Metformin helps protect eye nerve cells after blood flow injury and may keep vision stable in diabetic glaucoma patients

Retinal ischemia induces significant RGC loss [23, 25]. The animal model used in this study is characterized by a mechanical, pressure dependent occlusion of the central retinal artery, resulting in ischemia followed by reperfusion upon pressure release. This ischemia-reperfusion insult mimics some aspects of glaucomatous optic neuropathy, especially in patients with vascular dysregulation or impaired autoregulation of ocular blood flow. In this context, the model is useful for testing neuroprotective strategies that may mitigate RGC death regardless of IOP control, which is highly relevant for glaucoma patients whose disease progresses despite normalized IOP [2, 29, 30].

Body weight was daily monitored and glycemia was checked on the first and last day of the treatment period with no significant differences observed between metformin and vehicle treated animals (Fig. 1B, C).

Under this administration schedule, metformin reached effective concentrations in the retina as shown by the increased phosphorylation of AMPK, a well-characterized downstream target of the drug [12, 33, 34].

Metformin 10 or 50 mg/kg enhanced AMPK phosphorylation (Thr172) in control retinas and abrogated the difference in AMPK phosphorylation level between control and ischemic retinas observed in the vehicle treated group (Fig. 1D). Notably, only the 50 mg/kg dose significantly enhanced AMPK phosphorylation in ischemic retinas (Fig. 1D)

Histological evaluation revealed that retinas exposed to ischemia injury from mice treated with metformin 10 mg/kg displayed preserved laminar organization, ganglion cell layer (GCL) density, and fewer structural disruptions compared to vehicle-treated animals (Fig. 2C).

Metformin is recognized as a caloric-restriction mimetic and autophagy modulator [19, 35, 36].

To detect changes in autophagic structures, transmission electron microscopy (TEM) - the gold standard for autophagosome detection- was performed [37].

SQSTM1/p62 and optineurin are autophagy receptors acting as molecular bridges between the ubiquitinated cargo and the autophagy machinery. While SQSTM1/p62 is primarily involved in macroautophagy [38, 39], optineurin is an essential mitophagy adaptor serving as autophagy receptor for damaged mitochondria [40, 41]. Indeed, its down-regulation has been associated with reduced mitophagy and accumulation of damaged organelles [41, 42].

In our previous study we observed an increase in SQSTM1/p62 expression in the ischemic retinas 24 h post-insult, suggesting a reduction of autophagy efficiency [23]. Metformin did not reduce the accumulation of SQSTM1/p62 in the ischemic retina, but prevented the transcriptional upregulation of Sqstm1 mRNA observed in the vehicle-treated samples (Fig. 4B, C), suggesting that metformin may limit the stress response by mitigating tissue damage.

Under our experimental conditions, optineurin protein (Fig. 4D) and mRNA (Fig. 4E) levels were significantly reduced in the retina 24 h post-ischemia in vehicle-treated mice. This downregulation was prevented by metformin (10 mg/kg; Fig. 4D, E), suggesting preservation of mitophagy mechanisms.

Parkin recruitments and activation to damaged mitochondria is mediated by the activity of the serine/threonine kinase PINK1 [44].

24 h after ischemia PINK1 expression was significantly upregulated in the ischemic retinas of vehicle treated group compared to contralateral control retina (Fig. 5). Treatment with both 10 and 50 mg/kg metformin abrogated the differences in PINK1 expression between ischemic retina and contralateral control (Fig. 5C).

Western blot analysis was performed on the optic nerve (ON) 24 h post-ischemia.

Metformin treatment upregulated Parkin expression in the ischemic ON, supporting the hypothesis that mitophagy was sustained in both retina and ON by the treatment (Fig. 6).

A statistically significance was reported for 182 proteins and five different clusters (and two single proteins) were identified based on protein abundance distributions (Fig. 7A).

Interestingly, several proteins included in the cluster indicated in pink color were upregulated in both ischemic and contralateral control retinas from metformin treated mice as compared to vehicle-treated (Fig. 7A). This suggest a change in retinal protein distribution which is independent of the ischemic insult and may be driven by either local or systemic effects of the drug. Sixty-eight proteins were found to be significantly different between metformin and vehicle treated ischemic retinas (Fig. 7B); among these, 36 were upregulated (Fig. 7C) and 32 were downregulated (Fig. 7D). Although pathway enrichment analysis yielded no statistically significant results, several relevant targets were identified.

Proteins involved in the formation of mitochondrial respiratory chain complexes (NADH-ubiquinone oxidoreductase chain 5, NADH dehydrogenase [ubiquinone] 1 alpha subcomplex assembly factor 4) and enzymes involved in the ubiquitination machinery (Ubiquitin-like modifier-activating enzyme 6, Ubiquitin-conjugating enzyme E2 D3) were reported as significantly modulated by the treatment (Fig. 7C, D).

ADP-ribosylation factor 6 (Arf6) was found to be one of the foremost upregulated proteins. Arf6 is an ADP ribosylation factor (Arf), member of the small G protein family, acting as a positive modulator of autophagy via regulation of phosphatidylinositol 4-phosphate 5-kinase (PIP5K) [45, 46].

Another notable upregulated protein was form-binding protein 1 (FNBP1) recently identified as a nutrient stress-responsive golgiphagy receptor. FNBP1 localizes at the fragmented Golgi membrane and engages LC3 thereby sequestering the fragmented Golgi membrane in autophagosomes and preventing metabolic stress in neurons [47].

Gene products involved in the trafficking of proteins between the trans-Golgi network and endosomes, such as AP-1 complex subunit gamma-1 (Ap1g1) and Vacuolar Protein Sorting-Associated Protein 26B (VPS26B), were also found to be significantly upregulated in the metformin-treated ischemic retina.

Conversely, inositol polyphosphate 5-phosphatase K (Inpp5k), which takes part in the lysosomal recycling pathway [48], was among the most significantly downregulated proteins in metformin-treated ischemic retina.

In a pilot retrospective clinical analysis, visual field (VF) changes (the gold standard in the diagnosis, monitoring, and management of glaucoma) were evaluated in diabetic patients with glaucoma treated with metformin (16 eyes of 8 patients) or insulin (16 eyes of 8 patients) over a six-month period.

The Kolmogorov–Smirnov test indicated that age and IOP data were normally distributed, whereas VF parameters data were not; therefore, the paired or unpaired t-test was used for age and IOP comparisons and the Mann–Whitney U or Wilcoxon tests were applied for VF data, as appropriate. At T0, both groups were found to be comparable regarding age (p = 0.09) (metformin group 71.50 ± 7.06 years; insulin group: 71,38 ± 7.10 years), gender (p = 0.73), and glaucoma stage; as assessed by VF parameters including Mean Deviation (MD) (p = 0.75), Pattern Standard Deviation (PSD) (p = 0.93), and Visual Field Index (VFI) (p = 0.91). No intergroup differences emerged in terms of IOP at T0 and T6 (p = 0.75 and 0.86 respectively). No intragroup differences emerged from T0 to T6 (p = 0.40 and 0.31 in the metformin and insulin groups respectively); thus, ensuring that any differences observed during the study could more confidently be attributed to the treatment administered.

Over time, a notable distinction emerged between the two treatments: while patients in the insulin group experienced worsening of their visual function, those receiving metformin were stable.

Furthermore, no significant differences were found in terms of PSD at T6 compared to T0 in the metformin and insulin groups (p = 0.70 and p = 0.89; respectively) (Fig. 8A). PSD quantifies localized irregularities in the VF by measuring the variability of individual test point deviations from the expected hill of vision. A higher PSD suggests greater localized VF defects, which are typical in early stages of glaucomatous damage.

Finally, when VFI was analyzed, values were not significantly modified in the metformin glaucoma group (p = 0.38), while there was a significant worsening in the insulin group (p = 0.03) comparing T0 to T6 (Fig. 8B). VFI is a weighted index expressed as a percentage, where 100% represents a normal VF and lower values indicate more severe loss. It places greater emphasis on the central VF and is less affected by cataract or diffuse loss, making it useful for tracking glaucoma progression.

The short observation time in this study possibly further reinforced our results highly reducing the risk that other factors (such as cataract) may have induced VF alterations.

Male C57BL/6 J mice (25–30 g) were purchased from Charles River (Lecco, Italy) and housed with a 12 h light–dark cycle with ad libitum access to food and water. Animal care and experimental procedures were carried out in accordance with the guidelines of the Italian Ministry of Health (D.L. 26/2014), the European Communities Council Directive (2010/63/UE) and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The experimental protocol was approved by the Italian Ministry of Health (Rome; NIH license no. 1026/2016-PR and no. 455/2023-PR). All surgical procedures were performed under deep anaesthesia and efforts were made to minimize the number of animals used and their suffering. The sample size (3–6 animals per group) was chosen based on previous studies using the same animal model [23, 25]. This number was sufficient in earlier work to detect a biologically relevant effect on RGC survival and functional recovery, and aligns with ARRIVE guidelines for exploratory preclinical studies.

Retinal ischemia was induced in the right eye according to the method previously reported [23, 25]. Animals were deeply anesthetized by intraperitoneal injection (i.p.) of 15 mg/kg Xylazine (Sedaxilan®, Dechra Veterinary Products s.r.l., Turin, Italy) and 30 mg/kg Tiletamine-Zolazepam (Zoletil®, Virbac Srl, Milan, Italy) and laid on a heating pad to maintain the body temperature at 37 °C. Topical anaesthesia was induced by 0.4% Oxybuprocaine eye drops (Novesina®, Novartis Farma, VA, Italy). A 31-gauge infusion needle, connected to a 500 ml bottle of sterile saline, was inserted in the anterior chamber of the right eye, and the saline container was elevated to produce an intraocular pressure of 7080 mmHg for 60 min. IOP value (I, ischemic) was monitored by TonoLab (Icare, Finald). For each animal, the left eye was used as control (C, control). Body temperature was monitored and animals with values lower than 35.5 °C were excluded from the study. Mice were sacrificed, under deep anaesthesia, by cervical dislocation at 24 h or 7 days of reperfusion. To minimize the basal variations due to the circadian autophagy regulation [80], ischemia was induced between 8.00 am and 2.00 pm.

Metformin hydrochloride (PHR1084, Sigma-Aldrich, Milan, Italy) was dissolved in sterile water for injection and stored at −20 °C. Metformin (10 or 50 mg/kg) or vehicle were injected i.p. once a day. Transient ocular hypertension was induced on the fourth day of treatment and animals were killed by cervical dislocation under deep anaesthesia after 24 h or 7 days. Animals were randomly allocated to either the untreated control group or the treatment group to minimize selection bias and ensure balanced group distribution.

Eyes were enucleated and fixed in 4% paraformaldehyde at 4 °C for 2 h, the anterior cup and the lens removed, and the posterior cup cryopreserved in increasing concentrations of sucrose as follows: 10% sucrose for 1 h, 20% sucrose for 2 h, 30% sucrose overnight. Specimens were included in Optimal Cutting Temperature compound (Tissue-Tek®, Sakura Finetek Europe, The Netherlands) and frozen. 14 μm thick sections were cut with Cryostat (Leica), mounted onto Superfrost Plus glass slides (Thermo Fisher Scientific, Waltham, MA, USA) and stored at −20 °C until use. For cellular and subcellular localization of specific antigens, retinal sections were thawed, air-dried, post-fixed in 4% paraformaldehyde for 10 min and washed in 0.1 M PBS (pH 7.4). Sections were permeabilized with 0.3% Triton-X100 (Sigma-Aldrich, Milan, Italy) for 1 h, washed three times with PBS and blocked with 10% donkey serum (Sigma-Aldrich, Milan, Italy) at room temperature for 1 h. Slides were incubated with the anti-Parkin antibody (Santa Cruz, sc-32282; 1:50) in 5% donkey serum overnight, washed with PBS-0.1% Tween 20 and incubated with anti-mouse Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA) at room temperature for 1 h (1:1250). Sections were mounted with Vectashield mounting media with DAPI (Vector Laboratories, Burlingame, CA, USA) and images acquired using a confocal microscope (Olympus FV3000, Tokyo, Japan).

Total RNA was extracted with Triazol (Invitrogen, Carlsbad, CA). One microgram of total RNA was reverse transcribed in a final volume of 50 μl using the High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA); cDNA was diluted 1:3 in nuclease free water.

Quantitative PCR was performed using the following primer sequences: Specific primers for optineurin

Forward sequence 5'-CCAGCAGACTTACCTGTTTC-3'

Reverse sequence 5'- GCAGGAGTGAATCGGAATAC-3' Specific primers for p62

Forward sequence 5'- CACCAGAAGATCCCAATGTC-3'

Reverse sequence 5'- CCTCCATGTTCCACATCAAT-3' Specific primers for 18S

Forward sequence 5'-C GGCGACGACCCATTCGAAC-3'

Reverse sequence 5'- GAATCGAACCCTGATTCCCCGTC-3'

PCR reactions were performed in the QuantStudioTm 3 Real Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA) using 0.15 μmol/L of each primer. PowerUp™ SYBR™ Green MasterMix (Thermo Fisher Scientific, Waltham, MA, USA) with the dissociation protocol was used for gene amplification. Each sample was normalized to its 18S rRNA (18S) content. Results were expressed as n-fold differences relative to a calibrator and calculated using the ΔΔCt method.

Mice were sacrificed by cervical dislocation 24 h following retinal ischemia. Eyes were enucleated, fixed for 20 min in 4% PFA at 4 °C and the anterior cup and lens removed. Posterior cup was fixed in Karnowsky fixative (4% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4), and then postfixed with 1% osmium tetroxide in 0.1 M phosphate buffer for 1 h at room temperature. After standard ethanol dehydration, samples were embedded in an Epon-Araldite 812 mixture. Sections were obtained with a Reichert Ultracut S ultramicrotome (Leica, Nussolch, Germany). Semi-thin sections (600-nm-thick) were stained with crystal violet and basic fuchsin and then observed with an Eclipse Ni-U microscope (Nikon, Tokyo, Japan) equipped with a DS-SM-L1 digital camera (Nikon). Ultra-thin sections (70-nm-thick) were stained with uranyl acetate and lead citrate, and then observed with a JEM-1400Flash transmission electron microscope (TEM) (Jeol, Tokyo, Japan). TEM images were acquired with a Morada digital camera.

A morphometric analysis was performed on electron micrographs with Fiji (ImageJ, NIH, Bethesda, MD, USA). For each treatment, six TEM images were taken at 12,000x magnification; for each micrograph, a cytoplasmic area of 100 μm² was selected in the ganglion cell layer (GCL) and the number of autophagic compartments was counted.

RGCs were retrogradely labeled by stereotaxically injecting the fluorescent tracer FluoroGold (FG; Fluka, Sigma-Aldrich, Milan, Italy) into the superior colliculi. Four days after the ischemic insult, mice were deeply anaesthetized, immobilized in a stereotaxic device (Kopf 900, Analytical Control, Milan, Italy) and the positions of superior colliculi were identified by the stereotaxic coordinates reported in Paxinos and Franklin Mouse Brain Atlas (2019). The skull was exposed and 2 μl of 2% FG solution were injected on both sides of the skull 4 mm posterior to the bregma, 1 mm lateral to the sagittal suture and 1.6 mm ventral from the bone surface using a Hamilton Neuros-syringe with a 33-gauge needle (Hamilton Europe, Bonaduz, Switzerland). The skin was then sutured, and a 0.3% tobramycin ointment was applied (Tobral®, Alcon, Milan, Italy). Animals were killed 7 days after ischemia and eyeballs enucleated and fixed for 20 min in paraformaldehyde 4% in PBS. The anterior segment of the eye was removed and the posterior eyecup additionally fixed for 30 min. Isolated retinas were divided into four quadrants (superior, inferior, nasal and temporal) and mounted on slides using the ProLong Gold Antifade Mountant (Thermo Fisher Scientific, Waltham, MA, USA). Thirty-two images per retina (three from the peripheral, three from the middle, and two from the central retina for each quadrant) were acquired using a confocal microscope (Olympus FV3000, Tokyo, Japan) and subjected to cell count using ImageJ software (NIH, Bethesda, MD, USA). The total number of labeled cells in the eye subjected to retinal ischemia was compared with contralateral and expressed as percentage of RGC survival.

Proteins were digested according to the Protein Aggregation Capture (PAC) protocol [81]. Briefly, protein solutions were diluted in RIPA buffer to a volume of 30 ml (15 µg of total proteins). Reduction and cysteine alkylation proceeded by adding 4.5 ml of 100 mM DTT dissolved in Tris HCl pH 8.5 (1 h at 37 °C) and 3.6 ml of 200 mM iodoacetamide (1 h at 37 °C). 19.05 ml of the protein solution (10 mg) were processed by the PAC protocol. MagReSyn® Hydroxyl microparticles (Resyn Biosciences) were equilibrated twice in 70% acetonitrile, then resuspended in 70% acetonitrile in order to obtain the initial storage volume of the beads. Five µl of beads suspension were utilized per sample and placed in a 2 ml Eppendorf tube. Reduced and alkylated proteins (19.05 ml) were added to the beads. Then, 45 µl of acetonitrile were added to initiate protein precipitation. The solution was pipetted vigorously for 20 times and then incubated at 1100 rpm for 10 min. Beads were then separated using a magnet for a few seconds. The supernatant was removed, and beads were washed twice with 200 µl of 100% acetonitrile and one with 200 µl of 70% ethanol while still in the magnetic rack. The tubes were then removed from the magnetic rack. Trypsin solution (50 µl of 50 mM TEAB + 1 µl of trypsin concentrated 0.2 µg/µl) was added and digestion was allowed to proceed overnight at 37 °C and 1100 rpm. The following day, the supernatants containing the peptides were removed, and magnetic beads were washed with 50 µl of 0.1% FA (v/v). The two supernatants were pooled and the peptide mixtures were directly injected for nanoLC-MS/MS analysis. NanoLC-MS/MS analysis was performed on an EASY1200 LC system coupled to an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher). Peptides were separated using an in-house made 15 cm analytical column packed with 3 μm-C18 silica particles (Dr. Maisch) essentially as described in Prestagiacomo et al. [82]. Briefly, gradient elution of peptides was achieved at 300 nl/min using a 140 min gradient (from 3% B to 8% B in 40 min, then from 8% B to 32% B in 120 min, followed by column regeneration at 100% B for 20 min). Mobile phase A was 0.1% formic acid containing 2% acetonitrile, whereas mobile phase B was 0.1% formic acid containing 80% acetonitrile. The nanoLC effluent was directly electrosprayed into the mass spectrometer in positive ion mode (1800 V). The mass spectrometer operated in DIA mode, using 60 sequential acquisition windows covering an m/z range of 350–1000. Full MS scans (350–1000 m/z) at 60,000 resolution were followed by 60 sequential MS/MS windows: 45 scans in the range 350–710 having 8 m/z width, 9 scans in the range 710–870 having 16 m/z width, 5 scans in the range 870–1010 having 28 m/z width. In all cases, window overlap was 0.5 m/z, target was 500% (5e5), normalized collision energy was 25, resolution was 30,000, maximum injection time was 50 ms. For the data analysis, library generation was achieved in Spectronaut Pulsar (Biognosys, v.16.0) using default parameters. Details can be found in Prestagiacomo et al. [82]. MS/MS spectra were searched against the Mus musculus protein database accessed on May 23th 2022 (55.315 sequences). DIA data analysis was performed on the same platform (Spectronaut) using default parameters. The "Major group top N" parameter (number of peptides used for quantification) was between 1 and 10, data were imputed using a global imputation strategy and automatic normalization was adopted in Spectronaut. For statistical analysis of Spectronaut output, Perseus software was used as follows: the Protein Group Quantity from the Spectronaut analysis was imported. Data were transformed in logarithmic scale (log2). After filtering (at last two valid LFQ values in at least one group), two-sample t test has been used to assess the statistical significance of protein abundances using a 5% permutation-based FDR adjustment. A scaling factor was used for corrections 0:0.2.

For this pilot retrospective study, clinical notes of glaucomatous diabetic patients of the glaucoma department of the University Hospital of Rome Tor Vergata were reviewed.

All methods were carried out in accordance with the relevant guidelines and regulations, including the principles of the Declaration of Helsinki. Ethical approval for this study was obtained from the Independent Ethics Committee of University Hospital "Tor Vergata", Rome, Italy (Protocol name: INSUMET, Protocol number: 105.22). Informed consent was obtained from all participants involved in the study. All data were fully anonymized prior to analysis, and no images or other materials are included that could lead to the identification of individual participants.

Two groups of patients with glaucoma and diabetes, homogeneous in terms of age and gender, were extrapolated. The study was approved by the internal review board of the University Hospital of Rome "Tor Vergata".

The first group (metformin glaucoma group) included patients aged between 18 and 90 years. They must have a concomitant diagnosis of open-angle glaucoma and type 2 diabetes without signs of diabetic retinopathy; they must be under glaucoma treatment with IOP lowering medications and must be under glucose-lowering treatment with metformin. The clinical record must also report at least two 24-2 visual field (VF) carried out 6 months apart from each other.

The second group (insulin glaucoma group) included patients aged between 18 and 90 years. They must have a concomitant diagnosis of open-angle glaucoma and type 2 diabetes without signs of diabetic retinopathy; they must be under glaucoma treatment with IOP lowering medications and must be under glucose-lowering treatment with insulin. Metformin use was an exclusion criterion. The clinical record must also report at least two 24-2 VF carried out 6 months apart from each other.

Patients treated with brimonidine, citicoline, and coenzyme Q10 or affected by other neurological disorders were excluded.

The following data have been collected from these groups: Age Gender Type of glaucoma Topical and systemic therapy for glaucoma at the time of the exams (type of drugs and dosage) IOP at baseline (T0) and after 6 months (T6) 24-2 VF exam with particular attention to mean deviation (MD), pattern standard deviation (PSD), and visual field index (VFI).