Cortical neuronal cultures are an established model to study neuronal susceptibility to excitotoxic stimuli as a function of age. In our experimental setup, cortical neurons were cultured for 6 (6DIV) and 12 (12DIV) days in order to represent young/immature and adult/ mature fully differentiated neurons respectively [16,17,18].

Neuronal susceptibility to hypoxic insults was assessed by LDH assay (Figure 1A). Young neurons (6DIV) displayed resistance to 1% O₂ up to 24 h. Differently, adult neurons (12DIV) revealed a time-dependent increase in neuronal death (1.87 ± 0.01 fold increase after 6h, p < 0.001; 4.17 ± 0.012 fold increase after 24h, p < 0.001 compared to normoxic neurons) following hypoxic insult.

Following hypoxia, adult neurons displayed an increase in NMDA receptor protein levels as compared to younger-immature neurons. More specifically, Western blotting analysis confirmed that the NR1, NR2A and NR2B NMDA receptor subunits are higher in adult fully differentiated neurons as opposed to immature neurons (Figure 1B). A 0.59 ± 0.02, 3.32 ± 0.06, 1.53 ± 0.08 fold increase of NR1, NR2A and NR2B subunits’ protein levels respectively was observed (Figure 1C).

In addition, application of MK-801 (5 μM), a potent NMDA blocker, prevented hypoxia-induced neuronal death in adult neurons at both 6 h and 24 h (p < 0.001; Figure 1D) while having no effect on young ones (data not shown).

Increasing age leads to altered expression of specific stress-induced proteins. In vitro, increasing age led to an increase in the basal levels of the JNK signaling proteins, namely MKK7, MKK4 and JNK (Figure 2A). More importantly, the activation of the JNK cascade is more pronounced in mature compared to immature neurons (Figure 2A, Figure 2B, p < 0.05). On the other hand, Hif-1α levels are markedly reduced in adult hypoxic (1% O₂ for 3 h) neurons compared to young ones as shown by Western blot analysis (Figure 2C) and subsequent quantification (0.4 fold ± 0.086 decrease, p < 0.05, Figure 2D).

As shown in Figure 1D, hypoxia is an NMDA-mediated signaling pathway. In view of the fact that JNK plays a pivotal role in NMDA-mediated excitotoxicity, we investigated its regulation in our hypoxic model. Western blot analysis revealed hypoxic regulation of pJNK (Figure 3A). Quantification analysis (Figure 3B) showed that pJNK2/3 (54 kDa) levels decreased by 30 min of hypoxic exposure to reach maximal levels of activation by 1 h. Yet, pJNK1 (46 kDa) levels were induced by hypoxia both at 30 min and 1 h (p < 0.05). A decline in the phosphorylation of all JNK isoforms was observed by 2 h and 3 h (p < 0.05). Differently, Hif-1α expression gradually increased to reach maximal levels by 3 h (Figure 3C, 3D). At this time point pJNK levels were markedly decreased. These data suggest an inverse correlation between JNK phosphorylation and Hif-1α expression.

In order to identify a possible direct interaction between JNK and Hif-1α we initially performed sequence analysis. JNK phosphorylates some of its targets through the JNK binding domain (JBD): K/RX_0-2K/RX_0-4L/I-X-L/I. A number of transcription factors, such as c-jun, ATF-2 and Elk-1 are modulated by JNK via a JBD interaction and are thus called JDB-dependent substrates [19].

Sequence analysis of Hif-1α indicated the presence of a JBD motif within the N-terminal VHL (Von Hippel-Lindau) recognition site (Figure 4A). Although no protein in the PDB has good sequence homology with the whole human Hif-1α, some parts have low sequence similarity with known structures. Amongst them, the Murine Soluble Epoxide Hydrolase (MSEH) (pdb code 1CQZ) is the only one protein in the PDB that shares a local sequence similarity equal to 38.2%, with the human Hif-1α sequence containing the putative JBD (Figure 4B). Interestingly, the degree of similarity of 55.6% between the putative JBD and the corresponding sequence in MSEH is relatively high thus suggesting with reasonable confidence a structure similarity.

The structure of the MSEH shows that the sequence corresponding to the putative JBD forms a loop that is exposed to the solvent. Moreover, the secondary structures surrounding this sequence are sufficiently far not to disrupt the binding with other proteins, i.e., the MSEH folding allows the interaction with another partner (Figure 4C). These observations support the hypothesis that this sequences maybe a binding domain for JNK.

To elucidate the contribution of JNK signaling in hypoxic-induced neuronal death, neurons were treated with increasing D-JNKI1 concentrations and neuronal death was determined (Figure 5). Treatment with 2μM D-JNKI1 decreased neuronal death and 4μM D-JNKI1 led to a potentiation of this protective effect (p < 0.001). Application of SP600125, an ATP competitive and less specific JNK inhibitor, had no effect on the survival of the neurons.

We conclude that JNK signaling plays an important role in the early stages of hypoxia as underscored by the fact that inhibition of JNK led to a 55% survival against hypoxia as compared to hypoxic untreated controls.

Hif-1α is an important neuroprotective factor in hypoxic/ischemic injury and it is strongly induced in this model (Figure 3C, Figure 3D). Since Hif-1α has a JBD and D-JNKI1 prevents hypoxia-induced neuronal death we investigated whether D-JNKI1 affects Hif-1α expression.

Mature neurons were treated with D-JNKI1 and subsequently made hypoxic for 3h. Expression of Hif-1α protein levels were then determined by Western blotting. As expected Hif-1α expression was induced by 3 h of hypoxia. D-JNKI1 pre-treatment, led to a further increase in Hif-1-levels (0.61 ± 0.08 fold increase of Hif-1α induction in comparison to hypoxic controls, Figure 6A, Figure 6B). Treatment of neurons with increasing D-JNKI1 concentrations led to an additional increase of Hif-1α expression reaching a plateau at 4 μM (1.49 ± 0.18, p < 0.05), (Figure 6C, Figure 6D). Altogether, this data indicate that D-JNKI1 protects adult cortical neurons and this can partially modulate hypoxic-induced Hif-1α protein levels.

Primary neuronal cultures were obtained from the cerebral cortex of two days post-natal rats, incubated with 200 units of papain (Sigma Aldrich) for 30’ at 34 °C, then with trypsin inhibitor (Sigma Aldrich) for 45’ at 34 °C and subsequently mechanically dissociated. All experimental procedures on live animals were performed in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and all efforts were made to minimise animal suffering. Neurons were plated in 35 mm dishes (~7 × 10⁵ cells/dish) pre-coated with 25 μg/mL poly-D-lysine (Sigma Aldrich). The plating medium was B27/neurobasal (Life Technologies, Gaithersburg, MD) supplemented with 0,5 mM glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin. The experiments were performed 6 and 12 days from plating date.

Hypoxic experiments were performed in a glove-box chamber (In vivo 400, RUSKINN Technologies, Guiseley, UK) maintained at 37 °C with 5% CO₂. Neurons were exposed to 1% O₂ for 6 and 24 h and cell harvesting was performed within the chamber (i.e., no reoxygenation of samples). Control cultures were maintained at 21% O₂. Treatment with MK-801 (5 μM, Sigma), D-JNKI1 (2, 4, 6 μM, Xigen SA, Lausanne, Switzerland) or SP600125 (10 μM) was applied 45 minutes before hypoxic exposure.

Neuronal death was evaluated by a lactate dehydrogenase (LDH) assay. LDH released into the culture medium was measured using the Cytotox 96 no radioactive cytotoxicity assay kit (Promega, WI). All LDH assays were performed in triplicate.

Total protein extracts were obtained by scraping cells in lysis buffer [27]. Cells were washed twice in ice-cold PBS and lysed for 20’ at 4 °C in 1% Triton x-100 lysis buffer supplemented with proteases (1x CPIK, Roche, 10634200) and phosphatases (1 μM 4-NPP, Roche, 10030536) inhibitors.

10μg of total lysates were loaded and run onto 10% acrylamide gels and subsequently transferred to PVDF membranes. Incubation with primary antibodies was overnight at 4 ^oC: NMDA NR1 (1:2,000 Invitrogen), NMDA NR2A (1:2,000 Invitrogen), NMDA NR2B (1:2,000 Invitrogen). Hif-1α (1:1,000 NOVUS), pJNK (1:1,000 Santa Cruz), JNK (1:1,000, Cell Signaling), pMKK7 (1:1,000 Cell Signaling), pMKK4 (1:2,000 Cell Signaling) tMKK7 (1:10,000 BD Transduction Laboratories), tMKK4 (1:1,000 Upstate), actin (1:5,000, Chemicon). Blots were developed using horseradish peroxidase-conjugated secondary antibodies and the ECL chemiluminescence system. All blots were normalized to actin (1:50,000 Millipore) and three independent experiments were performed. Western blots were quantified by densitometry using Quantity One software (Biorad).

The human Hif-1α sequence (number accession: Q16665) was downloaded from the UniProt website [28]. Its analysis was performed using a home-made script for highlighting the presence of putative JNK binding sequences characterized by the following sequence: K/R-X_0-2-K/R-X_0-4-L/I-X-L/I [19].

To allow the interaction with another partner, a binding sequence must be at the surface of the protein and not into the protein core. The knowledge of the three-dimensional structure is then crucial to know the position of the putative JNK binding sequence. Unfortunately, the structure of human Hif-1α is not available in the Protein Data Bank (PDB) [29], indicating that its structure has not been yet solved. A BLASTp research on the human Hif-1α sequence has been carried out in the PDB [30] in order to obtain structural data about human Hif-1α protein and, in particular, about its putative JNK binding site. The structures were visualized using Pymol program.

Graphs and analyses were performed using GraphPad Prism software. Statistical comparison among groups was made using one-way and two-way ANOVA tests. A value of p < 0.05 was considered significant. All results are expressed as mean ± standard deviation.