c-Jun N-terminal kinases trigger both degeneration and neurite outgrowth in primary hippocampal and cortical neurons

Authors


Address correspondence and reprint requests to Thomas Herdegen, Institute of Pharmacology, Universitaetsklinikum Schleswig-Holstein Campus Kiel, Hospitalstrasse 4, 24105 Kiel, Germany. E-mail: herdegen.office@pharmakologie.uni-kiel.de

Abstract

We provide a comprehensive analysis on c-Jun N-terminal kinase (JNK) actions leading to death or differentiation in postnatal hippocampal and cortical neurons. Stimulation with glutamate or 6-hydroxy-dopamine caused activation of caspase-3 and apoptotic neuronal death which were both attenuated by JNK-inhibition. In cortical neurons, stress-induced nuclear JNK distribution was rather complex. We observed a decrease of activated and total JNK in the nucleus after stimulation, but an increase of the phosphorylated transcription factor c-Jun. Isoform-analysis revealed a nuclear translocation of JNK2, while nuclear protein levels of JNK1 decreased. This activation pattern differed from neurite formation. In hippocampal and cortical neurons, JNK activity continuously increased during neuritogenesis, whereas levels of phosphorylated c-Jun gradually declined. Despite these similarities, JNK inhibition by SP600125 only affected neurite outgrowth in hippocampal cells. Furthermore, experiments in JNK-deficient mice demonstrated that all JNK isoforms contributed to neuritogenesis. Summarizing, JNKs are involved in both neuritogenesis and death of primary neurons with differentially regulated nuclear translocation of specific isoforms after degenerative stress, while neuritogenesis is supported by all JNK isoforms.

Abbreviations used
6-OHDA

6-hydroxydopamine

ATF2

activating transcription factor 2

DIV

days (after plating) in vitro

IR

immunoreactivity

JNK

c-Jun N-terminal kinase

LDH

lactate dehydrogenase

MAP2

microtubule-associated protein-2

SDS

sodium dodecyl sulfate

WST

4-[3-(4-iodophenyl)-2-(4-nitrophenyl)-2H-5-teztrazolio]-1,3-benzene disulfonate

WT

wild-type

In neurons, injury-induced degeneration and repair are tightly linked. The family of c-Jun N-terminal kinases (JNKs) constitutes an essential signaling system for this responsive dichotomy [reviewed by Herdegen et al. (1997) and Waetzig and Herdegen (2005a)]. On the one hand, JNKs mediate neuronal degeneration and (apoptotic) death in response to numerous experimental stimuli (Yang et al. 1997; Kuan et al. 1999; Coffey et al. 2000, 2002; Borsello et al. 2003; Hunot et al. 2004; Brecht et al. 2005). On the other hand, JNKs are also relevant for neurophysiological processes, at which the signaling mechanism is far less understood. Recent data have demonstrated an important role for JNKs in injury-induced neurite outgrowth (Waetzig and Herdegen 2003; Brecht et al. 2005) and the requirement of all JNK isoforms for the correct development of the nervous system (Kuan et al. 1999; Sabapathy et al. 1999; Shoichet et al. 2006). However, isoform-specific functions in death and repair are rarely examined. Apart from JNK isoforms, there are also different JNK subpools with regard to cellular localization. These subpools form individual signalosoms mediating neuritogenesis or apoptosis (Coffey et al. 2000; Bjorkblom et al. 2005). Besides context- and isoform-specific JNK regulations, the outcome of JNK actions is highly dependent on the cell type. In pheochromocytoma cells (PC12) cells, dopaminergic MN9D cells, SHSY cells, or dorsal root ganglia JNK inhibition dramatically abrogated neuritogenesis (Lindwall and Kanje 2005a,b), while inhibition of JNK increased the number of processes in cerebellar granule cells (Coffey et al. 2000). This indicates that JNKs can either promote or attenuate neurite formation.

The present study investigated the contribution of JNK isoforms to excitotoxic and free radical stress on the one hand and neurite outgrowth as a morphological correlative for neuronal differentiation on the other hand. We demonstrate in cortical neurons that JNK2 and the amount of phosphorylated c-Jun increase and JNK1 decreases in the nucleus during cell death, while JNK activation increases and the amount of phosphorylated c-Jun decreases during neurite formation. Only on hippocampal cells, JNK inhibition affected neurite outgrowth.

Materials and methods

Chemicals

Gentamycin, glutamax, and B27 supplement were purchased from Gibco BRL (Carlsbad, CA, USA); transferrin: Calbiochem (San Diego, CA, USA); SP600125: Alexis (San Diego, CA, USA); lactate dehydrogenase (LDH), cell proliferation [4-[3-(4-iodophenyl)-2-(4-nitrophenyl)-2H-5-teztrazolio]-1,3-benzene disulfonate (WST)] assays: Roche, Germany.

Antibodies

Antibodies against glial fibrillary acidic protein, JNK2, activating transcription factor 2 (ATF-2), caspase-3 were purchased from: Santa Cruz Biotechnology (Santa Cruz, CA, USA); anti-phospho-c-Jun, anti-c-Jun, anti-total JNK, anti-cleaved caspase-3: Cell Signaling Technology (Beverly, MA, USA); anti-microtubule-associated protein-2 (MAP2): Chemicon (Temecula, CA, USA); anti-phospho-JNK: Promega (Madison, WI, USA); anti-JNK1: Pharmingen (San Diego, CA, USA); anti-JNK3: Upstate (Hauppauge, NY, USA); anti-β-actin: Sigma (Germany).

Cell culture experiments

Hippocampal or cortical cultures were grown from newborn (up to 24 h postnatal) Wistar–Kyoto rats or C57Bl6 mice or JNK knockout mice. The JNK ko strains and their wild-type (WT) controls were bred as described in detail previously (Brecht et al. 2005). In brief, hippocampus and cortex from newborn rats and mice were explanted and cleaned from meninges (Brecht et al. 2001). After mechanical and enzymatic dissociation in 3.3 mg/mL trypsin solution for 5 min, trypsin inhibitor was added to block the enzyme, and DNAse was added to break DNAs from dead cells. A series of trituration and mild centrifugation steps were included to disperse the neurons prior to resuspension in medium and to remove undissociated debris prior to plating in minimum essential medium supplemented with transferrin 0.1 mg/mL, insulin 25 μg/mL, glutamax 4 mmol/L, gentamycin 5 μg/mL and 10% fetal calf serum. Cells were plated onto 4-well plates containing poly-l-lysine coated coverslips at a density of 150 000 cells per wells. After 2 days, the serum was reduced to 5%, and 1% B27-supplement + 5 μmol/L cytosine-β-d-arabinofuranoside were added to the culture. The cells were fed every third day. After 5–6 days, the hippocampal and cortical cultures were treated with 250 μmol/L glutamate freshly prepared from a 5 mmol/L stock solution in the culture medium without serum and B27 supplement. Cortical cells were treated with 25, 50, and 100 μmol/L 6-hydroxydopamine (6-OHDA), freshly made from 100 mmol/L stock solution in ascorbic acid solution to prevent oxidation. The JNK inhibitor SP600125 (2, 5, 10 μmol/L) was added 30 min before the stimulation with glutamate or 6-OHDA. For determination of neurite length, neurites from MAP2 immunopositive cultures were marked by computer imaging and the length was automatically measured by LeicaQwin (Leica Comp., Germany).

Immunocytochemistry

Cells were fixed for 30 min with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100 in phosphate-buffered sodium chloride and blocked with 1% bovine serum albumin. The cells were incubated overnight at 4°C with mouse anti-MAP2 (1 : 10 000), rabbit anti-glial fibrillary protein (1 : 400), rabbit anti-phospho-JNK (1 : 500), rabbit anti-phospho-c-Jun (1 : 500), mouse anti-JNK1 (1 : 500), or mouse anti-JNK2 (1 : 500) antisera. The primary antibody binding was detected after four washes in phosphate-buffered saline (PBS) by species-matched biotinylated secondary antibodies for 1 h at 37°C. Immunolabeling was visualized by immunoperoxidase reaction with ABC reagents and diaminobenzidine/H2O2. The experiments were repeated at least three times.

Hoechst test

For quantification of apoptotic morphology, hippocampal, and cortical cells were fixed for 30 min with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, and stained for 5 min with 5 μg/mL Hoechst 33258 to uncover the nuclear condensation/aggregation. Cell death was determined by microscopical counting of the condensed or fragmented nuclei in four vision fields per well. The data were expressed as percentage of apoptotic cells from the total cell number. All experiments were repeated at least five times.

LDH release assay

Release of LDH into the culture medium was determined with cytotoxicity detection kit according to the manufacturer’s instruction. Briefly, the culture supernatant was collected in a 1.5 mL tube and centrifuged (2000 g; 10 min; 15–20°C). Subsequently, a 96-well tissue culture plate was filled with 90 μL/well PBS and 10 μL/well centrifuged culture supernatants. To determine LDH activity in the supernatants 100 μL reaction mixture was added to each well and incubated for 30 min at 15–25°C. During this incubation period the 96-well plates were protected from light. Finally, the absorbance of the samples was measured at 490 nm by ELISA reader (Bio-Rad, Germany). The experiments were repeated at least five times.

Cell proliferation assay

Cell viability was measured by WST-1 assay according to the manufacturer’s instruction. Briefly, the cells were cultured in 96-well tissue culture plate. Ten microliter per well of cell proliferation reagent was added to the cells already cultured in 100 μL/well. The cells were incubated for 1 h at 37°C and the absorbance of the samples was measured at 490 nm using an ELISA reader. All experiments were repeated at least five times.

Cell lysates and western blotting

Hippocampal and cortical neurons were washed with cold PBS. For whole cell extracts, cells were resuspended in denaturing lysis buffer [20 mmol/L Tris (pH 7.4), 2% sodium dodecyl sulfate (SDS), 1% phosphatase inhibitor Cocktail II and protease inhibitor complete], incubated at 95°C for 5 min, briefly sonicated, and centrifuged to remove the insoluble material (15 000 g for 15 min). For nuclear or cytoplasmic extracts, cells were lysed as described previously (Waetzig and Herdegen 2003). Protein extracts were stored at −80°C. The purity of nuclear and non-nuclear fractions was tested by visualization or absence of the nuclear ATF-2 transcription factor (data not shown). Whole cell, nuclear, and cytoplasmic lysates were separated on 15% or 12% SDS-polyacrylamide gels and transferred onto a polyvinylidene difluoride membrane. Western blotting was conducted as described previously (Brecht et al. 2005) with primary antibodies against rabbit anti-ATF2 (1 : 500), rabbit anti-caspase-3 (1 : 1000), rabbit anti-cleaved caspase-3 (1 : 1000), mouse anti-JNK1 (1 : 1000), mouse anti-JNK2 (1 : 1000), rabbit anti-JNK3 (1 : 1000), rabbit anti-total JNK (1 : 1000), rabbit anti-phospho-c-Jun, (1 : 1000), rabbit anti-c-Jun (1 : 1000), or rabbit anti-phospho-JNK (1 : 2500) antibodies. For normalization, anti-β-actin (1 : 5000) was used for cytoplasmic and whole cell extracts, ATF-2 (1 : 1000) was used for nuclear extracts. Between the stainings with phosphospecific antibodies, and total kinase, total transcription factor antibodies or different isoform-specific antibodies, blots were stripped in 2% SDS, 62.5 mmol/L Tris, and 100 mmol/L 2-mercaptoethanol for 30 min at 50°C, washed with tris-buffered saline with Tween-20, and blocked again. To normalize for the protein content of each lane and to confirm equal loading, all membranes were finally stained with Ponceau S. All experiments were repeated at least three times.

Statistical analysis

All data were obtained from four to five independent sets of experiments. Statistical analysis was performed with GraphPrism software (http://www.graphpad.com). Statistical significance of data from MAP2 immunoreactivity (IR), LDH, and WST assays were determined by Student’s t-test and mentioned in the text. Statistical significance of other assays were performed by one-way anova followed by the post hoc Bonferroni test. Significance was set at p ≤ 0.05 in all cases. Data were expressed as mean ± SD.

Results

Hippocampal neurons

Stimulation with glutamate

Cell death and survival. The first experiments addressed the role of JNKs as mediators of glutamate excitotoxicity in hippocampal neurons, an important feature of neuronal death, e.g. following seizure activity or ischemia [reviewed by Waetzig and Herdegen (2005a) and Waetzig et al. (2006)]. Five days after plating in vitro (DIV 5), mixed hippocampal neuronal cultures were stimulated with 250 μmol/L glutamate for 24 h, and stained with MAP2, a marker for the neuronal cytoskeleton. The counted density of neurons in untreated cultures was set as 100%. Glutamate treatment decreased the number of surviving MAP2 neurons by 34% (Fig. 1a), which was confirmed by WST assays (data not shown). Pre-incubation of the hippocampal cultures with 2 μmol/L SP600125, a specific pan-JNK inhibitor, rescued 61% of the otherwise dying neurons (Fig. 1a). Glutamate caused typical apoptotic features, such as nuclear fragmentation and condensation visualized by the DNA dye Hoechst 33258 (Fig. 1b). To which extent contribute JNKs to the glutamate-induced features of apoptotic death? Incubation of the hippocampal cultures with 2 and 5 μmol/L SP600125 rescued about 35% of the dying neurons. Two or 5 μmol/L SP600125 alone did not induce any apoptotic features (Fig 1b).

Figure 1.

 (a) MAP2 IR of rat hippocampal neurons from unstimulated controls (co) which were set as 100%, and following 250 μmol/L glutamate with or without SP600125 (each n = 6). (b) Death of rat hippocampal neurons 6 h following 250 μmol/L glutamate determined by counting of apoptotic nuclei visualized by Hoechst 33258 staining. Pre-incubation with 2 and 5 μmol/L SP600125 significantly diminished the number of apoptotic nuclei. Incubation with 2 and 5 μmol/L SP600125 alone for 6 h did not have any effect on apoptotic nuclei (each n = 3). **p < 0.01 compared to control; #p < 0.05 compared glutamate and glutamate with SP600125. (c) Western blots of the cleaved caspase-3 product p17 and total caspase-3 in unstimulated hippocampal controls (co) and 12 h following 250 μmol/L glutamate with or without SP600125.

c-Jun N-terminal kinases mediate the activation of caspase-3, an indispensable component for the formation of the apoptosome and the release of cyctochrome c [reviewed by Waetzig and Herdegen (2005a) and Waetzig et al. (2006)]. Therefore, we investigated whether JNKs were also involved in the activation of caspase-3 in primary postnatal hippocampal neurons. Western blotting with an antibody against the p17 subunit of activated caspase-3 demonstrated the appearance of activated caspase-3 12 h after glutamate stimulation, which was strongly attenuated following pre-incubation with the JNK-inhibitor SP600125 (Fig. 1c). Reprobing of the stripped membranes with an antibody against caspase-3 proved the normalization of the protein loading.

Distribution of JNK activity. The second set of experiments investigated the intracellular distribution of JNK activity following 250 μmol/L glutamate. In untreated cultures, IR of phosphorylated i.e. activated JNK was concentrated in the nucleus of neurons and astrocytes as well as in varicosities of the neurons (Fig. 2a). Four hours after glutamate treatment, the phospho-JNK IR had faded out in the varicosities and nuclei of neurons, whereas it did not change in astrocytes (Fig. 2b). In contrast, phosphorylated c-Jun IR which was hardly seen in untreated cells (Fig. 2c), strongly increased after glutamate treatment in the nuclei of both neurons and astrocytes (Fig. 2d). Western blotting experiments demonstrated that the basal JNK activation was prominent in the nuclear pool of untreated cultures and decreased after glutamate treatment, while phosphorylated c-Jun increased in the nuclear extracts. This activation of c-Jun could be attenuated by treatment with SP600125, which indicates that the reduced nuclear JNK activity was nevertheless sufficient for c-Jun phosphorylation (Fig. 2e). The paradox of enhanced c-Jun activation and simultaneously decreased JNK activation in the nucleus is addressed in detail under section 6-OHDA-induced death.

Figure 2.

 Immunoreactivity of phosphorylated JNK in rat hippocampal neurons in (a) untreated cultures and (b) 4 h following 250 μmol/L glutamate treatment. Immunoreactivity of phospho-c-Jun in rat hippocampal neurons in (c) untreated cultures and (d) 4 h following 250 μmol/L glutamate treatment. (e) Western blotting from nuclear extracts against phospho-JNK, phospho-c-Jun and ATF-2 in unstimulated controls (co) and 4 h following glutamate with or without SP600125.

Elongation of neurites

Previously, we and others have shown that JNKs are essential for the formation and elongation of neurites in PC12 cells (Leppa et al. 1998, 2001; Waetzig and Herdegen 2003; Brecht et al. 2005). Does this also hold true for hippocampal neurons? To answer this question, rat hippocampal cultures were exposed to 2 μmol/L SP600125 on the second, third, or fifth DIV i.e. following explantation. At DIV 6, the cultures were stained with MAP2 antibody, the lengths of neurites were measured and classified in four groups (< 40, < 80, < 120 and > 120 μm). The inhibition of JNKs between DIV 2 and DIV 3 effectively prevented neurite elongation in primary hippocampal neurons as shown by MAP2 immunocytochemistry (Fig. 3a and b). Measurement of neurite length displayed a significant shift from long to short neurites when SP600125 was applied after 48 or 72 h (Fig. 3c) with an increase from 65% to 90% of all neurites < 80 μm in length at DIV 3.

Figure 3.

 Hippocampal rat neurons were exposed to 2 μmol/L SP600125 at second, third, or fifth DIV. At DIV 6, cultures were stained with MAP2 antibody and the length of neurites was measured and classified. MAP2 immunoreactivity in (a) untreated cultures and (b) following application of SP600125 at DIV 2. (c) The distribution of the neurite length (μm) (each n = 3). (d) Hippocampal cultures were incubated with 2 μmol/L or 10 μmol/L SP600125 on the second day, and cultures cells were stained with the DNA dye Hoechst 33258 to visualize nuclear apoptotic morphology on the sixth day. *p < 0.05, **p < 0.01 compared to untreated controls.

To exclude the possibility that reduced neurite formation was because of a slowly occurring cell death in response to SP6001215, we stained hippocampal cultures, that were incubated with 2 and 10 μmol/L SP600125 on the second day, with the DNA dye Hoechst 33258 on the sixth day. Both concentrations of SP600125 did not affect the neuronal survival (Fig. 3d).

To clarify which JNK isoform(s) contribute(s) to the neurite elongation, neonatal hippocampal neurons were cultivated from JNK1 ko, JNK2 ko, and JNK3 ko mice as well as from their respective wild-type controls (WT1, WT2, WT3). As a positive control, the JNK inhibitor SP600125 was applied 24 h after explantation in WT cultures. At DIV 2 or DIV 6, the hippocampal cultures were fixed and stained with MAP2 antibody. The length of neurites was measured and classified in four groups (< 40, < 80, < 120, and > 120 μm). All three WT strains showed a very similar distribution with around 65% and 33% of neurites < 40 μm after 2 and 6 days, respectively (individual data not shown). Therefore, the data from all WT animals were pooled (WT pool in Fig. 4). Inhibition of JNK by SP600125 significantly shifted this distribution to the left hand side (i.e. shorter neurites) with 89% and 57% of < 40 μm neurites after 2 and 6 days, respectively; after 6 days, SP600125 reduced the proportion of > 80 μm neurites to 7% compared with 22% in untreated WT cultures (WT pool + SP600125 in Fig. 4a and b).

Figure 4.

 The effect of JNK isoforms on neurite elongation was analyzed in hippocampal neurons from JNK1 ko, JNK2 ko, and JNK3 ko mice and their respective wild-type WT1, WT2, and WT3. For transparency, the data from all WT animals were pooled (WT pool). All hippocampal cultures were stained with MAP2 and grown for (a) DIV 2 or (b) DIV 6. SP600125 was applied immediately after the explantation. *p < 0.05, **p < 0.01, ***p < 0.001 compared to the respective WT controls (each n = 3).

Contribution of JNK isoforms. After 2 days, the length of neurites from individual JNK knockout strains did not differ from the respective WT cultures. The failure of JNK ko, however, to develop an intact neuritogenesis became clearly visible after 6 days (Fig. 4b). Hippocampal neurons from all JNK ko mice similarly developed shorter neurites. Importantly, JNK3 ko, but not JNK1 ko or JNK2 ko, substantially interfered with the morphology of hippocampal neurons as neurons from JNK3 ko revealed an irregular cell shape and distorted neurites (data not shown). This observation corresponds to the recently defined role of JNK3 being relevant for the intact developmental of PC12 cells in vitro (Waetzig and Herdegen 2003; Shoichet et al. 2006), neurons in vivo and the human brain (Shoichet et al. 2006).

c-Jun N-terminal kinases are essential for neurite elongation after explantation as shown by JNK inhibition and also involved in axonal pathfinding guidance (Tararuk et al. 2006). What does this mean for the expression and activation of JNKs? Whole cell extracts from mixed rat hippocampal cultures at DIV 2, DIV 3, and DIV 6 were screened for total and activated JNK, JNK1, JNK2, JNK3, and for c-Jun as well as for its N-terminal phosphorylation (Fig. 5). The pool of activated JNKs substantially and continuously increased with a maximum at DIV 6, at which the shorter isoforms (46 kDa) were mainly activated. The amount of total JNK only moderately changed (Fig. 5a).

Figure 5.

 Western blotting against expression and activation of (a) total JNK and JNK isoforms, and (b) c-Jun from hippocampal neurons (whole cell extracts) 2, 3, and 6 days post-explantation. For normalization, β-actin was used for the JNK blots (a), while ATF-2 was used for c-Jun and phospho-c-Jun (b).

JNK isoforms. During neuritogenesis, the expression of JNK1 and JNK3 increased (Fig. 5a), while protein levels of JNK2 did not change distinctly, which suggests that particularly JNK1 and JNK3 might contribute to the increasing phosphorylation of the 46 kDa fraction (Fig. 5a).

The JNK substrate c-Jun plays a crucial role for axonal regeneration in peripheral nerve fibers (Raivich et al. 2004). Does this also hold true for neuritogenesis of hippocampal neurons? In contrast to the increasing JNK activity, expression and activation of c-Jun dramatically declined from its early maximal level at DIV 2; N-terminal phosphorylation became almost undetectable at DIV 6 (Fig. 5b). Thus, the increase of activated JNKs is not automatically associated with an elevation of c-Jun phosphorylation.

Cortical neurons

c-Jun N-terminal kinases mediate both glutamate-induced apoptosis and the differential elongation of neurites in hippocampal neurons. Is this functional dichotomy a general principle or does it depend on the applied stimulus and/or the neuronal subtype? To clarify this issue, we investigated the role of JNKs in (i) glutamate- or (ii) 6-OHDA-induced neurodegeneration as well as in (iii) neurite outgrowth of primary cortical neurons from neonatal rats and JNK-deficient mice.

Glutamate-induced death

Five and 11 days after plating in vitro (DIV 5 and DIV 11), the mixed cortical neuronal cultures were stimulated with 250 μmol/L glutamate for 24 h. Cell death was visualized with Hoechst 33258 staining. Untreated cultures showed a normal cellular morphology with uniformly stained nuclei; only few cells displayed fragmentation and condensation of the DNA (Fig. 6a) indicating that the dissociation and cultivation cause an ongoing DNA damage. After stimulation with 250 μmol/L glutamate, the proportion of apoptotic nuclei strongly increased up to 50% (Fig. 6a). Pre-incubation of cortical neurons with the JNK inhibitor SP600125 significantly decreased the fraction of apoptotic cells by 35% (Fig. 6a). These results were confirmed by WST assays (Fig. 6b). Application of glutamate in DIV11 cultures caused similar effects (data not shown).

Figure 6.

 Death of rat cortical neurons 24 h following 250 μmol/L glutamate determined by (a) counting of apoptotic nuclei visualized by Hoechst 33258 staining. (b) Cell viability quantified by WST assay. **p < 0.01 compared to untreated controls; #p < 0.05, ##p < 0.01 compared glutamate to glutamate with SP600125 (each n = 4).

6-OHDA-induced death

In hippocampal and cortical neurons, JNKs had a pro-degenerative role in glutamate-induced cell death. Are JNKs generally pro-apoptotic in neurons or is their effect stimulus-dependent? To answer this question, we examined the effect of the radical stressor 6-OHDA on cortical neurons. Application of 6-OHDA caused a dose-dependent cell death characterized by an increased proportion of apoptotic nuclei stained with Hoechst 33258. Following 25 μmol/L and 50 μmol/L 6-OHDA, the proportion of apoptotic nuclei substantially increased to 26% (data not shown) and 53% (Fig. 7a), respectively. WST and LDH assays confirmed the enhanced cell death following 6-OHDA (Fig. 7b and c). Pre-incubation of cortical neurons with the JNK inhibitor SP600125 significantly attenuated cell death by 40% as shown by Hoechst staining, WST assays, and LDH release (Fig. 7a–c). Similarly, the cleavage of caspase 3 could be reduced by pre-incubation with SP600125 (Fig 7d).

Figure 7.

 Death of rat cortical neurons 24 h following 50 μmol/L 6-OHDA determined by (a) counting of apoptotic nuclei by Hoechst 33258 staining, and cell viability quantified either by (b) WST assay or (c) LDH release. (d) Western blots of caspase-3 and its fragment p17 from unstimulated controls (co) and following 6-OHDA with or without SP600125. *p < 0.05, **p < 0.01 compared to untreated controls; #p < 0.05, ##p < 0.01, ###p < 0.001 compared 6-OHDA to 6-OHDA with SP600125 (each n = 6).

JNK distribution. Glutamate altered JNK distribution in hippocampal neurons. As the localization determines JNK functions (Coffey et al. 2002; Bjorkblom et al. 2005; Waetzig et al. 2005), we studied the cytoplasmic and nuclear presence of JNK isoforms after stimulation with 6-OHDA by western blotting (Fig. 8).

Figure 8.

 Western blotting against phosphorylated (phospho-) JNK, total JNK, and JNK isoforms as well as against phosphorylated c-Jun in nuclear and cytoplasmic extracts from unstimulated cortical neurons (co) and 1, 2, or 4 h following 50 μmol/L 6-OHDA. For normalization, ATF-2 was used for the nuclear extracts and β-actin was used for the cytoplasmic extracts.

As known for central neurons, we observed a high basal JNK activity. In the nucleus, the amount of phosphorylated JNK decreased 2 h after stimulation with 6-OHDA. A reduction of total JNK was detected 4 h after 6-OHDA application. As for the isoforms, JNK1 decreased, while JNK2 increased in the nucleus (Fig. 8). For both, a nuclear increase or decrease was associated with a respective decrease or increase in the cytoplasm. JNK3 slightly increased in the nucleus without any distinct effects on cytoplasmic levels (Fig. 8).

Despite a high basal JNK activity, phosphorylated c-Jun was almost absent in unstimulated cells. After stimulation with 6-OHDA, c-Jun phosphorylation substantially increased, while the amount of phospho-JNK decreased (Fig. 8). How can a reduced JNK activity be linked to an increased c-Jun phosphorylation? Recently, we have reported the similar paradox of increased c-Jun activation, while nuclear JNK activity decreased (Waetzig et al. 2005). Comparable to the situation in cortical neurons, isoform analysis demonstrated that JNK1 which was responsible for the basal nuclear JNK activation, disappeared and was replaced by a small pool of JNK2 (Waetzig et al. 2005). Especially the changes of JNK2 and JNK3 could be relevant, as they are involved in stress-induced c-Jun phosphorylation (Coffey et al. 2002). The present results also suggest that the nuclear pool of activated JNK isoforms changes in primary cortical neurons in response to cellular stress.

JNKs are not involved in neurite elongation of primary cortical neurons

JNKs play an essential role for neurite elongation of hippocampal neurons, but they inhibit neuritogenesis in embryonic E18 cerebellar neurons (Coffey et al. 2002). To investigate the involvement of JNKs in neurite formation of cortical neurons from neonatal rats, cultures were exposed to 2 μmol/L SP600125 on second, third, or fifth day in vitro (DIV). At DIV 6, the cultures were stained with MAP2 antibody and the length of neurites was measured. In contrast to rat hippocampal neurons, JNK inhibition by SP600125 did not significantly affect neurite elongation in cortical neurons (Fig. 9a), which was confirmed by experiments from JNK-deficient mice (Fig. 9b).

Figure 9.

 (a) Length of neurites of rat cortical neurons exposed to 2 μmol/L SP600125 at second, third, or fifth day post-explantation (DIV 2, 3, 5); at DIV 6, cultures were fixed and length of neurites was measured in MAP2 immunoreactive neurons (each n = 3; co = control). (b) The effect of deletion of JNK isoforms on neurite elongation was analyzed in neonatal murine cortical neurons from JNK1 ko and JNK3 ko and their respective wild-type animals (WT1, WT3) (each n = 1).

JNK and c-Jun activation during neuritogenesis

In contrast to hippocampal neurons, JNKs are not involved in neurite elongation of neonatal cortical neurons. Is this failure reflected by a different pattern of activated JNKs and/or c-Jun? Proteins were extracted from cortical cultures at DIV 2, DIV 3, and DIV 6 and screened for total and phosphorylated JNK as well as for JNK1, JNK2, and JNK3 (Fig. 10a) as well as for c-Jun and phospho-c-Jun (Fig. 10b) by western blotting. Surprisingly, the pattern of JNK activation resembled that of hippocampal neurons with a pronounced increase in JNK activation. We also observed a slightly elevated expression of JNK1, while protein levels of JNK2 and JNK3 remained unchanged. While c-Jun phosphorylation and expression was only attenuated in hippocampal neurons after 2 days in culture, it almost completely disappeared after 2 days in cortical neurons (Fig. 10b). In contrast to ATF-2 expression, phosphorylation of ATF2, another transcription factor and JNK substrate, also rapidly declined in cortical neurons between 48–72 h (Fig. 10b).

Figure 10.

 Western blotting against expression and activation of (a) total JNK and JNK isoforms, and (b) c-Jun and ATF-2 from cortical neurons (whole cell extracts) 2, 3, and 6 days post-explantation. Immunocytochemical analysis with antibodies against c-Jun at DIV2 (c) and DIV6 (d) and phospho-JNK at DIV2 (e) and DIV6 (f). As in western blots, staining experiments showed that c-Jun decreased, while phospho-JNK increased.

Immunocytochemistry for phosphorylated c-Jun and phosphorylated JNKs confirmed western blot experiments: while the signal for phospho-c-Jun decreased in stimulated cells (Fig. 10d) compared with control cells (Fig. 10c), the signal for phospho-JNK increased after stimulation (Fig. 10f).

Discussion

The c-Jun N-terminal kinases are generally considered as important mediators of neuronal death [reviewed by Borsello and Bonny (2004) and Waetzig and Herdegen (2005a)], but they also have crucial physiological functions in the brain including the development of the human brain [(Shoichet et al. 2006); reviewed by Waetzig et al. (2006)]. What determines these bipartite actions of JNKs? In the present study, we approached this question by analyzing the role of JNKs in the context of cell death and neuritogenesis in different primary neurons. We particularly focused on isoform-specificity and differential JNK distribution to find explanations for this unsolved problem.

In response to radical stress, we observed the activation of the transcription factor and JNK-high affinity substrate c-Jun and a nuclear translocation of JNK2, which seems to be responsible for nuclear stress-induced JNK activation. The differential analysis of JNK isoforms during neuritogenesis revealed that particularly protein levels of JNK1 and JNK3 were increased and the overall JNK activity was elevated in hippocampal neurons. Simultaneously, c-Jun expression and activation decreased. These data indicate that JNKs need to translocate to the nucleus and activate transcription factors to mediate cell death and remain cytosolic for neurite formation.

JNK-induced cell death

The inhibition of total JNK activity by the specific JNK inhibitor SP600125 significantly reduces the death of primary hippocampal and cortical neurons from neonatal rat brain following excitotoxic glutamate and the free radical stressor 6-OHDA, respectively. JNK-triggered cell death was shown in two ways: Hoechst 33258 stain of fragmented DNA and the activation of caspase-3, a central effect of JNK-mediated neurodegeneration [(Putcha et al. 2003; Eminel et al. 2005), reviewed by Borsello and Bonny (2004) and Wang et al. (2004)]. So far, only few data are available on the contribution of JNKs to excitotoxic death in postnatal primary neurons (Chi et al. 2005; Centeno et al. 2006), but all findings come to similar conclusions: (i) JNKs are activated in response to excitotoxic stress and (ii) JNK inhibition attenuates glutamate-induced cell death.

Our results suggest that among other factors nuclear JNK activity might have a crucial role for the induction of cell death in primary neurons. However, even in the nucleus, a small stress-sensitive pool of JNKs is responsible for substrate phosphorylation, as the total nuclear JNK activity decreases with a simultaneous increase of phosphorylated c-Jun. Previous findings have shown that c-Jun is central to neurodegeneration (Anderson et al. 1995; Ham et al. 1995; Behrens et al. 1999; Besirli et al. 2005), but have not addressed the question of nuclear JNK activity, yet. A similar ‘paradox’ of enhanced c-Jun phosphorylation during decreased JNK activation in the nucleus was described in embryonic cerebellar neurons (Coffey et al. 2000) and neonatal microglia (Waetzig et al. 2005). In both cases, the presence of JNK1, mediating the high basal nuclear activity, decreased in the nucleus, whereas JNK2 and/or JNK3 translocated into the nucleus-induced apoptosis (Bjorkblom et al. 2005). Our findings suggest a similar mechanism with nuclear translocation of JNK2 (and to a lesser extent of JNK3) and a decrease of JNK1 in postnatal cortical neurons. The observation that JNK1 not only contributes to the structural integrity of brain architecture, but also to cellular survival (Chang et al. 2003;Ham et al. 2003; Gao et al. 2005) and not to neural death (Coffey et al. 2000; Eminel et al. 2004) sustains the concept of JNK1 as a ‘physiologic’ isoform. These findings implicate that inhibition of JNKs aiming at neuroprotection should block individual isoforms rather than the total pool of JNKs as discussed in detail recently (Waetzig and Herdegen 2005a). Finally, we have shown that the reduction of caspase-3 activation by JNK inhibition also contributes to neuroprotection of postnatal hippocampal and cortical neurons (Singh et al. 2005; Guan et al. 2006).

JNK-induced neurite elongation

In the adult mammalian brain, JNKs are mainly considered as mediators of neurodegeneration [(Kuan et al. 1999; Borsello et al. 2003; Hunot et al. 2004; Brecht et al. 2005), reviewed by Borsello and Bonny (2004) and Waetzig and Herdegen (2005a)]. Physiological JNK functions, however, have been addressed by comparatively few studies; particularly, the role of JNK2 and JNK3 has hardly been examined. We have recently demonstrated that ‘apoptotic’ JNK3 is involved in neuronal differentiation of PC12 cells (Waetzig and Herdegen 2003) and contributes to the development of the human brain and its functionality (Shoichet et al. 2006). Morever, JNK2 activity is important for nerve growth factor-induced regenerative neurite outgrowth (Waetzig and Herdegen 2005b). Our observation that all JNKs are crucial mediators for neuritogenesis in primary hippocampal neurons adds a new aspect to the understanding that JNKs are not only involved in the induction of cell death.

The analysis of neurite formation in JNK3-ko mice was complicated by the finding that the absence of JNK3 severely disturbs the development of explanted neurons in vitro. In other words, it remains obscure whether JNK3 deficiency only affects neurite outgrowth and/or primary cell growth and adhesion. In any case, the sensitivity of hippocampal neurons for JNK3 deficiency correlates with our previous findings that a truncated JNK3 mutant with loss of the activation domain is not compatible with the survival of transfected neuroblastoma cells; importantly, this truncated JNK3 mutation provokes severe neurological symptoms in humans (Shoichet et al. 2006).

JNK activity underlying neurite outgrowth

The role of JNKs for neuritogenesis depends on the neuronal cell type as inhibition of JNKs by SP600125 did not influence neurite outgrowth of cortical neurons, which was confirmed by sprouting experiments in JNK-ko mice. Still, we and others detected a high basal JNK activity in primary neuronal cells indicating some kind of regulatory function, which might the control of the right number of processes as described for cerebellar granule cells (Coffey et al. 2000). Thus, JNKs have different roles in neurite outgrowth ex vivo depending on the neuronal phenotype and/or the stage of differentiation. In consequence, the functions of JNKs for neuritogenesis have to be defined for each individual type of neurons. At least for hippocampal neurons, JNKs provide a profound impact on neuritogenesis in both, nascent and mature neurons (Oliva et al. 2006).

As for isoform-specificity, expression and activation patterns of JNKs and the data from JNK-deficient mice demonstrate that all isoforms contribute to neurite formation in hippocampal neurons. This means that JNK2 and JNK3 also have physiological functions when the cells are not exposed to some kind of stress. In PC12 cells (which are devoid of JNK3), JNK2, but not JNK1, triggers neurite regrowth following injury (Waetzig and Herdegen 2005b). These observations demonstrate that similar as JNK3, JNK2 is involved in physiological and regenerative responses following injury.

What is the difference between neonatal hippocampal neurons and cortical neurons? In both cell types, the explantation induced a similar strong and ongoing JNK activation and a moderate increase in total JNK expression. Thus, the mere activation of JNKs in whole cell extracts is not indicative for its contribution to regrowth. However, the amount of activated c-Jun factor was distinctly smaller in the SP600125-insensitive cortical neurons as compared with the SP600125-sensitive hippocampal neurons. As c-Jun is an important regulator of axonal regeneration in vivo (Raivich et al. 2004) and can be linked to outgrowth of axons, dendrites, and neurites (Herdegen et al. 1998; Besirli et al. 2005; Brecht et al. 2005; Lindwall and Kanje 2005a,b), this difference of c-Jun expression and activation might be decisive for the role of JNKs during neuritogenesis. Furthermore, axon and dendrite outgrowth is greater and the polarity formation occurs earlier in cortical neurons compared to hippocampal neurons (Ko et al. 2005) so that JNKs might have been important in an earlier phase of development.

How to uncouple JNK-dependent death and survival

How can JNK signaling selectively activate either neuronal (apoptotic) death or regrowth/regeneration? It became apparent in recent studies that this dichotomy is not only a question of apoptotic or physiologic JNK isoforms, but that even a single isoform can confer both [reviewed by Waetzig and Herdegen (2005a)]. Our findings confirm this functional versatility, as all JNK isoforms are involved in neuritogenesis. In consequence, the search for a JNK inhibitor without side-effects appears to be difficult and requires different therapeutic strategies against the neurodegenerative signaling of JNKs (Brecht et al. 2005). Moreover, the mere inhibition of JNKs faces the problem that the otherwise dying neurons are rescued but cannot undergo regenerative programs, i.e. neurons stay alive but are not able to respond to neurotrophins as shown in embryonic chicken motoneurons (Newbern et al. 2007).

Summarizing, our data support the idea that neuroprotection does not require the selective inhibition of one defined JNK isoform, but rather the specific block of degenerative JNK actions (e.g. nuclear translocation with subsequent phosphorylation of particular transcription factors) which leaves the physiological-regenerative functions unaffected.

Acknowledgments

We thank Ms A. Dorst and E. Schroeder for excellent technical assistance. This project was supported by the Deutsche Forschungsgemeinschaft (SPP 1048, SFB 415) and the European Commission (STREP STRESSPROTECT).

Ancillary