Phosphoinositide 3-kinase couples NMDA receptors to superoxide release in excitotoxic neuronal death

Abstract

Sustained activation of neuronal N-methly D-aspartate (NMDA)-type glutamate receptors leads to excitotoxic cell death in stroke, trauma, and neurodegenerative disorders. Excitotoxic neuronal death results in part from superoxide produced by neuronal NADPH oxidase (NOX2), but how NMDA receptors are coupled to neuronal NOX2 activation is not well understood. Here, we identify a signaling pathway coupling NMDA receptor activation to NOX2 activation in primary neuron cultures. Calcium influx through the NR2B subunit of NMDA receptors leads to the activation of phosphoinositide 3-kinase (PI3K). Formation of phosphatidylinositol (3,4,5)-triphosphate (PI(3,4,5)P3) by PI3K activates the atypical protein kinase C, PKC zeta (PKCζ), which in turn phosphorylates the p47(phox) organizing subunit of neuronal NOX2. Calcium influx through NR2B-containing NMDA receptors triggered mitochondrial depolarization, NOX2 activation, superoxide formation, and cell death. However, equivalent magnitude calcium elevations induced by ionomycin did not induce NOX2 activation or neuronal death, despite causing mitochondrial depolarization. The PI3K inhibitor wortmannin prevented NMDA-induced NOX2 activation and cell death, without preventing cell swelling, calcium elevation, or mitochondrial depolarization. The effects of wortmannin were circumvented by exogenous supply of the PI3K product, PI(3,4,5)P3, and by transfection with protein kinase M, a constitutively active form of PKCζ. These findings demonstrate that superoxide formation and excitotoxic neuronal death can be dissociated from mitochondrial depolarization, and identify a novel role for PI3K in this cell death pathway. Perturbations in this pathway may either increase or decrease superoxide production in response to NMDA receptor activation, and may thereby impact neurological disorders, in which excitotoxicity is a contributing factor.

Figure 1

Excitotoxic neuronal superoxide production is mediated by NOX2 and triggered by NR2B-containing NMDA receptors. (a) NMDA-induced formation of oxidized dihydroethidium in cortical neurons is dose-dependent *P<0.05 versus control, (n=3). All NMDA exposures are 30 min. (b) NMDA-induced neuronal death is also dose-dependent, and roughly parallel to superoxide production *P<0.05 versus control, n=3). (c) Formation of 4HNE (red) in wild-type mouse cortical neurons after 30 min incubation with 100 μM NMDA. Neuronal processes are identified by MAP-2 (green), and nuclei by 4′,6-diamidino-2-phenylindole, DAPI (blue). NMDA-induced 4HNE formation was blocked by the Tat-conjugated NOX2 peptide inhibitor, gp91ds-Tat (5 μM), but not by the scrambled sequence peptide (Scr-Tat; 5 μM). 4HNE formation was also blocked by the NR2B inhibitor, Ro 25-6981 (0.5 μM), and not induced by ionomycin (3 μM). Scale bar 20 μm. (d) Similar results were obtained using oxidized Eth species ( red) as a marker of superoxide formation. Scale bar 10 μm. *P<0.01 versus NMDA, n=3. (e) Quantification of 4HNE staining shown in (c). *P<0.05 versus NMDA, n=3. (f) Quantification of Eth fluorescence shown in (d). *P<0.05 versus NMDA, n=3

Figure 2

NOX2-competent neurons produce oxidative stress and cell death in neighboring neurons after NMDA exposure. Neuronal cultures were prepared from p47^phox-deficient (NOX2-incompetent) mice and transfected with p47^phox-GFP (p47^phox, green) to reconstitute NOX2 function in a subset of the neurons. Control cultures were transfected with the GFP tag alone (empty vector; green) (a). Oxidized Eth species (red) in neurons treated with or without 100 μM NMDA. Neuronal processes are identified by Fura-2 (blue). Scale bar=10 μm. (b) Quantification of Eth-positive neurons. EV, transfected with empty vector (GFP only), p47^phox, transfected with GFP-tagged p47^phox. *P<0.01, n=3. (c) Dead neurons identified by propidium iodide staining (red) in neuronal cultures prepared from p47^phox-deficient mice transfected with either GFP alone (empty vector; green) or p47^phox-GFP (p47^phox, green) and treated for 30 min with 100 μM NMDA. Scale bar=10 μm. (d) Quantification of neuronal death. Where indicated neurons were treated with 100 μM NMDA together with 100 U superoxide dismutase-1, *P<0.01 versus NMDA, n=3

Figure 3

Excitotoxic neuronal death is mediated by NOX2 and triggered by NR2B-containing NMDA receptors. (a) Trypan blue staining of dead neurons 24 h after 30-min exposure to 100 μM NMDA, conditions as in Figure 1C. Scale bar 20 μm. (b) Quantified cell death. *P<0.05 versus NMDA, n=3. (c) Representative (of n=4) real-time calcium transients measured with Fura-2. Drugs were added after 5 min of baseline recording (arrow). (d) Representative trace of superoxide production as measured by Eth fluorescence in the same neurons as (c). Note Eth increase with NMDA but not ionomycin. (e) Quantified results; *P<0.01 versus NMDA, n=3. (f) Representative traces of mitochondrial membrane potential, as evaluated using TMRM fluorescence in a separate set of experiments. Note mitochondrial depolarization with both NMDA and ionomycin. (g) Quantified results; *P<0.01 versus NMDA, n=3

Figure 4

PI3K inhibition blocks NMDA-induced superoxide production and cell death. (a) Formation of 4HNE (red) in neurons after 30 min exposures to 100 μM NMDA. Neuronal processes are identified by MAP-2 (green), and nuclei by DAPI (blue). 4HNE formation was blocked in cultures treated with 10 μM wortmannin. Scale bar is 20 μm. (b) Quantification of 4HNE formation. *P<0.05 versus NMDA, n=3. (c) Phase contrast images of trypan blue staining shows that wortmannin prevents NMDA-induced neuronal death. (d) Quantification of neuronal death. *P<0.05 versus NMDA, n=3. (e) Phase contrast images show cell swelling following NMDA application is not blocked by wortmannin. Scale bar is 20 μm. (f) Quantification of neuronal swelling. *P<0.01 versus control, n=3. (g and h) Wortmannin had no significant effect on NMDA-induced calcium elevations as measured by Fura-2, or on mitochondrial membrane potential as measured by TMRM. Representative of n=4

Figure 5

Lipid products of PI3K and transfection with constitutively active PKCζ (PKM) reverse wortmannin inhibition of NMDA-induced superoxide formation. (a) Formation of oxidized dihydroethidium (red) in cortical neurons treated with NMDA is suppressed in cultures cotreated with wortmannin. Where indicated, neurons were also treated with the PI3K product, PI(3,4,5)P3 coupled to a cell permeable carrier tagged with BODIPY (green). The exogenously supplied PI(3,4,5)P3 restored NMDA-induced superoxide formation in the presence of wortmannin, but the PI3K substrate PI(4,5)P2 did not. Scale bar=20 μm. (b) Quantification of Eth-positive neurons. *P<0.01 versus control, n=3—6. (c) Formation of Eth (red) in neurons transfected with GFP-tagged PKM (green). Neurons were treated with NMDA (100 μM) alone or with wortmannin (10 μM). Scale bar 20 μm. (d) Quantification of Eth-positive neurons.*P<0.05 versus control; n=4

Figure 6

NMDA induces PI3K-dependent phosphorylation of PKCζ, and PKCζ-dependent phosphorylation of p47^phox. (a) Immunostaining for phospho-PKCζ (phosphorylated at Thr410/403; red). Neuronal processes are identified with antibody to MAP-2 (green). PKCζ phosphorylation was induced by NMDA, but not by the calcium ionophore ionomycin. PKCζ phosphorylation was blocked by both the NR2B inhibitor RO 25-6981 and the PI3K inhibitor wortmannin. Conditions as in Figure 1a. (b) Immunostaining for phospho-p47^phox (phosphorylated at Ser328; red). Phosphorylation was induced by NMDA, but not by ionomycin. p47^phox phosphorylation was blocked by both wortmannin and a Tat-conjugated peptide inhibitor of PKCζ. Scale bars are 20 μm. (c and d) Quantification of immunostaining. *P<0.05 versus NMDA; n=4

Figure 7

Signaling pathway linking NMDA receptor activation to NOX2 superoxide production. Calcium influx via NR2B-containing NMDA receptors induces PI3K to form PI(3,4,5)P3. PI(3,4,5)P3 activates PKCζ, which phosphorylates the p47^phox organizing subunit of NOX2. Phosphorylated p47^phox induces assembly of the active NOX2 complex at the cell surface and production of superoxide. Red lines indicate the targets of the probes used in these experiments