Valproic Acid–Mediated Neuroprotection and Regeneration in Injured Retinal Ganglion Cells

Abstract

Purpose.: Valproic acid (VPA) has been demonstrated to have neuroprotective effects in neurodegenerative conditions. VPA inhibits histone-deacetylases (HDAC) and delays apoptosis in degenerating neurons. The authors investigated whether VPA delays retinal ganglion cell (RGC) death and enhances axonal regeneration after optic nerve crush (ONC). Furthermore, potential molecular targets involved in VPA-mediated protection were analyzed.

Methods.: ONC was performed on the left eye of rats, which received VPA or Ringer's solution subcutaneously (SC; 300 mg/kg twice daily) or intravitreally (single postlesional injection). Densities of fluorogold-labeled RGC were analyzed in retinal flatmounts after 5 or 8 days. Retinal tissue was also harvested and processed to quantify axon growth in retinal explants; evaluate caspase-3 activity; analyze transcription factor cAMP response element binding protein (CREB); and determine acetylated histone 3 and 4, as well as phosphorylated extracellular signal-regulated kinase (pERK) 1/2.

Results.: Five and 8 days after ONC, 93% and 58% RGC survived after subcutaneous VPA treatment in comparison to Ringer's solution (62% and 37% viable RGC), respectively (P < 0.001). Likewise, a single intravitreal injection of VPA immediately after injury significantly delayed apoptosis in RGC (P = 0.0016). Injured RGC treated with VPA showed better regeneration of their axons in culture (196 axons/explant) than the crushed controls receiving Ringer (115 axons/explant). RGC axons of the right control eyes regenerated more after VPA treatment. VPA-mediated neuroprotection and neuroregeneration were accompanied by decreased caspase-3 activity, CREB induction, pERK1/2 activation, but not by altered histone-acetylation.

Conclusions.: VPA provided neuroprotection and axonal regrowth after ONC. Alterations were observed in several pathways; however, the precise mechanism of VPA-mediated protection is not yet fully understood.

Valproic acid (VPA), a short-chain fatty acid, has recently attracted attention as a potentially neuroprotective drug. VPA is a well-established, long-term therapy for treating epilepsy and bipolar disorders ¹ ; however, the underlying molecular mechanisms are still not fully understood. It has been suggested that the VPA action is linked to both inositol depletion ^2,3 and direct inhibition of histone deacetylases (HDAC), ⁴ causing histone hyperacetylation. 

The acetylation or deacetylation of histone N-terminal tails alter the interaction between histones and DNA in chromatin, and this chromatin remodeling has been identified as a key step in gene expression regulation. ^5,6 In general, hyperacetylation is associated with transcriptional activation, whereas hypoacetylation is associated with repression. The quantity and activity of histone acetyltransferases (HAT) and HDAC are finely balanced in neurons under normal conditions. This acetylation homeostasis is greatly impaired, shifting toward deacetylation in neurodegenerative diseases. ^7,8  

As an HDAC inhibitor, VPA is reported to be neuroprotective and neuroregenerative. In in vitro experiments, VPA protected neurons from glutamate-induced excitotoxicity, ⁹ oxygen-glucose deprivation injury, ¹⁰ and oxidative stress. ¹¹ It also prolonged the lifespan of cultured cortical neurons ¹² and promoted neuronal growth. ^13,14 Furthermore, in vivo investigations demonstrated that VPA protected neurons exposed to intracerebral hemorrhage ¹⁵ or ischemic stroke. ¹⁶ This VPA-induced neuronal protection probably involves, apart from HDAC inhibition, multiple mechanisms of action, including the activation of transcription factor CREB (cAMP response element binding protein), ¹⁷ the ERK (extracellular signal-regulated kinase) pathway, ¹³ increased protein levels of heat shock protein (HSP)-70, ¹⁸ the inhibition of pro-apoptotic molecules, ¹⁶ and microglia-mediated inflammation. ¹⁹  

Several ophthalmological diseases lead to optic nerve (ON) damage and retinal ganglion cell (RGC) loss by apoptosis. There are currently no reports on the effect of VPA on RGC, although some recent studies detected a neuroprotective effect of VPA on other central nervous system (CNS) neurons in neurodegeneration. Schwechter et al. ²⁰ recently demonstrated that another HDAC inhibitor, trichostatin A (TSA), caused significant differentiation and neuritogenesis of RGC-5 cells. In this study, we examined VPA's effect on injured RGC after optic nerve crush (ONC). We then investigated whether VPA encourages RGC axons to regenerate in a retinal culture model. Furthermore, we analyzed potential molecular targets involved in VPA-mediated protection, and scrutinized its role in modifying histone-acetylation levels in RGC. 

Materials and Methods

Animals

Adult male and female Sprague–Dawley rats (180–300 g bodyweight; Charles River, Sulzfeld, Germany) were used in this study. Animals were fed with a standard rodent diet ad libitum, while kept on a 12 hour light/12 hour dark cycle. All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and all procedures were approved by the Committee of Animal Care of the University of Freiburg. All types of surgery and manipulations were performed under general anesthesia with isoflurane/O₂. Body temperature was maintained at 37 ± 0.5°C with a heating pad and a rectal thermometer probe. After surgery, analgesic (0.5 mg/kg Temgesic; Essex Pharma, Germany) was applied intraperitoneally to treat pain. While recovering from anesthesia, the animals were placed in separate cages, and gentamicin ointment (Refobacin; Merck, Darmstadt, Germany) was applied on ocular surfaces and skin wounds. 

Retrograde RGC Labeling

Deeply anesthetized rats were placed in a stereotactic apparatus (Stoelting, Kiel, Germany), and the skin overlying the skull was cut open und retracted. The lambda and bregma sutures served as landmarks for drilling six holes. Fluorogold (FG) 7.8 μL (Fluorochrome; Denver, CO), dissolved in dimethylformamide (FG 3% in NaCl with 10% DMSO) were injected into both superior colliculi as described previously. ²¹ To ensure adequate RGC labeling, animals were allowed 6 days for retrograde transport of FG before further experimental intervention. 

Optic Nerve Crush and Drug Treatment

The left ON of rats was approached via orbitotomy on day 7 after stereotactical FG injection into both superior colliculi. After partial removal of the lacrimal gland and transsection of superior rectus and obliquus muscles, the ON was exposed by blunt dissection of the retractor bulbi muscle. The nerve was separated after splitting the meninges and as a whole mechanically crushed directly behind the eyeball for 10 seconds using a fine forceps without residual aperture. Intact retinal perfusion was confirmed within 3 minutes after the crush. Using this ONC technique, all RGC axons were irreversibly damaged and neuroprotection could be measured by analyzing the delay of apoptosis. 

Rats received either VPA (300 mg/kg dissolved in sterile water; Sigma-Aldrich, St. Louis, MO) or the same quantity of Ringer's solution (B. Braun Melsungen AG, Melsungen, Germany) subcutaneously (SC) every 12 hours, starting on the day before ONC until the day of execution after 5 or 8 days. The corresponding right eyes served as controls to detect potential side effects of VPA during SC therapy. To evaluate VPA's effect in a postlesional setting, rats received a single intravitreal (IVT) injection (5 μL) of VPA or Ringer's solution immediately after ONC. To reach an IVT VPA concentration of 50 mM (assuming an average vitreous volume of 40 μL in 60-day-old rats ²² ), 5 μL of VPA 450 mM were administered. RGC were quantified 5 days after the crush. Animals with lens injury were excluded from examination, as it prevents RGC death and promotes axonal regeneration. ²³  

RGC Quantification

Animals were killed by CO₂ inhalation 5 or 8 days after ONC with or without VPA treatment (n = 8 on each day). Retinal tissue was immediately harvested and further processed for wholemount preparation in ice-cold Hank's balanced salt solution. Retinas were carefully placed on a nitrocellulose membrane with the ganglion cell layer on top. After removing the vitreous body, retinas were fixed in 4% paraformaldehyde for 1 hour and then embedded in mounting media (Vectashield; AXXORA, Loerrach, Germany). The densities of FG-positive RGC were determined in blinded fashion using a fluorescence microscope (AxioImager; Carl Zeiss, Jena, Germany) and the appropriate bandpass emission filter (FG: excitation/emission, 331/418 nm), as previously described. ²⁴ Briefly, we photographed three standard rectangular areas (the size of this area was 0.200 mm × 0.200 mm = 0.04 mm²) at 1, 2, and 3 mm from the optic disc in the central regions of each retinal quadrant. Hence, we evaluated an area of 0.48 mm²/retina (12 × 0.04 mm²), which represents approximately 1% of the rat retina, assuming an average area per retina of approximately 50 mm² in rats. ²⁵ To receive cells/mm², we multiplied the number of analyzed cells/0.04 mm² by 25. Secondary FG stained activated microglia cells after RGC phagocytosis were separated by morphologic criteria and excluded from examination. All averaged data are presented as mean RGC densities (cells/mm²) ± SEM. 

Retinal Explants

The intrinsic ability of RGC to regrow axons after injuries has been previously demonstrated after ONC or axotomy. ^26,27 This regenerative ability was attributed to injury-induced upregulation of growth-associated proteins, ²⁸ molecules assembled to form growth cones and axons, ²⁹ and transcriptional activation. ^30,31 To evaluate whether VPA has an effect on the intrinsic ability of RGC to regrow axons after injury, retinal organ cultures were applied. Retinal tissue of SC treated rats was harvested 5 days after ONC and further processed for wholemount preparation (n = 6–9). Eight homogenous explants/retina (diameter 1.7 mm) were trepanned and cultured in serum-free S4 media ²⁶ (Astrocyte Microglia Growth Medium; Promocell, Heidelberg, Germany) on a laminin-1 coated substrate for 6 days. The numbers of regrowing axons from retinal explants were counted along the explant margin after 2, 4, and 6 days in vitro under an inverted phase-contrast microscope (Axiovert 135; Carl Zeiss) as described previously. ^26,32 Results are given as mean RGC axons/explant ± SEM. Immunohistochemistry was used following standard protocols to confirm the presence of RGC axons (anti-β-3-tubulin, red; Promega, Madison, WI) growing out of retinal explants after 6 days and to disclose migrating Müller cells (anti-glial fibrillary acidic protein [GFAP], green; LabVision, Fremont, CA). 

Fluorogenic Caspase-3 Activity Assay

To investigate possible molecular, anti-apoptotic mechanisms of VPA, we analyzed the activity of the effector caspase-3 in retinal tissue after ONC, with or without VPA treatment (n = 5). The untreated right eyes served as control. Total protein extracts (10 μg) of retinal cells were mixed with 90 μL assay buffer (100 mM HEPES [pH 7.5], and 2 mM dithiothreitol). The respective fluorogenic substrate for caspase-3, Ac-DEVD-AMC (1 μL; 60 μM; Alexis Corp.) was added and fluorescence measured at 30°C for 30 minutes in a microplate reader (Microplate Spectra Max Gemini XS; Molecular Devices, Sunnyvale, CA) at 380/460 nm. Results are given in arbitrary fluorescent units (RFUs) ± SEM. 

Electrophoretic Mobility Shift Assay (EMSA)

EMSAs were performed with a [γ-³²P]-dATP-labeled CREB oligonucleotide to determine whether ONC and/or VPA influence the DNA binding of CREB (n = 3). The kinase reaction consisted of 37 μL of purified water, 1 μL of CREB oligonucleotide (25 ng/μL; consensus sequence: 5′-AGA GAT TGC CTG ACG TCA GAG ACG TAG-3′; Promega, Madison, WI), 5 μL of γ-³²P deoxyadenosine triphosphate (Amersham International, Braunschweig, Germany), and 5 μL of kinase buffer and 1.5 μL of T4 kinase (New England Biolabs, Schwalbach, Germany) and was incubated for 30 minutes at 37°C. Equal amounts of protein were added to the specific antibodies for supershift reactions (i.e., sc-240x [CREB-1], sc-270x [ATF-1], sc-242x [ATF-2], sc-1694x [c-Jun], sc-372x [p65], and sc-1984x [PPAR-γ]; Santa Cruz Biotechnologies, Santa Cruz, CA) and incubated for 15 minutes. Afterward, the 20 μL EMSA reaction mixture containing 20 μg of bovine serum albumin, 2 μg of poly(dI-dC) (Roche, Mannheim, Germany), 2 μL of buffer D⁺ (20 mM HEPES [pH 7.9], 20% glycerol, 100 mM KCl, 0.5 mM EDTA, 0.25% Nonidet P-40, 2 mM DTT, and 0.1% PMSF), 4 μL of 5x Ficoll buffer (20% Ficoll 400, 100 mM HEPES, 300 mM KCl, 10 mM DTT, and 0.1% PMSF), 4 μL of double-distilled (dd) H₂O, and 1 μL of CREB ³²P-labeled oligonucleotide was added. These samples were incubated at room temperature for 20 minutes and then loaded on a 30% acrylamide gel. After the gel run, it was vacuum-dried (Gel Dryer 543; Bio-Rad, Hercules, CA) for 40 minutes on a 3 MM chromatography filter (Whatman, Maidstone, England) and exposed to radiograph film (Kodak Biomax MR, Stuttgart, Germany). EMSA autoradiographs were evaluated by volume quantification and local median of DNA binding and normalization against background using two-dimensional scanning (Personal Densitometer; Amersham). The results are given in relative densitometric units (mean ± SEM). 

Histone Extraction

Total retinal tissue was harvested with or without SC VPA treatment 5 days after ONC. After homogenization with an ultrasound douncer, histone extracts (n = 3) were isolated out of nuclear extracts by using the histone extraction kit as described by the manufacturer (Epiquik Histone (H)3 or H4 Assay kit; Epigentek, Brooklyn, NY). The protein content of histone extracts was determined with a Bradford-Assay system (Bio-Rad Laboratories, Munich, Germany). Histone extracts were used for Western blot analysis. 

Western Blot Analysis

We also extracted full cell lysates from retinal tissue 24 hours and 48 hours after ONC, with and without IVT VPA treatment. Eyes were enucleated, and retinas were dissected and immediately stored on ice in a tris-based lysis buffer containing a protease inhibitor cocktail (Complete tablets; Roche). After homogenization with an ultrasound douncer, the tissue was kept on ice for 30 minutes and protein concentration was determined using the Bradford Assay (Bio-Rad Laboratories). These protein samples and the histone extracts (see above) were used for proteomic analysis of acetylated histone proteins or MAP kinases. The proteins (30 μg) were separated with 12% sodium dodecyl sulfate (SDS)-acrylamid gel electrophoresis according to standard protocols. Proteins were transferred to a PVDF membrane (0.45 μm thick). After blocking nonspecific binding with 5% skim milk in tris-buffered saline-tween 20, the membrane was probed successively with primary antibodies to acetylated (Ac)H3 and AcH4, phosphorylated (p)ERK1/2 and total (t)ERK (all Cell Signaling, Danvers, MA), after which secondary antibodies labeled with horseradish peroxidase were added. Specific antigen and antibody binding was visualized with a chemiluminescence system (Amersham, UK) according to the method described by the manufacturer. 

Statistical Analysis

All averaged data are presented as means with their corresponding SEM. Statistical significance was assessed using unpaired t-test or ANOVA, followed by Tukey-Kramer post hoc testing for multiple comparison procedures (GraphPad InStat and GraphPad Prism software). Differences were considered significant at P < 0.05. 

Results

VPA Protected RGC from Death after ONC

RGC densities were compared between groups to determine whether VPA is neuroprotective after ONC. Subcutaneous administration of VPA twice daily significantly delayed cell death in crushed RGC. As in control retinas, almost all RGC stained FG positive 5 days after ONC in the VPA-treated group. Many crushed RGC receiving Ringer's solution died, and activated microglia cells (denoted with small arrows in Fig. 1A) stained FG-positive after RGC phagocytosis. 

Figure 1.

 

VPA exerts neuroprotective effects on RGC after ONC. (A) Representative pictures of retrograde-labeled RGC. As in control retinas, almost all RGC stained FG positive 5 days after ONC in the VPA-treated group. Many crushed RGC receiving Ringer's solution died, and activated microglia cells (arrows) stained FG-positive after their phagocytosis. Scale bar in picture 3 for all photographs: 100 μm. (B) Significantly more RGC were present after SC VPA treatment 5 and 8 days after ONC. ***P < 0.001. (C) A single postlesional injection of VPA intravitreally significantly protected RGC from death 5 days after ONC. **P = 0.0016. Data are presented as mean and SEM.

Five and 8 days after ONC, 1852 ± 57 and 1170 ± 102 RGC/mm² were counted after VPA treatment compared to Ringer's solution (1249 ± 63 and 738 ± 47), respectively (P < 0.001 on each day of analysis). In the control eyes, 2001 ± 49 and 2002 ± 45 RGC/mm² were counted after Ringer's or VPA treatment, respectively (Fig. 1B). Moreover, a single intravitreal injection of VPA immediately after injury significantly protected RGC from death 5 days after ONC compared to the Ringer-treated controls (1938 ± 85 vs. 1505 ± 65, respectively, P = 0.0016) (Fig. 1C). 

VPA Stimulated Adult RGC to Regenerate

We examined retinal cultures to evaluate whether VPA has an effect on the intrinsic ability of RGC to regrow axons after injury. Five days after ONC, VPA-treated RGC revealed a remarkable potential to regrow their axons on a laminin-1 coated dish (27 ± 11, 108 ± 27, and 196 ± 45 axons/explant) in comparison to the Ringer's-treated group (14 ± 4, 59 ± 13, and 115 ± 23 axons/explant); 2, 4, and 6 days in culture, respectively. In control retinas as well, VPA stimulated axonal growth (2 ± 0.4, 26 ± 4, and 71 ± 16 axons/explant) when compared to the Ringer's treatment (1 ± 0.4, 10 ± 4, and 27 ± 9 axons/explant) (Figs. 2A, 2B). The data were statistically analyzed using a three-factor ANOVA [multiplicative model, factor one: day of analysis, factor two: surgery (ONC), factor three: drug treatment (VPA)]. As expected, increasing numbers of axons grew out of the retinal explant over time (P < 0.001). As reported previously, ONC induces axonal regeneration in injured RGC in vitro (P < 0.001). ^26,27 Interestingly, VPA treatment in vivo enabled both crushed and control RGC to regenerate their axons in culture (P < 0.001) independent of ONC's induced sprouting response (ONC:VPA, P = 0.163). Multiple RGC axons stained β-3-tubulin–positive (red) in contrast to GFAP positive glial cell processes (green) in the VPA treated group (Fig. 2C). Fewer axons were present in the Ringer's-treated group (Fig. 2D). 

Figure 2.

 

VPA stimulated RGC to regenerate their axons. (A) A representative explant with numerous regenerating axons of a VPA-treated retina after ONC, 6 days in culture. (B) VPA (SC) exerts neuroregenerative effects on RGC axons in culture with or without ONC in comparison to the Ringer's-treated group. (C, D) VPA-treated RGC showed an enormous sprouting response after ONC (C), while Ringer's-treated RGC regenerated fewer axons after the crush (D). RGC axons stained β-3-tubulin positive (red), glial cells were GFAP-positive (green), cell nuclei were stained with DAPI (blue). Scale bar (C, D): 100 μm. Data are presented as mean and SEM.

Caspase-3 Activity Is Reduced in Retinal Tissue after VPA Treatment

To confirm the neuroprotective action of VPA and investigate whether VPA reduces ONC's apoptotic effects, we performed a caspase-3 activity assay with protein lysates of tissue samples using a specific fluorogenic caspase-3 substrate (7-amino-4-methylcoumarin, DEVD-AMC). Results are shown in Figure 3 in arbitrary fluorescent units (RFUs). In ONC+Ringer's-treated RGC, caspase-3 activity was significantly higher than in the contralateral controls (8241 ± 591 vs. 527 ± 39 RFU, respectively, P < 0.001). Animals exposed to ONC+VPA showed significantly less caspase-3 activity (4578 ± 414 RFU) than the ONC+Ringer's-treated animals (P < 0.001). Caspase-3 activity in VPA-treated controls did not differ significantly from Ringer's-controls (control+VPA = 631 ± 18 RFU) (Fig. 3). 

Figure 3.

 

Caspase-3 activity is reduced in retinal tissue after VPA treatment. DEVDase assay. Caspase-3 activity is significantly reduced in VPA-treated retinas in comparison to Ringer's-treated cells after ONC. ***P < 0.001. The results are given in arbitrary fluorescent units (RFUs). Data are presented as mean and SEM.

VPA Induces the DNA Binding of Transcription Factor CREB in RGC with and without ONC

To determine whether VPA and/or ONC have an influence on the DNA binding of CREB, EMSAs were performed. Figure 4A shows the densitometric analysis of EMSA lanes 2 to 5 (triplicates); Figure 4B shows a representative autoradiograph of EMSA. 

Figure 4.

 

VPA induces transcription factor CREB. (A) Densitometry of CREB-EMSA (lanes 2–5). The ONC procedure alone slightly induced the DNA binding of CREB (not statistically significant, P > 0.05), whereas VPA increased this effect independently from ONC, leading to a greater DNA binding of CREB. ***P < 0.001; *P < 0.05. The results are given in relative densitometric units (mean ± SEM). (B) EMSA (lanes 1–5) was performed with total protein lysates of rat retinas. Positive control of CREB was achieved in cell culture stimulated by Forskolin (2 hours, 30 μM). Supershift EMSA (lanes 6–11), indicating that among the contributing family members tested, CREB-1 and ATF-1 were part of the CREB complex (lanes 6 and 7), while ATF-2, c-Jun, p65, and PPAR-γ do not supershift. The supershift in lane 6 was characterized by the loss of the DNA-protein complex (lane 6 vs. lane 5) due to an interference of anti-CREB-1 with the protein binding spot on the labeled oligo, thus the complex could not be formed and vanished. The supershift in lane 7 (anti–ATF-1) could be seen as the loss of the original DNA-protein complex and the appearance at another spot earlier in the gel because of the higher weight and gain of molecular mass due to the binding of the antibody. To achieve evidence for the sensitivity of the binding oligo, a self competition with addition of unlabeled (nonradioactive) CREB oligo was performed. As shown in lane 12, the DNA-protein complex of CREB vanished, indicating that the used oligo really was CREB. Another control to provide sensitivity in this context could be seen by the addition of an oligo sequence (in our case HIF-1 alpha) which was not complementary to the protein of interest. Thus (as shown in lane 13) the DNA-protein complex of CREB was not affected. Anti, antibody; comp, competition.

In the Ringer's group, the DNA binding of CREB slightly increases after ONC compared to the control eye of the same animal (Figs. 4A and B, lane 2 vs. lane 3, P > 0.05). Pretreatment with VPA potentiated this effect independent from ONC leading to a higher DNA binding of CREB in the VPA group compared to the Ringer's treated group (Figs. 4A, 4B, lane 2 vs. lane 4, P < 0.05; lane 3 vs. lane 5, P < 0.001). We carried out supershift experiments by adding antibodies against CREB-1, ATF-1, ATF-2, c-Jun, p65, and PPAR-γ (lanes 6–11) to the protein extracts before EMSA to assess the specificity of the CREB complex and identify some of the contributing family members. Protein-antibody recognition is visualized by a decrease in the mobility of the DNA-protein complex or diminution of the CREB complex. Figure 4B indicates that, among the contributing family members tested, CREB-1 and ATF-1 were part of the CREB complex (lanes 6 and 7), while ATF-2, c-Jun, p65, and PPAR-γ do not supershift. 

VPA Does Not Raise Protein Levels of Acetylated H3+H4 Significantly

HDAC-inhibitor VPA has been shown to enhance the acetylation of histones. Thus we examined the acetylation levels of H3 and H4 after ONC in total retinal tissue and histone extracts with and without VPA treatment using Western blot technique. Five days after ONC and SC VPA treatment, the amount of AcH3 + AcH4 did not differ significantly from the Ringer's-treated group in histone extracts (Figs. 5A, 5B; nuclear extracts were corrected against loading control GAPDH). To exclude that earlier effects on acetylation had already elapsed on day 5 or had altered during histone extraction, we performed further analyses using full retinal protein extracts 24 hours and 48 hours after ONC and IVT administration of VPA or Ringer's (Figs. 5C, 5D). 

Figure 5.

 

VPA does not significantly increase protein levels of acetylated H3+H4. (A, B) Western blot analysis and densitometries of histone extracts 5 days after ONC. Five days after ONC and VPA treatment (SC), the amount of AcH3 (A) and AcH4 (B) did not differ significantly from the Ringer's-treated group in histone extracts. Densitometry data are presented as mean and SEM in relative densitometric units, nuclear extracts were corrected against GAPDH (loading control). (C, D) Western blot analysis of whole-cell lysates 24 hours and 48 hours after ONC. (C) Protein levels of AcH3 do not differ significantly in retinal tissue 24 hours or 48 hours after crush with or without VPA treatment (lane 2 vs. lane 4). At 48 hours after ONC, slightly more AcH3 was detected in both groups in comparison to their controls. (D) AcH4 levels did not alter 24 hours postlesional. At 48 hours after ONC, AcH4 seemed to be slightly upregulated in retinal tissue after Ringer's treatment (lane 2 vs. lane 1), whereas again, no significant differences were found in comparison to the ONC+VPA group (lane 2 vs. lane 4). All blots using whole-cell lysates were analyzed by densitometry; differences were not statistically significant (densitometry data not shown). β-actin served as loading control. AcH: acetylated histone.

As shown in Figure 5C, protein levels of AcH3 again did not differ significantly in retinal tissue 24 hours or 48 hours after crush with or without VPA treatment (lane 2 vs. lane 4). 48 hours after ONC, slightly more AcH3 was detected in both groups in comparison to their controls (Fig. 5C). AcH4 levels did not alter 24 hours postlesional (Fig. 5D); 48 hours after ONC it seemed to be slightly upregulated in retinal tissue after Ringer's treatment (lane 2 vs. lane 1), whereas again no significant differences were found in comparison to the ONC+VPA group (lane 2 vs. lane 4). All blots using whole-cell lysates were analyzed by densitometry; differences were not statistically significant (densitometry data not shown). 

VPA Prolonged the Activation of pERK1/2 in Injured RGC

In comparison to controls, the protein level of pERK1/2 was upregulated 24 hours after ONC with or without VPA treatment. There was no statistically significant difference between lane 2 and lane 4 (P > 0.05; Fig. 6A). The activation of pERK1/2 remained stable over 48 hours after ONC in the VPA-treated group, whereas protein levels fell rapidly in the Ringer's-treated group (P < 0.001, Fig. 6B). 

Figure 6.

 

VPA prolonged the activation of pERK1/2 in injured RGC. Representative western blots and corresponding densitometries of pERK1/2 and tERK in whole retinal tissue 24 hours (A) and 48 hours (B) after ONC and intravitreal drug treatment. (A) In comparison to controls, the protein level of pERK1/2 was upregulated 24 hours after ONC with or without VPA treatment. No statistically significant difference between lane 2 and lane 4; P > 0.05. (B) In the VPA-treated group, the activation of pERK1/2 remained stable more than 48 hours after ONC, whereas protein levels dropped rapidly in the Ringer's-treated group. ***P < 0.001. Data are presented as mean and SEM in relative densitometric units.

Discussion

This set of experiments provides evidence that VPA exerts a protective effect in acute experimental optic nerve injury. Both systemic and intravitreal VPA-administration enhanced RGC survival and regeneration by suppression of pro-apoptotic caspase-3 activity, induction of the DNA-binding of CREB, and pERK1/2 upregulation. However, modified histone-acetylation levels did not change in this investigation. 

Optic nerve injury triggers RGC death, ^33,34 which is primarily apoptotic. ³⁵ Numerous efforts have been made to decelerate the RGC loss triggered by optic nerve damage. So far, none of them have successfully delayed RGC death beyond 15 days postlesional. ^36–38 In the ONC model (total open ONC) used in this investigation, all RGC axons are irreversibly damaged and a neuroprotective effect can only express itself in a delay of apoptosis. Ringer's-treated cells died rapidly 5 or 8 days after ONC (62% or 37% viable RGC, respectively, P < 0.001), whereas VPA protected 93% or 58% of the damaged RGC 5 or 8 days after crush, respectively. VPA treatment (SC or IVT) delayed RGC death after acute ONC. 

This concurs with previous studies showing that VPA is an efficacious neuroprotective drug in other neurodegenerative diseases of the CNS. ^15,16 This protection was accompanied by the VPA-mediated suppression of pro-apoptotic molecules (caspase-3, -8, -9, and Bax) and the induction of anti-apoptotic factors (Bcl-2 and Bcl-XL). ^15,39 Caspase-3 is a key mediator of mammalian-cell apoptosis, and was upregulated in RGC after ONC. ³⁵ In this study, cell death induced by ONC correlated with caspase-3 activation, which was attenuated by VPA treatment, thus confirming its antiapoptotic potential. 

The neuroprotective action of VPA after ONC also involve other mechanisms of action at the molecular level. As a HDAC inhibitor, VPA may modulate the expression of downstream target genes by regulating the activities of hyperacetylated transcription factors. As alteration of gene expression occurs after ONC, ^31,35,40 the modification of transcription may represent a therapeutic option. The VPA dosage used was determined by previous studies ^5,15 based on the achievement of marked HDAC inhibition. We observed no statistically significant changes in the acetylation levels of H3+H4, although it has been shown that VPA increases histone acetylation in nervous tissue. ^9,11,12 Perhaps a methodical problem caused this result, as we were only able to use full retinal protein extracts for molecular investigations due to low protein content or tissue masses, although only the RGC were damaged via ONC (“signal-to-noise-ratio”). Another explanation is that HDAC inhibitors not only directly regulate gene expression via promoter hyperacetylation, they also selectively modulate transcriptional factors via hyperacetylation independently of histones. ^41,42 Therefore, further investigations are necessary to characterize an HDAC-inhibitory effect of VPA in RGC. 

CREB is a transcription factor known to mediate stimulus-dependent expression of genes critical to the plasticity, growth, and survival of neurons. ⁴³ CREB decreased with time after injury in RGC. ^30,44 In this investigation, the DNA-binding of CREB rose slightly after ONC (no statistical significance), whereas VPA treatment strengthened this effect independently from ONC, leading to twofold higher DNA binding of CREB, as shown via EMSA. Moreover, VPA treatment in vivo enabled both damaged and healthy adult RGC to regenerate their axons in vitro independent of ONC's induced sprouting response of injured RGC. In control eyes, SC VPA treatment led to a more than twofold higher axonal regeneration in comparison to Ringer's solution. Several studies have shown that the CREB transcriptional pathway regulates the expression of both bcl-2 and brain-derived neurotrophic factor (BDNF). ^45,46 Thus the neuroprotective and neuroregenerative effects of VPA on RGC may be mediated at least in part by the CREB-induced regulation of such neurotrophic factors. Additionally, histone acetyltransferases (HAT) like p300 and/or CREB-element binding protein (CBP) coactivate CREB. VPA can directly upregulate CBP at the transcriptional level and reverse degeneration-associated histone deacetylation in neurons. ¹¹  

It has to be discussed if the increase in axon outgrowth in the VPA-treated explants was a VPA-mediated action or just a result of healthier cells. An adult untreated rat retina does not noticeable regrow RGC axons in culture (in Fig. 2B the Control Ringer's group). This is in part due to a diminished intrinsic regenerative capacity of mature CNS neurons. We know from the fluorogold data and the caspase-3 data that the right control retinas of both Ringer's- and VPA-treated rats had similar RGC densities and the same low quantity of caspase activity. We thus assume that these cells have the same vitality. The finding that even the control retinal explants of VPA treated rats regenerated more axons in culture might be indicative of a direct action of VPA on neurite outgrowth. In damaged RGC, different interpretations are possible. After ON damage, RGC revealed a remarkable potential to regrow their axons. ³² Accordingly, stress exposed RGC regenerate their axons and two explanations are conceivable: 1) there is more over all regeneration, because RGC are healthier leading to a higher cell density in the VPA group; and 2) there is specific activation of regeneration. We favor the latter possibility for two reasons: 1) VPA was able to stimulate regeneration in isolated retina of previously healthy control eyes, and 2) the state of “health” is inversely correlated with the ability to regenerate as mentioned above. 

HDAC inhibitors have also been shown to activate the ERK pathway, ^13,47 thereby enhancing axonal regeneration and neuronal survival. We have shown that VPA maintained the activation of pERK1/2 over 48 hours after ONC, whereas protein levels decreased rapidly in the Ringer's-treated group. The role of the prolonged upregulation of pERK1/2 in VPA-treated RGC with regard to neuroprotection and neuroregeneration remains to be investigated in future studies. 

VPA is a well established substance for the long-term treatment of epilepsy and is generally well tolerated. The most common side effects are somnolence and fatigue, which were apparent to a lesser extent in the rats during sc therapy in this investigation. Changes in visual function have recently been studied extensively in animals and in epilepsy patients treated with different antiepileptic drugs. In the literature, VPA's effect on retinal function is controversially discussed. Goto et al. ⁴⁸ showed that long-term VPA treatment can suppress visual evoked potentials and electroretinogram waves in rats. Other reports stated that VPA can cause significantly impaired color perception in patients. ^49,50 One study reports a concentric visual field defect in a 22-year-old patient. ⁵¹ In contrast, Sorri et al. ⁵² found that VPA therapy was not associated with visual field defects after 8.4 ± 5.1 years of treatment. Furthermore, morphologic retinal alterations (modifications of retinal nerve fiber layer or macular thickness) have not been reported in patients after 1 year of VPA treatment. ⁵³ In accordance with our results, VPA treatment per se had no effect on RGC number even after 90 days of VPA treatment. ⁵⁴ If VPA will be considered in future for the treatment of a chronic retinal neurodegenerative disease such as glaucoma, possible effects of VPA on retinal function have to be monitored carefully. 

In summary, neuroprotective effects of VPA have been previously shown in the CNS; however, not in the retina or after ONC. We have demonstrated for the first time that VPA administered at a clinically-relevant dosage significantly delayed cell death in injured rat RGC and stimulated RGC axons to regrow in vitro. Alterations were observed in several pathways thought to be critical in VPA activity; however, the precise mechanism or pathway of VPA-mediated protection is yet unclear, as previously described VPA-mediated hyperacetylation via HDAC inhibition was not reproduced in this investigation. Further experiments needs to investigate which proteins and transcription factors are critically involved in VPA-mediated neuroprotection. Furthermore, studies in chronic optic nerve injury models (e.g., experimental glaucoma) are needed to evaluate VPA's effect at longer time points in vivo to strengthen the significance of the findings. 

Footnotes

 Supported by the Forschungskommission des Universitätsklinikums Freiburg, the Deutsche Ophthalmologische Gesellschaft, and the Verein Rheinisch-Westfälischer Augenärzte (JB); and the Deutsche Forschungsgemeinschaft (DFG; Grant Th 386-18-1) (ST).

Footnotes

 Disclosure: J. Biermann, None; P. Grieshaber, None; U. Goebel, None; G. Martin, None; S. Thanos, None; S. Di Giovanni, None; W.A. Lagrèze, None

The authors thank Sylvia Zeitler (University Eye Hospital Freiburg) and Mechthild Langkamp Flock (University Eye Hospital Muenster) for excellent technical assistance, and Michael Bach (University Eye Hospital Freiburg) for help with data analysis.