Nitric oxide regulates cell survival in purified cultures of avian retinal neurons: involvement of multiple transduction pathways

Authors

  • T. A. Mejía-García,

    1. Department of Neurobiology and Program of Neuroimmunology, Institute of Biology, Federal Fluminense University, Niterói, Brazil
    Search for more papers by this author
  • R. Paes-de-Carvalho

    1. Department of Neurobiology and Program of Neuroimmunology, Institute of Biology, Federal Fluminense University, Niterói, Brazil
    Search for more papers by this author

Address correspondence and reprint requests to Roberto Paes-de-Carvalho, Departamento de Neurobiologia, Instituto de Biologia, Caixa Postal 100180, Centro, Niterói, RJ 24001–970, Brazil.
E-mail: robpaes@vm.uff.br

Abstract

Nitric oxide (NO) is an important signaling molecule in the CNS, regulating neuronal survival, proliferation and differentiation. Here, we explored the mechanism by which NO, produced from the NO donor S-nitroso-acetyl-d-l-penicillamine (SNAP), exerts its neuroprotective effect in purified cultures of chick retinal neurons. Cultures prepared from 8-day-old chick embryo retinas and incubated for 24 h (1 day in culture, C1) were treated or not with SNAP, incubated for a further 72 h (up to 4 days in culture, C4), fixed, and the number of cells estimated, or processed for cell death estimation, by measuring the reduction of the metabolic dye 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT). Experimental cultures were run in parallel but were re-fed with fresh medium in the absence or presence of SNAP at culture day 3 (C3), incubated for a further 24 h up to C4, then fixed or processed for the MTT assay. Previous studies showed that the re-feeding procedure promotes extensive cell death. SNAP prevented this death in a concentration- and time-dependent manner through the activation of soluble guanylate cyclase; this protection was significantly reversed by the enzyme inhibitors 1H-[1,2,4]oxadiazolo-[4,3-a]quinoxalin-1-one (ODQ) or LY83583, and mimicked by 8-bromo cyclic guanosine 5′-phosphate (8Br-cGMP) (GMP) or 3-(5′-hydroxymethyl-2′-furyl)-1-benzyl indazole (YC-1), guanylate cyclase activators. The effect was blocked by the NO scavenger 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (PTIO). The effect of NO was also suppressed by LY294002, Wortmannin, PD98059, KN93 or H89, indicating the involvement, respectively, of phosphatidylinositol-3 kinase, extracellular-regulated kinases, calmodulin-dependent kinases and protein kinase A signaling pathways. NO also induced a significant increase of neurite outgrowth, indicative of neuronal differentiation, and blocked cell death induced by hydrogen peroxide. Cyclosporin A, an inhibitor of the mitochondrial permeability transition pore considered an important mediator of apoptosis and necrosis, as well as boc-aspartyl (OMe) fluoromethylketone (BAF), a caspase inhibitor, also blocked cell death induced by re-feeding the cultures. These findings demonstrate that NO inhibits apoptosis of retinal neurons in a cGMP/protein kinase G (PKG)-dependent way, and strengthens the notion that NO plays an important role during CNS development.

Abbreviations used
AKT

3-phosphoinositide-dependent protein kinase-1

BAF

boc-aspartyl (OMe) fluoromethylketone

CaMK

calmodulin-dependent kinase

Csp-A

cyclosporine-A

ERK

extracellular signal-regulated kinase

GSNO

S-nitrosoglutathione

KT5823

protein kinase G inhibitor

LY

LY294002

MEK

mitogen-activated or extracellular signal-regulated protein kinase

mPT

mitochondrial permeability transition pore

NBMPR

S-(nitrobenzyl)-mercaptopurine-riboside

NOS

nitric oxide synthase

ODQ

1H-[1,2,4]oxadiazolo-[4,3-a]quinoxalin-1-one

PD

PD98058

PI3-K

phosphoinositide-3-kinase

PKA

protein kinase A

PKG

protein kinase G

PTIO

2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide

RF

re-fed with fresh medium

sGC

soluble guanylate cyclase

SIN

1, 3-morpholinosydnonimine

SNAP

S-nitroso-N-acetylpenicillamin

Wor

Wortmannin

YC-1

3-(5′-hydroxymethyl-2′-furyl)-1-benzyl indazole

Zap

zaprinast

Comprehensive studies during the past two decades have shown that nitric oxide (NO), a free radical gas, although one of the simplest molecules, triggers a vast range of physiological and pathological responses, including neurotransmission, synaptic plasticity in the CNS, non-specific immunity, vascular tone regulation and the pathogenesis of several diseases (Hölscher 1997; Lane and Gross 1999). Generated in different cell types through the conversion of l-arginine to l-citrulline by nitric oxide synthase (NOS; EC 1.14.13.39), NO has a half-life of only a few seconds in vivo, but its solubility in both aqueous and lipid media is probably responsible for its pleiotropic effects in the cells. NO is also generated post-synaptically by Ca2+/calmodulin-dependent NOS and works as a retrograde messenger (Bredt and Snyder 1992).

The main targets of NO under physiological conditions are metal centered, particularly iron in iron-heme proteins. The most striking example is the formation of a nitrosyl complex with iron (II) of soluble guanylyl cyclase (sGC), inducing a conformational change of the enzyme and its activation to produce cyclic guanosine 5′-phosphate (cGMP) (GMP) (Ford and Lorkovic 2002), and subsequent activation of protein kinase G (PKG) and protein phosphorylation (Schmidt W. (1992)). Another major pathway by which NO mediates its biological effects (cGMP-independent pathway) is the S-nitrosylation of cysteine residues (Lipton et al. 1993) in multiple cell types and tissues (Benz et al. 2002; Gow et al. 2002). Alternatively, NO has been shown to mediate post-translational modification of proteins, such as ADP ribosylation (Duman et al. 1993), and nitration of tyrosine (Tyr) residues (Haqqani et al. 2002). These modifications may prevent normal interactions between proteins involved in pre-synaptic membrane-specific interactions occurring during exocytosis (Meffet et al. 1996), or in inhibition of cell death by apoptosis (Schonhoff et al. 2003).

The physiological levels of NO produced from NOS or NO donors can function as an endogenous regulator of apoptosis in primary cultured neurons (Farinelli et al. 1996) and PC12 cells (Kim et al. 1999). Recently, it has been demonstrated that the sustained inhibition of NO production triggers apoptosis in cerebellar granule neuronal cultures, accompanied by down-regulation of AKT/PKB kinase and the transcription factor CREB (cAMP-responsive element binding protein) (Ciani et al. 2002). Activation of AKT by phosphatidylinositol 3-kinase (PI3K) serves as a multifunctional regulator of apoptotic cell death, cell growth and glucose metabolism. With respect to neuronal cell function, AKT has been shown to be required for prevention of apoptosis and promotion of cell survival through the phosphorylation of the proapoptotic Bcl-2 family protein, Bad (Datta et al. 1997), and procaspase 9 (Cardone et al. 1998). However, the relationship between the anti-apoptotic action of the NO/cGMP and AKT pathways remains poorly understood.

Mitogen-activated protein kinases (MAPK) transduce extracellular signals from tyrosine kinase receptor and G protein-coupled receptor to cytoplasmic and nuclear effectors (Chang and Karin 2001). MAPK signaling pathways consist of sequentially acting serine-threonine protein kinases designated RAF, MEK and extracellular signal-regulated kinases (ERK) (Davis 2000). Dowstream targets of ERK1/2 include the 90 kDa ribosomal protein, S6 kinase (RSK), ets-like transcription factor-1 (Elk-1), CREB and the protein, Bad (Schaeffer and Weber 1999). Phosphorylation of Bad on Ser112 dissociates Bcl-2/Bad heterodimers and unmasks the anti-apoptotic effect of Bcl-2 (Jin et al. 2002).

Previous work showed the presence of nitric oxide synthase (NOS) in the retina (Ríos et al. 2000; Lee et al. 2003). In the chick retina, NADPH diaphorase activity, which could be inhibited by arginine analogs and stimulated by calcium, was found from the early stages of development in distinct cell types, including amacrine cells and photoreceptors (de Faria et al. 1996). NOS was also directly measured in the chick retina and could be stimulated by glutamate (Paes-de-Carvalho and Mattos 1996). Using mixed cultures of retinal neurons and glial cells, we were able to show the presence of a high affinity l-arginine uptake system in both types of cells, and that the amino acid could be released from glial cells and taken up by neurons (Cossenza and Paes-de-Carvalho 2000). The presence of NOS was also detected in neurons, but not in glial cells, using immunocytochemistry against nNOS or l-citruline (Cossenza and Paes-de-Carvalho 2000). More recently, we have shown that activation of NMDA receptors inhibits protein synthesis and promotes accumulation of l-arginine, which, in turn, is used for the synthesis of NO in the cultures (Cossenza et al. 2006). These results indicate that NO has important roles in the developing retina.

These previous results prompted us to exploit the same experimental model to confirm the link between NO and different signaling pathways in purified cultures of retinal neurons. Additionally, we aimed to find evidence for the involvement of these pathways in neuronal survival. We report that the NO donors SNAP or S-nitrosoglutathione (GSNO) suppress re-feeding-induced cell death, an effect prevented by inhibition of sGC, PKG, PI3K, protein kinase A (PKA) and calmodulin-dependent kinase (CaMK). Together, these findings show that NO has a pivotal role in regulating signaling pathways involved in neuronal survival in the retina.

Materials and methods

Reagents

Basal Eagle's medium (BME), penicillin, streptomycin and l-glutamine were purchased from Invitrogen (Carlsbad, CA, USA). KT5823, LY294002, PTIO, H89 and KN93 were from Calbiochem (San Diego, CA, USA). S-nitroso-N-acetyl-d,l-penicillamine (SNAP), S-(nitrobenzyl)-mercaptopurine-riboside (NBMPR), 1H-(1,2,4)-oxadiazole(4,3-a)quinoxalon-1-one (ODQ), zaprinast, poly l-ornithine, Wortmannin, LY58358 and S-nitroso-glutathione (GSNO) were from Sigma/RBI Chemical Co. (St. Louis, MO, USA). Boc-aspartyl(OMe)fluoromethylketone (BAF) was from ICN Biochemicals (Cleveland, OH, USA). Glutaraldehyde (25%) was from Fluka Chemie (Steinheim, Switzerland). Fertilized White Leghorn chicken eggs were obtained from a local hatchery and incubated at 38°C in a humidified atmosphere to the appropriate age. All other reagents were of analytical grade.

Preparation of purified cultures

Glial-free purified cultures of retinal neurons were prepared from 8-day-old chick embryos as previously described (Adler et al. 1984). In brief, retinas were dissected and incubated with 0.1% trypsin in CMF (Ca2+- and Mg2+-free Hank's balanced salt solution) for 10 min at 37°C. The cells were dissociated using a tapered Pasteur pipette, suspended in BME and seeded in 35 mm plastic tissue culture dishes pre-coated with poly l-ornithine for 24 h. The cell density was approximately 832 cells/mm2. Cultures were incubated in BME containing 2.5% fetal calf serum, 100 U/mL penicillin, 100 µg/mL streptomycin and 2 mm glutamine, for 4 days at 37°C in an atmosphere of 5% CO2/95% air.

Drug exposure and fixation procedure

Cultures incubated for 24 h (C1) were treated with different drugs at the indicated concentrations. At C3 (those days in culture), some cultures were re-fed with fresh medium containing the same drugs. Cultures were then incubated up to C4 and fixed for 1 h with a solution of 2.5% glutaraldehyde in phosphate buffer 0.2 m, pH 7.4. Control cultures were run in parallel with no re-feeding in the presence of the same drugs. Toxicity of the different drugs was first evaluated by testing different concentrations of each drug and verifying culture development. Concentrations chosen for use were the highest that did not promote any visible toxic effect.

Determination of neuronal number

Cells were counted in five, randomly selected microscopic fields of 0.114 mm2 under a 100× objective using an inverted microscope, and the mean number of viable cells per field was determined.

MTT assay

Cultures were incubated with MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) for 4 h at 37°C and washed in Hank's solution, pH 7.2. The dye produced by viable cells was dissolved in acid/alcohol solution (0.6% chloridric acid in isopropanol) and the optical density measured at 570 nm.

Measurement of nitrate/nitrite production

Samples were assayed for nitrate and nitrite (stable end-products of NO) after reduction of nitrate into nitrite by copper-plated cadmium, as described by Florquin et al. (1994). Briefly, samples from the culture medium or cell lysates were deproteinized with zinc sulfate, reduced with activated cadmium, and the nitrite content measured with Griess reagent.

Measurement of [3H]-adenosine uptake

Cultures at C1 were treated with SNAP or NBMPR for 30 min and, after addition of 0.5 µCi/mL [3H]-adenosine, were incubated for an additional 48 h. Then the medium was removed, the cultures washed, lysed with water, and the intracellular radioactivity determined by liquid scintillation.

Statistical analysis

Statistical analysis was performed using analysis of variance (anova) followed by the Bonferroni test, using the software Graphpad Prism.

Results

Neuroprotective effect of NO donors in purified neuronal cultures

Our previous work showed that re-feeding purified neuronal cultures with fresh medium causes intense cell death, an effect blocked by pre-treatment of cultures with adenosine and activation of A2a receptors (Paes-de-Carvalho et al. 2003). To evaluate the role of NO as a neuroprotective molecule we decided to verify its effects on the survival of retinal neurons in culture. Figure 1 shows the morphology of neurons in cultures incubated in the absence or presence of the NO donor, S-nitroso-acetyl-d-l-penicillamine (SNAP) (Figs 1a and c), and when cultures were re-fed with fresh medium in both situations (Figs 1b and d). Changes in cellular morphology can be observed and extensive cell death induced by re-feeding (Fig. 1b). However, when cells were pre-treated with SNAP, excellent neuroprotection was observed (Fig. 1d). In some cases, control cultures treated with SNAP showed an increase in cell number when compared with untreated cultures (compare Figs 1a and 1c). Interestingly, SNAP also produced an increase of neurite outgrowth in control and re-fed cultures (Figs 1c and d, arrows). Neuronal death promoted by re-feeding was about 65% in relation to the control. Another structurally different NO donor (GSNO, 300 µm) also inhibited neuronal death induced by re-feeding (Fig. 2a). Although with different chemical structures and different NO release capacities, no significant difference was observed among these NO donors. NAP, a product of SNAP after prolonged exposure to light, did not show any protective effect in our cultures (Fig. 2a). Other donors, such as 3-morpholinosydnonimine (SIN 1) or sodium nitroprussiate, were tested but found to be toxic to the cultures. l-arginine (1 mm), tested in order to determine whether stimulation of NOS and production of endogenous NO would be able to produce the survival effect, did promote an effect very similar to that promoted by SNAP (Fig. 2a).

Figure 1.

 Morphology of chick retina neurons in 4-day-old purified cultures (C4). (a, c) Control cultures incubated in the absence (a) or presence (c) of SNAP (100 µm) from C1 to C4. (b, d) Cultures incubated in the absence (b) or presence (d) of SNAP were re-fed with fresh medium at C3 and incubated up to C4. Notice the cell death and loss of cell processes induced by re-feeding medium (b), but the neuroprotection observed when cultures were pre-incubated with SNAP (d). Control cultures incubated with SNAP but not re-fed with fresh medium showed, in this case, an increase in cell number and extensive neurite outgrowth (c, arrows). Scale bar = 40 µm.

Figure 2.

 NO donors prevent cell death induced by re-feeding cultures with fresh medium. (a) Cultures were pre-treated at C1 with the NO donors SNAP (100 µm) or GSNO (300 µm), and also with NAP, a product of SNAP after NO release or l-arginine (1 mm). Some cultures were re-fed with fresh medium at C3 and further incubated up to C4, when they were fixed and the number of cells estimated. Control cultures were not re-fed and the number of cells was also estimated at C4. No significant effects of drugs were observed in control cultures not re-fed with fresh medium. The results obtained with each drug were normalized to 100%. Each value represents the mean ± SEM of four independent experiments performed in duplicate. Number of cells in control cultures was 103 ± 6 cells/mm2. (b) Cultures were pre-treated with SNAP as in (a) but processed for MTT assay. Each value represents the mean ± SEM of three independent experiments performed in quadruplicate; **p < 0.001 when compared with cultures re-fed with fresh medium (RF).

Neuroprotection by SNAP was also measured by the MTT reaction and, as observed in Fig. 2(b), produced essentially the same results, with a 50% reduction of dye detection in cultures re-fed with fresh medium and a complete recovery in cultures pre-treated with SNAP.

Kinetics of the effects of SNAP on cell survival

In order to determine whether the neuroprotective effect of SNAP on retinal neuronal cultures exposed to re-feeding medium was concentration-dependent, we constructed dose–response curves. C1 cultures were treated with different concentrations of SNAP (70, 90, 100, 120, 150, 170 and 200 µm). We observed that concentrations of 100 or 120 µm produced maximal neuroprotection (Fig. 3a). Interestingly, a decrease in cell number, 50% and 85% below control, was observed when cultures were treated with 170 or 200 µm, respectively, a phenomenon which became even more pronounced in concentrations above 250 µm (not shown). In the following experiments, we used treatments with 100 µm SNAP.

Figure 3.

 Neuroprotective effect of SNAP is concentration- and time-dependent. (a) Cultures were incubated with increasing concentrations of SNAP from C1 to C3, re-fed with fresh medium containing the same concentrations of SNAP, and then fixed. (b) Cultures were treated with SNAP (100 µm) for different periods of time before re-feeding at C3. SNAP (0 h) indicates that SNAP was added only in the re-feeding medium and SNAP (48 h) indicates that SNAP was added at C1, incubated up to C3 and re-fed with fresh medium containing SNAP. All cultures were fixed at C4, i.e. 24 h after re-feeding. Data represent the mean ± SEM of three independent experiments performed in duplicate. Number of cells in control cultures was 92 ± 1 cells/mm2; **p < 0.001 and *p < 0.05 when compared with cultures re-fed with fresh medium (RF).

We investigated the temporal kinetics of SNAP in mediating neuroprotection. The results shown in Fig. 3(b) indicate a total blockade of neuronal death in cultures pre-treated with SNAP for 24 or 48 h. Pre-incubation for 12 h caused a partial blockade, whereas incubation for 6 h, or SNAP treatment at the time of re-feeding (0 h), produced no neuroprotection.

Neuroprotective effect of SNAP is blocked by PTIO

It has been demonstrated that excessive production of NO (nitrosative stress) induces the formation of peroxinitrite (ONOO) through the reaction between NO and superoxide anion (inline image). In its turn, ONOO promotes nitration of proteins, producing cell damage (Radi et al. 2002) or neuroprotection (García-Nogales et al. 2003; Bolaños et al. 2004). To verify the direct participation of NO in our purified cultures, we decided to investigate the effect of 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (PTIO), a substance considered to be a NO ‘scavenger’. Purified cultures of neuronal cells were incubated with 50 µm PTIO, a concentration that does not produce cell death, 30 min before treatment with SNAP. As previously observed, SNAP blocked the cell death induced by re-feeding, but PTIO was able to suppress this neuroprotective action (Fig. 4). However, as carboxy-PTIO was able to inhibit some actions attributed to peroxinitrite as tyrosine nitration (Pfeiffer et al. 1997), we cannot rule out the possibility that the effect of SNAP is mediated by peroxinitrite. Although we were unable to use higher concentrations of uric acid, a peroxinitrite scavenger, because of its toxicity to retinal neurons, the concentration of 0.5 µm, which is near its Ki in some systems (Kooy et al. 1994), had no significant effect, suggesting that NO, and not peroxinitrite, is the neuroprotective agent.

Figure 4.

 NO scavenger blocks neuroprotection elicited by SNAP. C1 cultures were pre-incubated with 50 µm PTIO, a scavenger of NO, 30 min before addition of SNAP (100 µm). Number of cells in control cultures was 98 ± 2 cells/mm2. Data represent the mean ± SEM of four independent experiments performed in duplicate; **p < 0.001 when compared with cultures re-fed with fresh medium (RF). No significant difference was detected between cultures re-fed with fresh medium pre-incubated in the absence or presence of PTIO.

Measurement of nitrite production from SNAP

In order to evaluate the production of NO from SNAP, we performed experiments to measure nitrite production after incubating the culture medium with increasing concentrations of SNAP for different periods of time. Maximal nitrite levels were observed as early as 3 h after addition of 100 µm SNAP (Fig. 5a), and nitrite production was linear, as a function of SNAP concentration in the medium, after 24 h of incubation (Fig. 5b). Moreover, intracellular and extracellular nitrite levels were maximal after 30 or 60 min of incubation with 100 µm SNAP, respectively, and remained approximately constant up to 24 h of incubation (Figs 5c and d).

Figure 5.

 Measurement of nitrite production by SNAP in culture medium or cultured retinal cells. Culture medium in the absence of cells was incubated at 37°C for different periods of time with SNAP (100 µm) (a), or for 24 h with different concentrations of SNAP (b). In (c) and (d) SNAP (100 µm) was added to cultures at C1 and incubated for the indicated times. The medium was removed, the cells lysed, and the medium and cell extract processed for measurement of extracellular or intracellular nitrite concentrations, respectively. Data represent the mean ± SEM of three (a, c and d), or ± deviation from the mean of two (b) independent experiments in which nitrite concentrations were measured in triplicate. The error bars in (a) and (b) were smaller than the symbol size.

Prolonged exposure to SNAP blocks cell death by a pathway different from NBMPR

In the light of data from our laboratory showing that prolonged exposure to NBMPR (an adenosine transport inhibitor) also regulates the death produced by re-feeding retinal neuronal cultures (Paes-de-Carvalho et al. 2003), we decided to investigate whether the effect of NO would involve the regulation of adenosine uptake and whether there was some cross-talk between these two neuroprotective agents.

Cultures at C1 were incubated with or without [3H]-adenosine (0.5 µCi/mL) plus SNAP (100 µm) or NBMPR (5 µm) (Fig. 6). At C4, cultures incubated with [3H]-adenosine were lysed and the intracellular radioactivity determined by liquid scintillation. Cultures incubated without [3H]-adenosine were fixed and the number of cells determined. As expected, both NBMPR or SNAP promoted a protective effect (Fig. 6a). As shown in Fig. 6(b), SNAP did not block the uptake of [3H]-adenosine, opposite to the effect observed with NBMPR. These data suggest that the neuroprotection promoted by NO is not mediated by blockade of adenosine uptake.

Figure 6.

  (a) SNAP and NBMPR have similar neuroprotective effects. Cultures at C1 were pre-incubated with SNAP (100 µm) or NBMPR (5 µm) before re-feeding at C3, and the number of cells was determined at C4. Data represent the mean ± SEM of five independent experiments performed in duplicate; **p < 0.001 when compared with cultures re-fed with fresh medium (RF). (b) NBMPR, but not SNAP, blocks adenosine uptake. Cultures were pre-incubated with NBMPR (5 µm) or SNAP (100 µm) for 30 min before adding [3H]-adenosine (0.5 µCi/mL) and further incubated up to C4, when cultures were washed, lysed and the intracellular radioactivity estimated. Data represent the mean ± deviation from the mean from three independent experiments performed in duplicate; *p < 0.05 when compared with control. The symbols (•) without bars represent values in which the SEM was too small to be represented.

Neuroprotective action of NO is associated with cyclic GMP production

Previous work using NO donors has shown an increase of neurite outgrowth mediated by activation of soluble guanylate cyclase (sGC) and PKG in PC12 cells (Hindley et al. 1997). With the aim of characterizing the role of the NO/cGMP system, we incubated C1 cultures with zaprinast (10 µm), a selective inhibitor of phosphodiesterase type 5 (PDE5) that degrades cGMP, or the guanylate cyclase activators YC-1 or 8bromo cyclic guanosine monophosphate (8Br-cGMP). Pre-incubation for 48 h with each of these compounds before re-feeding the cultures blocks cell death in a similar way to SNAP (Fig. 7a), suggesting that the effect of NO is mediated by the production of cGMP and activation of sGC. To corroborate this possibility, we investigated the effects of two inhibitors of sGC, ODQ and LY83583. Neuroprotection induced by SNAP was inhibited by pre-incubation with both ODQ (1 µm) or LY83583 (1 µm) (Fig. 7b). On the other hand, treatment with SNAP plus zaprinast did not show any additive results (data not shown). We also tested the effect of KT5823 (0.4 µm), an inhibitor of the serine-threonine cyclic GMP-dependent kinase (PKG). As expected, pre-incubation with this compound 30 min before treatment with SNAP resulted in a complete blockade of the neuroprotective effect (Fig. 7b).

Figure 7.

 Neuroprotection by SNAP is mediated by the guanylate cyclase/PKG pathway. Cultures at C1 were pre-incubated for 30 min in (a) with 10 µm zaprinast (Zap), a cGMP phosphodiesterase inhibitor, 20 µm 8Br-cGMP, a permeable cGMP analog, or 15 µm YC-1, a guanylate cyclase activator, and in (b) with 1 µm ODQ or 1 µm LY83583, guanylate cyclase inhibitors, or 0.4 µm KT5823, a PKG inhibitor, before treatment with SNAP, and further incubated up to C3 before re-feeding with fresh medium containing the drugs. At C4, the number of cells was estimated. No significant difference was detected between cultures re-fed with fresh medium pre-incubated in the absence or presence of the inhibitors. (c) Time profile of the survival effect by 15 µm YC-1. These experiments were performed exactly as explained for SNAP in the legend of Fig. 3. Data represent the mean ± SEM of three independent experiments performed in duplicate. Number of cells in control cultures was 118 ± 9 cells/mm2 in (a) and (b) and 107 ± 4 cells/mm2 in (c); **p < 0.001 and *p < 0.05 when compared with cultures reed with fresh medium (RF).

One important question related to the survival effect of SNAP was whether its temporal profile (see Fig. 3b) was related to the kinetics of NO release from SNAP or to the time necessary for the activation of survival pathways. The kinetics of NO release indicated that NO was released from SNAP during the first 3 h of incubation (see Fig. 5). We then performed experiments to verify the temporal kinetics of the survival effect by the guanylate cyclase activator, YC-1, and compare it with that of SNAP. As observed in Fig. 7(c), the YC-1 profile was similar to that of SNAP, with a difference only at the time point of 6 h. The maximal effect with SNAP as well as with YC-1 was obtained with 24 h of pre-treatment (compare Figs 3b and 7c).

Protein kinases participate in the neuroprotective effect of SNAP

Evidence from the literature indicates the participation of PI3K and its downstream target, AKT, in cell death by apoptosis, and in the regulation of cell division, proliferation, differentiation and survival (Datta et al. 1999). Therefore, to investigate the participation of PI3K in the neuroprotective action of NO, we decided to study the influence of two inhibitors of this enzyme. Figure 8(a) shows that treatment with LY294002 (1 µm) or Wortmannin (0.8 µm), two potent inhibitors of PI3K, 30 min prior to SNAP blocked the neuroprotection by SNAP, suggesting that PI3K has a relevant role in cellular survival induced by this NO donor. The effect by YC-1 was also blocked by both PI3K inhibitors (Fig. 8b).

Figure 8.

 Neuroprotection by SNAP or YC-1 also involves PI3K, MEK, CAMK and PKA. Cultures at C1 were pre-incubated for 30 min in (a) and (b) with 1 µm LY294002 (LY) or 0.8 µm wortmannin (Wor), PI3K inhibitors, or in (c) with 10 µm PD98059 (PD), a MEK inhibitor, 1 µm KN93, a CAMK inhibitor, or 1 µm H89, a PKA inhibitor, before treatment with SNAP (100 µm) or YC-1 (20 µm) and further incubation up to C3 before re-feeding with fresh medium containing the drugs. At C4, the number of cells was estimated. No significant effects of the inhibitors were observed in control or re-fed cultures. Data represent the mean ± SEM of three independent experiments performed in duplicate. Number of cells in control cultures was 98 ± 1 cells/mm2; **p < 0.001 when compared with cultures re-fed with fresh medium (RF).

With the aim of testing the possibility of involvement of ERK1/2 and CaMK in the regulation of survival induced by SNAP, we decided to study the influence of inhibitors of these two kinases. Incubation of cultures with 10 µm PD98059, an inhibitor of MEK, or 1 µm KN93, an inhibitor of CaMK, did not show any cell death in control conditions (data not shown). However, a complete inhibition of the neuroprotective effect of SNAP was observed (Fig. 8c), raising the possibility that the PI3K, MEK and CaMK pathways are involved in the neuroprotective action of SNAP. Previous data from our laboratory showed that adenosine also prevented cell death induced by excitotoxicity of glutamate or re-feeding the cultures through activation of A2a adenosine receptors (Ferreira and Paes-de-Carvalho 2001; Paes-de-Carvalho et al. 2003). In order to verify the involvement of PKA in the regulation of cell survival induced by SNAP, we used H89 (1 µm), an inhibitor of this enzyme. This compound also blocked the neuroprotective effect of SNAP when cultures were re-fed with fresh medium (Fig. 8c), suggesting that cell survival induced by NO also depends on PKA activity.

Cell death produced by re-feeding with fresh medium involves apoptosis

Death by apoptosis involves the activation of caspases induced by oxidative stress or other stimuli, such as tumor necrosis factor α (TNF-α), withdrawal from neurotrophic factors or glucose deprivation (Ha et al. 2003). Oxidative stress also generates superoxide anions to form peroxynitrite, which, in turn, could form H2O2 in our experimental conditions. With the aim of characterizing the type of cell death promoted by re-feeding of cultures, we decided to investigate whether treatment with H2O2 (100 µm), an inducer of oxidative stress which leads to apoptosis (Lin et al. 2004), would be regulated by SNAP. Cultures were pre-incubated or not with SNAP and treated or not with H2O2 (100 µm) for 1 h, then fixed and the number of cells determined. As observed in Fig. 9(a), incubation with H2O2 did promote cell death comparable with that produced by culture re-feeding with fresh medium, and this effect was completely prevented by SNAP. Concentration curves for H2O2 indicated that an optimal concentration was 100 µm, which induced death of over 85% of the neurons (data not shown). Importantly, cell death induced by H2O2 was prevented by BAF, a pan- inhibitor of caspases (Fig. 9a).

Figure 9.

  (a) Cell death induced by hydrogen peroxide is prevented by SNAP or BAF. Cultures were pre-incubated at C1 with SNAP (100 µm) or BAF (50 µm) and treated at C3 with 100 µm hydrogen peroxide (H2O2) for 1 h. (b) Cyclosporin A or BAF block cell death induced by re-feeding. Cultures at C1 were pre-treated with 0.3 µm cyclosporin A (Csp-A) or BAF (50 µm), re-fed with fresh medium containing the drug and incubated up to C4 before fixation and estimation of number of cells. Data represent the mean ± SEM of four independent experiments performed in duplicate. Number of cells in control cultures was 98 ± 7 cells/mm2; **p < 0.001 and *p < 0.05 when compared with cultures re-fed with fresh medium (RF).

Several studies on the mechanisms of apoptosis have shown that the main organelles involved in life or death decisions are the mitochondria in situations of oxidative stress when pro-apoptotic molecules, such as cytochrome c and Bad, are liberated (Hansson et al. 2003). Cytochrome c release depends on the formation of a mitochondrial channel termed the mitochondrial permeability transition pore (mPT) (Desagher and Martinou 2000). To corroborate the hypothesis that NO blocks apoptosis produced by re-feeding the cultures, we used cyclosporine-A (Csp-A), an immunosuppressive agent which blocks the opening of the transition pore, or the caspase inhibitor, BAF. Cultures pre-incubated with Csp-A (0.3 µm) or BAF (50 µm) showed a very significant blockade of cell death (Fig. 9b). On the other hand, no additive effect was observed when cultures were treated with Csp-A plus SNAP (data not shown).

Discussion

NO is an important intercellular signalling molecule during CNS development, participating in phenomena such as neurite outgrowth, synaptic formation and cell proliferation (Cheng et al. 2003). Purified cultures of retinal neurons and photoreceptors appear to be a suitable model for the study of events occurring during development and in the search for the factors regulating neuronal survival. A simple change of medium at the appropriate stage of culture is able to promote extensive cell death. Previous work has shown that pre-incubation of cultures with adenosine A2a receptor agonists or permeable cyclic AMP analogs blocks this cell death (Paes-de-Carvalho et al. 2003). In the present work we have demonstrated that neuroprotection is also observed when cultures are pre-incubated with the NO donors SNAP or GSNO. This effect is inhibited by PTIO and guanylate cyclase or PKG inhibitors, strongly suggesting that it is mediated by the classical NO–guanylyl cyclase–cyclic GMP–PKG pathway. Additionally, in a search for possible pathways involved in this effect of NO, we found that the effect is time- and concentration-dependent and, interestingly, is also dependent on multiple pathways including AKT, ERK 1/2, PKA and CAM kinases. These results indicate that the production of NO during a critical period of development is an important signal for the regulation of neuronal survival.

Cell death by re-feeding cultures is by apoptosis?

The study of factors that regulate neuronal survival is of pivotal importance to understand the mechanisms of neurodegeneration that occur in several pathological conditions, as well as the shaping of neural connections during CNS development. It is well known that massive neuronal death takes place during a critical period of development, known as the period of programmed cell death (for review see Mey and Thanos 2000). This cell death occurs by apoptosis, a very complicated and multi-step process that involves a series of enzymatic processes triggered by the opening of a transition pore in the mitochondria (mTP), release of cytochrome c and activation of caspases (for review see Evans 1993; Desagher and Martinou 2000).

Our previous work (Paes-de-Carvalho et al. 2003) and this study shows that extensive cell death is promoted after re-feeding the cultures with fresh medium. The cause of this cell death is presently unknown. One interesting fact is that this death does not occur if we change the medium to a conditioning medium obtained from a ‘sister’ culture (Paes-de-Carvalho et al. 2003), suggesting that factors produced by the cultured cells and released in the medium are important for cell survival, as shown in many other systems (Abiru et al. 1998; Davies 2003). Pre-incubation of cultures for at least 24 h with adenosine (Ferreira and Paes-de-Carvalho 2001; Paes-de-Carvalho et al. 2003) or the NO donors, SNAP or GSNO (present work), was able to completely prevent the cell death. This pre-incubation time is necessary for the survival effect to take place, indicating the involvement of long-term metabolic events, probably protein synthesis. Involvement of PKG and other pathways in the effect of SNAP suggests the possibility that NO is inducing anti-apoptotic genes such as BCL2 that, in turn, are protecting the cells and inhibiting formation of the mTP (Takuma et al. 2001; Domanska-Janik et al. 2004).

Cell death induced by H2O2, which is known to be caused by apoptosis, could also be prevented by SNAP or cyclosporin A (Csp-A), an inhibitor of mTP opening (Takuma et al. 2001; Hansson et al. 2003). Indeed, Csp-A also promoted neuroprotection when cell death was triggered by re-feeding the cultures. Together with the data showing that this cell death is also blocked by BFA, a pan-inhibitor of caspases, these results strongly suggest that cell death induced by re-feeding is by apoptosis and involves the opening of the mTP. This hypothesis is also corroborated by our experiments using the MTT reaction, which is lost when cultures are re-fed and maintained when cultures are pre-treated with SNAP (Fig. 2b).

NOS, H2O2 production and apoptosis

In our present experiments, the concentration of SNAP used was important for neuroprotection, with concentrations above 120 µm producing cell death instead of neuroprotection. Several studies have shown that NO can be a neuroprotector or inducer of cell death (Moncada et al. 1991), depending on the NO donor used or its concentration (Nicotera et al. 1997; Murphy 1999; Figueroa et al. 2005; Martin et al. 2005). Interestingly, many authors also show that NOS inhibitors are neuroprotective (for example see Geyer et al. 1995). In many cases, generation of peroxinitrite by the reaction of NO with anion superoxide appears to be the causative agent of death. However, other studies indicate peroxinitrite as a neuroprotective agent (García-Nogales et al. 2003; Bolaños et al. 2004). The experiments discussed in the present work suggest that the neuroprotective substance is NO, and that the mechanism involves the guanylate cyclase–cyclic GMP–PKG pathway, as the protection is blocked by the NO scavenger PTIO, mimicked by the guanylate cyclase activators YC-1 and 8Br-cGMP, and inhibited by guanylate cyclase or PKG inhibitors. Further evidence that the protective effect is mediated by NO and not peroxynitrite is that the addition of SIN1, a superoxide donor that generates peroxinitrite, promotes only cell death in our cultures (not shown). Addition of l-arginine was also able to protect the cells, indicating that endogenous NO produced by NOS is able to produce the same effect as the addition of SNAP.

Long-term pre-treatments promote survival

As discussed above, the survival effect produced by SNAP is probably mediated by NO through the classical NO/sGC/PKG pathway. Indeed, the guanylate cyclase activator, YC-1, promotes a similar survival effect. The temporal profile of the survival effect of SNAP was slow and first observed when cultures were pre-treated for 12 h with the NO donor (Fig. 3b). This could be related to the kinetics of NO release from SNAP, as this compound has a half-life of 30–40 h depending on the cell system studied (Ferrero et al. 1999). However, our measurements of nitrite production showed that NO release from SNAP is complete after 3 h in the presence or absence of cells (Fig. 5). The temporal profile of YC-1 showed a pattern similar to SNAP, the maximal effect being obtained after pre-incubation for 24 h with this compound (Fig. 6c). These results strongly suggest that a long-term pre-treatment with NO is necessary for the survival effect in a way similar to that previously found for adenosine (Paes-de-Carvalho et al. 2003). However, the results of experiments measuring nitrite production also indicate that a short-term incubation with SNAP is sufficient to trigger the events that lead to cell survival. Why long-term pre-treatment is needed for the survival effects promoted by either NO or adenosine is presently unknown. Our hypothesis is that the phenomenon is related to the induction of gene transcription and accumulation of anti-apoptotic proteins.

A hypothetical direct pathway to CREB?

From our present experiments it is clear that the neuroprotection induced by SNAP or GSNO is mediated by the classical NO–guanylate cyclase–cyclic GMP–PKG pathway. Preliminary experiments in our laboratory also show that SNAP protects retinal cells in culture from death induced by exposure to glutamate, and that this effect is mimicked by cyclic GMP analogs (unpublished observations). NO also inhibits proliferation of chick retinal glial cells in mixed or purified cultures but in that case, the effect is mediated by a cyclic GMP-independent mechanism (Magalhães et al. 2006). In the case of the survival effect reported here, many intracellular kinase pathways appear to be involved, including PKA, PI3K/AKT, MEK/ERKs and CAMKs. As we have previously demonstrated a similar survival effect by activating adenosine A2a receptors or blocking adenosine uptake (Paes-de-Carvalho et al. 2003), we have tested whether SNAP could block adenosine uptake and promote the survival through this mechanism. Although the effect produced by SNAP is inhibited by H89, a PKA blocker, it does not involve inhibition of adenosine uptake as we have shown that this NO donor does not inhibit adenosine uptake in the cultures (Fig. 6). However, we cannot rule out the possibility that NO increases the release of adenosine, which is able to activate A2a receptors and stimulate neuroprotection. Involvement of CAMKs could also be explained in the same way, as we have recently demonstrated that the release of adenosine is regulated by these kinases in retinal cultures (Paes-de-Carvalho et al. 2005).

From our experiments it is also clear that the involvement of PI3K and ERK1/2 in the effect of NO as inhibitors of these kinases completely prevents the survival promoted by SNAP or YC-1. Our preliminary experiments indicate that SNAP promotes increases in the phosphorylation of AKT, ERK1/2 and CREB. Together with the data showing that anti-apoptotic genes are under the control of CREB (Ciani et al. 2002), we can propose a model in which NO first activates guanylate cyclase, then increases cyclic GMP levels and PKG activity. Subsequent activation of PI3K and AKT is possible, as shown in PC12 cells (Ha et al. 2003). ERK1/2 could also be subsequently activated, and it is well known that this pathway is able to activate CREB (Wu et al. 2005). In parallel, another pathway could involve the induction of adenosine release by NO, as shown for other neurotransmitters (Yu and Eldred 2005). Adenosine could then activate PKA and CREB to promote expression of anti-apoptotic genes (Das et al. 2005). NO was also shown to activate P/Q-type calcium channels and increase calcium influx (Chen et al. 2002). Calcium is then able to combine with calmodulin and activate CAMKs. It is also well known that CAMK IV phosphorylates CREB at distinct sites compared to PKA (Kornhauser et al. 2002). The involvement of calcium channels in the protective effect by NO is also suggested by our recent data showing that SNAP promotes a rapid increase of calcium influx in retinal cultures (not shown). As the effect of NO is blocked by inhibiting each of the different kinases, we propose that all three pathways end in the phosphorylation of CREB, induction of transcription of anti-apoptotic proteins, such as BCL2, and blocking of mTP opening, cytochrome c release and activation of caspases.

In conclusion, we found in the present work that the NO donors SNAP or GSNO promote neuroprotection of retinal cells in culture through a mechanism involving the classical NO–guanylate cyclase–PKG pathway, and also involving multiple kinase pathways including PI3K, MEK, PKA and CAMKs. Furthermore, the data strongly suggest that NO blocks apoptosis induced by re-feeding the cultures or by exposure of cells to H2O2.

Acknowledgements

We would like to thank Luzeli Ribeiro de Assis for technical assistance, Célio Lobo Jr for help with the initial experiments, and Dr Maurício Verícimo and Dr Armando P. do Nascimento for help with nitrite measurements. This work was supported by CAPES, CNPq, FAPERJ and PRONEX/MCT. TAM-G is the recipient of a graduate student fellowship from CAPES and RPC is a research fellow from CNPq.

Ancillary