Despite decades of research, diabetic retinopathy remains the leading cause of blindness in working age adults. While several inciting factors are likely involved, research has shown increased levels of inflammatory mediators in the diabetic retina [1–4]. Sortilin has emerged as a novel protein that can regulate inflammatory mediators [5]. Sortilin is a member of the type-I transmembrane receptor family containing a Vps10p domain [6]. All 5 members of this family have the Vps10 domain in the N-terminal region and play key roles in signal transduction and protein trafficking [6]. Sortilin can be bound by pro-neurotrophins, pro-nerve growth factor (NGF) and pro-brain derived nerve growth factor (proBDNF), along with P75NTR, to activate inflammatory and apoptotic pathways [7]. To date, sortilin has been associated with cellular trafficking of over 50 molecules, including lipoproteins, cytokines, and enzymes [5]. One function of sortilin is as a co-receptor for proNGF with P75NTR to mediate neuronal apoptosis [8]. In addition to binding proNGF, sortilin binds proBDNF to mediate neuropathic pain, as shown in sortilin knockout mice [9]. Sortilin plays a role in cardiovascular disease through the regulation of LDL [5]. It serves as a receptor for lipids, cytokines, and enzymes to mediate pathological changes involved in atherosclerosis and inflammation [10]. Sortilin also mediates endothelial cell dysfunction, leading to arterial hypertension [11].

In addition to inflammation, sortilin has also been linked to other pathological pathways associated with diabetic retinopathy, including autophagy. In hippocampal neurons treated with amyloid β1-42, Sortilin contributed to elimination of clustrin protein through targeting into lysosomes [12]. The mannose-6-phosphate receptor pathway is a mechanism by which sortilin targets cathepsins D and H into lysosomes for degradation [13], specifically in rat epididymal cells [14]. Sortilin also regulates macrophage accumulation of cholesterol by lysosomal degradation [15].

Some research has been done on sortilin in the retina and the diabetic retina. In a model of retinal ischemia, sortilin was localized to Müller cells with reduced expression in RGCs [16]. In studies of patients with proliferative diabetic retinopathy, significantly increased levels of sortilin were observed in retinal lysates of the patients [17,18]. A recent study showed increased sortilin levels in the human diabetic retina and in the retina of STZ-treated diabetic mice, an increase which was associated with the actions of proNGF/P75NTR in mice that were diabetic for 8 weeks [19].

Recently, a small molecule inhibitor of sortilin, AF38469, was shown effective when given orally in breast cancer studies [20]. AF38469 reduced metastases into the skin. Because sortilin is linked to inflammation, we wanted to evaluate whether inhibition of sortilin with AF38469 would reduce inflammatory mediators in diabetic mice. We hypothesized that inhibition of sortilin with AF38469 given in the drinking water would reduce inflammatory mediators and improve retinal function in 8-week diabetic mice.

Human subjects

Dr. Mohamed Al-Shabrawey (Oakland University) provided postmortem retinal protein samples from 4 non-diabetic donors and 5 donors with diabetic retinopathy (both type 1 and type 2). The eyes were obtained from the Ever sight Eye Bank (Ann Arbor, MI). Approval for these samples was obtained from Oakland University stating “This study followed the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Oakland University (5/9/23; IRB-FY2023-292)”. Samples were deemed not human research on May 9, 2023 for the study entitled “Molecular and cellular mechanisms of diabetic retinopathy.” All diabetic patients had disease for 10+ years. These samples were processed for protein detection by Western blotting, as we have done in the past [21].

Diabetic mice for sortilin protein assessment

C57BL/6 male mice were purchased from Jackson Laboratories at 8 weeks of age. Diabetes was induced as we have done in the past [22]. Glucose levels >250 mg/dl were considered diabetic. At 2 months of diabetes, 5 control and 5 diabetic mice were sacrificed for analyses [23]. Mice were allowed free access to water and food and kept at a constant temperature for all experiments. Euthanasia was done via CO₂ overdose and cervical dislocation. All mouse experiments were approved by the Wayne State University IACUC committee (IACUC 24-02-6569) and adhere to the rules provided by the Association for Research in Vision and Ophthalmology (ARVO).

Diabetic mice for AF38469 treatment

Additional 8-week-old male C57BL/6 mice were made diabetic with STZ (60 mg/kg over 5 days). Once blood glucose was >250 mg/dl, a subset of the diabetic mice was given oral AF38469 (10 ug/kg) [20] for 6 weeks in their drinking water. AF38469 was dissolved in saline to make a 10 ug/kg dose and then placed into drinking water for the mice. Water intake was not measured directly. All mice were given free access to water.

Electroretinogram (ERG)

ERG was done on mice after 6 weeks of AF38469 treatment. Mice were dark-adapted overnight. The following morning, mice were anesthetized with ketamine/xylazine. Anesthetized animals were given eye drops of topical anesthesia (proparacaine, Alcon, Fort Worth, Texas), and pupils dilated with topical drops of tropicamide 1% (Alcon Laboratories, Fort Worth, TX). DTL ERG thread electrodes (Diagnosys LLC.) were placed on the cornea of both eyes to allow simultaneous recording using a Celeris machine. A plastic contact lens custom-made using Aclar film (Ted Pella) was used to help hold the electrodes in place. Tests will be carried out sequentially at light intensities of 0, 10, 100 mcd •s/m². A five-minute recovery period will be allowed between different intensities. Each response will be recorded for 250 ms. Responses to 25 light flashes will be averaged to produce an ERG recording at each intensity. Analysis of ERG graphs was carried out using ERGView software to extract values for the b-wave amplitudes, as well as oscillatory potential amplitudes.

Fluorescein angiography

After 6 weeks of diabetes and AF38469 treatment, some mice were subjected to fluorescein angiography. Briefly, the pupil of the mice was dilated with tropicamide ophthalmic solution. Fifteen minutes later, mice were anesthetized by ketamine and xylazine. After mice were deeply anesthetized without toe pinch reflex, 150 µl of AK-FLUOR (1% W/V, AKORN, INC, Lake Forest, IL) was injected intraperitoneally. Retinal vessel leakage was analyzed and photographed using the Micron IV (Phoenix Research Labs, Pleasanton, CA).

Western blotting

The Western blotting on the human and diabetic mouse retinal protein samples was done as we have previously published [24]. After blocking in TBST (10 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl, 0.1% Tween 20) and 5% (w/v) BSA, the membranes were treated overnight with primary antibodies against sortilin (ab263864, Abcam) or beta actin (sc-47778, Santa Cruz Biotechnology, Santa Cruz, CA) followed by incubation with secondary antibodies (anti-rabbit-HRP, Promega, Madison, WI) for 0.5–1 hour. Antigen-antibody complexes were detected with a chemiluminescence reagent kit (Thermo Scientific, Pittsburgh, PA) and data were acquired using an Azure C500 imager (Azure Biosystems, Dublin, CA). Western blot data were assessed with Image Studio Lite software.

For the western blots for Figures 2–6 on diabetic mouse samples treated with AF38469, we used Capillary Electrophoresis SDS (CE-SDS), also known as Simple Westerns. It is an automated assay performed by a machine, Abby (Biotechne) that allows for quantitative protein analysis. The total protein for each sample is determined before running and diluted to equal concentrations. Samples are then prepared to manufacture specifications and subjected to CE-SDS. Abby’s CE-SDS allows for the migration of proteins based on size within the capillary, which are then immobilized to the sides of the sample’s respective capillary. After the immobilization, the resolving matrix is removed, blocked, and treated with subsequent antibodies. Detection is done via chemiluminescent and images at varying predetermined time points. Light intensity is recorded and then quantified. After the first run, the capillaries can be stripped and re-probed for further proteins. Results can be normalized to either a housekeeping protein or a total protein. Due to the automated nature of the assay, consistency, reproducibility, and efficiency are greatly improved. It also removes problematic steps that can hinder data acquisition. Due to the combination of processes, samples can be treated as truly quantifiable and compared to conventional Western blots. Antibodies used for these studies were the same sortilin and beta actin antibodies, as well as HMGB1 (ab18256, Abcam), IL-1β, (2805, Invitrogen) proBDNF (ab68151, Abcam), and LAMP2 (66301-1-lg, Proteintech).

Statistics

All experiments were repeated a minimum of three times, and the data are presented as mean ± SEM. Data were analyzed by a non-parametric Kruskal-Wallis 1-way ANOVA, followed by a Dunn’s test or 2-tailed t-test with p values <0.05 considered statistically significant. A representative blot is shown.

Sortilin is increased in the diabetic retina

We used retinal protein samples from human diabetic patients and 2-month diabetic mice to measure protein levels. We found that diabetes significantly increased sortilin levels in both human and mouse retina (Figure 1). These findings suggest diabetes affects sortilin actions.

Sortilin inhibitor AF38469 significantly reduced sortilin levels

To determine whether AF38469, a small molecule inhibitor for sortilin, could reduce sortilin levels in whole retina from diabetic mice, we made mice diabetic and then gave a subset of mice the AF38469 dissolved in their drinking water for 8 weeks. Figure 2 shows that diabetes increased sortilin levels, which was significantly decreased by oral delivery of AF38469, even relative to untreated mice.

AF38469 improves retinal electrical activity

To determine whether inhibition of sortilin could protect the diabetic retina, we measured the electroretinogram (ERG) from control, diabetic, and diabetic+ oral AF38469 mice. We found that diabetes decreased the B-wave and OPs in the retina, which has been reported by others [25]. Figure 3 shows that, diabetic mice treated with oral AF38469 had improved ERG waveforms compared to diabetes alone.

Sortilin inhibition improves retinal leakage

Since diabetes increases retinal permeability [26], we wanted to determine if inhibition of sortilin could reduce vascular leakage. Figure 4 shows that diabetes has increased permeability (arrows), which is substantially reduced in diabetic mice that received oral AF38469.

AF38469 reduced inflammatory mediators in diabetic mice

We have previously reported that diabetes significantly increased inflammatory mediators [22]. We found that diabetes increased HMGB1 and IL-1β levels in the mice. Figure 5 shows that oral AF38469 significantly reduced protein levels of both HMGB1 and IL-1β.

Sortilin regulates proBDNF and LAMP2 Levels

Literature suggests that sortilin should regulate proBDNF and autophagic proteins [15,27]. To investigate this in the retina, we measured protein levels of proBDNF and LAMP2 in the diabetic retina. Figure 6 shows that diabetic mice have significantly increased proBDNF and LAMP2 levels, which were significantly reduced by oral AF38469. These data suggest that inhibition of sortilin may offer a newer therapeutic option for diabetic retinopathy.

Our findings of increased sortilin in the diabetic retina of human patients and mice agree with the literature [18,19]. Jakobsen et al., showed improved neuronal responses using a nerve growth factor (NGF)-binding antibody [19]. For our studies we chose to use the small molecule inhibitor of sortilin (AF38469), which had been shown effective in reducing metastases in breast cancer models [20]. Using the same dose as in the breast cancer model, we treated some diabetic mice with AF38469 in the drinking water. After mice consumed the water for 6 weeks, we measured ERG and performed angiography on the mice. We and others have previously reported that diabetes causes a decrease in the B-wave and oscillatory potentials in the ERG [28–30]. These studies confirmed a loss of B-wave amplitude, which was significantly improved by oral AF38469. Additionally, diabetes leads to increased retinal leakage [31]. Oral AF38469 reduced vascular leakage in diabetic mice.

While we saw improvement in retinal function in diabetic mice treated with oral AF38469, we also wanted to investigate potential pathways that are often associated with diabetes. We have spent the past decade investigating inflammatory pathways in the diabetic retina [21,22,32]. These studies showed increased HMGB1 and IL-1β levels in the diabetic retina that was significantly reduced by oral AF38469. Our findings of a role for sortilin in the retina follows literature in other models [5,11]. In addition to inflammatory mediators, we found that oral AF38469 reduced pro-BDNF and LAMP2 levels. These findings also match literature for sortilin actions in other models [12,27].

While these studies demonstrate the potential for AF38469 as a therapeutic for diabetic retinopathy, additional studies need to be completed for the chronic effects of AF38469 treatment. Side effects of oral drug delivery need to be further studies. We also need to measure local retinal levels of sortilin and the AF38469 inhibitor. We also acknowledge that the human samples were post-mortem, and the collection times/disease duration could be different. Control mice will also need to be studied for toxicology of AF38469 alone. Future studies will be designed to address these concerns.

In conclusion, these early studies show that sortilin is significantly increased in the diabetic retina. This was associated with increased lysosomal proteins and inflammatory mediators. Delivery of an oral sortilin antagonist, AF38469, significantly improved the ERG and angiography measures in diabetic mice. These mice also had reduced inflammatory mediators and LAMP2 protein levels.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

LL did all mouse work; YJ performed the Western blotting work; MA provided the human samples; JJS designed the experiments and wrote the text.

These studies were funded by R01EY030284 (JJS), P30EY04068 Core grant (LDH, PI of Core grant), an unrestricted grant from Research to Prevent Blindness, and R01EY030054 (MA).