All experimental procedures used in the present study were approved by the Ethics Committee of the 3rd Xiangya Hospital of Central South University in accordance with the National Institutes of Health (NIH) Guidelines for the Care and Use of Laboratory Animals. Primary retinal neuronal cultures were prepared from 1-day-old neonatal Sprague-Dawley (SD) rat as described previously (Li et al., 2016; Morgan-Warren et al., 2016; Wang S. et al., 2017). In brief, the eyes were removed aseptically, and then the corneas, crystalline lens, and pigment epithelium layer were removed, leaving the retinae. The retinae were placed in Dulbecco’s modified Eagle’s medium (DMEM, GE Healthcare, Logan, UT, United States) containing 0.02% papain and digested in a 5% CO₂ incubator at 37°C for 10 min. The tissue was then triturated 20 times with a Pasteur pipette and filtered with a 70-mm nylon cell strainer. The cells were counted and plated at a density of 6 × 10⁵ cells/mL. Cells were cultured in 5% CO₂ incubator at 37°C. Four hours after plating, the medium was replaced with neurobasal medium (Thermo Scientific, Waltham, MA, United States) supplemented with B27 (Thermo Scientific). Culture media were changed every 2 days. On the 7th day, the cultures were treated with 50 μM glutamate (Sinopharm, Beijing, China) for 2 h, and then allowed to recover for 0, 2, 4, 6, 12, or 24 h (Aihara et al., 2014; Kim et al., 2016).

The RGC-5 cell line was provided by the Department of Ophthalmology, Second Hospital of Jilin University in China (Ding et al., 2015). RGC-5 cells were cultured in DMEM (Thermo Scientific) supplemented with 10% FBS (Thermo Scientific) and 1% Penicillin-Streptomycin (Thermo Scientific). The cells were grown in 5% CO₂ incubator at 37°C. The RGC-5 cells used in the experiment was within 2–3 passages post-thawed. The density of RGC-5 cells was approximately 80% in the T25 flask before the experiment. The RGC-5 cells were treated with 10, 20, 30, 40, or 50 mM glutamate for 1 h (Maher and Hanneken, 2005; Fatma et al., 2008).

The SD rat glutamate model was prepared following the previously reported protocol (Fang et al., 2010). Rats were anesthetized using an intraperitoneal injection with a 1:1 mixed solution (5 mL/kg) of 10% chloralic hydras and 25% urethane. A drop of chloramphenicol eye drop was administered to the left conjunctiva sac. The pupils were dilated with tropicamide and a single dose of 5 μL of 100 mM glutamate in 0.01 M PBS was injected into the vitreous cavity using a 32-gauge Hamilton needle and syringe (Sisk and Kuwabara, 1985). A sham operation was performed as a control. Animals were allowed to survive for 6 h. Each group was composed of four animals.

At each sacrificed time-point, animals were anesthetized with 10% chloralic hydras (5 mL/kg) and perfused transcardially with 0.9% sodium chloride, followed by 4% paraformaldehyde (PF) in 0.1 M PB. After perfusion, the eyes were enucleated, anterior segments removed, and posterior eyecups were post-fixed in 4% PF overnight at 4°C. Then, the eyes were placed sequentially in 15 and 30% sucrose in 0.1 M PB at 4°C. Next, the eyecups were embedded in Tissue-Tek optimal cutting temperature medium and frozen in liquid nitrogen. After, cryosections were cut, at a thickness of 6 mm, using a microtome (Thermo Scientific). The sections that included the optic nerve were stored at -20°C until use.

Juglone (i.e., a specific pin1 inhibitor) was dissolved in dimethyl sulfoxide (DMSO, Sigma–Aldrich, St. Louis, MO, United States) as a stock solution. The stock solution was further diluted in culture medium to achieve DMSO concentrations below 0.1%. Juglone was used at a concentration of 10 μM. The drug solution was administered directly to primary retinal neuronal cultures or by intravitreal injection to rats, 2 h before the glutamate treatment.

The siRNA kit against Pin1 was obtained from RIBO-Biology (Guangzhou, China). In companion control experiments, RGC-5 were transfected with the same amount of either a targeting or non-targeting control. Another group with Lipo2000 (Thermo Scientific) alone was used as a normal control. The transfection was performed by following the manufacturer’s instructions. Briefly, the siRNA and transfection reagent were first separately diluted in opti-MEM for 5 min. Then, the diluted siRNA and transfection reagent solutions were mixed together for another 20 min to form the complex. Finally, the cell medium was replaced with opti-MEM, and the transfection complexes were added dropwise to the cultures. The culture medium was replaced with normal medium 6 h after transfection. The cultures were collected for the next experiments 24 h after transfection. Knockdown of calpain2 or Pin1 was validated using western blot.

For in vitro experiments, the LDH cytotoxicity assay kit (Beyotime, Shanghai, China) was used to measure LDH released from necrotic cells into the extracellular space/supernatant upon the rupture of plasma membrane (Xiong et al., 2016; Shang et al., 2017) after the different treatments. Cell-free culture supernatants were collected from 96-well microtiter plate and incubated with the appropriate reagent mixture according to the manufacturer’s instructions at room temperature (RT) for 30 min. For the in vivo experiment, the LDH cytotoxicity assay kit was purchased from Jiancheng institutes (Nanjing, China) and used according to the manufacturer’s instructions. Briefly, the retinae were homogenized by sonication in 0.86% NaCl and incubated with the appropriate reagent mixture at 37°C for 30 min. The intensity of the red color that formed in the assay, which was measured at a wavelength of 490 or 450 nm, was proportional to both the LDH activity and percentage of necrotic cells. The percentage of necrotic cell death was calculated by the color intensity of treated cells minus control cells/LDH releasing reagent treated cells minus control cells, from four independent experiments.

PI staining was used to identify cells undergoing necrosis (Rosenbaum et al., 2010; Pietkiewicz et al., 2015). Cell cultures on the coverslips were washed twice with PBS and incubated with PI at RT for 20 min in the dark. For animal experiment, 5 μL PI were intravitreally injected, 30 min prior to sacrifice (Huang et al., 2013). Rats were euthanized at the indicated time-point. The eyes were frozen in nitrogen vapor and cryosectioned. The coverslips or retinal sections were washed three times and covered with an anti-fading mounting medium with DAPI (Vector Laboratories, Burlingame, CA, United States). Images were captured using a fluorescence microscope (Olympus, Tokyo, Japan). The percentages of cells were analyzed using Image J software (NIH, Baltimore, MD, United States). The number of PI-positive cells was calculated in every intact captured image.

Cell cultures on the coverslips were fixed for 20 min with 4% PF and the rat retinal sections were recovered to RT. The coverslips or retinal sections were washed three times for 5 min in ice-cold 0.01 M PBS. Subsequently, the coverslips or retinal sections were blocked for 1 h in blocking buffer, i.e., PBS containing 5% normal bovine serum and 0.3% Triton X-100, and then incubated with combinations of the primary antibodies against the following targets, overnight at 4°C: tAIF (1:500, sc-13116, Santa Cruz), CAST (1:50, sc-2-7799, Santa Cruz, Dallas, TX, United States) and calpain2 (1:100, ab39165, Abcam, Cambridge, United Kingdom). The coverslips or retinal sections were shifted to RT for 30 min, washed three times as described above, and incubated with Alexa-conjugated secondary antibodies (1:200, Jackson Immuno Research, West Grove, PA, United States) for 2 h. After washing three times in PBS, the coverslips or retinal sections were covered with Vectashield mounting medium containing DAPI. The coverslips or retinal sections were stained in parallel and images were acquired, with a fluorescence microscope, using the same settings.

Total RNA was extracted using TRIzol Reagent (Thermo Scientific). For mRNA levels detection, cDNA was synthesized using the Revert Aid First Strand cDNA Synthesis Kit (#K1621) (Fermentas, Canada). RT-qPCR was performed using the Power SYBR Green PCR Master Mix on a MiniOpticon^TM Real-Time PCR System (Bio-Rad, United States). The data were analyzed using the 2^-ΔΔC_t method, with CFX Manager Software package (Bio-Rad). GAPDH was used as an internal control.

For western blot analyses, cell cultures and rat retinal tissues were lysed in RIPA buffer, which contained 1% protease inhibitors (CWBIO, Beijing, China). The extracts were incubated at 4°C for 45 min and then centrifuge at 12,000 g for 20 min at 4°C. The supernatant was collected carefully. The protein concentration was determined using a BCA assay and 10 μg of each protein sample was loaded per lane. Proteins were separated using 4–6 or 10% SDS–PAGE gel and transferred to a nitrocellulose membrane (GE Healthcare). The membranes were blocked with blocking buffer containing 5% non-fat milk in Tris-buffered saline with 0.1% Tween 20 for 1 or 3 h, at RT, before being incubated overnight at 4°C in blocking buffer containing the following primary antibodies: CAST (1:200), calpain2 (1:500), Pin1 (1:100, 3722S, Cell Signaling, Danvers, MA, United States), tAIF (1:50), PTEN (1:50, BA1377, Boster, Wuhan, China), GAPDH (1:5,000, AF0006, Beyotime), and β-Actin (1:5,000, AA128, Beyotime). The membranes were shifted to RT for 30 min. After washing three times, the membranes were incubated in blocking buffer containing HRP-conjugated anti-rabbit or anti-mouse IgG (1:2,000, Beyotime) for 2 h, at RT. After washing three times, the immunoreactive bands were visualized by low or high sensitivity chemiluminescence reagent (CWBIO). GAPDH was used as an internal reference control. Integrated density values of specific proteins were quantified using Image J software and normalized to the GAPDH or β-Actin values.

The protocol was similar with Western blot, except for the differences outlined below. The phosphatase inhibitor was added to the tissues prepared for proteins extracts. The phos-tag^TM (Wako, 192-18001-EA-S, Richmond, VA, United States) and Mn²⁺ were added to the SDS–PAGE gel at a concentration of 50 and 100 μM individually. Before transferring onto membrane, the gel was washed with transferring buffer containing 1 μM EDTA for 15 min, and then with transferring buffer without EDTA for 15 min.

The RGC-5 cells were trypsinized, followed by one wash. Cells were then re-suspended in 500 μL of binding buffer, and 5 μL of PI were added. After being incubated for 10 min at RT in the dark, the cells were washed and analyzed using FACSCalibur^TM (Becton, Dickinson Company, Franklin Lakes, NJ, United States). The percentages of cells in each quadrant were analyzed using FlowJo software (FlowJo LLC, Ashland, OR, United States). Statistical analyses of flow cytometry results were conducted by calculating the number of PI-positive cells (Yang et al., 2017).

The three dimensional structure of proteins were built based on available protein crystal diffraction structure and frozen electron microscopic structure by homologous modeling (Swiss Model) (Schwede et al., 2003). Protein surface electrostatic potential energy were calculated using APBS (figures were produced using PyMOL) (Baker et al., 2001).

Cultured retinal neurons were lysed with non-denaturing lysis buffer, protease and phosphatase inhibitor mixture. Then, the extracted protein was incubated with CAST antibody (1:50, Cell Signaling) or Pin1 (1:100, 10495-1-AP, Proteintech, Wuhan, China) coupled to protein A/G Agarose beads, overnight at 4°C. Finally, the bound proteins were analyzed using western blot.

The RM6240 system (Chengdu Instrument Factory, Chengdu, China) was used for fERG recording. After the drug treatment, the rats were dark-adapted for 6 h. Under dim red illumination, the rats were anesthetized using an intraperitoneal injection with a 1:1 mixed solution (5 mL/kg) of 10% chloralic hydras and 25% urethane. Then, a recording electrode was inserted into the anterior chamber. The reference and ground electrodes were placed on the subcutaneous layer of the forehead and tail base, respectively. A bandpass filter of 10 Hz was used and the lash luminance was 1.6 cd/s/m². Each eye was exposed to flashes, three times, at a 5-min interval. When one eye was recorded, the contralateral eye was covered. All procedures were repeated at least four times. The amplitude of b wave was calculated from the bottom of a wave to the peak of b wave.

Figure panels were assembled using Photoshop CC (Adobe Systems Incorporated, San Jose, CA, United States). The measurement data were presented as the mean ± standard deviation (SD). One-way analysis of variance and independent sample t-tests were used to analyze the data, with GraphPad Prism 5 software (GraphPad Software, Inc., San Diego, CA, United States). Statistical significance was set at p < 0.05.

We first examined whether necrosis occurred in primary retinal neurons following glutamate treatment. In these experiments, cultured retinal neurons were treated with 50 μM glutamate for 0, 2, 4, 6, 12, or 24 h, respectively. At our observation time-points, the number of PI-positive cells gradually increased from 2 h and peaked at 6 h following glutamate treatment (Figures 1A,B). We also conducted the LDH cytotoxicity assay to evaluate the percentage of necrosis cells. When compared with 0 h, the increased percentage of necrosis cells were significant at all observed time-points (Figure 1C). These results demonstrated a glutamate-dependent increase in necrosis in neuronal retinal primary cultures.

Possible interaction between Pin1 and CAST was investigated first using a computer. The CAST inhibitory domain 1 (the whole structure data of CAST were not showed) had negative electrostatic potential on most of its surface (Figure 2A). Pin1, however, displayed positive electrostatic potential, which could be properly docked to the CAST (Figure 2A). The four inhibitory domains of CAST were similar to previous reports. Therefore, all CAST inhibitory domains had the potential to bind Pin1 by electrostatic interaction. Further, we performed Co-IP, which showed that Pin1 also interacted with endogenous CAST (Figure 2B).

Next, RT-qPCR was performed to analyze the changes of Pin1 and CAST mRNA levels. As shown in Figure 2C, glutamate significantly upregulated the level of Pin1 mRNA at 0, 2, or 4 h, but downregulated the CAST mRNA level at 2, 4, 6, 12, or 24 h (Figure 2C). Western blot was used to detect the proteins expression. We first showed one complete blot of each antibody used in Supplementary Figure S1 to evaluate specific binding of antibodies to proteins. Western blot analysis further showed that Pin1 expression increased after 4, 6, 12, or 24 h of glutamate treatment (Figures 2D,G). The expression of CAST, however, decreased after 2, 4, or 6 h of glutamate treatment (Figures 2D,G). Western blot results further demonstrated that the calpain2 expression levels at 4, 6, 12, or 24 h were increased after glutamate treatment (Figures 2D,G). The tAIF (a calpains cleave target) protein levels demonstrated a similar upregulation with calpain2 at the different time-points (Figures 2D,G). In addition, the level of PTEN, which is regulated by specific proteolysis of calpain2, was used to represent the degree and duration of calpain2 activation. The increased expression of calpain2 was also associated with the activation of calpain2, as indicated by the decrease in the expression levels of PTEN (Figures 2D,G). These results showed that glutamate treatment could increase Pin1 and calpain2 activity, but decrease CAST expression in retinal neuronal cultures.

It is well-known that phosphorylation of the N-terminal of CAST L-domain reduces its interaction with calpain (Salamino et al., 1997; Liu et al., 2011). Next, using phos-tag SDS–PAGE, we assessed whether the phosphorylation of CAST changed after glutamate treatment. Interestingly, the levels of phosphorylated CAST (p-CAST) were almost same at every time-point (Figures 2E,F). However, the relative p-CAST compared with total CAST increased at 2, 4, and 6 h after glutamate treatment (Figures 2E,F). This finding suggested that the level of CAST phosphorylation was enhanced after the glutamate treatment in retinal neuronal cultures.

To examine the role of Pin1 in CAST/calpain2 pathway after glutamate treatment, we investigated the effect of juglone, which can covalently inactivate and degrade Pin1 (Baik et al., 2015). The Co-IP results (Figure 3A) showed that the interaction between Pin1 and CAST was increased in the glutamate groups compared with control groups. Additionally, the interaction between the Pin1 and CAST could be decreased through pretreatment with juglone (Figure 3A). In the RT-qPCR experiments, we chose the 4-h glutamate time-point as our primary intervention time-point, since the changes of Pin1 and CAST mRNA were significant compared with CTL group. Compared with the 4-h glutamate groups, juglone treatment did not change the mRNA levels of Pin1 (Figure 3B). Juglone, however, did decrease the protein levels of Pin1, when compared with the 6-h glutamate group. Compared with that in the CTL group, the Pin1 and CAST protein levels were significantly changed at this time-point (Figures 3C,F). This indicated that although juglone may degrade Pin1 directly, it has no effect on the transcription of Pin1 (Baik et al., 2015). Meanwhile, after Pin1 inactivation and degradation by juglone, the decreased the levels of CAST mRNA and protein induced by glutamate was almost recovered, as shown by RT-qPCR and western blot results (Figures 3B,C,F). Moreover, the juglone treatment decreased the expression of calpain2 and tAIF compared with that in the glutamate group (Figures 3C,F). In addition, the expression of PTEN was nearly recovered in the juglone-treatment group (Figures 3C,F). These results indicated that Pin1 inhibition by juglone may regulate the CAST/calpain2 pathway through the upregulation of CAST transcription and expression.

Next, we assessed the change of CAST phosphorylation after Pin1 inhibition. The level of p-CAST was unchanged after juglone application (Figures 3D,E). However, relative p-CAST compared with total CAST in the juglone group was decreased than in the glutamate group (Figures 3D,E). These results indicated that the phosphorylation effects were blocked by the inhibition of Pin1 due to the juglone application.

Double immunofluorescence was used to further evaluate the regulatory role of Pin1 on CAST/calpain2. The immunostaining results showed that CAST was predominantly located in the cytoplasm and nucleus, and slightly distributed in the dendrite (Figure 4A). The immunofluorescence intensity of CAST showed a large decrease in the cytoplasm, nucleus, and dendrites after glutamate treatment (Figure 4A). However, the decrease in CAST was almost recovered after the juglone application (Figure 4A). Calpain2 immunostaining was mainly present in the cytoplasm and nucleus. Calpain2 expression was increased within both the cytoplasm and the nucleus after the 6-h glutamate treatment (Figure 4A). Calpain2 immunostaining in the dendrites of neurons was also slightly enhanced after 6 h. The increase in the level of calpain2, however, was decreased after the juglone application (Figure 4A). A dramatic increase of tAIF immunofluorescence was also observed at 6 h after glutamate treatment. Furhtermore, the tAIF translocated into the nucleus showed by double labeling with DAPI (Figure 4B). After the juglone application, the nuclear translocation and enhanced immunofluorescence intensity of tAIF, induced by glutamate, were no longer observed (Figure 4B).

We evaluated the effect of juglone-induced Pin1 inhibition on retinal neuronal necrosis. As shown in PI staining and LDH assay, there was a reduction in necrosis in juglone-treated retinal neurons, following the 6-h glutamate treatment (Figures 5A–C). These results indicated that inhibition of Pin1 by juglone application can decrease the necrosis following glutamate treatment in retinal neurons.

After initial investigation of primary cultured retinal neurons treated with Pin1 inhibitors following glutamate treatment, we used siRNA to investigate the regulatory role of Pin1 in glutamate-induced necrosis in the RGC-5 cell line. RGC-5 cells were subjected to different concentrations (10, 20, 30, 40, or 50 mM) of glutamate for 1 h. There was a gradient increase in the number of necrotic cells with increasing concentrations of glutamate (Data not shown); the 40 mM glutamate concentration was chosen for subsequent experiments. In this section, we used glutamate concentrations in the millimolar range to induce necrosis in RGC-5, whereas a micromolar concentration range was used for primary cultures because RGC-5 requires higher glutamate concentrations to display similar features to primary neurons (Maher and Hanneken, 2005; Fatma et al., 2008).

In our present study, we investigated whether Pin1 knockdown can reduce the necrotic cells during glutamate excitotoxicity. RGC-5 cells were transfected with Pin1 siRNA for 24 h. Scrambled siRNA and transfection-reagent treated cells were compared. The western blot results showed a significant reduction in Pin1 protein levels in RGC-5 cells treated with Pin1 siRNA (Figures 6A,B). Western blot results further indicated that, cells subjected to Pin1 siRNA had an augmented decrease in CAST and PTEN levels, while the increase in calpain2 and AIF levels was reduced (Figures 6C,D). Meanwhile, RT-qPCR demonstrated that CAST was reduced after the glutamate treatment. Further, the decrease in CAST was augmented after Pin1 siRNA application (Figure 6J). The level of p-CAST, however, remained unchanged after glutamate and Pin1 siRNA application. The relative level of p-CAST compared with total CAST in the siRNA group was lower than in the glutamate treatment group (Figures 6E,F). This further indicated that CAST is further downstream of Pin1 and that the phosphorylation of CAST is modulated by Pin1.

Finally, flow cytometry (Figures 6G–I) and LDH assay (Figure 6L) were used to study cell necrosis. Statistical analysis of both flow cytometry and LDH detection (Figures 6K,L) results demonstrated that the ratio of RGC-5 cell necrosis increased when exposed to glutamate (Figure 6H), than in the control group (Figure 6G). Further, the percentage of necrotic cells was decreased after the Pin1 siRNA pretreatment (Figure 6I).

Taken together, this study provided further evidence that Pin1 plays a role in glutamate injury in retinal neurons and that Pin1 may be an upstream regulatory molecular of the CAST/calpain pathway.

First, PI were intravitreally injected to examine necrotic cells in the rat retina. Double immunofluorescence of DAPI and PI showed an increased in the PI-positive necrotic cells in the GCL and INL in the glutamate groups (Figures 7E,F) than that in the control (Figures 7A,B) and sham groups (Figures 7C,D). Then, we investigated the role of juglone in glutamate-induced necrosis in vivo. PI staining results demonstrated a significant reduction in necrotic cells with juglone pretreatment (Figures 7G,H). Figure 7M shows quantitative analysis of PI-stained cells in the GCL and INL. An in vivo LDH cytotoxicity assay was also conducted. Compared with the control and sham groups, LDH release was increased in the glutamate group (Figure 7N). However, there was a decrease in the increased LDH release in the juglone pretreatment group (Figure 7N). Taken together, these results indicated that juglone could provide a protective role to retinal cells in the GCL and INL against glutamate-induced necrosis.

We then performed fERG to evaluate visual function (Piano et al., 2016). The fERG results showed a significant decrease in the amplitudes of b-wave in the glutamate group than in the control and sham groups (Figures 7I–K). However, the amplitudes of b-wave were all enhanced in the juglone-treated group than in the glutamate group (Figure 7L). Figure 7O shows the quantitative analysis of the b-wave amplitudes. These results demonstrated that the protective effects of juglone on visual function following glutamate excitotoxicity.

Together these results suggested that Pin1 plays an important role in glutamate-induced necrosis and dysfunction in vivo.