Treatment of diabetic retinopathy through neuropeptide Y‐mediated enhancement of neurovascular microenvironment

Abstract Diabetic retinopathy (DR) is one of the most severe clinical manifestations of diabetes mellitus and a major cause of blindness. DR is principally a microvascular disease, although the pathogenesis also involves metabolic reactive intermediates which induce neuronal and glial activation resulting in disruption of the neurovascular unit and regulation of the microvasculature. However, the impact of neural/glial activation in DR remains controversial, notwithstanding our understanding as to when neural/glial activation occurs in the course of disease. The objective of this study was to determine a potential protective role of neuropeptide Y (NPY) using an established model of DR permissive to N‐methyl‐D‐aspartate (NMDA)‐induced excitotoxic apoptosis of retinal ganglion cells (RGC) and vascular endothelial growth factor (VEGF)‐induced vascular leakage. In vitro evaluation using primary retinal endothelial cells demonstrates that NPY promotes vascular integrity, demonstrated by maintained tight junction protein expression and reduced permeability in response to VEGF treatment. Furthermore, ex vivo assessment of retinal tissue explants shows that NPY can protect RGC from excitotoxic‐induced apoptosis. In vivo clinical imaging and ex vivo tissue analysis in the diabetic model permitted assessment of NPY treatment in relation to neural and endothelial changes. The neuroprotective effects of NPY were confirmed by attenuating NMDA‐induced retinal neural apoptosis and able to maintain inner retinal vascular integrity. These findings could have important clinical implications and offer novel therapeutic approaches for the treatment in the early stages of DR.


| INTRODUC TI ON
Diabetic retinopathy (DR) is the leading cause of visual loss in adults aged 20-74 years. 1 The incidence of DR is expected to rise further due to the increasing prevalence of diabetes, ageing of the population and increase in the life expectancy of individuals with diabetes. 2 Conventional clinical assessment and classification is based on classical microvascular features including the following: haemorrhage, lipid exudate, cotton wool spots and neovascularization, all observed predominantly in the inner retina. 3 However, pre-clinical studies also demonstrate that components of the neurovascular unit, including the inner and outer neurosensory retina, are disrupted in diabetes. 4,5 Perturbed neuronal function reflected by impaired glutamatergic and dopaminergic neurotransmitter signalling, 6 altered dendritic fields 7 and reduced synaptic protein expression has been documented. 8 These changes ultimately leading to apoptosis of neurons alongside persistent uncontrolled diabetes. 9 Additional diabetic changes include altered glial cell activation, demonstrated by impaired interconversion of glutamate and glutamine, 10 regulation of potassium channels 11 and subsequent expression of the glutamate-aspartate transporter and intermediary filament proteins such as glial fibrillary acidic protein (GFAP). Diabetes also induces changes to retinal astrocytes, which located in the retinal nerve fibre layer and aligned with blood vessels provide contact with synapses, 12 and where connexin expression is reduced in the early course of diabetes prior to astrocyte loss. 13 Collectively, diabetes causes components of the retinal neurovascular unit to 'dis-integrate', and accordingly, DR should be recognized as a neurovascular degeneration and not solely a microvascular disease. 14,15 Nevertheless, recognition that disruption of the neurovascular unit presents opportunities for new therapeutic strategies and/or molecular targets that may offer potential for treating in the early stages of disease.
Neuropeptide Y (NPY) is one of the most abundant peptides in the mammalian central nervous system (CNS). 16 NPY is a highly conserved 36 amino acid peptide which binds to a family of G proteincoupled receptors Y1-6, 17 expressed in the retina of several species including mice. [18][19][20][21] Studies demonstrate putative neuroprotective effects of NPY in different CNS regions, 22 including inhibition of glutamate release in rat hippocampus and striatum. 23 Selective activation of NPY receptors has also been shown to protect mouse hippocampal cells from excitotoxic lesions. 24 Specific activation of Y2 receptor is effective in a transient middle cerebral artery occlusion model of ischaemia, 25 and NPY receptor signalling suppresses glutamate-induced necrosis and apoptosis in retinal neural cells, 26 modulates retinal ganglion cell (RGC) physiology and elicits neuroprotective effects in vitro. 27 Although no direct evidence in retinal endothelium, NPY can induce migration and proliferation of endothelial cells via its receptors and regulate their function. 28 Conversely, a recent study demonstrates that NPY treatment in a model of acute retinal ischaemia reduced retinal function and tissue damage. 29 In light of the reported decrease in mRNA and protein levels for NPY in diabetic retina, 30 as well as observations of NPY function, we examined whether administration of NPY would extend and offer protection of the neurovascular multicellular complex in the diabetic retina. Using in vitro platforms, we first assessed whether neuroprotective actions of NPY could reduce the sensitivity of retinal neurons to glutamate-induced excitotoxicity. Additionally, we evaluated the potential intracellular signalling pathways of NPY suppression of vascular endothelial growth factor (VEGF)-induced vascular permeability in retinal endothelial cells. Finally, we established an in vivo model of DR, permissive to accelerated glutamate excitotoxicity and VEGF-induced vascular leakage to evaluate whether NPY is neuroprotective, attenuating retinal neural apoptosis and able to maintain inner retinal vascular integrity.

| Cell culture and viability
Primary Retinal Microvascular Endothelial cells from mouse (MRMECs) or human (HRMECs) were purchased from company (Generon) and were cultured in Complete Mouse and Human Endothelial Cell Medium, respectively (Generon). We used the MTT assay to assess cell viability.

| In vitro vascular permeability assay
Permeability visualization experiments were performed on 18 × 18 mm glass coverslips as previously described. 31 For assays, serum-free culture medium supplemented with the treatment (combinations of NPY, N-methyl-D-aspartate [NMDA], VEGF) was added for a further 24 hours. Then, fluorescein-streptavidin was directly added to the culture medium for 5 minutes and washed twice with PBS before cell fixation with 3.7% paraformaldehyde (PFA) in room temperature (RT) subjected to immunofluorescence staining according to the manufacturer's instructions.

| Retinal explant culture ex vivo model
Retinal explant culture was performed as previously described. 31,32 Healthy Han Wistar rat's retina (20-old-day) was dissected into four equal-sized pieces, and the explants were separately transferred onto 12-mm-diameter filters (0.4 μm pore, Millipore) with the RGC side facing up. The filters were placed into the wells of a 24-well plate, and each contained 800 μL of culture media. Retinal explant cultures were maintained in humidified incubators at 37°C and 5% CO 2 . Half of the media was refreshed on day 1 and every second day thereafter.

| Immunofluorescence staining
Eyes from 6-month diabetic mice and cultured rat retinal explants were fixed with 4% PFA and snap-frozen in OCT (VWR Chemicals).

| Quantitative real-time RT-PCR
Total RNA extraction was carried out using TRIzol™ Reagent (Life Technologies, UK) according to the manufacturer's instructions.

| TUNEL assay
The terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay was performed according to the manufacturer's instructions (Roche). In brief, retinal explants were fixed with 4% PFA and washed in PBS, and retinas were permeabilized with 0.1% Triton X-100 for 1 hour and rinsed three times with PBS, before incubation with the TUNEL reaction mixture for 1 hour at 37°C. Retinal explants were mounted using antifading mounting medium, and apoptotic cells were observed using Leica SP5-AOBS confocal laser microscope. Adult (6-8 weeks) C57BL/6J male mice were obtained from Charles River Laboratories. Type I diabetes was induced by five consecutive days intraperitoneal injections with streptozotocin (STZ) (Sigma-Aldrich; 40 μg/g bodyweight in 0.1 mol L −1 citrate buffer). 33 Control mice were injected with citrate buffer alone. Diabetes was confirmed by measuring urine glucose (14 days after the onset of diabetic induction, >300 mg/dL considered diabetic; overall disease incidence 88%).

| Intravitreal administration of NPY, VEGF and NMDA
In vivo retinal vascular permeability was assessed as previously described. 34 Briefly, the pupils of 3-month-old diabetic mice were dilated using topical 1% tropicamide before induction of anaesthesia.
Mice were killed 10 minutes later, and eyes were fixed with 4% PFA for 2 hours. Retinal flatmounts were mounted in antifading medium, and images were captured by Leica SP5-AOBS confocal laser microscope and processed with ImageJ.
NMDA (Sigma-Aldrich) was used to induce excitotoxicity of RGC in the retina, 31 under experimental conditions described above. A dose-response was undertaken previously to ascertain the optimal dosing of NPY (5 to 20 nmol); then, mice received the following intravitreal injections (2 µL) with a 33-gauge needle into one eye: NMDA (10 nmol); NPY (10 nmol); NMDA with NPY together; and PBS vehicle. Mice were killed in 2 days following NMDA injection, and eyes and optic nerves were isolated. Eyes were fixed in 2% PFA for 2 hours for immunofluorescence staining.

| Electroretinography (ERG) and optical coherence tomography (OCT) measurements
In vivo imaging was performed in anaesthetized mice using the Micron IV retinal imaging system (Phoenix Research Laboratories). Pupils were dilated with tropicamide (1%) eye drops and positioned on a regulated temperature pad at 37°C to maintain constant body temperature.
Subdermal needle electrodes were inserted underneath the skin at the base of the tail (ground electrode) and between the eyes on the forehead (reference electrode). Corneas were lubricated with coupling gel, the objective/electrode was advanced near the corneal surface, and deep red illumination was used to focus on the retina. Responses to focal light stimuli (1mm diameter; spot size D) were recorded at luminance ranging from −0.9 to 3.9 log cd*sec/m 2 . Data are displayed as the mean amplitude (in µV) of the a-wave (as a measure of photoreceptor function) and b-wave (as a measure of bipolar cell function).

| Statistics
Results are therefore presented as means ± standard deviation (SD).
Differences between groups were analysed by Student's t test or analysis of variance (ANOVA). All the analysis was performed using GraphPad Prism 6 (GraphPad Software, version 6.01). The significant differences were considered at P ≤ .05.

| Diabetes drives neurovascular dysfunction, including Müller cell activation and ganglion cells loss
To understand how diabetes influences the neurovascular unit in the eye, we first confirmed what alterations to vasculature and neural retina occur in STZ-induced diabetic model. In vivo monitoring of STZ injected mice and controls permitted longitudinal assessment of retinal function by ERG. There was a steady decline of ERG function in diabetic eyes from 6 to 25 weeks, and by 6 months a significant suppression of the b-wave, indicative of dysfunctional Müller and bipolar cells ( Figure 1A,B). Ex vivo assessment of vascular integrity using retinal flatmount demonstrates leakage of Evans blue dye is significantly increased in the diabetic retina (219.6 ± 29.2%) compared to control retina (100.0 ± 9.4%, P ≤ .001; Figure 1C,D). The number of Brn3a + cells, a specific marker of RGC, was quantified on tissue sections and was significantly decreased by 43.6 ± 12.8% of normal controls ( Figure 1E,F). In the healthy retina, Müller cells do not express high levels of GFAP, a common marker of reactive gliosis 35 ; however, in the diabetic retina an increase MFI of 76.7 ± 17.1% of GFAP expression was observed ( Figure 1G,H). Next, we assessed whole retinal tissues for altered expression of gene transcripts encoding NPY and vasoproliferative factors, VEGF and ANGPTL-4.
In 6-month diabetic retinas, there was significantly reduced NPY mRNA levels detected ( Figure 1I), with corresponding increased expression of the pro-angiogenic VEGF and ANGPTL-4 ( Figure 1J). with VEGF resulted in a significant (43%) reduction, whereas NPY increased expression by 40% compared to untreated controls. When cells were pre-treated with NPY, applied 6 hours before VEGF, there was a significant increase in ZO-1 expression (33% higher than VEGF alone) equivalent to expression of the control cells ( Figure 2E). To understand the potential mechanism for the protective effect of NPY, we next examined expression of mitogen-activated protein kinase (MAPK), involving in regulating the expression of tight junction proteins. 36 Our results indicate that treatment of MRMECs with NPY increased the expression of ZO-1 and down-regulated MAPK isoforms p38-MAPK and p44/p42 MAPK (P ≤ .05; Figure 2F breakdown integrity of tight junction following VEGF treatment, which was similarly partially prevented by NPY pre-treatment ( Figure 2H).

| NPY regulates the tight junctions of retinal endothelium through MAP kinase
To assess whether NPY modulation of murine endothelial cells could be extended to human and thus bring a translational understanding of the role of NPY, experiments were also performed using HMRECs, demonstrated similar protective effects and maintained ZO-1 expression ( Figure S1A). To model microvascular disturbance observed in the hyperglycaemic retina, 37 we determined the angiogenic potential of HRMECs in response to NPY under high glucose (HG) conditions.
Utilizing the in vitro tube formation assay, quantitative analysis showed that HRMEC tube length was significantly reduced with HG exposure (50 mmol L −1 ) compared with control group (5.5 mmol L −1 glucose), whilst NPY co-treatment under HG conditions maintained the angiogenic capacity (P < .05; Figure 1B,C).

| NPY inhibits VEGF-induced vascular permeability in diabetic mice
The in vitro data demonstrate that NPY can modulate VEGF-induced vascular changes, so we next investigated whether this protective effect on vascular integrity translated to an in vivo setting. Groups

| NPY protects retinal ganglion cells against apoptotic cell death induced by NMDA ex vivo and in vivo
As NPY receptors are expressed in the inner retina and recognition that RGC are perturbed in diabetes, 38 Figure 4A, B). NPY treatment alone results in comparable low TUNEL + cells in comparison with control retinal explants.
To further elucidate with an in vivo correlate, we analysed flatmounts of diabetic eyes exposed to NMDA for 24 hours as a model of induced neuronal excitotoxic damage. A dose-response was tested prior to confirm the optimal dosing of NPY (5 to 20 nmol), which showed that a 10 nmol NPY dose provided the optimal protective effect in perturbing NMDA-induced RGC apoptosis model ( Figure S2A,B). In diabetic eyes, administration of NMDA alone reduced the overall total retinal thickness by 23% compared to disease controls. However, the combined administration of NPY and NMDA the extent of cell loss is significantly reduced by 16% compared to NMDA alone (P < .05; Figure 4C,D). Similarly, ERG responses demonstrate that eyes receiving a combined NPY and NMDA injection have improved functional responses, with increased a-wave and b-wave, compared to NMDA alone, although these were not statistically significant in b-wave ( Figure 4E,F). Retinal wholemount assessment from diabetic eyes receiving NMDA shows a 7.9-fold increase in TUNEL + cells compared to control. The combined NPY injection leads to a significant reduction in TUNEL + cells (from 7.9-fold to 4.1fold of control; P < .05; Figure 4G,H).

| NPY restored neurovascular function in diabetic retinas in vivo
To determine whether the positive effects of NPY extended to offer long-term protection in the diabetic retina, NPY was administered via intravitreal injection at 3 and 5 months post-induction of diabetes.
Clinical assessment by OCT to measure retinal thickness and ERG to determine changes in retinal function was performed at regular intervals throughout the experimental time course. Representative clinical assessment at 4 months ( Figure S3) showed no significant changes to the fundal appearance, total retinal thickness or ERG response between NPY-injected eyes and age-matched control eyes. Whilst the total retinal thickness was slightly reduced in the diabetic eyes, this was not significant. At 6 months following induction of diabetes, there was no difference in retinal thickness in the NPY-injected eyes compared to age-matched control eyes. However, the inner plexiform layer thickness in DM groups was reduced by 42% compared to control group and the inner plexiform layer thickness in the NPY treated group was increased by 47% compared to DM group. Cell loss occurred predominantly in the inner plexiform layer (IPL; Figure 5A-C). ERG assessment also demonstrates that NPY administration increased a-wave and b-wave amplitudes to levels equivalent to age-matched non-diabetic controls (P < .05; Figure 5D). These data support a notion that diabetic eyes injected with NPY had sustained neuronal cell function.
To confirm whether the protective effect of NPY was evident in ganglion cells and astrocytes, retinal flatmounts were prepared for ex vivo immunohistochemistry assessment. Quantification of Brn-3a + RGC numbers in the diabetic retina shows a 66% reduction in cell numbers compared to age-matched controls. By contrast, DM eyes treated with NPY demonstrate a significant increase in Brn-3a + cells (from 34% to 59% of control) (P < .05; Figure 5E,F). Retinal glial coverage was evaluated by staining with GFAP, showing widespread loss of glial coverage was observed in diabetic retina ( Figure 5G). The relative glial coverage in the diabetic retina was significantly reduced by 55% compared to control retina. The diabetic eyes treated with NPY demonstrated a significant increase by 49% in glial cell coverage compared to untreated diabetic eyes (P < .05; Figure 5H). Vascular integrity assessment by extravasation of EB dye showed a 3-fold increase in the diabetic retina as compared to control eyes, and this was remarkably reduced in the NPY-treated group (P < .05; Figure 5I,J).

| D ISCUSS I ON
In this study, we demonstrate a protective role of NPY in the retina, preventing loss of RGC and vascular leakage in diabetic mice in vivo. Increased vascular permeability in diabetes is recognized as a complex process involving multiple signalling pathways, mediated principally by VEGF which disrupts the integrity of endothelial cellcell junctions. 45 Among the various protein components of tight junctions, ZO-1 is a phosphoprotein that participates in multiple protein-protein interactions regulating tight junction integrity. 46 VEGF disrupts tight junctions by altering phosphorylation of ZO-1 and occluding through a Src-dependent pathway. 47 In our experiments, and in accordance with previous work, 48 50 In addition, NPY has been shown to inhibit nuclear translocation of NF-κB in microglia challenged with IL-1β, thus preventing IL-1β-induced nitric oxide release and a further mechanism to protect and regulate the endothelial barrier function. 51,52 Our previous studies using in vitro vascular permeability assay clearly demonstrate VEGF mediated increase in retinal endothelial F I G U R E 4 NPY protects retinal ganglion cells against apoptotic cell death induced by NMDA. A, B, Representative images showing TUNEL + cells in rat retinal explants exposed to NMDA (10 µmol L −1 ), NPY (10 µmol L −1 ) or pre-treated with NPY 6 h before NMDA exposure, showing TUNEL+ (red) and cell nuclei stained with DAPI (blue). Scale bar, 50 µm. Quantification of TUNEL + cells expressed as percentage of control. Groups of 3-month-old diabetic mice received intravitreal injections of NPY (10 nmol), NMDA (10 nmol) and NPY in combination with NMDA (n = 6 per group). C, D, At 48 h post-injection, OCT images show intravitreal NMDA results in reduced retinal thickness and NPY partially protects. ONH = optic nerve head. Data are representative of two measurements per retina. E, F, ERG a-wave and b-wave responses represented by mean values of amplitudes in scotopic conditions. G, H, Representative images of retinal whole mounts prepared for TUNEL staining and ImageJ analysis of confocal images. *P < .05; **P < .01; ***P < .001; ****P < .0001, statistical analysis was performed with one-way ANOVA with Dunn's test for multiple comparison, and EGF was analysed using two-way ANOVA for multiple comparison  Taken together, these studies highlight the regulatory and homeostatic role of the neurovascular unit that is disrupted in DR. The current data support that adjunctive NPY therapy sustains neuronal health and can attenuate at least in models the downstream effects of diabetic retinal vasculopathy, notably vascular leakage and neuronal death, as well as the loss of Glia cells.

ACK N OWLED G EM ENTS
This work was supported by The Dunhill Medical Trust [grant number: R474/0216] and China Scholarships Council (CSC).

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interests regarding the publication of this paper. Mertsch, Stefan Schrader, Lei Liu and Andrew D Dick drafted the F I G U R E 5 NPY restores neurovascular function in diabetic retinas. Groups of diabetic mice received intravitreal injections (NPY; 10 nmol) at 3 and 5 mo. A-C, Representative fundal and OCT images captured at 6 mo, showing no difference in retinal thickness between NPY-injected eyes and age-matched control eyes. The total retinal thickness is significantly reduced in the diabetic retina, primarily in IPL. D, ERG a-wave and b-wave responses represented by mean values of amplitudes in scotopic conditions. E, F, Representative confocal images of retinal whole mounts and quantification of Brn-3a immunostaining, showing that NPY prevents the loss of retinal ganglion cells in diabetic mice. G, H, Representative images and quantification of GFAP immunostaining, showing that NPY also protects against the loss of retinal astrocytes. I, J, Representative images 48 h following treatment demonstrate differences in the vasculature (leakage of Evans blue dye, which appears as red). Scale bar = 50 μm. The fluorescent intensity of Evans blue was quantified by ImageJ. *P < .05; **P < .01; ***P < .001; statistical analysis was performed with one-way ANOVA with Dunn's test for multiple comparisons, and EGF was analysed using two-way ANOVA for multiple comparison manuscript. David A Copland, Sofia Theodoropoulou, Lei Liu and Andrew D Dick provided critical suggestions and revised the manuscript. All the authors had the approval of the submitted and published versions.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support findings of the study are available from the corresponding author upon request.