading control.After that, the blots were incubated in horseradish peroxidase-conjugated donkey anti-rabbit IgG secondary antibody (1:2000 dilution, Jackson Labs, West Grove, PA, USA) for 2 hours at room temperature.The immunosignals were visualized with enhanced chemiluminescence (Thermo Fisher Scientific) and analyzed on ChemiDoc XRS System with Image Lab software (Bio-Rad, Hercules, CA, USA).

Labeling and counting of RGCs
The procedures of immunocytochemistry refer to our previous work, 81 with minor modifications.In brief, isolated retinas from both eyes were immediately fixed in fresh 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 2 hours at room temperature.Then the whole-mount retinas were blocked in 0.1 M phosphate buffered-saline (PBS, pH 7.4) containing 6% donkey serum and 1% Triton X-100 for 2 hours at room temperature.Mouse anti-Brn3a antibody (1:50 dilution, 3 days at 4 C, Millipore) was used to label RGCs. 45Immunoreactivity was detected with donkey anti-mouse IgG tagged with Alexa Fluor 488 (1:200 dilution, 2 hours at room temperature, Invitrogen, Carlsbad, CA, USA).
The number of Brn3a-positive (Brn3a+) nuclei was counted by an investigator who was blinded to the study treatments using a fluorescence microscope (Axioskop 40, Carl Zeiss Inc., Oberkochen, Germany) under a 403 objective lens.Each retina was equally divided into 4 quadrants including nasal (N), dorsal (D), temporal (T) and ventral (V) under a dissection microscope.As shown in Figures 4A-4C, twelve 256 3 256 mm areas (centered at 1, 2 and 3 mm from the optic nerve head in four quadrants) per whole mount were imaged and cell counting was conducted with ImageJ software.RGC densities (cells/mm 2 ) were grouped by retinal eccentricity (central, middle and peripheral) and expressed as the mean G SD.For the measurement of the retinal area, a series of photomicrographs for each retina was captured automatically by a fluorescence microscope (Eclipse Ni-U, Nikon, Tokyo, Japan) under a 203 objective lens, and the whole retina image was digitally reconstructed by NIS-Elements imaging software (Nikon).The whole-retina area was delineated using the Polygon tool in ImageJ software.

Retrograde labeling of RGCs
The detailed procedure refers to our previous work. 48,82Briefly, after the rats were deeply anaesthetized, 20% rhodamine-labelled fluorescent latex microsphere (LumaFluor, Durham, NC, USA) was injected into the superior colliculus bilaterally.Following a survival time of 2-3 days, RGCs were labelled clearly for electrophysiological recording.
For recording Ca 2+ currents, isolated cells were perfused with Ba 2+ Ringer's solution, in which 5 mM Ba 2+ was used to substitute for Ca 2+ , containing (in mM) NaCl 120, KCl 2.5, BaCl 2 5, CsCl 5, HEPES 15, TEA-Cl 15, glucose 10 and tetrodotoxin (TTX) 0.4 mM, adjusted to pH 7.4 with NaOH. 83Ba 2+ can go through Ca 2+ channels as a current carrier, thus making the currents larger in size without modifying the intracellular Ca 2+ -dependent processes. 84Patch pipettes were made by pulling borosilicate glass capillaries (BF150-86-10, Sutter Instrument Co., Novato, CA, USA) on a micropipette puller (P-97, Sutter Instrument Co.).The pipettes (4-6 MU) were filled with internal solution containing (in mM): CsCl 128, CaCl 2 1, MgCl 2 2, EGTA 10, HEPES 10, ATP-Mg 2, GTP-Na 0.4, phosphocreatine 10, adjusted to pH 7.2 with CsOH and osmolality to 290-300 mOsm/L with sucrose.Whole-cell membrane currents of RGCs were recorded using a patch amplifier (EPC 9/2; HEKA Elektronik, Lambrecht/Pfalz, Germany).Fast capacitance was fully canceled, and cell capacitance was partially canceled as much as possible by the circuits of the amplifier.Seventy percent of the series resistance of the recording electrode was compensated.High-voltage-activated Ca 2+ currents of RGCs were induced by a series of 150 ms depolarizing pulses from À40 mV (the holding potential) up to +10 mV in increments of 10 mV, and the average current amplitudes of the voltage pulses during 135-140 ms were detected and analyzed.Data were acquired at a sample rate of 20 kHz, filtered at 2 kHz, and then stored for further analysis.A fast solution exchanger (RSC-160, Bio-Logic, Claix, France) based on a stepping motor, was used for solution transport, with a solution exchange time being $5 ms.All experiments were performed at room temperature (20-25 C).

MEA recording
The procedures for MEA refer to our previous work, 11,85 with minor modifications.Briefly, rats were dark-adapted overnight (> 12 h) and anesthetized by isoflurane under dim red light.Retinas were dissected free from the eyecup and one quarter of the retina was placed photoreceptor-side down on a piece of Anodisc filter membrane (Whatman, Piscataway, NJ, USA).The mounted retina was transferred into the recording chamber of an MEA chip (60MEA200/30iR-ITO-gr, Multi-Channel Systems GmbH, Reutilingen, Germany) and the GCL faced the array.The retina was continuously superfused with oxygenated bicarbonate-buffered Ames' medium (5-6 mL/min) and maintained at 30 G 2 C using a temperature controller (TC-324B, Warner Instruments, Hamden, CT, USA).The retina sat in the chamber for 40 min in the dark before beginning recordings to permit stabilization of spike amplitudes.Voltage data were digitized at 10 kHz, amplified and acquired using the USB MEA60 Inv BC System and MC Rack software (Multi Channel Systems).Signals were high-pass filtered at 200 Hz and stored on a personal computer.Retinas were full-field stimulated by 1-s 500-nm light flashes generated using a custom-modified fiber optic LED illuminator (Model 66991, DiCon Fibersoptics Inc., Richmond, CA, USA) with stimulus timing controlled by a logic circuit integrated in the illuminator.The intensity of light stimulation ranged from 3.42 3 10 6 to 5.86 3 10 10 photons/cm 2 /s by introducing different neutral density filters (Edmund Optics Inc., Barrington, NJ, USA) into the light path.Inter stimulus intervals increased progressively within the series, ranging from 1 min between the first three dim stimuli to 5 min between the last two brightest ones.The recording of each stimulus was kept for 3 s, with 1 s before, during and after light stimulation.
Spike sorting of the raw recording data followed a protocol previously described 85 using Offline Sorter software (Plexon Inc., Dallas, TX, USA).The spike detection threshold was set for each channel at 3-4 times the standard deviation of the voltage.The detected spike waveforms were subjected to cluster analysis using the first three principal components, and the resulting clusters were manually corrected for clustering errors.RGCs were classified into three types by their spiking pattern to light stimuli, which was defined by the response dominant index (RDI) according to previous studies. 86,87RDI is computed using the equation: RDI = (R ON ÀR OFF )/(R ON +R OFF ), where R ON and R OFF are the spike number during the 1-s light stimulus and 1 s after the light offset respectively, minus the spike number of 1 s preceding the light onset.Cells with an RDI smaller than À0.6 or larger than 0.6 at more than half of light intensities were defined as OFF-and ON-RGCs, respectively; cells with an RDI between À0.6 and 0.6 were defined as ON-OFF RGCs.The total spike number that is R ON (for ON RGC) or R OFF (for OFF RGC) in the equation was analyzed.

Chemicals
Ex-4, nimodipine, Ex-9-39 and BayK-8644 were purchased from MedChemExpress (Monmouth Junction, NJ, USA), while TTX, Ryanodine, W-7 and U-73122 were from Tocris Bioscience (Ellisville, MO, USA).NF-449 was purchased from Cayman Chemical (Ann Arbor, MI, USA) and C2 Ceramide from Santa Cruz (Dallas, TX, USA).The others were purchased from Sigma-Aldrich.Nimodipine, Bis-IV, U-73122, ryanodine, W-7, FK-506, OA, and C2 Ceramide were initially dissolved in DMSO for stock.The final concentration of DMSO was less than 0.1%, with no effects on the calcium currents of RGCs.All other drug solutions were prepared in ion-free water, stored at À20 C and freshly diluted to the final concentrations using extracellular or intracellular solutions.For topical ocular administration, Ex-4 and Ex-9-39 were dissolved in saline, and BayK-8644 was dissolved in 50% PEG-400 diluted with saline.

QUANTIFICATION AND STATISTICAL ANALYSIS
All data were analyzed using Excel (Microsoft, Redmond, WA, USA), Pulsefit 8.62 (HEKA Elektronik), and GraphPad prism 8 (GraphPad Software, San Diego, CA, USA).To identify significant differences, paired t test, unpaired t test, one-way ANOVA followed by post hoc Tukey's multiple comparisons tests and two-way repeated measures (RM) ANOVA followed by post hoc Sidak's multiple comparisons tests were used.Unless otherwise specified, p values represented the results of the paired or unpaired t test.p < 0.05 was considered statistically significant.

Figure 1 .
Figure 1.Ex-4 suppressed I LCa in RGCs via GLP-1R (A) Real-time quantitative RT-PCR analysis showing a significant decrease in GLP-1 mRNA levels in DM retinas (n = 6, ****p < 0.0001 by unpaired t test).Beta-2microglobulin was used as a house-keeping gene.(B) Representative current recordings from an acutely isolated RGC.The cell was held at À40 mV, and the high-voltage activated Ca 2+ currents were induced by depolarization voltage pulses from À30 mV to +10 mV in increments of 10 mV.The currents were almost completely suppressed by 10 mM nimodipine.(C) Bar chart summarizing the changes of current amplitudes after nimodipine application (n = 8).(D) Representative I LCa traces of an RGC, showing that extracellular application of Ex-4 (200 nM) suppressed the current amplitudes.(E) Current-voltage (I-V) relationship curves showing that Ex-4 voltage-dependently suppressed the I LCa amplitudes (n = 9).(F and G) Current traces and bar chart showing the effect of Ex-4 on I LCa recorded at À10 mV (n = 9).(H) Ex-4 suppressed I LCa in a dose-dependent manner.Only one concentration of Ex-4 was tested for a single cell.Each cell was recorded before Ex-4 perfusion (control) and during Ex-4 perfusion, and each Ex-4 response was then normalized against its corresponding control response, which is always 100%.Cell numbers are marked inside the bars.(I) Representative currents showing that 100 nM Ex-9-39 blocked the Ex-4 induced suppression of I LCa .(J) Bar chart summarizing the effects of Ex-9-39 and Ex-4 on I LCa obtained at À10 mV (n = 8).Data are presented as mean G SD. n.s., p > 0.05, *p < 0.05, **p < 0.01 and ****p < 0.0001 by paired t test.

Figure 2 .
Figure 2. Ex-4 inhibited hyperglycemia-induced increases in I LCa densities of RGCs (A) Representative I LCa traces, recorded in isolated RGCs with similar membrane capacitance from control, and diabetic retinas before (DM) and after perfusion of Ex-4 (DM + Ex-4).The holding potentials of RGCs were set at À40 mV, and the currents were evoked in the range of À30 mV to +10 mV in increments of 10 mV.(B) Comparison of the I-V relationship curves of I LCa in the RGCs of the control (n = 54), DM (n = 57) and DM + Ex-4 retinas (n = 20) (**p < 0.01 and ***p < 0.001 for DM vs. control; yp < 0.05 and yyp < 0.01 for DM vs. DM + Ex-4 by Sidak's multiple comparisons test after two-way RM ANOVA).(C) Bar chart summarizing the changes of I LCa densities at À10 mV from control, DM and DM + Ex-4 retinas.(D) Timeline of the experimental procedure and data collection.On the 17th day after STZ injection, DM rats received eye drops of Ex-4 or saline twice daily for two weeks.After 2 weeks of eye drops, all rats were sacrificed to conduct patch-clamp recording.(E) Representative I LCa from an RGC of a DM rat treated with saline (DM + Saline) and an RGC of a DM rat treated with Ex-4 eye drops (DM + Ex-4).(F) Comparison of the I-V relationships of I LCa obtained in RGCs from rats in DM + Saline group and DM + Ex-4 group (*p < 0.05 by Sidak's multiple comparisons test after two-way RM ANOVA).(G) Bar chart summarizing the changes of I LCa densities obtained at À10 mV from DM + Saline group and DM + Ex-4 group.Data are presented as mean G SD. See also Figure S2.

Figure 5 .
Figure 5. Involvement of cAMP-PKA signaling pathway in the Ex-4-induced suppression of I LCa (A) Current traces obtained from an RGC at À10 mV showing that during internal dialysis of 3 mM GDP-b-S, addition of Ex-4 failed to suppress the I LCa .(B) Bar chart summarizing the Ex-4 effect on the amplitudes of the currents in the presence of GDP-b-S (n = 11).The data obtained for each cell were normalized to the amplitudes obtained at least 8 min after GDP-b-S infusion.(C and D) Representative recordings of an RGC showing that Ex-4 did not change the I LCa of the RGC pre-incubated with NF-449 (10 mM), and summary data are shown in (D) (n = 7).(E and F) Current traces recorded in an RGC at À10 mV showing that perfusion of 8-Br-cAMP (500 mM) suppressed the I LCa (E), and summary data are shown in (F) (n = 8).(G and H) Sample traces recorded from another RGC at À10 mV showing that perfusion of 50 mM Rp-cAMP blocked the Ex-4-induced reduction of the I LCa (G), and summary data are shown in (H) (n = 8).Data are presented as mean G SD, n.s., p > 0.05 and **p < 0.01 by paired t test.

Figure 6 .
Figure 6.Involvement of intracellular Ca 2+ , CaM, calcineurin and PP1 in suppression of I LCa by Ex-4 (A and B) Representative recordings from an RGC at À10 mV, showing that during internal infusion of Ca 2+ -free solution (containing 10 mM BAPTA; control), Ex-4 failed to suppress the I LCa , and summary data are shown in (B) (n = 6).(C and E) Current traces of two RGCs, showing that during internal dialysis of 5 mg/mL heparin (C), but not ryanodine (50 mM, E), the Ex-4-induced suppression of I LCa was seen.(D and F) Bar charts summarizing the effects of Ex-4 on the I LCa amplitudes in the presence of heparin (n = 7) (D) or ryanodine (n = 7) (F).(G, I, and K) Representative recordings obtained from three different RGCs showing that Ex-4 no longer reduced the I LCa amplitudes during the internal infusion of 100 mM W-7 (G), 50 mM FK-506 (I) and 1 mM OA (K) to block CaM, calcineurin and PP1, respectively.(H, J, and L) Bar charts summarizing the results regarding the effects of W-7 (n = 7) (H), FK-506 (n = 8) (J) or OA (n = 8) (L).(M and N) Current traces of an RGC, showing that perfusion of C2 Ceramide (500 nM) suppressed the I LCa (M), and summary data are shown in (N) (n = 6).Data are presented as mean G SD, n.s., p > 0.05, **p < 0.01 and ***p < 0.001 by paired t test.(O) Representative immunoblots showed changes in the expression of L-VGCC subunits Cav1.2 and p-Cav1.2 in membrane components in retinas from the control, DM rats treated with or without Ex-4 eye drops.Na + -K + -ATPase served as loading control.