The present study used 8-week-old male Wistar rats and was approved by Ethics Committee of the Institute of Experimental Animals of St. Marianna University School of Medicine. The rats were maintained in the controlled rooms (23±1°C; humidity at 55±5%; light on 06:00 to 18:00).

Single i.p. administration of physiological saline solution (PSS) or 60 mg/kg streptozotocin (STZ; Wako Pure Chemical Industries, Ltd., Osaka, Japan) was carried out for the normoglycemic (NG) rats or the hyperglycemic (HG) rats, respectively. The plasma glucose levels were measured using a glucometer (Johnson & Johnson, Tokyo, Japan) 4 days after intraperitoneal injection. We only included the individuals as the HG group when plasma glucose exceeded 250 mg/dl. The plasma glucose levels of HG groups were 441.5±71.3, 435.0±32.3, and 432.1±73.3 mg/dl, immunoblot analysis, immunohistochemical analysis, and morphometric analysis studies, respectively.

Intravitreal administration of 10 ng TNF (2 µl) was carried out into the right eye of rats under anesthetization with a combination of ketamine and xylazine. The left eye was received an intravitreal administration of phosphate-buffered saline (PBS). These intravitreal administrations were carried out 4 days following i.p. injection of PSS or STZ. In the HG group, a simultaneous intravitreal administration of 50 pmol Beclin-1 siRNA (Cell Signaling Technology, Inc., Danvers, MA, USA) with TNF was carried out into the right eyes.

Thirty-six rats (NG: 18 rats; HG: 18 rats) were euthanatized 1 week after intravitreal administrations for immunoblot analysis. Four mm optic nerves from immediately behind the eye ball were homogenized in protein extraction buffer. Since the optic nerve pieces were small, each sample included two optic nerves. Equal amount of proteins (3 µg) determined by the Bradford assay was applied and loaded. Then, samples were transferred to PVDF membranes. After blocking, membranes were exposed with the primary antibodies: Anti-Beclin-1 antibody for overnight (1:200; Medical & Biological Laboratories, Co., Ltd., Nagoya, Japan) or anti-β-actin antibody for 2 h (1:500; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). After three times washing, membranes were reacted with the secondary antibodies: Rabbit IgG or mouse IgG. Immunoblots were evaluated with the Amersham ECL detection system (GE Healthcare Life Sciences, Little Chalfont, UK).

Six rats were used for immunohistochemical analysis. One week following intravitreal administration, optic nerve samples were immersed in 10% neutral-buffered formalin. Paraffinized cross sections were made in 2 µm thick and incubated with 1% bovine serum. The primary antibodies were anti-Beclin-1 antibody (1:100; Medical & Biological Laboratories, Co., Ltd.), glial fibrillary acidic protein (GFAP, a marker of astrocytes; 1:200; Agilent Technologies, Inc., Santa Clara, CA, USA), and neurofilament-L (a marker of neurons; 1:100; Agilent Technologies, Inc.). The secondary antibodies were FITC-labeled and rhodamine-labeled IGG. The slides were mounted in 4′,6-diamidino-2-phenylindole-including medium (Vector Laboratories, Ltd., Peterborough, UK).

Fourteen rats (NG: 4 rats; HG:10 rats) were euthanatized 2 weeks after intravitreal administration for axon morphometric analysis (10,27). Optic nerve samples were immersed in Karnovsky's solution for overnight. After embedded in plastic blocks, cross thin sections were made beginning 1 mm from the eye ball. The sections were stained with 1% paraphenylen-diamine (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). Five images (center and periphery in quadrant per optic nerve) were acquired and quantified using an image-processing software (Aphelion, ADCIS, Hérouville Saint-Clair, France). The average of axon number in each optic nerve was expressed as the number per mm².

Data are presented as mean ± standard error of the mean. Differences among groups were analyzed using one-way analysis of variance with Dunnett's post hoc test. JMP v12.0.1 software (SAS Institute, Inc., Cary, NC, USA) was used for statistical analyses. P<0.05 was considered to indicate a statistically significant difference.

The current study found that HG condition markedly increases Beclin-1 protein levels in optic nerves (Fig. 1). These significant increments were seen in both PBS-treated and TNF-treated eyes (Fig. 1). However, intravitreal administration of TNF did not alter the Beclin-1 expression in both NG and HG conditions.

In cross sections, immunohistochemical study revealed abundant colocalization of Beclin-1 and GFAP in the optic nerves in NG group (Fig. 2A-C). Although immunoreactivity pattern of Beclin-1 is different from that of neurofilament (Fig. 2D-F), some immunoreactivities of Beclin-1 were apparently colocalized with those of neurofilament in the HG group (Fig. 2G-L). These findings suggest that Beclin-1 is present mainly in glial cells, but partially in neurofilament, and that the expression of Beclin-1 may be upregulated by HG.

Consistent with our previous findings (10), the current study showed that STZ-induced HG condition appeared a significant ameliorative effect on axonal loss caused by TNF (Fig. 3B, C, E). Since we found a significant upregulation of Beclin-1 protein level in the HG group, we examined whether Beclin-1 siRNA alters this protective effect. Noticeable degenerative changes were seen in the HG-TNF with Beclin-1 siRNA treatment group (Fig. 3D). Although no significant difference in the axon number was seen in between the HG-TNF group and the HG-TNF with Beclin-1 siRNA treatment group, no significant difference was also seen in between the NG-TNF group and the HG-TNF with Beclin-1 siRNA treatment group (Fig. 3E). These observations suggested that the protective effect of HG was only partially suppressed by Beclin-1 siRNA.