Address correspondence and reprint requests to Santosh R. D’Mello, Department of Molecular and Cell Biology, University of Texas at Dallas, 2601N. Floyd Road, Richardson, TX 75083, USA. E-mail: firstname.lastname@example.org
GW5074 a brain-permeable 3′ substituted indolone, protects neurons from death in culture and in an in vivo paradigm of neurodegeneration. Using low potassium (LK) induced apoptosis of cerebellar granule neurons, we report here that the protective action of GW5074 is mediated through the activation of B-Raf. Over-expression of a kinase-dead form of B-Raf blocks the ability of GW5074 to neuroprotect, whereas over-expression of active forms of B-Raf protect even in the absence of GW5074. Although mitogen-activated protein kinase kinase (MEK) and extracellular signal-regulated protein kinase (ERK) are activated by GW5074, pharmacological inhibition of MEK-ERK signaling by U0126 or PD98059 does not reduce neuroprotection suggesting that B-Raf signals through a non-canonical signaling pathway. GeneChip microarray analyses identified activating transcription factor-3 (ATF-3) as a gene whose expression is induced by LK but that is negatively regulated by GW5074. Forced inhibition of ATF-3 expression using siRNA protects neurons against LK-induced apoptosis, whereas the over-expression of ATF-3 blocks GW5074-mediated neuroprotection. Not unexpectedly, expression of active B-Raf inhibits the apoptosis-associated increase in ATF-3 expression. We extended our work to include three other 3′ substituted indolones – a commercially available inhibitor of RNA-dependent protein kinase and two novel compounds designated as SK4 and SK6. Like GW5074, RNA-dependent protein kinase inhibitor, SK4, and SK6 all inhibited c-Raf in vitro but activated B-Raf in neuronal cultures. All four compounds also inhibited ATF-3 expression. Taken together our results indicate that all four indolones mediate neuroprotection by a common mechanism which involves B-Raf activation, and that a downstream target of B-Raf is ATF-3.
Neurodegenerative diseases disrupt the quality of the lives of patients and often lead to their death prematurely. A common feature of these pathologies is the abnormal degeneration of neurons, which results from an inappropriate activation of apoptosis. Drugs that inhibit neuronal apoptosis could thus be candidates for therapeutic intervention in neurodegenerative disorders. Moreover, identifying the molecular targets of such neuroprotective drugs and understanding the signal transduction pathways that are utilized in their action would lead to the development of more effective therapeutic strategies. In 2004, we reported the identification of a pharmacological compound, GW5074 that completely inhibits neuronal cell death in vitro (Chin et al. 2004). GW5074 also prevents striatal degeneration and improves behavioral performance in mice administered with 3-nitropropionic acid (3-NP), a commonly used in vivo paradigm of Huntington’s disease. GW5074 could therefore have therapeutic value against neurodegenerative pathologies and understanding how it acts is therefore of significance.
GW5074 is a potent inhibitor of c-Raf when tested in vitro (Lackey et al. 2000; Chin et al. 2004). Surprisingly, however, the treatment of cultured neurons with GW leads to c-Raf activation (Chin et al. 2004). The phosphorylation of mitogen-activated protein kinase kinase (MEK) and extracellular signal-regulated protein kinase (ERK), downstream effectors of Raf kinases, is also stimulated by GW5074. Such a paradoxical activation of c-Raf has also been reported to occur in cell lines treated with other structurally distinct c-Raf inhibitors (Hall-Jackson et al. 1999a,b). Although the molecular basis for this paradoxical activation is not understood, it is likely that cells respond to GW5074 treatment by triggering compensatory mechanisms, such as the accumulation of activating modifications on c-Raf, which are revealed in assays performed in the absence of the inhibitor.
GW5074 is a 3′ substituted indolone. We recently showed that several other commercially available 3′ substituted indolones designed to inhibit a variety of other protein kinases are also capable of preventing neuronal apoptosis (Johnson et al. 2005). In this study, we extended our investigation to include another commercially available 3′ substituted indolone sold as an inhibitor against RNA-dependent protein kinase (PKR) designated in this report as PKRi (Jammi et al. 2003). To identify chemical groups within 3′ substituted indolones that contribute to neuroprotective efficacy we initiated a preliminary structure activity relationship (SAR) analysis using GW5074 as the parent compound. This analysis led to the identification of two other compounds SK4 and SK6.
Here, we report that PKRi, SK4, and SK6 are all neuroprotective. Like GW5074, these compounds inhibit c-Raf in vitro. We show that all four compounds also activate B-Raf and that their neuroprotective action is mediated by B-Raf. Finally, we report that neuroprotection by B-Raf does not involve the well-described MEK–ERK signaling pathway but acts through the suppression of activating transcription factor-3 (ATF-3) of the ATF/cAMP response element binding protein family.
Materials and methods
Unless specified otherwise, all chemicals, including GW5074, were purchased from Sigma Chemicals (St Louis, MO, USA). PKRi, PD98059, and U0126 were purchased from Calbiochem (La Jolla, CA, USA). SK4, SK6, and SK9 were synthesized by Edward Biehl (Department of Chemistry, Southern Methodist University, Dallas, TX, USA). Enhanced polyvinylidene difluoride membrane was from Bio-Rad (Hercules, CA, USA). Enhanced chemiluminescence was from GE Healthcare (Piscataway, NJ, USA). Phospho-MEK, MEK, and Phospho-ERK antibodies were purchased from Cell Signaling Technology (Beverly, MA, USA); c-Raf, B-Raf, c-Jun, and ATF-3 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA); c-myc, Flag, and tubulin were purchased from Sigma (St Louis, MO, USA). A/G PLUS-Agarose was also obtained from Santa Cruz Biotechnology. All cell culture media used in these experiments were purchased from Invitrogen (Carlsbad, CA, USA). Trizol and Thermoscript reverse transcriptase PCR (RT-PCR) system were also from Invitrogen. Terminal transferase dUTP nick end labeling (TUNEL)-staining was performed using the DeadEnd Fluorometric TUNEL system from Promega (Madison, WI, USA). The B-Raf constructs, myc-B-Raf KD, myc-B-Raf CAT, and myc-B-Raf V600E, are kind gifts from Michael White’s lab at the University of Texas-Southwestern Medical Center (Dallas, TX, USA). The ability of these constructs to increase or block B-Raf activity has been tested in 293T (Supplementary Fig. S1). The expression plasmid encoding kinase-dead MEK (GST-MEK1 K97M) was also a kind gift from Melanie Cobb, the University of Texas-Southwestern Medical Center. The ATF-3 gene promoter-luciferase reporter plasmid (pGL3-ATF3-Luc) was provided by Seung Baek, University of Tennessee (Knoxville, TN, USA). ATF-3 siRNA was purchased from Ambion (Austin, TX, USA). p-TARGET™ vector and PCR master mix were from Promega.
Neuronal culture, treatments, and viability assay
Cerebellar granule neurons were obtained from dissociation of 7- to 8-day-old rats as described previously (D’Mello et al. 1993). The cells were plated in basal minimal Eagle’s medium, supplemented with 10% fetal bovine serum (FBS), 25 mM KCl, 2 μM glutamine, and 0.2% gentamycin (D’Mello et al. 1993), in 24-well dishes (1 × 106 cells/well for viability assay), or 60-mm dishes (12 × 106 cells/dish for immunoprecipitation and western blotting). Cytosine arabinoforanoside (10 μM) was added to the culture medium 18–22 h after plating to prevent replication of non-neuronal cells. Unless indicated otherwise, treatments were performed 6–7 days after plating. For treatments, the cultures were switched to serum-free basal minimal Eagle’s medium in the absence [low potassium (LK) medium] or presence of 25 mM KCl [high potassium (HK) medium]. Purchased or synthesized compounds (GW5074, PKRi, SK4, SK6, and SK9) were added to LK medium at the time the medium was switched. After 24 h treatment, cells were fixed and stained with DNA fluorescent dye, 4′6′-diamidino-2-phenylindole hydrochloride. Cell viability was quantified by visualizing the morphology of the nuclei. Apoptotic cell death is revealed by nuclei condensation and/or fragmentation.
Cortical cultures were obtained from the cerebral cortex of Wistar rats (day 17 of gestation) as described previously (Murphy et al. 1990). Cortical neurons were treated with 1 mM homocysteic acid (HCA) for 24 h to induce oxidative stress 1 day after plating or 5 μM camptothecin for 6 h 3 days after plating. HCA was made as 150 mM stock solution that was adjusted to pH 7.5.
Equivalent amounts of protein were mixed with 6x sodium dodecyl sulfate (SDS) sample buffer (375 mM Tris–HCl, pH 6.8 at 25°C, 12% SDS, 60% glycerol, 300 mM dithiothreitol, and 0.012% bromophenol blue). This was followed by heating at 95°C for 5 min, subjected to SDS–polyacrylamide gel electrophoresis, and transferred electrophoretically to polyvinylidene difluoride membrane. After staining with Ponceau S to verify uniformity of protein loads/transfer, the membranes were blocked. Incubation with primary antibodies was performed overnight at 4°C or for 1 h at 25°C followed by secondary antibodies for 1 h at 25°C. Immunoreactivity was developed by enhanced chemiluminescence and visualized by autoradiography.
ATF-3 cloning, transfection, and immunocytochemistry
Flag-ATF-3 was cloned to p-TARGET™ vector with reverse primer 5′-TAAGCTCTGCAATGTTCCTTCTT-3′ and forward primer 5′-ATGATGCTTCAACACCCAGG-3′ from an embryonic human brain cDNA library. Plasmids were purified by performing CsCl equilibrium centrifugation. Cerebellar granule neurons are able to be transfected after 4–5 days following plating by using a well-established calcium–phosphate protocol (Dudek et al. 1997). Briefly, cell media was replaced with serum-free Dulbecco’s modified Eagle’s medium (DMEM) for 1 h. Original conditioned media was saved at 37°C in a CO2 incubator. For each well a mixture was prepared as follows: 15 μL of 2x HEPES-buffered saline buffer, pH 7.05, was combined with 15 μL of a 0.25 M CaCl2 solution containing 3 μg of plasmid DNA or 40–80 pmole of ATF-3 siRNA. The mixtures were vortexed and incubated at 25°C for 30 min. The mixture was added to the cells drop-wise and allowed to incubate for another 45 min. After washing cells three times with serum-free DMEM, original medium was added back to cultures. One day following transfection, cerebellar granule neurons were treated with HK, LK, or LK plus drugs for 24 h for immunocytochemistry or 3 h for western blotting. Immunocytochemistry was performed by fixing cells for 20 min with 4% formaldehyde. After blocking in 0.1 M phosphate-buffered saline containing 5% bovine serum albumin and 5% goat serum for 30 min, immunostaining was performed by incubation with the primary antibody for 90 min and the secondary antibody for 45 min at 25°C. During washing of the secondary antibody, nuclei were stained with 4′6′-diamidino-2-phenylindole hydrochloride for 10 min at 25°C.
Immunoprecipitation and kinase assay
Two hundred and fifty micrograms of protein was incubated with 1.0 μg of primary antibody and 12 μL of Protein A/G PLUS-Agarose beads overnight. Immunoprecipitates were collected by centrifugation at 3000 g for 30 s and washed twice with lysis buffer, twice with lysis buffer supplemented with 350 mM NaCl, and twice with kinase buffer (25 mM HEPES, pH 7.4, and 10 mM MgCl2). Purified recombinant GST-MEK1 K97M protein was added as a substrate in kinase buffer supplemented with 85 μM ATP for 35 min at 30°C. For in vitro kinase assays, drugs were added in kinase buffer and incubated for 5 min at 30°C prior to kinase reaction. The kinase reactions were stopped by addition of 6x SDS sample buffer and boiled for 5 min. Proteins were resolved by SDS–polyacrylamide gel electrophoresis and subjected to western blotting. The level of kinase activity was detected by a phospho-MEK antibody.
GeneChip microarray analysis
Total RNA was extracted from cultures treated with or without 1 μM GW using Trizol according to the manufacturer’s instructions. Integrity of the RNA was confirmed by gel electrophoresis. To reduce false-positives resulting from technical variance and potential biological variation between samples, three independent sets of RNA samples were prepared from three different sets of cerebellar granule neuron cultures and separately subjected to microarray analysis. Probe preparation and hybridizations were performed through the NIH MicroArray Consortium at the Translational Genomic Research Institute (TGRI, Tempe, AZ, USA) on a fee-for-service basis using the Affymetrix GeneChip Rat Expression 230 2.0 Array (Affymetrix, Santa Clara, CA, USA), which permits analysis of the expression of over 31 000 transcripts and variants from over 28 000 well-substantiated rat genes. The qvalue program from GeneSpring R-integration package (Agilent Technologies, Santa Clara, CA, USA) was used for data analysis.
RNA preparation and semi-quantitative RT-PCR
RNA was extracted from cultured neurons by using Trizol according to the manufacturer’s instructions. Three micrograms of total RNA from each sample was reverse transcribed by Thermoscript reverse transcriptase PCR (RT-PCR) system (Invitrogen) according to the manufacturer’s instructions. PCR was performed with PCR master mix (Promega). The primers used for PCR amplification were as follows: Rat glyceraldehyde 3-phosphate dehydrogenase, forward 5′-CCATCACCATCTTCCAGGAG-3′ and reverse 5′-CCTGCTCACCACCTTCTTG-3′; Rat β-Adducin, forward 5′-GGCGATGCAGATACCAAAGAT-3′ and reverse 5′-ATCTCCTCTTGCCCTCCTTG-3′; Rat insulin-related protein-2, forward 5′-GTGTGGTTTCAGAACAAGCGTT-3′ and reverse 5′-TTACAAGGGCCGGAGATTTT-3′; Rat SH3/ankyrin domain gene 3, forward 5′-TAAGCTCCATCCACGTCACTC-3′and reverse 5′-ATCTCTGTGGGGTCTATCACAA-3′; Rat Jun, forward 5′-TGAAGCAGAGCATGACCTTG-3′ and reverse 5′-GACACTGGGCAGCGTATTCT-3′; Rat ATF-3, forward 5′-ATGCTTCAACATCCAGGCCA-3′ and reverse, 5′-CGCCTCCTTTTTCTCTCATCT-3′; Rat stathmin-like 4, forward 5′-GGCTGCAAGAGAAGGACAA-3′ and reverse 5′-TGACAATGAACAGACGCACA-3′; Rat tumor necrosis factor (ligand) superfamily member 11, forward 5′-TATGATGGAAGGTTCGTGGCT-3′ and reverse 5′-AGTACGTCGCATCTTGATCCG-3′.
In vivo 3-nitropropionic acid treatment
3-Nitropropionic acid administration and analysis of brain sections was performed as previously described (Chin et al. 2004). Briefly, mice were injected intraperioneally with 3-NP at 50–55 mg/kg twice a day for 4 days. The following day, the mice were deeply anesthetized, brains removed, washed in phosphate-buffered saline, and rapidly frozen in Cryo-Stat embedding medium (StatLab Medical Products, Lewisville, TX, USA). Fifty micron coronal sections were cut on a cryostat and stained for Nissl substance with cresyl violet (Chin et al. 2004). For western blotting analysis, striatum from control or 3-NP injected animal were homogenized in RIPA buffer. Then, protein was extracted after centrifugation for 15 min at 15 000 g. Protein was also extracted from the rest of the brain (lacking striatum) and used for western blotting.
N1E-115 cell culture and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay
The N1E-115 cell line was purchased from ATCC (Manassas, VA, USA) and cultured in DMEM with 4.5 g/L glucose (without sodium pyruvate) supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin. N1E-115 cells were plated into 24-well tissue culture dishes at 5 × 104 per well 1 day before 6-hydroxydopamine (6-OHDA) treatment. Cell death was determined 24 h later by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as previously described (Koulich et al. 2001).
N1E-115 cells were cultured in antibiotic-free DMEM media containing 10% FBS 1 day before transfection. Transient transfection was performed using Lipofectamine 2000 (Invitrogen). Cultured/transfected media was switched to regular N1E-115 culture media 2 h after transfection. The next morning, cells were treated with or without 6-OHDA for 6 h. Cells then were harvested and luciferase activity determined by using BD Monolight Enhanced Luciferase Assay Kit (BD Biosciences, San Jose, CA, USA) following the manufacturer’s instructions.
GW5074 activates both B-Raf and c-Raf kinase activity in cultured cerebellar granule neurons
GW5074 inhibits c-Raf when tested in vitro on purified enzyme (Lackey et al. 2000; Chin et al. 2004). As we have previously reported, however, when cultured cerebellar granule neurons induced to undergo apoptosis are treated with GW5074, the activity of endogenous c-Raf is robustly activated (Chin et al. 2004; Fig. 1a). This paradoxical activation of c-Raf has also been observed by other investigators with other structurally distinct c-Raf inhibitors (Hall-Jackson et al. 1999a,b; Chin et al. 2004). Not surprisingly, when activated c-Raf that is immunoprecipitated from GW5074-treated neuronal cultures is exposed to GW5074 readded in vitro, the kinase activity of c-Raf is again inhibited (Chin et al. 2004; Fig. 1a). In addition to c-Raf, B-Raf activity is stimulated by GW5074 in neuronal cultures switched to LK-treatment (Fig. 1b). To examine how quickly c-Raf and B-Raf activation occurs, we immunoprecipitated these enzymes from LK-treated neuronal cultures exposed to GW5074 for 10 min, 20 min, 1 h, and 2 h, and evaluated their activation status in vitro. As shown in Fig. 1c and d, activation of both c-Raf and B-Raf is clearly discernible at 10 min. The rapidity at which c-Raf and B-Raf activation occurs suggests a more direct effect of GW5074 on enzyme activity as opposed to a feedback-loop mechanism activated as a consequence of enzyme inhibition.
B-Raf activity is required for GW5074-mediated neuroprotection
As the activation of B-Raf and c-Raf is an early response to GW5074 treatment, we hypothesized that either B-Raf or c-Raf activation is important for GW5074-mediated neuroprotection. To test this hypothesis we performed over-expression experiments in which active or kinase-dead forms of the two kinases were introduced into neurons which were then switched to HK or LK medium. Preliminary studies indicated that the over-expression of c-Raf constructs do not have a substantial effect on the ability of GW5074 to neuroprotect (data not shown). Over-expression of a kinase-dead form of B-Raf on the other hand, blocked GW5074-mediated neuroprotection (Fig. 2a). Furthermore, the over-expression of either of two different constitutively active forms of B-Raf prevented LK-induced neuronal death even in the absence of GW5074 (Fig. 2b). These results suggest that B-Raf activity is necessary for GW5074-mediated neuroprotection.
ATF-3 is a downstream mediator of neuroprotection by GW5074 and B-Raf
Although Raf kinases typically act through MEK and ERK, this pathway is not necessary for GW5074-mediated neuroprotection (Chin et al. 2004). To gain insight into downstream regulators of GW5074, we performed GeneChip microarray analyses. These experiments identified 12 mRNAs that were up-regulated in neurons by GW5074 treatment and 13 mRNAs that were down-regulated (Table 1). Altered expression of a subset of these mRNAs was validated by RT-PCR analyses (Fig. 3). Among the genes whose expression was confirmed to be down-regulated by GW5074 is the one encoding ATF-3. ATF-3 has been implicated in the promotion of neuronal death in some experimental paradigms (Hai et al. 1999; Vlug et al. 2005). In cerebellar granule neuron cultures, ATF-3 is induced after switching to LK media within 3 h (Fig. 4a and b). Previous studies have demonstrated that in this paradigm, neurons are irreversibly committed to death within 4–6 h of LK treatment (Galli et al. 1995; Schulz et al. 1996; Nardi et al. 1997; Borodezt and D’Mello 1998). Thus, ATF-3 expression occurs prior to the point at which the neurons are committed to dying. More long-term analysis of ATF-3 indicates that its expression remains elevated for at least 12 h after LK treatment before declining to normal levels (Fig. 4c). A similar profile of expression is observed with c-Jun. In comparison, caspase 3 activation, as judged by generation of the 17 kDa proteolytic product, begins later and is more sustained (Fig. 4c). It is noteworthy that ATF-3 has been shown to play an essential role in stress-induced caspase 3 activation (Lu et al., 2007).
Table 1. Genes that are regulated by GW5074 treatment of cerebellar granule neurons
Three independent sets of RNA were prepared from CGNs treated with LK or LK plus 1 μM GW5074 for microarray analysis. Candidate genes were identified as those exhibiting 1.5-fold or greater change with GW5074 treatment compared with LK alone. A list of genes was created with q < 0.05 using Student’s t-test.
Tumor necrosis factor (ligand) superfamily, member 11
To examine if the stimulation of ATF-3 played a causal role in LK-mediated neuronal death, we over-expressed it in neurons. As shown in Fig. 5a, the over-expression of ATF-3 reduces neuronal survival in HK and blocks the ability of GW5074 to neuroprotect in LK medium. The pro-apoptotic effect of ATF-3 on HK and GW5074-treated neurons was confirmed by TUNEL staining (Supplementary Fig. S2). This finding suggests that ATF-3 promotes apoptosis in LK and that GW5074 and HK maintain neuronal survival by inhibiting the increase in ATF-3 expression. To verify this conclusion, we knocked down ATF-3 expression using siRNA. As shown in Fig. 5b, the forced down-regulation of ATF-3 expression reduces neuronal death induced by LK medium.
We looked at ATF-3 expression in two other well-established paradigms of neurodegeneration. Camptothecin is a DNA damaging agent which induces apoptosis in cultured cortical neurons (Morris and Geller 1996). In contrast to LK-induced death which is mediated in a p53-independent mechanism, neuronal death by camptothecin is p53-dependent (Enokido et al. 1996; Xiang et al. 1998; Morris et al. 2001). As shown in Fig. 5c, ATF-3 is also up-regulated in cortical neurons primed to die by camptothecin treatment. HCA treatment of cortical neurons is also used widely as a paradigm to study the mechanisms underlying neuronal death. By causing glutathione depletion, HCA induces oxidative stress leading to apoptosis (Murphy et al. 1990; Ratan et al. 1994a,b). As shown in Fig. 5d, HCA treatment induces ATF-3 expression in cortical neurons. We extended our studies to an in vivo paradigm of neurodegeneration. Administration of 3-NP to mice causes selective striatal neurodegeneration and is widely used as an experimental model of Huntington’s disease. As shown in Fig. 5e, ATF-3 expression is stimulated in the degenerating striatum of 3-NP administered mice (also see Supplementary Fig. S3 for evidence of striatal degeneration). In contrast, the expression of ATF-3 is not increased in the non-striatal brain tissue, which does not display significant neurodegeneration. These results suggest that ATF-3 stimulation contributes to cell death in multiple neuronal types and in response to other apoptotic stimuli.
We next proceeded to examine whether the increased expression of ATF-3 was transcriptionally mediated or mediated post-transcriptionally. As primary neurons are poorly transfected, we utilized the dopaminergic N1E-115 cell line to investigate this issue. Treatment of N1E-115 with 6-OHDA induces apoptosis (Fig. 6a and b). As observed in primary neurons, ATF-3 expression is up-regulated by 6-OHDA (Fig. 6c). Using a promoter construct containing 514 bp of upstream sequence of the ATF-3 gene fused to a luciferase reporter we find that 6-OHDA results in a small but significant increase in transcriptional activity (Fig. 6d).
To investigate whether the suppression of ATF-3 by GW5074 (Fig. 7a) was mediated by B-Raf, we over-expressed B-Raf in neuronal cultures. The over-expression of an active form of B-Raf reduced the apoptosis-associated stimulation of ATF-3 expression even in the absence of GW5074 (Fig. 7b). Another molecule that is known to play a pivotal role in neuronal apoptosis is the c-Jun transcription factor. The need for c-Jun activation in the induction of LK-mediated death of cerebellar granule neurons as well as other paradigms of neurodegeneration is well-established (Estus et al. 1994; Ham et al. 1995; Watson et al. 1998). GW5074 inhibits the apoptosis-associated stimulation of c-Jun expression (Fig. 7a). As shown in Fig. 7b, the expression of active B-Raf also blocks c-Jun induction by LK.
Other neuronal protective compounds also utilize similar mechanisms while mediating neuronal survival
In a previous study, we reported the identification of other 3′ substituted indolones which were also capable of preventing neuronal apoptosis (Johnson et al. 2005). In this study, we investigated PKRi, an inhibitor of PKR first described by Jammi et al. (2003) and that has a 3′ substituted indolone core. As shown in Fig. 8, treatment with PKRi prevented the death of cerebellar granule neurons induced by LK. In addition to PKRi, two other 3′ substituted indolones, SK4 and SK6, synthesized as part of an ongoing SAR analysis were also found to be neuroprotective (Fig. 8). In contrast to SK4 and SK6, a closely related 3′ substituted indolone designated as SK9, fails to block LK-induced neuronal death at any of the concentrations tested (Fig. 8). A description of the SAR analysis along with other compounds analyzed will be described elsewhere.
Given their structural similarity, it was possible that PKRi, SK4, and SK6 exerted their neuroprotective action by a common mechanism. We first examined whether the three compounds inhibited c-Raf in vitro. As shown in Fig. 9a, all three neuroprotective compounds (SK4, SK6, and PKRi) inhibit the activity of c-Raf that had been immunoprecipitated from neurons. In comparison, SK9 is less effective. While SK6 and SK4 also produced the paradoxical activation of c-Raf in intact neuronal cultures, PKRi did not. This finding suggested that the in vivo activation of c-Raf is not an essential feature for neuroprotection by 3′ substituted indolones (Fig. 9b). In contrast, PKRi, SK4, and SK6 all activated B-Raf kinase activity while treated in neuronal culture (Fig. 9c). PKRi, SK4, and SK6 also inhibited the apoptosis-associated stimulation of ATF-3 and c-Jun expression that is triggered in cerebellar granule neurons (Fig. 9d and e).
Since its discovery in 2003, PKRi has been used in a number of studies and a variety of biological effects produced by this compound have been attributed to its ability to inhibit PKR (Page et al. 2006; Shimazawa and Hara 2006; Ingrand et al. 2007). As described above, PKRi does not activate c-Raf in response to GW5074 treatment. PKRi also inhibits c-Jun and ATF-3 expression much more substantially that GW5074, SK4, and SK6 (Fig. 9). It was therefore conceivable that although sharing structural similarity with the other three indolones, PKRi mediates neuroprotection via a distinct mechanism. To address this issue, we examined more rigorously whether B-Raf is necessary for neuroprotection by PKRi. As shown in Fig. 2a, over-expression of the kinase-dead form of B-Raf blocks neuroprotection by PKRi. Over-expression of ATF-3 also inhibits PKRi-mediated neuroprotection (Fig. 5a) as was observed with GW5074.
Our results indicate that GW5074 exerts its neuroprotective action by activating B-Raf. Indeed, the over-expression of a kinase-dead form of B-Raf blocks neuroprotection whereas expression of an active form of B-Raf promotes neuronal survival even in the absence of GW5074.
Our conclusion that B-Raf has neuroprotective effects is consistent with the finding that sensory neurons and motor neurons cultured from B-Raf-deficient embryos (but not from c-Raf or A-Raf deficient embryos) fail to survive in response to neurotrophic factors (Wiese et al. 2001). In response to growth factor stimulation, activated B-Raf directly phosphorylates MEK, a serine–threonine kinase that then phosphorylates and activates ERK (Baccarini 2002; Hindley and Kolch 2002). Not unexpectedly, we find that MEK and ERK are stimulated by all four neuroprotective compounds (Supplementary Fig. S4a). Surprisingly, the inhibition of MEK–ERK signaling has no effect on neuroprotection (Supplementary Fig. S4b). It is noteworthy that while MEK remains the most widely accepted substrate, recent studies have identified a number of other proteins including Bad, p53, and Rb that can be directly phosphorylated by Raf (Bonni et al. 1999; Baccarini 2002; Hindley and Kolch 2002). Moreover, Raf is able to prevent apoptosis by interacting with the anti-apoptotic molecule, Bag-1(Rudolf Götz et al. 2005), or sequestering the pro-apoptotic molecule apoptosis signal-regulating kinase-1 (Chen et al. 2001) through a MEK and ERK independent pathway. Although further work is needed to clarify the mechanism by which it occurs, we find that neuroprotection by B-Raf involves the inhibition of the ATF-3 transcription factor whose expression is induced in neurons primed to undergo apoptosis. Over-expression of active B-Raf blocks the apoptosis-induced increase in ATF-3 expression. A relationship between B-Raf and ATF-3 has been described in actively proliferating cell lines (Nilsson et al. 1997). In these cells, B-Raf has been described to actually stimulate ATF-3 expression. Whether ATF-3 is induced or suppressed by B-Raf may therefore be cell and context specific.
Activating transcription factor-3 is a member of the ATF/cAMP response element binding protein family of basic leucine zipper-type transcription factors which can form both homo- and heterodimers with other basic leucine zipper-type transcription factors (Hai et al. 1999). While ATF-3 homodimers are believed to inhibit gene transcription, heterodimers can activate transcription. The most commonly described dimerization partners for ATF-3 are members of the Jun family (c-Jun, Jun B, and Jun D). ATF-3 expression is generally induced by cellular damage and a variety of stress stimuli in different tissue types (Hai et al. 1999). In the nervous system, ATF-3 has been implicated in neuronal degeneration following spinal cord damage (Tsujino et al. 2000; Averill et al. 2004; Huang et al. 2006), cerebral artery occlusion (Ohba et al. 2004), and neurotoxin-induced oxidative stress (Holtz et al. 2006). Elevated ATF-3 expression also precedes the death of spinal motor neurons in a transgenic mouse model of amyotrophic lateral sclerosis (Vlug et al. 2005) and of retinal ganglion cells after crushing of the optic nerve (Takeda et al. 2000). In addition to cerebellar granule neurons, we find that ATF-3 expression is increased in cortical neurons induced to die by oxidative stress and DNA damage, and in the degenerating striatum of 3-NP administered mice. In some cases however, ATF-3 has been reported to promote neuronal survival. For example, ATF-3 over-expression protects hippocampal neurons from kianic acid-induced degeneration (Francis et al. 2004) and sympathetic neurons from nerve growth factor deprivation-induced apoptosis (Nakagomi et al. 2003). Whether ATF-3 kills or promotes neuronal survival may be dependent on whether it acts as a homo- or heterodimer or on other proteins with which it associates within the specific cell type. Other mechanisms such as differential post-translational modifications on ATF-3 may also contribute to the different roles that ATF-3 displays. Initial indications based on the utilization of an ATF-3 promoter-luciferase reporter construct is that the increase in ATF-3 expression during apoptosis is mediated at least in part at the transcriptional level. The small increase in promoter activity we observe in 6-OHDA-treated N1E-115 cells does not, however, explain the substantial increase in ATF-3 protein expression. It is possible that the 514 bp of 5′ genomic sequence that we have utilized does not contain all the transcriptional elements necessary for full induction. Alternatively, post-transcriptional mechanisms such as increased mRNA stability might contribute to increased expression. Post-transcriptional regulation of ATF-3 expression has been suggested in other studies (Lu et al., 2007).
In cerebellar granule neurons, ATF-3 induction is causally involved in promoting apoptosis. Indeed, the over-expression of ATF-3 in healthy neurons is sufficient to induce apoptosis whereas the forced suppression of ATF-3 reduces the extent of LK-induced neuronal death. The partial protection is likely to reflect the incomplete suppression of ATF-3 expression by the siRNA used in this study. It is also possible that suppression of ATF-3 alone is not sufficient for complete protection and that other molecules cooperate with ATF-3 to promote neuronal death. One such molecule is c-Jun. Several different investigators have established the importance of c-Jun in promoting neuronal death (Estus et al. 1994; Ham et al. 1995; Herdegen et al. 1998; Watson et al. 1998; Crocker et al. 2001). Interestingly, the apoptosis-associated stimulation of c-Jun expression is also inhibited by active B-Raf.
Another 3′ substituted indolone utilized in this study is PKRi, a compound identified based on its ability to inhibit PKR activity in vitro (Jammi et al. 2003). We have discovered that PKRi protects against LK-induced death of cerebellar granule neurons. While this study was in progress other investigators also reported neuroprotective actions of PKRi in other experimental paradigms of neuronal death (Page et al. 2006; Shimazawa and Hara 2006; Ingrand et al. 2007). Some investigators have also reported that PKR activity is elevated in certain tissue culture models of neurodegeneration and in the brains of patients with Alzheimer’s diseases raising the possibility that activation of PKR may contribute to neurodegeneration (Bando et al. 2005; Page et al. 2006; Eley et al. 2007; Shimazawa et al. 2007). We find that PKRi stimulates B-Raf activity and that its neuroprotective effect in cerebellar granule neurons is dependent, at least in part, on B-Raf. Furthermore, PKRi also blocks the apoptosis-associated induction of ATF-3. As observed with GW5074, over-expression of ATF-3 blocks the ability of PKRi to be neuroprotective. At the present time we do not know whether activation of PKR is necessary for apoptosis in the paradigms we have investigated comprising cerebellar granule neurons or cortical neurons. Preliminary work in cerebellar granule neurons indicates that PKRi treatment does not inhibit the phosphorylation of eIF2α, which is a downstream target of PKR (Chen and D’Mello, unpublished observation). Further studies are necessary to determine whether PKR inhibition is involved in the neuroprotective action of PKRi, or whether this compound acts through separate effects on B-Raf activity. In addition to PKRi, we have studied two other 3′ substituted indolones which were identified in a SAR screen. Both SK4 and SK6 activate B-Raf and suppress ATF-3 expression.
In summary, we identify B-Raf and ATF-3 as important players in the regulation of neuronal survival. These proteins can serve as molecular targets in the development of neuroprotective therapeutic strategies. Furthermore, our studies identify 3′ substituted indolones as a core chemical structure that can be used as a starting point for the development of therapeutic drugs against neurodegenerative disorders. At the present time we do not know whether activation of PKR is necessary for apoptosis in the paradigms we have investigated comprising cerebellar granule neurons or cortical neurons. Perliminary work in cerebellar granule neurons indicates that PKRi treatment does not inhibit the phosphorylation of eIF2α, which is a downstream target of PKR.
This work was supported by a grant from the National Institute of Neurological Diseases and Stroke (NS047201) to SRD. We acknowledge the help of Edward Biehl and Kamila Sukanta (Southern Methodist University, Dallas, TX, USA) who synthesized some of the compounds used in this study. We thank Michael White (University of Texas Southwestern Medical Center, Dallas, TX, USA) for the c-Raf and B-Raf expression constructs and Seung Baek (University of Tennessee, Knoxville, TN, USA) for the ATF-3 gene promoter-luciferase reporter plasmid (pGL3-ATF3-Luc).