Prevention of NMDA-induced death of cortical neurons by inhibition of protein kinase Cζ


Address correspondence and reprint requests to Jari Koistinaho at Department of Neurobiology, A.I.Virtanen Institute for Molecular Sciences, University of Kuopio, Neulaniementie 2, P.O. Box 1627, FIN-70211 Kuopio, Finland. E-mail:


Excitotoxicity through stimulation of N-methyl-d-aspartate (NMDA) receptors contributes to neuronal death in brain injuries, including stroke. Several lines of evidence suggest a role for protein kinase C (PKC) isoforms in NMDA excitotoxicity. We have used specific peptide inhibitors of classical PKCs (α, β, and γ), novel PKCs δ and ε, and an atypical PKCζ in order to delineate which subspecies are involved in NMDA-induced cell death. Neuronal cell cultures were prepared from 15-day-old mouse embryos and plated onto the astrocytic monolayer. After 2 weeks in vitro the neurons were exposed to 100 µm NMDA for 5 min, and 24 h later the cell viability was examined by measuring the lactate dehydrogenase release and bis-benzimide staining. While inhibitors directed to classical (α, β, and γ) or novel PKCs (δ or ε) had no effect, the PKCζ inhibitor completely prevented the NMDA-induced necrotic neuronal death. Confocal microscopy confirmed that NMDA induced PKCζ translocation, which was blocked by the PKCζ inhibitor. The NMDA-induced changes in intracellular free Ca2+ were not affected by the peptides. In situ hybridization experiments demonstrated that PKCζ mRNA is induced in the cortex after focal brain ischemia. Altogether, the results indicate that PKCζ activation is a downstream signal in NMDA-induced death of cortical neurons.

Abbreviations used

α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid


intracellular free Ca2+ concentration




days in vitro


embryonic day


fetal bovine serum heat-inactivated


horse serum heat-inactivated


lactate dehydrogenase


middle cerebral artery


minimum essential medium


nuclear factor kappa-B




protein kinase C



Glutamate is an essential amino acid neurotransmitter in the mammalian central nervous system, but when its extracellular concentration increases above normal level, overstimulation of N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA)/kainate receptors results in sustained Ca2+ influx initiating a toxic metabolic cascade, which eventually leads to neuronal death (Choi 1994). This phenomenon, known as excitotoxicity, contributes to acute brain injuries, including brain ischemia, as antagonists of both NMDA and AMPA/kainate receptors provide protection in rodents when applied prior to or immediately after occlusion of the middle cerebral artery (MCA; Gill et al. 1992; Iijima et al. 1992). Even though toxic activation of glutamate receptors after acute brain insults occurs too soon to have clinical relevance, the activated pathways downstream to the glutamate receptors within the damaged neuron may involve enzymes, which are crucial to excitotoxicity and which may be important for understanding the mechanisms of neuronal death in acute brain diseases. In particular, both tyrosine kinases and serine/threonine kinases, such as protein kinase C (PKC) appear to play such a role.

Conflicting roles for PKC in neuronal (Hara et al. 1990; Mattson 1991; Durkin et al. 1997; Villalba 1998) and non-neuronal (Emoto et al. 1995; Li and Pierce 1996; Pongracz et al. 1999; Shizukuda and Buttrick 2002) cell death have been reported. The exact role of PKC clearly depends on the characteristics of the cell type and the type of external stimulation, but most importantly, the functions appear to be PKC subspecies-specific. The PKC family consists of at least a dozen structurally related phospholipid-dependent serine/threonine protein kinases, which are classified into the three distinct subgroups based on their activation requirements: the conventional (cPKC), novel (nPKC), and atypical (aPKC; Nishizuka 1986, 1995; Tanaka and Nishizuka 1994; Hofmann 1997). The cPKC subgroup contains α, βI, βII, and γ isoforms, which can be activated by calcium (Ca2+), phosphatidylserine (PS), other phospholipids and diacylglycerol (DAG) or phorbol esters. The nPKC isoforms δ, ε, θ, and η/L (mouse/human) are Ca2+-independent isozymes; however, they require PS and DAG or phorbol esters for efficient activation. The subgroup of aPKCs ζ and λ/ι (mouse/human) isoforms (Selbie et al. 1993; Akimoto et al. 1994) require PS for activation. Importantly, this aPKC subfamily is also Ca2+-independent, like nPKCs, but does not respond to DAG or phorbol esters.

Of the multiple PKC subspecies, at least α (Okuda et al. 1999), β (Pandey et al. 2000), δ (Emoto et al. 1995; Villalba 1998; Pongracz et al. 1999), and ζ (Shizukuda and Buttrick 2002) subspecies have been shown to promote cell death under certain experimental conditions. Especially PKCδ, which is a substrate of caspase-3 (Emoto et al. 1995; Ghayur et al. 1996), a proapoptotic enzyme, has been previously implicated in apoptotic death of non-neuronal (Emoto et al. 1995; Koriyama et al. 1999; Pongracz et al. 1999) and neuronal cells (Villalba 1998; Anantharam et al. 2002) and is induced after acute brain ischemia (Miettinen et al. 1996; Koponen et al. 2000). On the other hand, signaling through PKCζ turns on nuclear factor kappa-B (NF-κB; Chang et al. 2002), a transcription factor regulating neuronal cell death. Because there is little understanding of which PKC subspecies may mediate excitotoxicity, we studied the role of cPKCs, PKCδ, PKCε, and PKCζ in NMDA-induced neuronal death by using specific inhibitor peptides of these subspecies. While inhibition of other PKC subspecies had no effect, the peptide inhibiting PKCζ activation completely prevented the neuronal death without altering the NMDA-induced calcium influx when measured with fura-2 cellular fluorescence imaging. Because mRNA expression of PKCζ was also induced in mouse cerebral cortex after focal ischemia, PKCζ may mediate neuronal death in diseases involving NMDA excitotoxicity.

Experimental procedures

Primary cortical cell cultures

All experimental protocols were approved by the Animal Care and Use Committee of Kuopio University and follow the NIH guidelines for animal care. The neuronal cell cultures were prepared from the cortices of 15-day-old mouse embryos and plated onto a previously established astrocytic monolayer (Boulton et al. 1992). Briefly, astrocytic cultures were established from the cerebral cortices of newborn mouse pups (CD1, Stanford University, Palo Alto, CA, USA; Balb/C, University of Kuopio, Kuopio, Finland). After aseptic dissection, the cortices were mechanically dissociated and passed through the 70-µm and 40-µm Cell Strainers (Falcon, Becton Dickinson Labware, Franklin Lakes, NJ, USA) in order to obtain cultures free of vascular elements. The astrocytes were incubated in minimum essential medium (MEM) 1× (Gibco-BRL, Life Technologies, Rockville, MD, USA) in presence of 20% heat-inactivated fetal bovine serum (FBS-HI; Gibco-BRL, Life Technologies) at cell density 3 × 104 viable cells/cm2 at 37°C in a humidified 5% CO2 atmosphere. The fresh cell culture medium with MEM/20% FBS-HI was replaced after 3 days. Thereafter, the astrocytic cultures were fed two to three times per week with MEM/10% FBS-HI for 2 weeks.

To plate neuronal cells on top of the astrocytic cell layer, minced cortices were incubated in 0.25% trypsin (Gibco-BRL, Life Technologies) diluted in Earle's balanced salt solution (Gibco-BRL, Life Technologies) for 2 min at 37°C and the trypsinization reaction was stopped by adding 1/10 of total volume heat-inactivated horse serum (HS-HI, Gibco-BRL, Life Technologies). The cortices were centrifugated for 10 min at 620 g (Function Line Labofuge 400e, Heraeus Instruments, Osterode, Germany) and thereafter, the cell pellet was resuspended into 10 mL of MEM. The cell suspension was filtrated through 70-µm and 40-µm Cell Strainers. The cell density was counted with Burker hemocytometer and diluted in the density of approximately 3–4 × 105/mL in MEM supplemented with 10% FBS-HI and 10% HS-HI. The diluted cell suspension was plated on top of the astrocytic monolayer. The 24-well plates (0.5 mL/well) were used for lactate dehydrogenase (LDH) release measurements, 6-well plates (2 mL/well) containing coverslips for calcium-imaging, and 2-well chamber slides (2 mL/well) were used for confocal microscopy experiments. After 5 days in vitro (DIV) the neurons were treated with MEM/10% FBS-HI + 10% HS-HI + 10 µm cytosine β-d-arabinofuranoside (Sigma Chemicals, St Louis, MO, USA) to eliminate any mitotic cells, such as microglia or oligodendrocytes. In addition, at 7 DIV and 11 DIV, the fresh MEM/10% HS-HI was replaced. Neurotoxicity was studied at 14 DIV.

NMDA neurotoxicity experiments

The dose–response curve of NMDA neurotoxicity was determined by exposing the cultures for 5 min to various concentrations of NMDA (50, 100, 200, 500 and 1000 µm; Sigma), which was added onto the cells in a HEPES-buffered solution containing 120 mm NaCl, 5.4 mm KCl, 0.8 mm MgCl2, 1.8 mm CaCl2, 20 mm HEPES, 15 mm glucose, and 0.01 mm glycine. Neuronal cell death was assessed by measuring the release of LDH 24 h after the NMDA exposure. Results were presented as a percentage of total neuronal death (100%), which was achieved by exposing the culture to 300 µm NMDA for 24 h. In three separate trials, the dose triggering about 50% of neuronal cell death without causing any glial damage was found to be 100 µm NMDA (Fig. 1), which was then chosen for further experiments to study the effect of PKC inhibitor peptides on NMDA neurotoxicity.

Figure 1.

The dose–response curve of NMDA neurotoxicity. A 5-min exposure to 100 µm NMDA triggered about 50% neuronal death without causing glial damage in the experiments. LDH release was measured 24 h after the 5-min exposure. The total neuronal death (100%) was achieved by incubating cultures with 300 µm NMDA for 24 h. Results are presented as percentage of the total neuronal death. The experiment was repeated at least three times.

PKC inhibitor peptide treatments in NMDA neurotoxicity experiments

The PKC inhibitor peptide was administered to the mouse mixed cortical culture 20 min prior to NMDA exposure in MEM/10% HS-HI or simultaneously with NMDA in HEPES buffer to block the action of the particular PKC isozyme. The PKC inhibitor peptide was present during the 5-min NMDA exposure as well as during the 24-h post-incubation with MEM/10% HS-HI containing 0.01 mm glycine.

The following PKC inhibitor peptides were used: cPKC inhibitor, βC2-4, a nonapeptide (Ser-Leu-Asn-Pro-Glu-Trp-Asn-Glu-Thr) corresponding amino acids 218–226 in the C2 domain of PKCβ contains the RACK binding site on cPKCs (PKCα, β and γ; Ron et al. 1995). δPKC inhibitor, δV1-1, a dekapeptide (Ser-Phe-Asn-Ser-Tyr-Glu-Leu-Gly-Ser-Leu) designed from the RACK-binding site of PKCδ (and corresponds to amino acids 8–17 within the V1/C2 domain of the isozyme; Chen et al. 2001). PKCε inhibitor, εV1-2, an octapeptide (Glu-Ala-Val-Ser-Leu-Lys-Pro-Thr) designed from the RACK-binding site of PKCε and corresponds to amino acids 14–21 within the V1/C2 domain of the isozyme (Johnson et al. 1996). PKCζ inhibitor, a 13-amino acids peptide (Ser-Ile-Tyr-Arg-Arg-Gly-Ala-Arg-Arg-Trp-Arg-Lys-Leu) designed from the pseudosubstrate sequence of human PKCζ corresponding the amino acids 113–125 (Berra et al. 1993; Laudanna et al. 1998). Scrambled PKCβC2.4 (sc) was used as the control peptide, as it does not contain the RACK-binding sequence of any PKC isoforms (Ron et al. 1995). All peptides were cross-linked via a cystein S-S bond to a Tat carrier peptide for delivery across biological membranes as described (Chen et al. 2001).

Bis-benzimide and immunocytochemical stainings

The mixed cortical cultures were fixed 24 h after the exposures with 4% paraformaldehyde in 0.1 m phosphate buffered saline (PBS) for 25 min and stained with bis-benzimide (5 µg/mL; Hoeschst 33342, Sigma) in 0.1 m PBS for 5 min at room temperature (RT) for confirming the LDH results and possible detection of apoptotic nuclear morphology.

For confocal microscopy studies the cultures were fixed with 4% paraformaldehyde 5 min after the exposure and stained with PKCζ polyclonal (1 : 300, rabbit; Santa Cruz Biotechnology, Santa Cruz, CA, USA) or neuron-specific microtubule-associated protein-2 (MAP-2) monoclonal antibody (1 : 300, mouse; Boehringer-Mannheim, Philadelphia, PA, USA) or double-stained with PKCζ and MAP-2 antibodies. Immunocytochemical localization of the PKCζ isozyme was determined using confocal microscopy (Leica CLSM, Leiz Diaplan, Heidelberg, Germany) using either cyanine-conjugated anti-mouse (1 : 60, Jackson ImmunoResearch, West Grove, PA, USA) or fluorescein isothiocyanate-conjugated anti-rabbit (1 : 60, Jackson ImmunoResearch) as secondary antibodies.

Intracellular calcium measurements

The effect of PKC inhibitor peptides on NMDA-induced changes in intracellular calcium concentration was monitored using the fluorescent Ca2+-indicator fura-2 (fura-2/AM, TEF Laboratories, Austin, TX, USA; Grynkiewicz et al. 1985). A 2-([2-hydroxy-l,1-bis{hydroxymethyl}ethyl]amino) ethane sulfonic acid (TES) buffered medium consisting of 137 mm NaCl, 5 mm KCl, 10 mm glucose, 1 mm CaCl2, 0.44 mm KH2PO4, 4.2 mm NaHCO3, 1.2 mm MgCl2 × H2O, 20 mm TES adjusted to pH 7.4 with NaOH was used in calcium imaging experiments. The cultures on coverslips were incubated with 4 µm fura-2/AM in 2 mL of TES-buffered medium for 30 min at 37°C in humified 5% CO2 atmosphere. The PKC inhibitor peptides were added to the TES-buffered medium during this pre-incubation and were present throughout the experiment. Subsequently, the coverslips were attached to the bottom of a thermostated (37°C) perfusion chamber and fixed with a rubber ring and a screw cap (Shariatmadari et al. 2001). The volume of the chamber was 200 µL and the cells were constantly perfused at a rate of 2 mL/min. The cells were excited by alternating wavelengths of 340 and 380 nm using narrow-band excitation filters and the fluorescence was measured through a 430-nm dichroic mirror and a 510-nm barrier filter with an InCytIm IM Fluorescence Imaging Systems (Cincinnati, OH, USA). One rationed image was acquired per second.


Data are expressed as mean ± SD. The comparisons were made with one-way analysis of variance followed by Newman–Keuls post-hoc test. Two-group comparisons were evaluated by two-tailed Student's t-tests.

Focal brain ischemia

The focal brain ischemia was produced by MCA occlusion using intraluminal nylon thread introduction method (Koizumi et al. 1986) with following modifications. The 3-month-old male mice (CD1) were anesthetized with 4% isoflurane (70% N2O and 30% O2) and isoflurane concentration was reduced to 2% during the surgery. The mouse temperature was maintained between 36.5 and 37.5°C with a heating pad. After a midline skin incision, the left external carotid artery was exposed and its branches were electrocoagulated. An 11-mm long surgical monofilament nylon suture (5–0, Dermalon, Davis + Geck, Seoul, Korea) was introduced into the left internal carotid artery through the external carotid artery stump and left at the origin of the MCA for 12 h. The animal was decapitated at 12 h and brain tissues were processed for in situ hybridization.

In situ hybridization

In situ hybridization was performed as previously described (Koponen et al. 1999). Briefly, the 10 µm frozen brain sections were cut on a cryostat (Leica CM 3000) at − 20°C, and placed on SuperFrost Plus slides (Fisher Scientific, Pittsburgh, PA, USA). The following PKC isoform specific oligonucleotide probes were used for in situ hybridization: 5′-CGGGGCCCAGCTTGGCTTTCTCGAACTTCTGCCTG-3′ (PKCα); 5′-CCTTGGTACCTTGGCCAATCTTGGCTCTCT-3′ (PKCβ); 5′-GAATGGGAGAGGAAGAGGGGCCCATCCGCACTCTC-3′ (PKCγ); 5′-AGACAGCTGTCTTCTCTCGAATCCCTGGTATATT-3′ (PKCδ); 5′-TAGACGACGAGGCTCGGTGCTCCTCTCCTCGGTTG-3′ (PKCε); 5′-GTCTGGGTGGCCAGCATCCCTCTCTGGCTGCTTGC-3′ (PKCζ).

A control probe possessed the corresponding GC-ratio and the length as the PKC probes, but had no homology with any known gene sequence.

The oligonucleotide probes were end-labeled with 35S-ATP using terminal deoxynucleotidyl transferase (MBI, Fermentas, Vilnius, Lithuania) and purified using a Probe Quant G50 column (Pharmacia Biotech, Uppsala, Sweden). For overnight incubation the following hybridization solution was used; 10 × 106 cpm/mL probe, 40 µL of 5 m dithiothreitol, 50 µL of salmon sperm DNA (10 mg/mL), and 900 µL of hybridization cocktail [50 mL of formamide, 20 mL of 20 × standard saline citrate, 2 mL of 50 × Denhardt's reagent, 10 mL of 0.2 m sodium phosphate buffer (pH 7.4), 10 g of dextran sulfate, 4 mL of 25% sarcosyl]. After hybridization, the sections were washed in 1 × standard saline citrate at 55°C for 2 h, rinsed for 2 × 5 min in deionized water at RT, dehydrated for 30 s in 60% and 90% ethanol, air dried, and covered with Kodak XAR-5 film (Kodak, Rochester, NY, USA) for 21 days.


The influence of specific PKC inhibitor peptides on NMDA neurotoxicity

Neuronal cortical cultures were challenged with NMDA for 5 min to induce cell death (Fig. 2). To test the role of PKC isozymes, selective PKC inhibitor peptides were added at 1 µm concentration to cultures 20 min before 100 µm NMDA exposure for 5 min and kept in the medium throughout the NMDA exposure and the 24-h follow-up time. When LDH release was quantified at 24 h, only the PKCζ inhibitor peptide had a protective effect on NMDA-induced neuronal cell death (Fig. 2a, p < 0.01, anova). The PKCζ inhibitor peptide prevented the neuronal death even when the NMDA concentration was as high as 1000 µm (Fig. 2b, p < 0.05, anova). No other PKC peptide studied had any significant effect on neuronal survival (Fig. 2a). However, when the PKCζ inhibitor peptide was administered immediately after the onset of NMDA exposure and kept in the medium throughout the 24 h follow-up time, no protection was achieved (45.2% and 66.1% neuronal death without and with the inhibitory peptide, respectively), indicating that either PKCζ is activated immediately after stimulation of NMDA receptors or penetration of the inhibitory peptide through the plasma membrane takes too long to be able to inhibit the target enzyme. In addition, the PKCζ inhibitor peptide was not able to protect neurons against NMDA excitotoxicity, which lasted for 24 h, possibly due to degradation of the peptide during the long incubation (Fig. 2b).

Figure 2.

The effect of PKC inhibitor peptides on NMDA neurotoxicity. (a) Cortical cultures were treated with 100 µm NMDA for 5 min with and without adding PKC inhibitor peptides (1 µm) as described in the text. Cell death was determined after 24 h by measuring LDH release. PKCζ inhibitor peptide completely prevented the NMDA-induced neuronal cell death (*p < 0.01, anova), while other PKC peptides had no significant effect on neuronal survival. (b) PKCζ inhibitor peptide prevents the neuronal cell death induced by 1000 µm NMDA exposure for 5 min (*p < 0.05, anova) but not the neuronal death induced by 300 µm NMDA exposure for 24 h cPKC, peptide inhibitor of classical PKCs. δ, Peptide inhibitor of PKCδ; ε, peptide inhibitor of PKCε; PKCsc, scrambled PKC peptide which served as control.

The bis-benzimide and immunocytochemical stainings

The possible changes in the morphology of neuronal nuclei in response to 100 µm NMDA were studied by bis-benzimide staining 24 h after the 5-min exposure. As expected, the number of vital and round neuronal nuclei appeared to be similar in the PKCζ inhibitor peptide + 100 µm NMDA-treated wells (Fig. 3d), the control wells treated with MEM (Fig. 3a), and the control wells treated with HEPES-buffer (Fig. 3b). On the contrary, the density of healthy-looking round neuronal nuclei was clearly decreased in the cultures treated with 100 µm NMDA alone (Fig. 3c). The condensed nuclei or apoptotic fragments referring to apoptosis were not seen after 100 µm NMDA exposure (Fig. 3c).

Figure 3.

Photomicrographs of bis-benzimide stained neurons in control and NMDA-exposed cultures with and without adding PKCζ inhibitor peptide. (a) Treatment with MEM for 5 min. (b) Treatment with HEPES buffer for 5 min. (c) Treatment with 100 µm NMDA for 5 min. (d) Treatment with PKCζ inhibitor peptide followed by 5-min exposure to 100 µm NMDA. Note that no apoptotic fragmentation of the neuronal nuclei can be detected after NMDA exposure. The faint gray nuclei seen under the bright neuronal layer belong to the astrocytic monolayer. Bar = 50 µm.

In order to investigate the possible changes in PKCζ localization in response to neurotoxic NMDA exposure, we performed immunocytochemical analysis of PKCζ using confocal microscopy. Co-localization of MAP-2 with PKCζ antibody (Fig. 4a) confirmed that PKCζ immunoreactivity is exclusively expressed in neurons and has a diffuse cytosolic location in neuronal cell bodies in control cultures. The 100-µm NMDA treatment induced relocation of PKCζ immunoreactivity from diffuse distribution in the cell bodies to cytosolic clumps, which was detected both in the cell bodies and neurites (Figs 4b and c). However, pre-treatment with PKCζ inhibitor peptide 20 min before the 100-µm NMDA exposure prevented the translocation of PKCζ, suggesting that the activation of PKCζ isozyme was blocked, most likely due to the binding of PKCζ inhibitor peptide (Fig. 4e).

Figure 4.

Confocal microscopy photographs showing the localization of PKCζ immunoreactivity in the mouse cortical neurons. The cultures were stained with polyclonal PKCζ (red) antibody (a, b, and d) or monoclonal MAP-2 (green) antibody (a and c) after 5-min exposure to HEPES-buffer with or without 100 µm NMDA. In control culture (a), PKCζ immunoreactivity is diffusely distributed in the cytoplasm of neurons. After the 5-min exposure to 100 µm NMDA PKCζ immunoreactivity is seen in cytosolic clumps both in the neuronal cell body and neurites (b). The localization of the PKCζ immunoreactivity in neurons is confirmed by double labeling of PKCζ (b) and MAP-2 (c, the same field as in b). When the cultures are pre-treated with PKCζ inhibitor peptide for 20 min before the 5-min exposure to 100 µm NMDA, PKCζ immunoreactivity is diffusely localized to the cytosol, similar to control cultures (d).

Intracellular calcium measurements

To examine whether the protective role of the PKCζ peptide could be due to interference with NMDA receptor activation, the effects of the peptides on NMDA-induced Ca2+ elevation were tested using fura-2, a fluorescent calcium indicator. Treatment with 100 µm NMDA caused a rapid increase in [Ca2+]i, which reached a fairy stable plateau (p < 0.05, t-test). When NMDA was removed from the medium there was only a partial reversal of the response and [Ca2+]i remained elevated (Fig. 5a). As shown in Fig. 5(b), the inhibitor peptides did not affect the magnitude of NMDA-induced elevation of [Ca2+]i or the residual Ca2+ elevation 15 min after the NMDA challenge (p > 0.05, t-test). This suggests that the neuroprotective effect of PKCζ takes place downstream to the NMDA receptors (Fig. 5).

Figure 5.

Effect of PKC inhibitor peptides on NMDA induced elevation of [Ca2+]i. See Experimental procedures for details. (a) Mouse mixed cortical cultures were exposed to 100 µm NMDA and 10 µm glycine as shown. The recording is an average from 6 cells (± SD). (Basal calcium level varies between 50 and 100 nm.) (b) The fura-2 peak response to NMDA/glycine alone (n = 31 cells) is compared to the response in the presence of the scrambled (sc) (n = 20 cells), the ζ (n = 12 cells) or δ (n = 14 cells) inhibitor peptides from several independent experiments. The white column demonstrates the maximal peak response to NMDA/glycine and the gray column the residual response 15 min after the NMDA challenge. The double arrows above the recording in (a) denote the time points used for analysis.


Ischemic stroke is one of the brain diseases in which excitotoxicity is thought to be involved (Gill et al. 1992; Iijima et al. 1992) To see whether PKCζ could play a role in ischemic brain damage, we performed in situ hybridization analysis of PKC isoforms after permanent focal ischemia of the mouse. The permanent focal brain ischemia induced expression of PKCζ mRNA in the cortex adjacent to the infarcted core (Fig. 6a). In addition, PKCδ mRNA was induced in the perifocal cortex (not shown), as previously reported (Miettinen et al. 1996). However, PKCα, β, γ, and ε isoforms were not induced at mRNA level (Figs 6b–e).

Figure 6.

In situ hybridization photomicrograph showing the induction of PKCζ mRNA (a) in the cortex (asterisk) adjacent to the infarction after 12 h permanent focal ischemia. PKCα (b), PKCβ (c), PKCγ (d) or PKCε mRNA (e) were not induced in these animals. The ischemic core area is separated by the dashed line.


This study demonstrates that NMDA-induced neuronal cell death can be blocked by specifically preventing the activation of PKCζ subspecies in mouse cortical neurons. In addition, focal ischemia induces expression of PKCζ (this study) and PKCδ (Miettinen et al. 1996) mRNA in the cortex. Even though PKCδ has been previously reported to promote apoptotic cell death both in the peripheral and neuronal systems (Villalba 1998; Koriyama et al. 1999; Pongracz et al. 1999; Anantharam et al. 2002), inhibition of PKCδ had no effect on NMDA-induced neuronal death. Altogether, our results suggest that PKCζ may mediate excitotoxin-induced necrotic cell death, whereas the execution function of PKCδ might be limited to apoptotic cell death. This hypothesis is supported by our other finding that both PKCδ and PKCζ mRNA are induced by focal ischemia, which triggers both necrotic and apoptotic neuronal death.

Previous studies have suggested that PKC plays a role in excitotoxic neuronal death, but conflicting results concerning effects of activation (Durkin et al. 1997; Tremblay et al. 1999) or inhibition of PKC (Favaron et al. 1990; Mattson 1991; Jiang et al. 2000) have been reported. Because several PKC subspecies with different and even opposite functions are expressed in each neuron (Zheng et al. 1999; Lu et al. 2000; Lan et al. 2001; MacDonald et al. 2001), inhibition of the total PKC activity does not reveal specific functions of individual PKC subspecies, but instead, the potentially beneficial effects of one inhibited PKC subspecies may be masked by harmful effects of inhibiting another. Taking an advantage of previously characterized peptides, which inhibit PKC translocation acting as subspecies-selective competitors of PKC-RACK binding and function (Ron et al. 1995; Johnson et al. 1996), we showed that inhibition of cPKCs, PKCε or PKCδ had no effect on NMDA neurotoxicity, whereas inhibition of PKCζ, by a selective inhibitor of the catalytic activity of this isozyme, completely prevented the NMDA-induced LDH release and loss of bis-benzimide-stained neurons. We also confirmed a role of PKCζ in NMDA excitotoxicity by showing that NMDA exposure translocates PKCζ immunoreactivity from diffuse distribution in the cell body to cytosolic clumps in the cell body and neurites, and that this translocation is prevented by the peptide PKCζ inhibitor. Because we did not detect PKCζ immunoreactivity in non-neuronal cells, it can be concluded that activation of PKCζ in neurons is required for NMDA-induced excitotoxic neuronal death.

We found that PKCζ inhibitor provided protection against NMDA-induced neuronal death only when it was administered before the toxic insult, suggesting that activation of PKCζ takes place immediately after stimulation of NMDA receptors. While we cannot exclude the possibility that penetration of this 13-amino acid peptide through the plasma membrane to reach its target takes several minutes, in vivo studies with non-peptide inhibitors may be needed to determine whether PKCζ activation can be sufficiently prevented with a clinically relevant time window.

One of the mechanisms by which PKC regulates NMDA receptor-mediated signaling is a potentiation of NMDA receptor activity (Liao et al. 2001; Wagey et al. 2001). To address the question whether inhibition of PKCζ directly blocked the function of NMDA receptors and thereby protected the neurons, we quantified the changes in Ca2+ influx by fura-2 fluorescence imaging. However, administration of PKCζ inhibitor peptide had no effect on NMDA-induced neuronal Ca2+ influx, indicating that PKCζ acts downstream of NMDA receptor-stimulated Ca2+ elevation. This finding is also consistent with a previous report that it is the βΙ PKC subspecies that potentiates the function of the NR1/NR2A-containing NMDA receptors complex (Wagey et al. 2001). In addition, because PKCζ does not require Ca2+ for activation, it is likely that PKCζ activation is mediated by Ca2+-activated kinase different from the Ca2+-activated PKC.

Stimulation of NMDA receptors results in activation of different pathways of MAP kinases, which regulate cell stress and survival-related signals by phosphorylating intracellular enzymes and transcription factors (Cobb 1999). Extracellular signal-regulated protein kinase (ERK) mediates p21Ras pathway in response to NMDA-induced generation of nitric oxide (Yun et al. 1998; Iida et al. 2001), and this pathway has been shown to activate NF-κB in a PKCζ-dependent manner in non-neuronal systems (Anrather et al. 1999). Also, p38 MAP kinase subtype, which is strongly activated upon NMDA receptor stimulation (Kawasaki et al. 1997; Legos et al. 2002), may activate NF-κB by a mechanism involving PKC (Herlaar and Brown 1999). As inhibition of ERK or p38 MAP kinases reduces excitotoxic injury (Stanciu et al. 2000; Legos et al. 2002), and prevention of NF-κB activation is protective in some in vivo (Schneider et al. 1999) and in vitro (Grilli et al. 1996) systems, it is conceivable that PKCζ may be involved in sustained activation of MAP kinase pathways after NMDA stimulation.

Inhibition of PKCζ not only prevented the NMDA-induced neuronal death but also increased the neuronal survival above that of the control cultures, which were not challenged with NMDA. This ‘superprotection’ may be simply because an NMDA receptor-dependent loss of the neurons occur at low but stabile level in vitro after the first 13 days (S. Koponen et al., unpublished data). An alternative explanation is that PKCζ in involved in several death mechanisms of neurons, thereby improving the overall survival of neurons in compromised growth conditions in vitro.

Inhibition of PKCδ isozyme did not provide protection in our excitotoxicity model even though a body of evidence has indicated that PKCδ is a target for caspase-3 and that the generated PKCδ cleavage product is needed for apoptocic cell death in various cellular models (Emoto et al. 1995; Ghayur et al. 1996). In addition, PKCδ has been reported to be involved in oxidative death of dopaminergic neurons (Anantharam et al. 2002) and to increase the cardiac ischemic damage both in vitro and in vivo (Chen and Mochly Rosen 2001; Chen et al. 2001). Expression of PKCδ is induced in ischemic (Miettinen et al. 1996; Koponen et al. 2000) and kainic acid-induced excitotoxic (Kaasinen et al. 2002) brain injury. Because it has been reported that NMDA can cause either apoptotic or necrotic neuronal death, depending on the extent of the insult (Bonfoco et al. 1995), it is possible that our failure to detect protection by inhibiting PKCδ is due to the lack of apoptosis and caspase-3 activation in the present model. Our observation that no obvious apoptotic nuclei were detected 24 h after the insult supports this notion. The NMDA-induced signaling pathway, however, also involves the activation of PKCδ, as it has been shown previously that activation of NMDA receptors is required for depolarization and ischemia-induced activation of PKCδ (Miettinen et al. 1996; Koponen et al. 1999, 2000; Kurkinen et al. 2001).

We have found that expression of both PKCδ (Miettinen et al. 1996) and PKCζ mRNA (this study) is induced in the cortex after focal brain ischemia. Considering that only PKCδ of these two is activated after potassium chloride-induced pre-conditioning (Koponen et al. 1999; Kurkinen et al. 2001), which provides tolerance to a subsequent ischemia in the cortex (Kobayashi et al. 1995), and that only PKCζ of these two mediates NMDA-induced death of cortical neurons, we hypothesize that these PKC isozymes have different functions in brain ischemia: activation of PKCζ-mediated excitotoxic neuronal death, whereas activation of PKCδ, when not cleaved, may be protective and when cleaved, may mediate apoptotic neuronal death. Because both PKCζ (Martin et al. 2002) and PKCδ (Leitges et al. 2001; Miyamoto et al. 2002) knockout mice have been made, it will be interesting to verify the role of these PKC subspecies in excitotoxic and ischemic neuronal death in vivo.