Heat shock protein 70 induction by valproic acid delays photoreceptor cell death by N-methyl-N-nitrosourea in mice

Authors

  • Yoshiki Koriyama,

    Corresponding author
    1. Department of Molecular Neurobiology, Graduate School of Medicine, Kanazawa University, Kanazawa, Japan
    2. Graduate School and Faculty of Pharmaceutical Sciences, Suzuka University of Medical Science, Suzuka, Japan
    Current affiliation:
    1. Graduate School and Faculty of Pharmaceutical Sciences, Suzuka University of Medical Science, 3500-3 Minamitamagaki, Suzuka 513-8670, Japan
    • Address correspondence and reprint requests to Yoshiki Koriyama, Department of Molecular Neurobiology, Graduate School of Medicine, Kanazawa University, 13-1 Takaramachi, Kanazawa 920-8640, Japan. E-mail: koriyama@suzuka-u.ac.jp

    Search for more papers by this author
  • Kayo Sugitani,

    1. Division of Health Sciences, Graduate School of Medical Science, Kanazawa University, Kanazawa, Japan
    Search for more papers by this author
  • Kazuhiro Ogai,

    1. Department of Molecular Neurobiology, Graduate School of Medicine, Kanazawa University, Kanazawa, Japan
    2. Wellness Promotion Science Center, Institute of Medical, Pharmaceutical and Health Sciences, Kanazawa University, Kanazawa, Japan
    Search for more papers by this author
  • Satoru Kato

    1. Department of Molecular Neurobiology, Graduate School of Medicine, Kanazawa University, Kanazawa, Japan
    Search for more papers by this author

Abstract

Retinal degenerative diseases (RDs) are a group of inherited diseases characterized by the loss of photoreceptor cells. Selective photoreceptor loss can be induced in mice by an intraperitoneal injection of N-methyl-N-nitrosourea (MNU) and, because of its selectivity, this model is widely used to study the mechanism of RDs. Although it is known that calcium-calpain activation and lipid peroxidation are involved in the initiation of cell death, the precise mechanisms of this process remain unknown. Heat shock protein 70 (HSP70) has been shown to function as a chaperone molecule to protect cells against environmental and physiological stresses. In this study, we investigated the role of HSP70 on photoreceptor cell death in mice. HSP70 induction by valproic acid, a histone deacetylase inhibitor, attenuated the photoreceptor cell death by MNU through inhibition of apoptotic caspase signals. Furthermore, HSP70 itself was rapidly and calpain-dependently cleaved after MNU treatment. Therefore, HSP70 induction by valproic acid was dually effective against MNU-induced photoreceptor cell loss as a result of its anti-apoptotic actions and its ability to prevent HSP70 degradation. These findings might help lead us to a better understanding of the pathogenic mechanism of RDs.

image

Retinal degenerative diseases are characterized by the loss of photoreceptor cells. We proposed the following cascade for N-methyl-N-nitrosourea (MNU)-induced photoreceptor cell death: MNU gives rise to cleavage of heat shock protein 70 (HSP70); HSP70 induction by valproic acid (VPA) is dually effective against MNU-induced photoreceptor cell loss because of its anti-apoptotic actions and its ability to prevent HSP70 degradation. We hope that the present study heralds a new era in developing therapeutic tools against retinal degenerative diseases.

Abbreviations used
4HNE

4-hydroxy-2-nonenal

AcH3

acetylation of histone H3

ALLN

N-Acetyl-Leu-Leu-norleucinal

DAPI

4′-6-diamidino-2-phenylindole

GCL

ganglion cell layer

HSP70

heat shock protein 70

INL

inner nuclear layer

IPL

inner plexiform layer

LY

LY294002

MNU

N-methyl-N-nitrosourea

ONL

outer nuclear layer

OPL

outer plexiform layer

PI3K

phosphoinositide 3-kinase

PI

propidium iodide

VPA

valproic acid

Retinal degenerative diseases such as retinitis pigmentosa and age-related macular degeneration are major causes of blindness (Margalit and Sadda 2003; Hartong et al. 2006; Yang et al. 2007). Although it is estimated that at least 50 million people have these diseases (Yang et al. 2007), no effective drugs have been discovered. It is difficult to choose an appropriate genetic model for these diseases because there are many causative genes (Pennesi et al. 2012; Rossmiller et al. 2012); for example, more than 30 genes and more than 100 mutations in rhodopsin have been reported to be involved in retinitis pigmentosa. However, the various pathogenetic mechanisms of these retinal degenerative diseases have a common end stage: photoreceptor cell death. One model uses N-methyl-N-nitrosourea (MNU), an alkylating agent, because of its selective photoreceptor cell death through an apoptotic mechanism (Yoshizawa et al. 2000); this model has been studied extensively in rodents (Smith et al. 1988; Kiuchi et al. 2003). MNU-induced photoreceptor cell death is accompanied by up-regulation of Bax and down-regulation of Bcl-2, leading to activation of the caspase cascade (Yang et al. 2007). MNU induces accumulation of intracellular calcium ions in the retina and induces calpain-dependent photoreceptor cell loss after intraperitoneal MNU injection (Oka et al. 2007). It has been also reported that calpain activation promotes photoreceptor cell death via caspase-3 activation (Mizukoshi et al. 2010; Tsubura et al. 2010). However, the mechanism of MNU-induced photoreceptor cell death is not fully understood.

Heat shock proteins (HSPs) are a family of stress-activated proteins that participate in protein folding and repair (Snoeckx et al. 2001). HSP70 plays a key role in protecting cells against various types of environmental stresses. A few reports show the effect of HSP70 on photoreceptor cell death (Kayama et al. 2011). For example, hyperthermic preconditioning induces HSPs that contribute to retinal protection against excitatory amino acid-induced neurotoxicity (Kwong et al. 2003) and light-induced photoreceptor degeneration (Barbe et al. 1988). A recent report has also shown that inducible HSP70 is a critical mediator of the phosphoinositide 3-kinase (PI3K)/Akt pathway, which becomes a key molecule for cell survival and the inhibition of apoptosis (Kayama et al. 2011) through inactivation of caspase-3 (Rashmi et al. 2004). However, the role of HSP70 in photoreceptor cell death by MNU has not yet been reported. Thus, we investigated the protective effect of HSP70, and its mechanism, using a well-known HSP70 inducer, valproic acid (VPA, Ren et al. 2004; Marinova et al. 2009). VPA is a short-chain fatty acid that has been used to treat seizures and bipolar mood disorder (Tunnicliff 1999; Phiel et al. 2001). VPA inhibits histone deacetylases, resulting in histone hyperacetylation (Zhang et al. 2012), and loosens the interaction of histone with DNA. This leads to relaxation of the nucleosome structure which facilitates transcription factor binding to DNA elements (Berger 1999). It has been reported that VPA induces HSP70 through histone H3 acetylation in neural cells (Xuan et al. 2012), including retinal neurons (Marinova et al. 2009; Zhang et al. 2012). Moreover, histone deacetylase activity is causally linked to inherited photoreceptor cell death in the rd1 mouse, one of the most studied animal models of human homologous retinitis pigmentosa (Sancho-Pelluz et al. 2010). Therefore, VPA could act as a histone deacetylase inhibitor to increase photoreceptor cell survival in retinitis pigmentosa. In this study, we investigated the effect of VPA-induced HSP70 on MNU-induced photoreceptor cell death in mice.

Materials and methods

Materials

MNU was purchased from Toronto Research Chemicals Inc (North York, Canada). VPA and calpain inhibitor, N-acetyl-Leu-Leu-norleucinal (ALLN), were purchased from Sigma-Aldrich (St. Louis, MO, USA). ALLN is a specific calpain inhibitor that interacts with the calcium-binding site of calpain (Debiasi et al. 1999). The PI3K inhibitor LY294002 (LY, Stein 2001), a competitive inhibitor of ATP binding in the PI3K kinase domain, was purchased from Sigma-Aldrich. Heat shock protein inhibitor I (HSP inhibitor), which was used to specifically suppress HSP70 gene expression by inhibiting heat shock transcription factor-heat shock element enhancer binding activity (Manwell and Heikkila 2007; Nagashima et al. 2011) was obtained from Calbiochem (Darmstadt, Germany).

Animals

All animals were maintained and handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, the guidelines of the Declaration of Helsinki and the Guiding Principles in the Care and Use of Animals and Kanazawa University's guidelines for animal experiments. Male C57BL/6 mice (8–9 weeks old; Japan SLC, Inc., Shizuoka, Japan ) were reared in clear plastic cages and kept under a 12 h light–dark cycle at 23°C. Mice were anesthetized by intraperitoneal injection of sodium pentobarbital (30–40 mg/kg body weight). VPA was intraocularly injected 1 day before MNU treatment by intraperitoneal injection. Inhibitors (HSP inhibitor, LY and ALLN) and VPA were intraocularly injected using a Hamilton microsyringe with a 30G needle. The volume of injection was set at 5 μL of total volume after removal of the same volume of vitreous fluid. Inhibitors were intraocularly injected 6 h before VPA treatment.

Morphological observation

Fixation and cryosection of retinal samples have been described elsewhere (Koriyama et al. 2013). In brief, mouse eyes were enucleated and fixed overnight in 0.1 M phosphate buffer (pH7.4) containing 4% paraformaldehyde and 5% sucrose. They were then incubated in 30% sucrose overnight at 4°C. All eyes were then embedded in OCT compound (Sakura Finetek, Tokyo, Japan) and frozen in liquid nitrogen. Retinal sections in central areas (within 300 μm of the optic disc) were cut at 12 μm thickness and mounted onto silane-coated slides. Hematoxylin and eosin staining of transverse sections was used to evaluate the thickness of the inner retina [outer nuclear layer (ONL), outer plexiform layer, inner nuclear layer (INL), inner plexiform layer or ganglion cell layer]. The thickness of each retinal layers was measured using Scion Image software (Scion Corporation, Frederick, MD, USA).

Immunohistochemistry

After washing and blocking with Blocking One (Nacalai Tesque, Kyoto, Japan), retinal sections were incubated (at 4°C overnight) with primary antibodies for rabbit anti-recoverin (1 : 500, recoverin, Chemicon, Millipore Corporation, CA, USA), rabbit anti-HSP70 (1 : 500; Cell Signaling Technology, Tokyo, Japan), mouse anti-HSP70 (1 : 500; Abcam, Cambridge, MA, USA), rabbit anti-p-Akt (1 : 500, Ser473; Cell Signaling Technology), rabbit anti-AcHistone H3 (K9) (1 : 500; Cell Signaling Technology) and activated caspase-3 (1 : 200; Calbiochem, La Jolla, CA, USA). The sections were then incubated with Alexa Fluor anti-IgG (1 : 1000; Molecular Probes, Eugene, OR, USA) at 23°C. To quantify immunoreactivity, high-resolution images of two areas 1500 μm away from the optic disc were obtained (×200 magnifications, five mice in each group) as representative data. We randomly chose the two areas (100 × 50 μm2) per section of ONL and assessed the fluorescence intensity by using ImageJ software (Wayne Rasband, NIH, Bethesda, MD, USA). As Gaub et al. (2011) reported, the intensity values of each area were normalized to the 4′-6-diamidino-2-phenylindole or propidium iodide (PI) signal and mean values of intensities were calculated for each animal (five mice in each group).

Western blot analysis

Retinas were isolated and aliquots containing 30 μg of protein were subjected to polyacrylamide gel electrophoresis using a 12.5% gel as described previously (Koriyama et al. 2011). The separated proteins were transferred to a nitrocellulose membrane and sequentially incubated with primary and secondary antibodies. By using primary antibodies, signals for the HSP70, recoverin and caspase-3 protein bands were detected using a BCIP/NBT Kit (KPL, Gaithersburg, MD, USA). An antibody against rabbit β-actin (1 : 500; GeneTex, San Antonio, TX, USA) was used as an internal standard. Protein bands isolated from retinal samples were densitometrically analyzed using Scion Image Software (Scion Corporation). All experiments were repeated at least three times.

Terminal transferase-mediated dUTP nick-end labeling (TUNEL) staining

After fixation and cryosection, retinal sections were incubated in 0.1% Triton X-100 and 0.1% sodium citrate for 15 min. Sections were then further incubated with 20 μg/mL proteinase K in 10 mM Tris/HCl (pH 7.4) at 37°C for 20 min and rinsed in phosphate-buffered saline. DNA fragmentation of cells undergoing apoptosis was detected using an in situ cell death detection kit (Roche, Mannheim, Germany) containing terminal transferase and fluorescence dUTP. Retinal sections were incubated in this reaction mixture overnight at 37°C and rinsed twice in phosphate-buffered saline. The number of TUNEL-positive cells was counted by fluorescent microscopy (×200 magnification; E600, Nikon, Tokyo, Japan) in the middle region of the dorso-ventral retinal sections [including the midpoint between the optic disc and the edge of retina (750–1250 μm from the optic disc)]. We showed the percentage of TUNEL-positive cells in PI-positive ONL cells.

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assays were performed using a ChIP Assay Kit (Upstate/Millipore Corporation, Temecula, CA, USA). Briefly, proteins and DNA were cross-linked with formaldehyde, and cells were lysed in sodium dodecyl sulfate-lysis buffer and then sonicated. To reduce non-specific background, the sheared chromatin was incubated with Protein A agarose/salmon sperm DNA. The remaining chromatin was immunoprecipitated with IgG (control) or AcHistone H3 (K9) antibodies (1 : 500; Cell Signaling Technology). DNA–protein complexes were eluted from the antibody with elution buffer containing 1% sodium dodecyl sulfate and 0.1 M NaHCO3, as well as formaldehyde-reversed cross-links by 5 M of NaCl and heating at 65°C for 4 h. DNA was purified and PCR was performed using primers that spanned the HSP70 promoter site (Hspa1b) (Zhang et al. 2012). The primers used were: 5′- CCCAGCCCCTAAAGTTTGTT -3′ (forward) and 5′- GGGGATAGGGCTGATTAAGATT -3′ (reverse). A 1.5% agarose gel with ethidium bromide was used to separate and examine the PCR products.

Dot blotting analysis for 4-hydroxy-2-nonenal

We performed dot blotting analysis for 4-hydroxy-2-nonenal (4HNE), a marker of lipid peroxidation. After 0.5–1 day of treatment with MNU, in the presence or absence of VPA and/or inhibitors (HSP inhibitor, ALLN), retinal samples were isolated and homogenized. Equal amounts of protein (30 μg) were applied to a Hybri-SLOT apparatus (Gibco BRL, Rockville, MD, USA) and transferred to a nitrocellulose membrane (Whatman, Whatman Int. Ltd., Maidston, UK) by vacuum filtration. After blocking with 3% bovine serum albumin for 1 h at 23°C, the samples were incubated with an anti-4HNE antibody (1 : 100; Nikken SEIL Co, Shizuoka, Japan) at 4°C overnight, followed by incubation with anti-mouse IgG antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 1 h at 23°C. Antibody-bound protein bands were detected using a BCIP/NBT Kit and analyzed densitometrically as described above. All experiments were repeated at least three times.

Visual cliff test

A visual cliff test was used to evaluate the gross visual ability in visual acuity and depth perception (Fox 1965). The custom apparatus comprised an open-top box (75 × 75 × 15 cm) made of clear panels of polymethylmethacrylate. The central 30 × 30 cm was covered from outside with a black-and-white checkerboard (each square 1 × 1 cm) pattern as the ‘board’. The box was raised 100 cm above the ground to give the animal a sense of height, whereas the edge of the checkerboard created the edge of a ‘cliff’. The mouse was placed on the central checkerboard in the box and the total time spent on the board within 2 min was measured. The elapsed time to the first step off the board was measured as latency. To reduce the effect of learning and memory, the apparatus was turned 90 degrees for each trial. The plastic surface was cleaned thoroughly between trials to prevent mice from discovering visual clues about depth. Behavior was evaluated by an investigator blind to each treatment.

Statistics

All results were reported as mean ± SEM for three to five experiments. Differences between groups were analyzed using one-way anova, followed by Dunnett's multi-comparison test with PASW Software (SPSS Inc., Chicago, IL, USA). p values less than 0.05 were considered statistically significant.

Results

MNU induces selective and progressive loss of photoreceptor cells

Hematoxylin and eosin staining in MNU (60 mg/kg)-treated mouse retinal sections (Fig. 1) (a: 0 day, b: 3 days, c: 5 days, d: 7 days) showed that the ONL and outer plexiform layer became significantly thinner by 3 days when compared with control retinas (Fig. 1e). These changes in thickness became more severe by 7 days. The INL and inner plexiform layer became significantly thinner from 5 to 7 days (Fig. 1c, d). No differences were seen in the ganglion cell layer within 7 days of MNU treatment.

Figure 1.

Selective and progressive loss of photoreceptors by N-methyl-N-nitrosourea (MNU). (a–d) Microscopy images of retinas following injection of MNU (60 mg/kg) at 0 days (a) scale bar = 50 μm, 3 days (b), 5 days (c), 7 days (d). (e) Retinal degeneration was evaluated by measuring the thickness of each layer of mice retina. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer in this figure and subsequent figures. *p < 0.01 versus 0 day (n = 20). (f–h) MNU-induced photoreceptor cell death in ONL revealed by TUNEL staining. (f) Vehicle control, scale bar = 50 μm, (g) 2 days after MNU treatment. (h) Graphical representation of percentage of TUNEL-positive cells in the ONL per visual field (×200 magnification). *p < 0.01 versus 0 day (n = 8).

Western blot analysis of recoverin, a photoreceptor marker protein (Nagar et al. 2009), showed that levels were significantly decreased from 3 days after MNU treatment (data not shown). TUNEL-positive cells were observed in the ONL from 1 day after MNU treatment, but not in any other layers (data not shown) when compared with the vehicle control (Fig. 1f). The percentage of TUNEL-positive cells dramatically increased in the ONL at 1–3 days after MNU treatment (Fig. 1g and h).

VPA induces histone H3 acetylation and transcriptional activation for HSP70

VPA directly inhibits histone deacetylase, resulting in histone hyperacetylation (Göttlicher et al. 2001), and facilitates transcriptional activity (Berger 1999). For example, VPA induces the expression of HSP70 in neurons, which has neuroprotective effects (Ren et al. 2004; Wu et al. 2008; Marinova et al. 2009). The acetylation level at lys 9 of histone H3 (AcH3), an index of histone deacetylase inhibition, was measured 1 day after intra-ocular VPA injection. In the vehicle control (Fig. 2a–c and g), a slight increase in acetylation levels of histone H3 was seen in the INL and ONL. After treatment with 5 mM VPA, intensive staining of AcH3 was seen in the INL and ONL for 1 day (Fig. 2d–f and g). Fluorescent intensity of AcH3 by VPA increased in a dose-dependent manner in the ONL (Fig. 2g). To determine whether VPA treatment increases HSP70 promoter activity, we performed ChIP. This revealed that VPA markedly increased histone H3 acetylation in the HSP70 promoter, compared with the vehicle control (Fig. 2h), in a dose-dependent manner (2.5 mM and 5 mM). Next, we studied the effect of VPA on HSP70 protein expression by western blotting. Intra-ocular injection of VPA significantly increased HSP70 protein levels 1.7-fold at 1 day after treatment when compared with no treatment (Fig. 2i). HSP70 expression levels returned to control levels by 3 days after treatment; levels of β-actin expression did not change during this period. To determine the localization of HSP70 expression by VPA treatment, we performed immunohistochemistry of HSP70 in retinas treated with VPA for 1 day. VPA at 5 mM increased the levels of HSP70 in the ONL and, to a lesser extent, in the INL (Fig. 2m–o) compared with the vehicle control (Fig. 2j–l). Pre-treatment of HSP inhibitor suppressed the expression of HSP70 by VPA (data not shown). Fig. 2p, quantifies HSP70 fluorescence intensity. VPA increased the levels of HSP70 immunoreactivity and HSP inhibitor significantly suppressed this increase.

Figure 2.

HSP70 induction by valproic acid (VPA) through histone H3 acetylation. (a–g) Histone H3 acetylation by VPA treatment. (a, d) Immunohistochemistry for acetylation of histone H3 (AcH3) in the mouse retina at 0 day (a) and 1 day (d) after treatment with 5 mM VPA. (b, e) PI staining of (a) and (d). (c, f) Merged images of VPA and PI. Scale bar = 50 μm. (g) Levels of AcH3 expression were quantified by analysis of fluorescence intensity. *p < 0.05, **p < 0.01 versus 0 mM (n = 8). (h) Retinas were treated with VPA for 1 day and chromatin was prepared and sheared by sonication. The protein/DNA complex was then incubated with AcH3 antibody for chromatin immunoprecipitation (ChIP) assay. Representative gel electrophoresis images of amplification of HSP70 gene promoter are shown. **p < 0.01 versus 0 mM (n = 3). (i) HSP70 protein expression after 5 mM VPA treatment was quantified by western blot analysis. *p < 0.05, **p < 0.01 versus 0 day (n = 3). (j–p) HSP70 induction by 5 mM VPA treatment. Immunohistochemistry for HSP70 in the mouse retina at 0 day (j) and 1 day (m). (k, n) PI staining of (j, m). (l, o) Merged image of HSP70 and PI. Scale bar = 50 μm. (p) HSP70 expression quantified by analysis of fluorescence intensity. HSP inhibitor (10 μM) was intraocularly injected 6 h before VPA (5 mM) injection. *p < 0.01 versus vehicle control (n = 8).

HSP70 induction in the ONL by VPA activates PI3K/Akt signaling

It has been reported that inducible HSP70 regulates the activity of Akt kinase (Koren et al. 2010). Akt is a key molecule in cell survival pathways through inhibition of caspases and activation of the proapoptotic protein Bad. Intra-ocular VPA significantly increased p-Akt levels during the same period of HSP70 induction (Fig. 3a). p-Akt returned to control levels by 3 days after VPA treatment; the levels of total Akt and β-actin expression did not change during this period. VPA at 5 mM increased the levels of p-Akt in the ONL and, to a lesser extent, in the INL (Fig. 3c, g, k) as compared with the vehicle control (Fig. 3b, f, j). Pre-treatment of PI3K inhibitor LY294002 (LY) suppressed the activation of Akt by VPA (Fig. 3d, h, l). Pre-treatment of HSP inhibitor also decreased the immunoreactivity of p-Akt by VPA (Fig. 3e, i, m). Fig. 3n, quantifies p-Akt fluorescence intensity under various conditions. Neither LY nor HSP inhibitor alone significantly affected Akt activation in the vehicle control. To confirm the colocalization of Akt activation and HSP70 induction, we did double staining of p-Akt (Fig. 3o) and HSP70 (Fig. 3p) using VPA-treated retinal sections at 1 day after treatment (Fig. 3q).

Figure 3.

Akt activation by valproic acid (VPA) mediated by HSP70 expression. (a) Activation of Akt was quantified by western blot analysis. *p < 0.05, **p < 0.01 (n = 3). (b–m) Akt activation by VPA. (b–e) Immunohistochemistry for p-Akt in the mouse retina of the vehicle control (b), 5 mM VPA (1 day) (c), VPA + 10 μM LY (d), and VPA + 10 μM HSP inhibitor (e). (f–i) PI staining of (b–e). (j–m) Merged image of p-Akt and PI. Scale bar = 50 μm. (n) p-Akt expression quantified by analysis of fluorescence intensity. *p < 0.01 versus vehicle control, +p < 0.01 versus VPA alone (n = 8). (o–q) Colocalization of p-Akt (o) and HSP70 (p) at 1 day VPA-treated retina (q: Merged image). Scale bar = 50 μm.

HSP70 induced by VPA protects photoreceptor cell death by MNU

Next, we evaluated the effect of HSP70 on the MNU-induced change in ONL thickness. At 3 days, the ONL was significantly thinner in MNU-treated retinas (Fig. 4b, g) when compared with vehicle controls (Fig. 4a, g). Pre-treatment of VPA significantly reduced this MNU-induced thinning of the ONL (Fig. 4c, g). Furthermore, HSP inhibitor completely canceled the effect of VPA (Fig. 4d, g). Quantitative evaluation of ONL thickness showed that HSP inhibitor significantly suppressed the protective effect of VPA against MNU-induced photoreceptor cell loss. To confirm the recovery effect of VPA on MNU-induced photoreceptor cell loss, we performed a western blot for recoverin. Recoverin levels were significantly reduced at 3 days after MNU treatment (Figure S1). Pre-treatment of VPA significantly increased the levels of recoverin in MNU-treated retina. HSP inhibitor completely canceled the rescue effect of VPA on the MNU-induced decrease in recoverin expression. The quantitative changes in recoverin protein levels induced by MNU with VPA or HSP inhibitor matched the data on ONL thickness. At 7 days of treatment, the changes in ONL thickness became more severe (Fig. 4e, h) when compared with the vehicle control (Fig. 4a). Pre-treatment of VPA slightly inhibited photoreceptor cell loss by MNU (Fig. 4f, h). HSP inhibitor canceled the protective effect by VPA against MNU-induced photoreceptor cell loss (Fig. 4h). VPA or HSP inhibitor alone did not affect the ONL thickness when compared with the vehicle control at 7 days (Fig. 4h) or 0 days (data not shown).

Figure 4.

Valproic acid (VPA) attenuated retinal degeneration by N-methyl-N-nitrosourea (MNU) in mice. (a–f) Microscopical images of retina treated with vehicle (a), MNU (3 days) (b), MNU + VPA (3 days) (c), and MNU + VPA + HSP inhibitor (3 days) (d), MNU (7 days) (e), MNU +VPA (7 days) (f). Scale bar = 50 μm. (g) Vertical thickness of outer nuclear layer (ONL) in mice retina (3 days). *p < 0.01 versus vehicle control, +p < 0.01 versus MNU, #p < 0.01 versus MNU +VPA (n = 20). (h) Vertical thickness of ONL in mice retina (7 days). *p < 0.01 versus vehicle control, +p < 0.01 versus MNU, #p < 0.01 versus MNU +VPA (n = 20).

HSP70-dependent cell survival against MNU-induced photoreceptor cell loss

We evaluated the effect of HSP70 induction by VPA on MNU-induced photoreceptor cell death. Numerous TUNEL-positive cells were observed in the ONL 3 days after MNU treatment (Fig. 5b and f) when compared with the vehicle control (Fig. 5a). Treatment of VPA 6 h after MNU injection showed a smaller protective effect when compared with treatment of VPA 6 h before MNU injection (Figure S2). Thus, we used pre-treatment of VPA in this study. Pre-treatment of VPA (6 h before MNU treatment) clearly inhibited the MNU-induced cell death in the ONL (Fig. 5c and f). HSP inhibitor canceled the protective effect of VPA (Fig. 5d and f). As calpain activation is involved in MNU-induced photoreceptor cell death, calpain inhibition effectively protects against photoreceptor cell loss in mice (Kuro et al. 2011). In the present study, the calpain inhibitor ALLN also clearly inhibited photoreceptor cell death by MNU treatment (Fig. 5e and f). The percentage of TUNEL-positive cells by VPA treatment alone did not show any significant difference when compared with the vehicle control at 7 days (data not shown).

Figure 5.

HSP70 induction by valproic acid (VPA) protected N-methyl-N-nitrosourea (MNU)-induced photoreceptor cell death. (a–e) VPA reduced MNU-induced TUNEL-positive cells mediated by HSP70 expression. Fluorescence images of vehicle control (a), MNU (3 days) (b), MNU + VPA (c), MNU + VPA + HSP inhibitor (d), and MNU + N-acetyl-Leu-Leu-norleucinal (ALLN) (e). Scale bar = 50 μm. (f) Graphical representation of percentage of TUNEL-positive cells in outer nuclear layer (ONL) per visual field (×200 magnification). *p < 0.01 versus vehicle control, +p < 0.01 versus MNU, #p < 0.01 versus MNU +VPA (n = 8).

MNU induced 4HNE production and HSP70 cleavage before photoreceptor cell death

Tsuruma et al. (2012) reported that oxidative stress is involved in photoreceptor cell death in MNU-treated mouse retinas. 4HNE is generated by free radical attack on omega polyunsaturated fatty acids and is largely responsible for pathogenesis during oxidative stress. Using whole-retina samples, we used dot blot analysis to evaluate 4HNE levels in the MNU-treated retina. Levels of 4HNE clearly increased after MNU treatment in a time-dependent manner at 1 day (Fig. 6a, d, g) and 2 days (Fig. 6a, e, g) compared with the vehicle control (Fig. 6a and c). It appears that 4HNE production is much faster than photoreceptor cell loss, because pre-treatment of VPA, HSP inhibitor or ALLN did not affect 4HNE production (Fig. 6b, f, g). Proteomic analysis has identified HSP70 as a key substrate of calpain in the retina (Nakajima et al. 2006). In the vehicle control, intact HSP70 (~70 KDa) were slightly cleaved, as shown by a band of ~30 KDa. After 1 day of MNU treatment, the intact bands (~70 KDa) decreased and the cleaved bands (~30 KDa) of HSP70 dramatically increased (Fig. 6h–j). ALLN significantly suppressed the cleavage of HSP70 that arose from MNU treatment. VPA dramatically returned intact HSP70 to control levels (Fig. 6h, i). MNU induced caspase-3 activation in the retina (Fig. 6k, l). Both VPA and ALLN completely blocked caspase-3 activation of MNU (Fig. 6k, l). However, in the presence of HSP inhibitor, VPA lost its ability to block MNU-induced caspase-3 activation.

Figure 6.

Valproic acid (VPA) supplemented HSP70 against N-methyl-N-nitrosourea (MNU)-induced cleavage and inhibited caspase-3 activation. (a, b) MNU-induced accumulation of 4HNE. 4HNE production was measured by dot blotting analysis with an anti-4HNE antibody. Graphical representation of 4HNE bands density in the blot. The orders of the bands consist with that of the bar graph. *p < 0.01 versus vehicle control (n = 3). HSP inh.: HSP inhibitor, AL: N-acetyl-Leu-Leu-norleucinal (ALLN). (c–g) Immunohistochemistry for 4HNE in the vehicle control with nuclear 4′-6-diamidino-2-phenylindole (DAPI) staining (c), MNU, (1 day, d), MNU 2 day (e), and MNU + VPA (2 day, f). Scale bar = 50 μm. (g) 4HNE production quantified by analysis of fluorescence intensity. *p < 0.01 versus vehicle control (n = 8). (h–j) VPA induced HSP70 against cleavage of HSP70 by MNU. Density of HSP70 (i) and cleaved HSP70 (j) in the blot. *p < 0.01 versus vehicle control (n = 8) M: MNU, V: VPA, AL: ALLN. (k and l) VPA suppressed caspase-3 activation by MNU through HSP70 induction (k). Graphical representation of activated caspase-3 bands density of in the blot (l). *p < 0.01 versus vehicle control, +p < 0.01 versus MNU (n = 3).

Induction of HSP70 by VPA delayed visual function loss by MNU

We tested visual behaviors to evaluate the functional consequences of photoreceptor degeneration. We evaluated depth perception using a visual cliff test. Mice were placed on a checkerboard in the visual cliff apparatus and allowed to roam freely for 2 min. Generally, normal mice hesitated at the checker board and did not walk off the cliff (Krishnamoorthy et al. 2008; de Lima et al. 2012). Most control mice spent approximately 80 s in the central zone (checkerboard) (Fig. 7a). This value progressively decreased and they spent only 30 s in the central zone by 7 days after MNU treatment. Treatment with VPA delayed this loss of visual function at 3–5 days after MNU. Furthermore, HSP inhibitor canceled the delaying action of VPA at 3 days after treatment of MNU (Fig. 7b). Latency (the time it takes for mice to first cross the visual cliff) was also shortened by 3 days after MNU treatment (Fig. 7c). HSP70 induction by VPA significantly increased latency when compared with the MNU-treated mice at 3–5 days and HSP inhibitor canceled this increase at 3 days (Fig. 7d).

Figure 7.

HSP70 induction by valproic acid (VPA) delayed visual function loss by N-methyl-N-nitrosourea (MNU). (a) Data of time positioned on checker board. *p < 0.01 versus 0 day, +p < 0.01 versus MNU (n = 5). (b) HSP-dependent recovery of visual function (time positioned on checker board). HSP inhibitor (10 μM) was intraocularly injected 6 h before VPA treatment. **p < 0.01, *p < 0.05 versus 0 day, +p < 0.01 versus MNU, #p < 0.01 versus MNU + VPA (n = 5). (c) Latency analysis. *p < 0.01 versus 0 day, +p < 0.01 versus MNU (n = 5). (d) HSP-dependent recovery of visual function (latency). HSP inhibitor (10 μM) was intraocularly injected 6 h before VPA treatment. **p < 0.01, *p < 0.05 versus 0 day, +p < 0.01 versus MNU, #p < 0.01 versus MNU + VPA (n = 5).

Discussion

This study has four salient findings: HSP70 induction by VPA pre-treatment shows a protective effect against MNU-induced photoreceptor cell death; HSP70 induction delays the visual function loss caused by MNU; Akt activation by VPA is HSP70-dependent; and MNU gives rise to calpain-dependent cleavage of HSP70.

HSP70 induction by VPA protects MNU-induced photoreceptor cell death

There has been an increasing need to establish an animal model for human retinal degenerative disorders. Herrold originally reported that MNU causes loss of photoreceptors in golden hamsters within one week (Herrold 1967; Tsubura et al. 2011) in both males and females (Nakajima et al. 1996). In previous studies, injection of 60 mg/kg MNU into rats or mice evoked photoreceptor cell loss and depletion within 7 days of injection (Kiuchi et al. 2002; Nagar et al. 2009). Tsubura's group found that TUNEL-positive cell death peaked at 3 days after MNU treatment and almost all photoreceptor cells were lost by 7 days (Kuro et al. 2011). In the present study, we also showed a similar time course of photoreceptor cell death. MNU selectively damaged photoreceptor cells; no other cells in the retina were TUNEL-positive. Several reports have been published on the mechanism of MNU-induced photoreceptor apoptotic cell death: a decrease in antiapoptotic Bcl-2 protein, an increase in proapoptotic Bax protein, and activation of caspases (Yoshizawa et al. 1999, 2000). Therefore, therapies to prevent MNU-induced photoreceptor cell loss have been tried using caspase-3 inhibitor (Yoshizawa et al. 1999; Uehara et al. 2006), nicotinamide (Petrin et al. 2003a), poly (ADP-ribose) polymerase inhibitor (Miki et al. 2007) and X-linked inhibitor of apoptosis protein (Petrin et al. 2003b). VPA induces the expression of HSP70 in various cells mediated by histone hyperacetylation (Ren et al. 2004; Marinova et al. 2009; Zhang et al. 2012). In this study, we provide compelling evidence that HSP70 induction by VPA, with hyperacetylation of histone H3, protects photoreceptors against MNU-induced cell death. It has been reported that HSP70 has multiple anti-apoptotic effects both upstream and downstream of caspase cascades (Garrido et al. 2003). For instance, up-regulation of HSP70 increases expression of the major anti-apoptotic protein Bcl-2 (Kelly et al. 2002), interferes with the function of apaf-1 thereby preventing the formation of the apoptosome (Yenari et al. 2005), and inhibits the proapoptotic Jun-N terminal kinase (Gabai et al. 1997). In the present study, MNU caused degeneration of photoreceptor cells by activating caspase-3. HSP70 induction by VPA attenuated caspase-3 activation, whereas HSP inhibitor canceled this attenuation. These data suggest that HSP70-induction by VPA protects MNU-induced photoreceptor cell death and inactivates caspase-3. Yoshizawa et al. (2000) reported that caspase-3 inhibitor was effective in protecting against MNU-induced photoreceptor cell death.

Based on the recent report that HSP70 can modulate Akt kinase activity (Koren et al. 2010), we investigated the effect of HSP70 on Akt activity by VPA. We found that VPA induced activation of Akt, and HSP inhibitor suppressed it. Activation of Akt in neurons leads to inhibition of cell death machinery proteins such as proapoptotic Bad and members of the caspase family (Kayama et al. 2011). Interestingly, HSP70 directly interacts with p-Akt and inhibits the caspase cascade, thereby preventing photoreceptor cell loss in an animal model of retinal detachment (Koren et al. 2010). Moreover, HSP70 is associated with rictor, one of the cofactors that make the mammalian target of rapamycin complex 2 (Kudchodkar et al. 2006; Martin et al. 2008). This association leads to Akt phosphorylation at serine 473, resulting in full activation of Akt (Sarbassov et al. 2005). These mechanisms might contribute to HSP70-induced photoreceptor cell survival after MNU treatment. We used VPA because it is a well-studied inducer of HSP70 in the central nervous system. Other HSP70 inducers, such as geranylgeranylacetone (Kayama et al. 2011), arimoclomol and celastrol (Kalmar and Greensmith 2009), might also protect against MNU-induced photoreceptor cell death.

Possible mechanism of MNU-induced photoreceptor cell loss through cleavage of HSP70

Another novel finding of this study is that MNU cleaved HSP70 before inducing photoreceptor cell death through 4HNE production. There are several reports on the mechanism of photoreceptor cell death caused by MNU (Tsubura et al. 2010). Relative to its size, the eye requires three to four times more oxygen than the brain and, consequently, produces more of reactive oxygen species. In addition, as retinal neurons are highly enriched in lipids containing polyunsaturated fatty acids (Fliesler and Anderson 1983), they can produce 4HNE from polyunsaturated fatty acids during oxidation from aging. Tsuruma et al. (2012) reported that MNU induced oxidative radicals and production of 4HNE within 12 h of treatment. Oka et al. (2007) further reported that total calcium ion in MNU-treated retinas is significantly increased, and calpain activity is dramatically increased, from 1 day after MNU administration. Moreover, calpain inhibitor rescued photoreceptors from MNU-induced cell death (Kuro et al. 2011). In 1998, Yamashima et al. (1998) proposed the ‘calpain-cathepsin hypothesis’ to explain the cell death in hippocampal neurons after ischemic insult. They suggest that the lysosomal membrane is disrupted by activation of calpain, which causes the release of lysosomal protease, in particular, cathepsin. Recently, they further found that the key event in cell death by the calpain-cathepsin hypothesis is HSP70 cleavage through carbonylation of HSP70 by 4HNE (Yamashima 2012). Proteomic analysis using monkey and human retina has identified HSP70 as a key substrate of calpain (Nakajima et al. 2006). Calpain-mediated cleavage of HSP70 after 4HNE production may be possible in the MNU-induced photoreceptor cell death seen in the present study. As carbonylated proteins such as carbonyl HSP70 cannot be repaired, they are removed by proteolytic degradation or accumulation as damaged or unfolded proteins in the cell (Stadtman and Berlett 1998).

In our study, calpain inhibitor suppressed HSP70 cleavage and subsequent photoreceptor cell death by MNU. Furthermore, induction of HSP70 by VPA prevented both HSP70 cleavage and the MNU-induced photoreceptor cell death through caspase-3 inactivation. HSP inhibitor canceled this rescue effect. The production of 4HNE by MNU must be an early event because neither VPA nor calpain inhibitor affected it. Thus, photoreceptor cell death by MNU might be mediated by the calpain-cathepsin hypothesis through production of 4HNE and the dysfunction of HSP70 (Figure S3). HSP70 induction by VPA significantly delayed visual function loss at 3 days, but not at 7 days after MNU treatment (Fig. 7). However, HSP70 induction significantly protected photoreceptor cell death both at 3 days and 7 days after MNU treatment (Fig. 4). This difference in the time course of photoreceptor cell death and visual function loss might be because more intact photoreceptor cells are required for depth perception. As HSP70 expression by VPA is transient (Fig. 2i), another genetic probe is required to get continuous HSP70 protein expression. We do not know whether the results from this MNU model can be applied to human inherited retinitis pigmentosa and age-related macular degeneration. We do know, however, that oxidative stress, 4HNE accumulation and calpain activation are well-known risk factors in these human diseases (Shen et al. 2005; Paquet-Durand et al. 2006). Further studies are needed to confirm these possibilities and to clarify the possible mechanisms of pathogenesis and interaction between HSP70 cleavage and photoreceptor cell death. Taken together, these data strongly indicate that HSP70 is a key molecule both for pathogenesis and protection in MNU-induced photoreceptor cell death. We hope that the present study heralds a new era in developing therapeutic tools against retinitis pigmentosa and age-related macular degeneration. We need to use genetic models for RDs to prove these hypotheses in the future.

Acknowledgments and conflict of interest

We thank Mr Takashi Baba, Ms Sachiko Higashi and Tomoko Kano for their administrative and technical assistance. We thank Dr Tetsumori Yamashima of Kanazawa University for his critical comments. This work was partly supported by research grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Nos 25462753 to YK and 25640008 to SK) and KAKENHI. The authors have declared that no competing interests exist.

All experiments were conducted in compliance with the ARRIVE guidelines.

Ancillary