The Synthetic Microneurotrophin BNN27 Affects Retinal Function in Rats With Streptozotocin-Induced Diabetes

Diabetic retinopathy (DR) is a serious complication of diabetes that leads to loss of visual acuity and blindness. It has been estimated that by 2030 the number of people with DR globally will rise to 191 million from 127 million in 2010 (1). The healthy retina provides a homeostatic equilibrium between all its cellular components, namely, neurons, vascular cells, and glial cells (Müller cells, astrocytes, and microglia). An imbalance of this homeostatic equilibrium may play a major role in the pathophysiology of DR (2).

Proliferative DR is defined as a microvascular disease characterized by neovascularization. An earlier study (3) reported the increase of apoptotic cells in retinas of rats 1 month after treatment with streptozotocin (STZ). TUNEL-positive (TUNEL⁺) cells were detected in the ganglion cell layer (GCL) but not in vascular endothelial cells (3). These findings suggested for the first time that DR has a neurodegenerative component, since retinal neural cell death occurred in diabetes prior to the vascular insult. Subsequent studies contributed to this principle (4–6). Also, electroretinograms in patients with diabetes and in diabetic rats showed abnormalities that preceded the vascular features associated with DR (7).

The imbalance between prosurvival neurotrophic factors and inflammatory components is believed to lead to apoptosis and proinflammatory responses (2,8). Indeed, proinflammatory markers were detected in the vitreous of patients with proliferative DR (9,10). Standard therapies for DR, such as photocoagulation and intraocular injection of anti-VEGF agents, are efficacious but target only the neovascular component of the disease. New research strategies are essential in order to identify new therapeutic targets to combat the neurodegenerative and proinflammatory components of DR (11).

The neurosteroid dehydroepiandrosterone (DHEA) (12) has provided neuroprotection in different central nervous system paradigms, including excitotoxicity, ischemia, and ischemia-reperfusion injury (13–15), and against serum deprivation in PC12 cells (16). Most recently, DHEA was shown to mimic the actions of the prosurvival neurotrophin nerve growth factor (NGF) in protecting neuronal cell types from apoptosis (17).

NGF binds with high affinity to the prosurvival tropomyosin-related kinase (Trk)A receptor (TrkAR) and with a lower affinity to neurotrophin receptor p75 [p75^NTR], member of the TNF receptor superfamily. DHEA was shown to activate both TrkAR and p75^NTR via which it mediates its neurotrophic effects (17). DHEA also binds to the other two mammalian Trk receptors (TrkB and TrkC), as well as their ancestral isoforms, with high affinity (18).

Neurosteroids are expressed in the retina (19), as are NGF and its receptors, TrkAR and p75^NTR (20). NGF prevented ganglion cell death when administered to diabetic rats (21). More recent studies showed that topical ophthalmic administration of NGF protected retinal ganglion cells (RGCs) in an animal model of glaucoma, in patients with glaucoma (22), and in the STZ model of DR (23). DHEA was also shown to protect rat retinal neurons from AMPA excitotoxicity, mimicking the actions of NGF via an NGF TrkAR mechanism (24).

These studies support the use of NGF and DHEA as a therapeutic in neurodegenerative retinal disease. However, NGF is a large peptidic molecule (26.5 kDa) not able to cross the blood-brain barrier (BBB) and susceptible to proteolysis (25). DHEA is also restricted owing to its metabolism to androgens and estrogens. Its long-term use in humans runs the risk of increasing hormone-dependent tumors, particularly in genetically predisposed patients (26).

Recently, BNN27, a novel C17-spiroepoxide derivative of DHEA, that has no affinity for the classic steroid receptors, was synthesized. This agent is a small, highly lipophilic neurosteroidal molecule that crosses the BBB (27,28). BNN27 was recently reported to activate TrkAR signaling in neuronal (sympathetic and sensory primary neurons [PC12 cell line]) and microglial (BV2 mouse cell line) cells and to protect TrkA-positive and NGF-dependent sympathetic and sensory neurons from apoptosis. It had no effect on TrkB or TrkC receptors (29). We recently reported that BNN27 reversed the serum deprivation–induced apoptosis of cerebellar granule neurons (that do not express TrkA) by activating p75^NTR (30). These data strongly suggest that BNN27 mediates its pharmacological prosurvival actions by activating specifically the NGF TrkAR and p75^NTR, thus being more advantageous compared with DHEA, which binds to all Trk receptors (18).

We hypothesized that smaller-than-NGF, highly lipophilic molecules that cross the BBB/blood retina barrier (BRB), such as BNN27, that mimic the neurotrophic prosurvival properties of NGF, may be more efficacious as therapeutics in DR. The aim of this study was to investigate how BNN27 administration affects the diabetic retina. Our findings support that BNN27 reverses the diabetes-induced retinal damage by activating the NGF TrkA receptor, reducing p75^NTR, cleavage of caspase-3, glial activation, and anti-inflammatory cytokine levels.

Animal experiments followed the Guide for the Care and Use of Laboratory Animals, 8th edition (2011), and were in compliance with Greek national legislation (Animal Act, P.D. 160/91). Both male and female Sprague-Dawley rats (180–300 g) were used in the current study. The animals were maintained on a 12-h light-dark cycle at 22–25°C. Food and water were available ad libitum.

Diabetes was induced by a single dose of STZ (70 mg/kg i.p.; Sigma-Aldrich, Tanfkirchen, Germany) dissolved in sodium citrate (0.1 mol/L) buffer (diabetic group) after a fasting period of 8–12 h. Animals with blood glucose levels >350 mg/dL, at 72 h post–STZ injection, were considered diabetic (31). Both male and female Sprague-Dawley rats were used in the current study. Higher morbidity and mortality were observed in male STZ-administered rats. Specifically, 70% of the STZ-injected animals died during the first week postinjection. Of these animals, 76.2% were male and the rest (23.8%) were female. Male pancreatic islet β-cells are more prone than female to STZ-induced cytotoxicity (32). Therefore, male rats may die as a result of islet β-cell necrosis. Lower mortality rates were observed at later time points after STZ injection. Glucose levels were >400 mg/dL in all experimental rats 7 days after the STZ injection and >600 mg/dL the following weeks and until the end of the experiments. No differences were observed in glucose levels between male and female rats. Body mass was monitored every week after the STZ injection. Loss of 11.47 ± 1.47% (mean ± SEM) of body mass was observed in male rats, while female rats lost 1.75 ± 4.0% of body mass at day 7 after the STZ treatment. Small fluctuations of body mass were observed the following weeks in both male and female rats. The sex difference (in mortality and body mass) noted in the present study and in other studies (33) has not been explicitly defined; yet, estrogens are known to have beneficial effects on hyperglycemia and islet β-cell functions in the STZ-induced diabetic rat and mouse models (34,35).

Three experimental groups were used, namely, a control group, a diabetic group, and a diabetic treated group in two different paradigms: in paradigm A, BNN27 (2, 10, and 50 mg/kg i.p.) was administered daily for 7 days at 4 weeks post–STZ injection, and in paradigm B, BNN27 (10 mg/kg i.p.) was administered 48 h after STZ injection every second day for 4 weeks. The intraperitoneal route was selected to assure active levels of the above agent in the retina. BNN27 was dissolved in absolute ethanol (10% v/v in water for injection). Appropriate vehicles were administered to control and diabetic nontreated animals.

For examination of the involvement of the NGF TrkAR in the actions of BNN27, a specific TrkAR inhibitor (10⁻³ mol/L, category no. 648450; Calbiochem) was administered intravitreally (flow rate of 1 μL/min for 5 min) at days 1, 4, and 6 in conjunction with BNN27 administration (10 mg/kg i.p.) (24). The TrkAR inhibitor concentration was chosen taking into account the calculated molarity of the BNN27 dose of 10 mg/kg i.p. DMSO (10% v/v in PBS [50 mmol/L K₂HPO₄/NaH₂PO₄ and 0.9% NaCl, pH 7.4]) was used as the solvent for the inhibitor and as vehicle in the other experimental groups. Both eyes received the same treatment.

Glucose levels were monitored in the diabetic treated animals to ascertain the putative effect of BNN27 (10 and 50 mg/kg i.p. for 7 days [paradigm A]). The results obtained did not suggest any BNN27 effect on glucose levels (diabetic, mean ± SEM 573 ± 14 [n = 23]; diabetic+BNN27 [10 mg/kg], 575 ± 13 [n = 23]; and diabetic+BNN27 [50 mg/kg], 564 ± 29 [n = 10]). No statistical significant differences were observed between the diabetic treated and nontreated animals or between treated groups (P > 0.05). Control animals did not show any significant alteration in glucose levels during the 5-week period.

Animals were euthanized with ether inhalation 24 after the last treatment, their eyes were removed, and retinas were obtained and prepared for immunohistochemical, Western blot, ELISA, and high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS) studies.

The eyes were fixed by immersion in 4% paraformaldehyde in 0.1 mol/L phosphate buffer for 1 h at 4°C. After fixation and cryoprotection, tissues were embedded in optimal cutting temperature compound (Prolabo, Leuven, Belgium) and frozen in isopentane for 1 min. The eyecups were sectioned vertically, and the initial 1.5 mm tissue was discarded. Serial 10-μm-thick sections were placed onto six slides, and nine sections were collected on every slide. This way, every slide contained a representative part of retinal tissue, including the optic nerve head (24).

Immunohistochemical studies using antibodies raised against retinal cell markers for ganglion cell axons (nerve fiber layer [NFL]) and brain nitric oxide synthetase (bNOS)- and tyrosine hydroxylase (TH)-expressing amacrine cells were performed to assess the effect of the STZ on retinal cell viability in rats 1–5 weeks after its administration (24). Moreover, antibodies raised against glial fibrillary acidic protein (GFAP), ionized calcium-binding adaptor molecule-1 (Iba-1), cleaved caspase-3, and p75^NTR were also used (Supplementary Table 1). Cryostat sections were treated overnight with the appropriate primary antibody in 0.1 mol/L tris-buffered saline containing 0.3% Triton-X-100 and 0.5% normal goat serum and subsequently with a fluorescence secondary antibody for 1–2 h. Slides were cover slipped using EverBrite Mounting Medium with DAPI (Biotium).

The terminal deoxynucleotidyl transferase (TDT)-mediated TUNEL assay (Roche, Grenzach-Wyhlen, Germany) was employed to assess retinal cell death in control and diabetic retinas 4 and 5 weeks post–STZ administration, as well as at the onset of diabetes. Retinal tissues were incubated with the TUNEL reaction mixture containing Label and Enzyme Solutions according to the directions of the manufacturer. Slides were cover slipped using EverBrite Mounting Medium with DAPI (Biotium). Colocalization studies (bNOS immunoreactivity [bNOS-IR] and TUNEL) were performed in order to examine the viability of bNOS-immunoreactive cells in retinas of diabetic animals and diabetic animals treated with BNN27 (10 mg/kg).

Retinal lysates were prepared in Tris-HCl buffer (50 mmol/L, pH 7.5) containing NaCl (150 mmol/L), NP40 (1%), sodium deoxycholate (0.1%), and phenylmethylsulfonyl fluoride (0.1 mmol/L) and protease/phosphatase inhibitor cocktail (Thermo Fisher Scientific, Waltham, MA). For immunoprecipitation, retinas were suspended in lysis buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1% Triton X-100, and protease inhibitor cocktail [pH 7.5]). Lysates were centrifuged (10,000g for 20 min) and the supernatants incubated with TrkA monoclonal antibody (Millipore) (1:60 dilution) overnight at 4°C. Protein G PLUS-Agarose (Santa Cruz Biotechnology) beads were added (1:8 dilution) and incubated for 2 h at 4°C. Beads were collected by centrifugation (2,000g for 2 min) and pellets were washed in lysis buffer and resuspended in SDS sample buffer. Precipitates and lysates were analyzed by SDS-PAGE (12.5% and 7.5% acrylamide, respectively) and incubated with specific antibodies (Supplementary Table 1). The GAPDH antibody was used to normalize protein content in lysates. Two to three different retinas for each group were run on each gel. The proteins were visualized using an ECL Western blotting kit (Thermo Scientific, Rockford, IL), and quantitative densitometry of the protein bands was performed using Image Lab, version 5.0, software (Bio-Rad) (36).

ELISA kits (Abcam) were used to evaluate the levels of TNFα, IL-1β, IL-10, and IL-4 in retinas according to the manufacturer’s instructions. The samples were analyzed in duplicates using an ELISA reader (450 nm, model 680; BioRad). Protein concentration was determined by NanoDrop 2000 (Thermo Fisher Scientific).

Rats were administered BNN27 (10 mg/kg) for 7 days. This dose was chosen because it was the lowest dose that led to the activation/phosphorylation of the TrkAR. To each retina, methanol (495 μL) and 5 μL d6-DHEA (2,500 ng/μL) were added, and the samples were homogenized for 1.5 min on ice, sonicated for 30 s, and centrifuged at 13,500 rpm for 1 h at 4°C. The supernatant was stored at −20°C until HPLC-MS/MS analysis (37). Chromatographic separation was performed on a UniverSil UHS18 column (150 × 2.1 mm, 3 μm) with a gradient elution system consisting of methanol and water, both with 0.1% formic acid (total flow rate 200 μL/min). Tandem mass spectral analysis was performed with a Thermo Fisher Scientific TSQ triple quadrupole mass analyzer equipped with an electrospray ionization source operating in positive mode using selected reaction-monitoring detection. For BNN27, 315.3–297.1 ion pair was used for quantitation, and 315.3–255.2 was used for confirmation. d6-DHEA was used as the internal standard (selected reaction monitoring detection 295.3–277.2).

The total number of bNOS- and TH-IR somata was manually counted in three sections/retina. The density area starting from the GCL until the inner plexiform layer (IPL) for NFL and p75^NTR and from GCL until the outer nuclear layer (ONL) for GFAP and the inner nuclear layer (INL) for cleaved caspase-3 was delineated in each image (2 images from 3 sections/retina) using the public domain ImageJ, version 1.43m, software. The mean gray value [integrated density (fluorescence density)/delineated area] of this region was calculated and expressed as a percentage of the mean gray value of the control or diabetic retina (100%). The number of Iba-1–positive cells was manually counted, with data normalized to the total counting area (from GCL to INL) and expressed as percentage of the control retinas (100%). TUNEL⁺ cells and DAPI nuclei were counted from the ONL to GCL (total retinal thickness) and individual layers (ONL, INL, and GCL) using ImageJ 1.43m software as described above. Data were normalized and expressed as the number of TUNEL⁺ cells per area (mm²). The percentage of the ratio of TUNEL to DAPI cells was calculated in retinas of animals 4 weeks post–STZ injection. Each experiment was replicated a minimum of three times. Images were obtained using an Axioskop Plan-Neofluar (Carl Zeiss, Oberkochen, Germany) or HC PL Fluotar, Leica DMLB (×20 or ×40/0.75 lens), and Leica SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany).

Data are expressed as mean ± SEM or mean ± SD (scatter plots). Statistical analyses were performed using GraphPad Prism, version 5.0 (GraphPad, San Diego, CA), and differences between groups were evaluated by one-way ANOVA with Tukey post hoc analysis or unpaired t test (two tailed). Statistical significance was set at P < 0.05. The number of animals used in each group (n values) is shown in the relevant figures or in the legend.

Qualitative immuhistochemical studies were performed to assess the effect of the STZ administration on retinal cell viability in rats 1–4 weeks after STZ administration. No differences were observed between the control and diabetic retinas in the first 3 weeks in the staining pattern of bNOS-IR localized in amacrine cell somata in the INL, displaced amacrine cell somata in the GCL and amacrine cell processes in the IPL, TH-IR localized in amacrine cell somata in the INL, or NFL-IR localized in ganglion cell axons in the GCL and in the IPL (data not shown). However, 4 weeks post–STZ injection, a significant decrease in NFL-, bNOS-, and TH-IR was observed in the diabetic retina (Supplementary Fig. 1A). Quantitative data analysis of the images confirmed the immunohistochemical data (Supplementary Fig. 1B–D).

TUNEL analysis was performed to examine retinal cell death after 4 weeks of STZ injection (Supplementary Fig. 2A). Quantification of the TUNEL⁺ cells per area (mm²) in the total retinal thickness showed an approximate 10-fold increase (2.9 ± 0.6, n = 6, compared with the number of apoptotic retinal cells found in control retinas [0.3 ± 0.1, n = 6] [P < 0.01]) (Supplementary Fig. 2B). The TUNEL-to-DAPI ratios (Supplementary Fig. 2C) show the percent of retinal cell death in total retina and its layers (GCL, INL, and ONL), with the highest in the GCL.

BNN27 administered intraperitoneally 4 weeks post–STZ injection, daily for 7 days (paradigm A), reversed in a dose-dependent manner the diabetes-induced reduction in the NFL-, bNOS- and TH-IRs (Fig. 1A). Quantitative analysis of the immunohistochemical data gave the following results for the control and diabetic retinas: NFL-IR (mean gray area/image), control 25,837 ± 1,373 and diabetic 15,613 ± 2,077; bNOS-IR (cell somata/section), control 85 ± 2 and diabetic 40 ± 1; and TH-IR (cell somata/section), control 3.5 ± 0.3 and diabetic 1.4 ± 0.3. BNN27 reversed the diabetes-induced decrease in the expression of the three retinal markers at the doses of 10 and 50 mg/kg (P < 0.001, P < 0.01, and P < 0.05 compared with the nontreated diabetic retina and P > 0.05 compared with control) (Fig. 1B–D). No statistically significant difference was observed at the dose of 2 mg/kg (P > 0.05 compared with the diabetic nontreated retinas).

The effect of TrkAR inhibitor (10⁻³ mol/L) on the BNN27-induced actions on NFL bNOS- and TH-IR and in the diabetic retina was examined (Fig. 2). Quantitative analysis of the immunohistochemical images (not shown) suggests that the intravitreal administration of TrkAR inhibitor reversed the BNN27 (10 mg/kg) actions (P < 0.01, P < 0.05, and P < 0.05, respectively, for NFL-IR, bNOS-IR, and TH-IR, compared with diabetic retinas treated with BNN27 alone). The TrkAR inhibitor had no effect when administered alone (P > 0.05 compared with the diabetic retinas). These results suggest that the NGF TrkAR mediates the restoration of the phenotype of bNOS-, TH-, and NFL-positive cells in the diabetic retina.

BNN27 (10 and 50 mg/kg) increased TrkAR phosphorylation on Y⁴⁹⁰ residue to similar levels (P < 0.05 compared with the diabetic nontreated retinas, with no statistically significant difference between the two doses [P > 0.05]). No statistically significant effect was observed at the dose of 2 mg/kg (P > 0.05 compared with the diabetic nontreated retinas) (Fig. 3A). The phosphorylated-to-total TrkA ratio in retinal samples showed no statistically significant difference between control and diabetic nontreated animals (0.18 ± 0.02 and 0.20 ± 0.03 phosphorylated-to-total TrkA ratio, respectively, P > 0.05). The TrkAR inhibitor reversed the BNN27 -dependent (10 mg/kg) increase in phosphorylation of TrkA^Y490 (P < 0.001 compared with the diabetic treated retina and P < 0.001 compared with the diabetic retina) (Fig. 3B).

Representative images from immunohistochemical studies depict an increase in p75^NTR expression in diabetic retinas (Fig. 4A). Quantitative analysis of the images show that BNN27 (10 and 50 mg/kg i.p.) attenuated the diabetes-induced increase in p75^NTR-IR (P < 0.001 compared with control and P < 0.01 compared with the diabetic nontreated retina) (Fig. 4B). p75^NTR expression was increased in the diabetic retina (P < 0.001 compared with control), and BNN27 (10 mg/kg i.p.) led to its reversal (P < 0.001) (Fig. 4C).

TUNEL⁺ cells were observed in the 5-week diabetic retinas (paradigm A). With the use of confocal microscopy, we observed TUNEL⁺ cells (light and darker stain) colocalized with DAPI (Supplementary Fig. 3). We considered as TUNEL⁺ cells all cells that were observed to be colocalized with DAPI. BNN27 had no effect on the total number of TUNEL⁺ cells in the diabetic retina (Fig. 5A), but there was a statistical significant difference of the diabetic and diabetic+BNN27 compared with control (P < 0.05). No colocalization of TUNEL or one of the retinal markers (bNOS) was observed (Fig. 5B).

TUNEL⁺ cells were observed in retinas of animals that were administered BNN27 (10 mg/kg) every other day for 4 weeks (paradigm B) (Supplementary Fig. 4). In this paradigm, as in paradigm A, BNN27 was not able to reduce the number of TUNEL⁺ cells per area (mm²) (P > 0.05 compared with the diabetic retina). However, BNN27 did reverse the diabetes-induced attenuation of the bNOS and TH immunoreactive cells (data not shown).

For assessment of the effect of BNN27 on caspase-3–induced apoptosis, immunohistochemical studies were performed in diabetic and diabetic+BNN27 retinas in both paradigms (Fig. 6A and C). Quantitative analysis of the images showed that BNN27 induced a statistically significant decrease (paradigm A, P < 0.01 compared with diabetic, and paradigm B, P < 0.05 compared with diabetic) in cleaved caspase-3–IR (Fig. 6B and D).

The effect of BNN27 was also examined on macroglia and microglia markers for proteins, GFAP and Iba-1, respectively, known to play an important role in inflammation. Diabetes induced an increase in macroglial activation (GFAP-IR, 149.0 ± 8.0 percent of mean gray area, compared with control retinas, 100.0 ± 4.0 percent of mean gray area; P < 0.001) (Fig. 7A and B), whereas BNN27 prevented glial activation (10 mg/kg, 111.0 ± 4.1 percent of mean gray area; 50 mg/kg, 109.5 ± 4.2 percent of mean gray area; P < 0.001 compared with diabetic nontreated retinas). BNN27 also reduced the number of Iba-1–positive cells per area (mm²) (10 mg/kg, 67 ± 20; 50 mg/kg, 113 ± 34; P < 0.001 and P < 0.01, respectively, compared with diabetic nontreated retinas [255 ± 91], and P < 0.001 compared with control [100 ± 29]) (Supplementary Fig. 5).

Diabetes induced an increase in the levels of proinflammatory cytokines (TNFα and IL-1β). BNN27 (10 mg/kg i.p. [paradigm A]) reversed this increase in TNFα (P < 0.001 compared with the diabetic), but only the higher dose of 50 mg/kg reduced the diabetes-induced increase in IL-1β levels (P < 0.05). BNN27 also increased the levels of anti-inflammatory cytokines IL-10 and IL-4 (P < 0.05 compared with the levels observed in the diabetic retina) (Fig. 8 and Supplementary Table 2).

HPLC-MS/MS analysis of retinal samples of control animals (nondiabetic) that received BNN27 (10 mg/kg i.p. for 7days) showed that BNN27 levels reaching the retina were in the order of 227 ± 95 nmol/L (n = 6).

In the current study, we identified a putative therapeutic for the treatment of DR, the neurosteroidal microneurotrophin BNN27. This novel spiroepoxy derivative of DHEA was shown to reverse the diabetes-induced retinal damage by activating the NGF TrkAR and reducing the expression of p75^NTR and glial activation.

Hyperglycemia was detected as early as 1 day post–STZ injection, with glucose levels remaining high for up to 5 weeks (>600 mg/dL). Its effect on retinal neurons was evident 4 weeks post–STZ injection, as shown by the reduction of retinal markers bNOS, TH, and NFL (Supplementary Fig. 1). A 10-fold increase in the number of TUNEL⁺ cells per area (mm²) was observed across the whole retina thickness, with significant increases in apoptotic cells in layers GCL, INL, and ONL (% TUNEL/DAPI) (Supplementary Fig. 2). These results are in agreement with other studies that used the experimental STZ model of DR (38,39).

BNN27 administered (intraperitoneally) for 7 days, 4 weeks post–STZ injection, reversed the diabetes-induced attenuation of NFL-, bNOS-, and TH-IR in a dose-dependent manner, suggesting a restorative role of this agent against the diabetic insult. The TrkAR inhibitor reversed the BNN27-induced protective effects on bNOS- and TH-IR amacrine cells and NFL-IR ganglion cell axons but had no effect when administered alone (Fig. 2). Thus, BNN27 restores the phenotype of these retinal cells.

The TrkAR is located in ganglion cells, in the INL, and processes of the IPL (40). Therefore, the neuroprotection of ganglion cell axons by BNN27 may be due to the direct activation of the TrkAR located in these neurons. However, the identity of the amacrine cells in the INL expressing the TrkAR has not been reported. The activation of the TrkAR in the GCL may initiate a cascade of events and provide neuroprotection to amacrine cells via an indirect mechanism.

The aforementioned effects of BNN27 are mediated by the activation/phosphorylation of the TrkA^Y490 in the diabetic retina (Fig. 3A). NGF induces the phosphorylation of several tyrosine residues located intracellularly, leading to the activation of downstream prosurvival signaling cascades. Among the three residues described to mediate TrkAR activation (490, 674/675, and 785) (41), only Y490 was phosphorylated in our paradigm. The expression of the TrkAR was not altered in the diabetic retinas, which is in agreement with the report of Ali et al. (42).

In a recent report, BNN27 was shown to bind and activate solely the TrkAR receptor and not TrkB or TrkC receptors and to protect TrkA-positive and NGF-dependent sympathetic and sensory neurons from apoptosis (29). These results complement the present data that show that the BNN27-induced phosphorylation of TrkAR is reversed by the intravitreal injection of the TrkAR inhibitor (Fig. 3B). Therefore, we conclude that the TrkAR is responsible for the BNN27-induced neuroprotection of the retina against the diabetic insult.

The p75^NTR is also located on ganglion cells in the INL and Müller cells. Therefore, the attenuation of its expression by BNN27 may also affect its downstream signaling and provide neuroprotection to ganglion and amacrine cells. In the current study, our data support that BNN27 downregulates the expression of the p75^NTR death receptor. BNN27 significantly attenuated the diabetes-induced increase of p75^NTR protein expression in a dose-dependent manner. These findings suggest that the pharmacological actions of BNN27 involve both NGF receptors (TrkA and p75^NTR), in agreement with the report of Lazaridis et al. (17), who showed that the parent molecule DHEA decreased p75^NTR levels in serum-deprived PC12 cells.

While the above-mentioned data suggest a neuroprotective role of BNN27 in the DR model, the TUNEL analysis performed on the tissues obtained from paradigm A showed that BNN27 had no effect on the total number of TUNEL⁺ cells in the retina. In order to ascertain whether the amacrine cells in the INL shown in the current study to be affected by diabetes undergo apoptosis, we performed costaining for TUNEL and one of the two retinal markers used (bNOS). TUNEL⁺ cells and bNOS staining were clearly not colocalized. These results suggest that the reduction of specific neural markers in the diabetic retina is not due to retinal cell death but may be a result of diabetes-induced protein synthesis or metabolism dysfunction, in which BNN27 plays a restorative role. The correlation, and lack thereof, between a neuronal marker and caspase-3–IR or TUNEL⁺ cells at different time points has been also reported in a model of focal ischemia in human hippocampus (43). It was also reported that patterns of retinal cell immunoreactivity in transmissible spongiform encephalopathies are not correlated with retinal degeneration (44). The mechanism via which this phenomenon is mediated has not been elucidated, but it may be a result of protein synthesis or metabolism dysfunction. This may hold true in our paradigm, with BNN27 playing an important role in restoring retinal cell function.

To examine the preventive properties of BNN27 on DR, we administered 10 mg/kg i.p. every other day for 4 weeks starting at the onset of diabetes (paradigm B). BNN27 had no statistical significant effect on TUNEL staining, but it reversed the diabetes-induced decrease of bNOS and TH immunoreactive cells (data not shown) similarly to what was observed in paradigm A.

At this point, we chose to investigate the effect of BNN27 on cleaved caspase-3, since cleavage of caspase-3 more than TUNEL labeling alone is suggested to be a more specific marker for early apoptosis in tissue sections (45). BNN27 reduced the diabetes-induced increase of cleaved caspase-3 in both paradigms, suggesting that it provides neuroprotection to the diabetic retina specifically against caspase-3–mediated cell death. Thus, BNN27 seems to selectively prevent the caspase-3–mediated cell death, having no effect on other apoptotic or necrotic pathways, in contrast to its lack of effect on TUNEL staining that labels all dead cells.

Diabetes was shown to induce an increase in peroxynitrite concentration leading to TrkAR tyrosine nitration, the attenuation of its activation by NGF, and the increase in the expression of p75^NTR. These events reversed the NGF/TrkA prosurvival actions leading to retinal neurodegeneration (38). Investigations using glaucoma and optic nerve transection models (in vivo RGC degeneration) reported that NGF activation of p75^NTR and increases in glia activation led to RGC toxicity. This was shown to be mediated by increases in TNFα and TNFα₂-microglobulin levels (46). In the current study, we show that BNN27 reduces the diabetes-induced increase in 1) p75^NTR-IR and its expression and 2) macroglial (GFAP-IR) and microglial (Iba-1–IR) activation. Increase in Iba-1⁺ cells does not necessarily indicate microglia activation, since macrophages may also be present in the DR retina (BRB is not intact in DR [8]) and are also Iba-1⁺. However, the morphology of the Iba-1⁺ cells indicates that they are in an activated (or, rather, intermediate) state. The actions of BNN27 on GFAP and Iba-1 expression suggest its role as an anti-inflammatory agent. This is further strengthened by its ability to reduce the levels of the proinflammatory agents TNFα and IL-1β in the diabetic retina and increase the levels of the anti-inflammatory cytokines, IL-10 and IL-4 (Fig. 8 and Supplementary Table 2).

Basal levels of NGF in rat retina were reported to be 147 ± 52 pg/g tissue (47). Topically applied NGF (10 μL drop of 200 μg/mL) resulted in a twofold increase in NGF levels in retina 6 h after treatment (47). NGF was shown to protect RGCs in an animal model of glaucoma and improved visual function in three patients with advanced glaucoma, as well as in the STZ model of DR (22,23). In the current study, BNN27 levels (227.3 ± 95.7 nmol/L) were detected in control rat retinas, after its intraperitoneal administration, with HPLC MS/MS analysis. These results suggest that BNN27 crosses the BRB and provides neuroprotection to the retina, in agreement with Bennett et al. (28), who reported that BNN27 crosses the BBB.

In total, these findings suggest that BNN27 protects the retina from the STZ-induced diabetic insult acting as a microneurotrophin. Sato et al. (48) reported that DHEA improved hyperglycemia in the STZ model of DR by activating glucose metabolism–related signaling pathways in skeletal muscle. In our paradigm, BNN27 had no effect on rat hyperglycemia at the doses examined. (See research design and methods.) Thus, the BNN27 effects in the diabetic retina are clearly independent of any improvement of hyperglycemia. On the contrary, and most importantly, BNN27 exerts its protective effects despite sustained high glucose levels.

BNN27 can be recommended as a therapeutic for DR, having the following advantages over NGF. BNN27, like NGF, binds to the high-affinity receptor (TrkA) in a nanomolar concentration (29), propagating prosurvival signaling. Moreover, it attenuates the expression of p75^NTR death receptor in the diabetic retina. In addition, BNN27 unlike NGF is a small, lipophilic molecule that crosses the BRB.

In conclusion, BNN27 has the pharmacological profile of a therapeutic for DR, since it targets both the neurodegenerative and inflammatory components of the disease. More preclinical data are essential for further assessment of the pharmacodynamic and pharmacokinetic properties of BNN27 that will recommend its investigation at the clinical level. Studies are in progress examining the effect of BNN27 on proNGF/NGF levels and proNGF/ p75^NTR signaling in the diabetic retina.

Acknowledgments. The authors thank Smaragda Poulaki and Maria Boumpouli, Department of Pharmacology, School of Medicine, University of Crete, for their assistance in various phases of this project; Ioannis Dalezios, Department of Physiology, School of Medicine, University of Crete, for advice on statistics; and Petros Chatzakis, Department of Ophthalmology, School of Medicine, University of Crete, for assistance in measuring cage illumination during the 12-h light/dark cycle.

Funding. This study was cofinanced through the Operational Program “Education and Lifelong Learning” of the National Strategic Reference Framework - Research Funding Program ARISTEIA II by a grant to K.T. from the European Union (European Social Fund-ESF) and Greek National Funds.

Duality of Interest. A.G. is the cofounder of Bionature E.A. LTD, proprietor of compound BNN27 (patented with the WO 2008/1555 34 A2 number at the World Intellectual Property Organization). I.C. has patent WO 2008/1555 34 A2 with royalties paid. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. R.I.-A. performed experiments, analyzed and interpreted data, revised the manuscript, and was instrumental in the resubmission. S.L. performed experiments, analyzed and interpreted data, and contributed to the drafting and revision of the original manuscript. N.M., E.K., P.I., M.F., S.P., and A.S. performed experiments, analyzed and interpreted data, and revised the manuscript. D.K. performed experiments, analyzed and interpreted data, revised the manuscript, and was instrumental in the resubmission. A.K. contributed to the design, performed the HPLC-MS/MS analysis of BNN27, and interpreted data. H.E.K. contributed to the design and interpretation of the HPLC-MS/MS data and revised the manuscript. A.G. contributed to the conception of the study and revised the manuscript. I.C. contributed to the study design, interpreted data, and revised the manuscript. K.T. conceived and designed the experiments, analyzed and interpreted data, wrote and revised the manuscript, and supervised the project. All authors read the final version of the manuscript. K.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the American Society for Neuroscience 44th Annual Meeting, Washington, DC, 15–19 November 2015; the Association for Research in Vision and Ophthalmology Annual Meeting, Denver, CO, 3–7 May 2015; the European Society for Neurochemistry conference Molecular Mechanisms of Regulation in the Nervous System, Tartu, Estonia, 14–17 June 2015; and the Federation of European Neuroscience Societies Featured Regional Meeting, Thessaloniki, Greece, 7–10 October 2015.