Apoptosis inducing factor mediates caspase-independent 1-methyl-4-phenylpyridinium toxicity in dopaminergic cells

Abstract

Parkinson’s disease is a debilitating neurodegenerative disease characterized by loss of midbrain dopaminergic neurons. These neurons are particularly sensitive to the neurotoxin 1-methyl-4-phenylpyridinium (MPP⁺), the active metabolite of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), which causes parkinsonian syndromes in humans, monkeys and rodents. Although apoptotic cell death has been implicated in MPTP/MPP⁺ toxicity, several recent studies have challenged the role of caspase-dependent apoptosis in dopaminergic neurons. Using the midbrain-derived MN9D dopaminergic cell line, we found that MPP⁺ treatment resulted in an active form of cell death that could not be prevented by caspase inhibitors or over-expression of a dominant negative inhibitor of apoptotic protease activating factor 1/caspase-9. Apoptosis inducing factor (AIF) is a mitochondrial protein that may mediate caspase-independent forms of regulated cell death following its translocation to the nucleus. We found that MPP⁺ treatment elicited nuclear translocation of AIF accompanied by large-scale DNA fragmentation. To establish the role of AIF in MPP⁺ toxicity, we constructed a DNA vector encoding a short hairpin sequence targeted against AIF. Reduction of AIF expression by RNA interference inhibited large-scale DNA fragmentation and conferred significant protection against MPP⁺ toxicity. Studies of primary mouse midbrain cultures further supported a role for AIF in caspase-independent cell death in MPP⁺-treated dopaminergic neurons.

Parkinson’s disease is a debilitating neurodegenerative movement disorder characterized by loss of monoaminergic neurons, particularly dopaminergic neurons of the ventral midbrain. The neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetra-hydropyridine (MPTP) causes a parkinsonian pattern of neuron loss in humans and other mammals. The active metabolite of MPTP, 1-methyl-4-phenylpyridinium (MPP⁺), is internalized by the dopamine transporter. Although MPP⁺ acts as a complex I inhibitor, other mechanisms may also contribute to toxicity (Dauer and Przedborski 2003). The observations that transgenic mice overexpressing anti-apoptotic Bcl-2 (Offen et al. 1998; Yang et al. 1998) or inhibitor of apoptosis protein (Eberhardt et al. 2000), and that knockout mice lacking pro-apoptotic bax (Vila et al. 2001), are all protected from MPTP toxicity suggests a role for regulated pathways of active cell death in MPTP/MPP⁺ toxicity.

The classic apoptotic pathway of regulated cell death ultimately involves cleavage and activation of effector caspases such as caspase-3, often following mitochondrial cytochrome c release and formation of apoptotic protease activating factor 1 (Apaf-1)/caspase-9 complexes. Although apoptotic morphology, cytochrome c release and cleaved caspase-3 expression have been described in animal models (Tatton and Kish 1997; Yang et al. 1998; Turmel et al. 2001) and some culture models of MPP⁺ toxicity (Mochizuki et al. 1994; Dodel et al. 1998; Eberhardt and Schulz 2003), the ability of caspase inhibitors to confer protection has been inconsistent (Lotharius et al. 1999; Hartmann et al. 2001; Bilsland et al. 2002; Han et al. 2003; Yang et al. 2004). These observations suggest that alternative death pathways can be activated by MPP⁺.

Apoptosis inducing factor (AIF) is a more recently characterized apoptogenic factor that mediates caspase-independent cell death in other systems (Susin et al. 1999; Daugas et al. 2000; Cregan et al. 2002). AIF normally resides in the intermembrane space of mitochondria, and is ubiquitously expressed in brain tissues (Cao et al. 2003). AIF release from mitochondria precedes large-scale DNA fragmentation (50 kbp) (Daugas et al. 2000). Nuclear translocation of AIF is elicited by transient cerebral ischemia (Cao et al. 2003) or traumatic brain injury in rats (Zhang et al. 2002). AIF also mediates poly(ADP-ribose) polymerase-1-dependent forms of cell death (Yu et al. 2003). Caspase inhibitors do not affect mitochondrial release of AIF (Daugas et al. 2000; Cao et al. 2003). Thus, we hypothesized that AIF release may contribute to MPP⁺-elicited cell death.

We studied the potential role of AIF in MPP⁺ toxicity using a dopamine-producing midbrain-derived neuronal cell line and primary midbrain neuronal cultures. Our data indicate that MPP⁺ elicited nuclear translocation of AIF. Whereas caspase inhibitors were unable to reduce MPP⁺ toxicity, inhibiting the cellular expression of AIF using RNA interference [small interfering RNA (siRNA)] conferred significant protection against MPP⁺ toxicity. These data indicate an important role for AIF in dopaminergic neuronal cell death, suggesting that multiple pathways must be considered when developing neuroprotective therapies.

Materials and methods

Cell lines, culture and treatments

MN9D is a mouse midbrain-derived dopaminergic cell line (Choi et al. 1991) that recapitulates cell death responses of primary dopamine neurons to different stimuli (Choi et al. 1999; Lotharius et al. 1999). For culturing, MN9D cells (provided by Dr Alfred Heller, University of Chicago) were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (Hyclone, Logan, Utah, USA) on Primaria culture dishes (Becton Dickson, Lincoln Park, NJ, USA) at a density of not more than 80% confluence. The cells were plated to various configurations needed for the experiments at 40% confluence before undergoing differentiation, and all experiments were performed on differentiated MN9D cells. The differentiation process was achieved by adding n-butyrate to a final concentration of 1 mM (Byrd and Alho 1987). Typical duration required to complete the differentiation process is 5–7 days.

Measurement of cell viability by sodium 3,3′-{1-[(phenylamino) carbonyl]-3,4-tetrazolium}bis(4-methyoxy-6-nitro)benzene sulfonic acid) (XTT) assay

The viability of MN9D cells was determined by measuring the oxidation of XTT using the Procheck cell viability assay kit (Intergen, NY, USA). In brief, cells plated in 24-well culture dishes were washed once with medium and incubated at 37°C. At the appropriate sampling time points, the ProCheck cell viability reagents were diluted into the medium at a ratio of 1 : 5, followed by a further 2-h incubation at 37°C. Cell viability was reflected by changes in the optical density at 425 nm, measured using a spectrophotometer microplate reader (SpectraMax 340; Molecular Devices, Sunnyvale, CA, USA). All the values in the figures were calculated from three independent experiments, with each experiment containing at least three replicates for each experimental condition.

Assessment of caspase activity

Differentiated MN9D cells were washed three times with ice-cold PBS and then lysed by adding 1 mL lysis buffer (1% NP40, 50 mM Tris.HCl pH 7.6, 5 mM EDTA), followed by a 10-min incubation on ice. The samples were sonicated for 16 s and then centrifuged for 30 min at 4°C, 16 000 g. After centrifugation, protein concentrations of the supernatants were ascertained using the Bio-Rad Protein Assay Kit (Bio-Rad, Hercules, CA, USA). Samples containing 100 μg protein were mixed with 20 μM of the fluorogenic substrate Acetyl-Asp-Glu-Val-Asp-7-amido-4-trifluoromethyl-coumarin (Chemicon International, Temecula, CA, USA), followed by a 2-h incubation at 37°C. The changes in fluorescence were quantified every 20 min using a luminescence spectrometer (Winlab; Perkin Elmer, Shelton, CT, USA) (excitation 400 nm, emission 505 nm). TUNEL(Terminal deoxynucleotidyl transferase biotin-dUTP Nick End Labelling) was performed as previously described (Chen et al. 1997).

Immunocytochemistry

Triple-label staining was performed to examine AIF translocation after MPP⁺ treatment in MN9D cells. At 12–24 h after MPP⁺ treatment, MN9D cells cultured in a 16-well chamber were incubated with MitoTracker Red (Molecular Probes, Eugene, OR, USA) at 50 nM for 15 min, washed three times with phosphate-buffered saline (PBS), and then fixed with 4% paraformaldehyde for 10 min. Cells were permeabilized using 0.5% Triton X-100 in PBS for 30 min, and were blocked with goat serum for 1 h at 4°C. The cells were incubated for 2 h at room temperature (22–24°C) with rabbit anti-AIF antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) diluted 1 : 100. After three washes with PBS, the cells were incubated for 1 h at room temperature with Alex488-conjugated goat anti-rabbit antibody (Jackson ImmunoResearch Inc., West Grove, PA, USA) diluted 1 : 2000. After another three consecutive washes with PBS, the cells were stained with Hoechst 33258 to visualize changes in nuclear morphology. Finally, cells were mounted with Fluoromount-G (Southern Biotechnology Associates, Birmingham, AL, USA), and observed and imaged with a confocal microscope (LSM 510; Zeiss, Jena, Germany). Digital images were captured with the LSM 510 3.2 software (Zeiss, Gottingen, Germany).

Adeno-associated virus (AAV) transfection of Apaf-1-interacting protein (AIP)

For the infection of AIP in MN9D cells, the AAV-hemagglutinin (HA)-AIP vector (Cao et al. 2004) was added to the serum-free medium at a particle/cell ratio of 1 × 10⁵: 1 and incubated for 6 h, and then the cells were incubated in vector-free normal medium for 3 days. Expression of the transgene products in neurons was verified by western blotting and immunocytochemistry using the antibody against the HA tag.

AIF knockdown using siRNA

To reduce the expression of AIF, a mammalian expression plasmid that directs the transcription of siRNA-like transcripts was constructed. This expression plasmid uses the pcDNA3.1(+) backbone. In its SpeI/EcoRV cloning sites, it contains the H1-RNA promoter followed by the AIF-specific targeting sequence (designated as pH1-AIF). Upon construction, the inserted cDNA in pH1-AIF was verified by sequencing of both strands (University of Pittsburgh Sequencing Service Facility). To generate a stable cell line with AIF depletion, pH1-AIF was transfected into MN9D cells using LipofectAMINE reagent (Invitrogen, Carlsbad, CA, USA). Forty-eight hours after the transfection, the cells were passaged at a ratio of 1 : 3, and G418 (GibcoBRL) was added the next day at a concentration of 450 ng/mL. Cells were maintained in G418 for 1 month to ensure selection of a stable cell line. To confirm the AIF knockdown effect of pH1-AIF in transfected MN9D cells, cells were grown to confluence, collected and subjected to protein extraction. Immunoblotting and immunocytochemistry was performed using the anti-AIF antibody as described previously (Cao et al. 2003).

Pulse-field gel electrophoresis

Cells were harvested after the indicated treatments and the cell pellet (40–50 mg) was suspended in 80 μL TE buffer (10 mM Tris-Cl, pH 7.5, 1 mM EDTA) with the aid of a 1-mL pipette tip. After mixing with an equal volume of 1.5% low-melting-point agarose at 37°C, the mixture was cast into a 1-mL syringe and set at 4°C for 5 min. The plug was then immersed in lysis buffer (10 mM Tris, 125 mM EDTA, 1.0% N-lauryl sarcosine pH 9.0, 0.1 mg/mL proteinase K) and incubated with rotation at 37°C for 20–24 h. The plug was rinsed in TE and further incubated with TE containing 33 μg/mL Rnase A at 4°C for 1 h. After a rinse in TE, the plug was incubated with 1 mM 4-(2-aminoethyl)benzensulfonylfluoride hydrochloride in TE at 4°C for 1 h. It was then rinsed twice in TE at 4°C for 30 min each, pulled back into a 1-mL syringe and stored at 4°C. High molecular weight DNA was separated using a pulse-field gel electrophoresis system (CHEF Mapper systems; Bio-Rad). A 0.02-mL slice of prepared cast agarose plug was loaded on to a well in 0.8% agarose gel in TBE electrophoresis buffer (8.9 mM Tris-Cl, 8.9 mM Boric acid, 2 mM EDTA-Na2) and sealed in place with 1% low-melting-point agarose (FMC). The gel was subsequently run on a Q-life autobase PFG system with software-assisted ROM card 3 (resolution 8–500 kbp; Q-Life, Kingston, ON, Canada) in TBE buffer at 14°C. After staining with ethidium bromide, the gel was destained and photographed under UV transillumination.

Primary midbrain neuronal cultures

The ventral midbrain was dissected from 15-day C57BL/6 mouse embryos (Hilltop Laboratory Animals, Inc., Scottdale, PA, USA). Dissected tissues were incubated with 100 units papain (Worthington Biochem. Corp. Lakewood, NJ, USA) in Hank’s Balanced salt solution for 30 min at 37°C, then mechanically dissociated using a flame-polished Pasteur pipette in Dulbecco’s modified Eagle’s medium/Ham’s F12 1 : 1 (v/v; GIBCO/Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine seum (Cambrex Biological Science, Walkersville, MD, USA), 5 mg/L insulin (Invitro Life Technologies) and 30 mM D-glucose (Sigma, St Louis, MO, USA). Cells were collected by centrifugation at 200 g for 5 min, resuspended in culture medium and plated in poly-L-lysine coated 16-well chamber slides (Nunc Laboratory-Tek, Fisher Scientific, Agawam, MA, USA) at a density of 2 × 10⁵ cells/cm². Cultures were maintained at 37°C in 5% CO₂. After 3 days, fresh medium containing 2 μM cytosine arabinoside was applied for 72 h to inhibit the proliferation of glia.

Primary cultures were treated at 7 days in vitro with 5 μM MPP⁺ iodide for 48 h. This dose of MPP⁺ elicited degeneration of about 50% of tyrosine hydroxylase (TH) neurons without affecting non-TH neurons. To assess the effect of caspase inhibitors on MPP⁺-induced neuronal cell death, the caspase-3 inhibitor Acetyl-Asp-Glu-Val-Asp-chloromethylketone or a broad-spectrum caspase inhibitor, Boc-aspartyl(OMe)-fluoromethyl ketone (Boc) (Calbiochem, San Diego, CA, USA) was added with MPP⁺ to yield final doses of 50–100 μM.

Immunofluorescence and data analysis of primary cultures

Cells from the different treatment conditions were fixed in 3% paraformaldehyde and stained with antibodies for TH, the rate-limiting enzyme for DA synthesis, and for microtubule-associated protein 2 (MAP2), a general neuronal cell marker. Immunocytochemistry was performed using rabbit anti-TH (1 : 2000; Chemicon) and mouse anti-MAP2 (1 : 1000; Sternberger Monoclonals Incorporated, Lutherville, MD, USA). To address the activation of caspase-3 and distribution of AIF, multilabel immunocytochemistry was peformed using the following antibodies: rabbit anti-cleaved caspase-3 (1 : 600; Cell Signaling Technology, Beverly, MA, USA); goat anti-AIF (1 : 100; Santa Cruz Biotechnology) or mouse anti-TH (1 : 4000; Calbiochem). Non-specific binding sites were blocked using Protein Blocking Solution (Dako, Carpinteria, CA, USA), then primary antibodies were added at 4°C overnight, followed by incubation with appropriate Cy™ 3-conjugated (Jackson Immuno-Research Laboratories) and Alexa Fluor® 488-conjugated (1 : 500; Molecular Probes) antibodies. The cells were counterstained with the nuclear marker DAPI (4′,6-diamidino-2-phenylindole, dihydrochloride) (Molecular Probes). The slides were observed using an Olympus Provis fluorescence microscope (Olympus America Inc., Melville, NY, USA) equipped with three filter cubes: FITC (excitation 490/emission 520), TRITC (excitation 541/emission 572 nM) and DAPI (excitation 350/emission 470 nM).

The total number of TH neurons in control or treated mesencephalic cultures and the percentage of TH neurons showing cleaved caspase-3 expression was quantified in at least four wells per treatment. We routinely obtain 500–650 TH-positive cells per control well (0.4 cm²). For AIF staining, 9–16 fields from each experimental condition were acquired using the same fluorescence threshold settings, and assigned a randomly generated number. Digital images in each channel were acquired separately to minimize bleed-through. All TH neurons with a clear nuclear contour were classified independently by two individuals in a blinded fashion according to the following staining patterns: normal punctate (mitochondrial), clustered perinuclear, nuclear or uninterpretable owing to overlap. Occasionally perinuclear clustering was observed in conjunction with diffuse nuclear staining, which was scored as nuclear staining. For caspase-3 and AIF studies, data were expressed as the percentage of countable TH neurons from each experiment, and the results from at least three independent experiments were averaged.

Statistical analysis

All results are expressed as mean ± SEM from at least three independent experiments. Two-group comparisons were performed using Student’s t-test. Multiple-group comparisons were performed using ANOVA. Post hoc testing used the Student’s t-test with Bonferroni correction, and p < 0.05 was accepted as statistically significant.

Results

MPP⁺ elicits an active form of caspase-independent cell death in MN9D cells

Differentiated MN9D cells demonstrated a dose- and time-dependent toxic response to MPP⁺ treatment (Figs 1a and b), with an LD₅₀ of ~30 μM. The addition of a broad-spectrum caspase inhibitor had no effect on MPP⁺ toxicity at any concentration (Fig. 1c). As cytochrome c release and caspase-3 activation has been observed in other cells treated with MPP⁺ (Kaul et al. 2003), we further investigated the potential role of the mitochondrial pathway using a dominant negative inhibitor for Apaf-1/caspase 9. AIP has been previously shown to suppress neuronal apoptosis elicited by several pro-apoptotic stimuli in vitro and in vivo (Cao et al. 2004). However, AIP overexpression had no effect on MPP⁺-elicited toxicity to MN9D cells (Fig. 1d).

Fig. 1.

MPP⁺ induces caspase-independent programmed cell death in MN9D cells. (a) Neuronal differentiated MN9D cells were treated with MPP⁺ at the indicated concentrations. Cell survival was measured using the XTT assay 24 h after the insult. (b) Cells were treated with vehicle or MPP⁺ (30 μM) and cell viability was measured at the indicated time points. (c) Cells were treated with MPP⁺ in the presence (30 or 100 μM) or absence of the caspase inhibitor z-VAD.fmk [Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone] (z-VAD). (d) Cells transfected with HA-tagged AIP, a dominant negative inhibitor for Apaf-1/caspase-9 (AAV-AIP) (Cao et al. 2004), or the empty vector (AAV) were treated with MPP⁺ and cell viability was measured at 24 h. (e, f) Cells were treated with MPP⁺ (30 μM) in the presence of the indicated concentrations of actinomycin D or cycloheximide, and cell viability (e) and the percentage of condensed nuclei (f) were measured at 24 h. All data are expressed as mean ± SEM from three independent experiments with at least nine replicates per data point. *p < 0.05, **p < 0.01 versus vehicle treatment.

In order to determine whether MPP⁺-elicited cell death represented active or passive cell death, we used the eukaryotic protein synthesis inhibitor cycloheximide and the transcriptional inhibitor actinomycin D. Inhibition of either transcription or translation conferred significant protection from MPP⁺ toxicity as assessed by both metabolic and morphologic criteria (Figs 1e and f). These data indicate that MPP⁺ elicits an active form of cell death in MN9D cells characterized by chromatin condensation.

To further determine the involvement of caspase-3 activation in MPP⁺ toxicity, MN9D cells were treated with 30 μM MPP⁺ for 3, 6 or 24 h, and then subjected to caspase-3 measurement using western blot analysis and the DEVDase activity assay (Figs 2a and b). MPP⁺ elicited a minimal amount of caspase-3 activation, which was detectable only at the 24-h time point by the DEVDase activity assay (2-fold increase vs. control). In control experiments, a robust activation of caspase-3 was observed in cells treated with staurosporine (STS), indicating that neuronal differentiated MN9D cells possess the intrinsic capacity for caspase-3 activation.

Fig. 2.

MPP⁺ elicits minimal caspase-3 activation in MN9D cells. Neuronal differentiated MN9D cells were treated with MPP⁺ (30 μM) for 3, 6 or 24 h, and then subjected to caspase-3 measurement. (a) Western blot analysis using antibodies against caspase-3 (recognizing both pro- and cleaved caspase-3, top panel) or cleaved caspase-3 (middle panel). MPP⁺ failed to induce cleaved caspase-3 at any time point tested. In control experiments, STS (0.5 μM for 3 h) induced the 17-kDa active fragments of caspase-3; z-VAD.fmk (z-VAD; 100 μM) abolished the formation of the active 17-kDa fragments induced by STS, but allowed partial processing of caspase-3 to its 19-kDa fragments. (b) DEVDase activity assay. MPP⁺ elicited a mild 2-fold increase in caspase-3-like activity at 24 h, but not at earlier time points. In control experiments, STS induced robust increases in caspase-3-like activity, which were inhibited by z-VAD.fmk or z-DEVD.fmk [Benzyl-oxycarbonyl-Asp(OMe)-Glu(OMe)-Val-Asp(OMe)fluoromethylketone] (z-DEVD). Data are expressed as mean ± SEM from three independent experiments. *p < 0.05, ***p < 0.001 versus control.

AIF is released from mitochondria and undergoes nuclear translocation in MPP⁺-treated MN9D cells

To determine the potential role of AIF in MPP⁺ toxicity, we performed subcellular fractionation studies at different time points. Both cytochrome c and AIF were released into the cytoplasmic fractions (Fig. 3). Cytochrome c was released at earlier time points than AIF, which displayed nuclear translocation before the major drop in viability observed between 12 and 24 h (see Fig. 1b).

Fig. 3.

AIF is released to cytosolic and nuclear fractions during MPP⁺ toxicity. Western blot analysis was performed on MPP⁺-treated cells (30 μM) following subcellular fractionation. Cytochrome c release (increases in the cytosolic fraction) occurred as early as 1 h after MPP⁺ exposure, whereas AIF release (increases in both nuclear and cytosolic fractions) occurred in a delayed fashion. The graph summarizes the increases in nuclear AIF and cytosolic cytochrome c following MPP⁺ toxicity, based on optical density measurement of blots from three independent experiments. Values are mean ± SEM. *p < 0.05 versus vehicle treatment.

Multilabel immunofluorescence of vehicle-treated control cells showed the normal punctate cytoplasmic staining pattern for AIF, which co-localizes with mitochondrial markers and is excluded from the nucleus (Figs 4a, c and g). In MPP⁺-treated cells, AIF was diffusely distributed within the nuclear contour in cells that concurrently exhibited chromatin condensation (Figs 4b, f and h).

Fig. 4.

AIF is released and translocated to the nucleus during MPP⁺ toxicity. Confocal microscopic images showing AIF translocation during MPP⁺ toxicity. Panels (a), (c), (e) and (g) are images showing localization of AIF, MitoTracker Red and nuclei (Hoechst) in vehicle-treated MN9D cells. Panels (b), (d), (f) and (h) are images showing the presence of AIF immunofluorescence in the nucleus (arrows) of MPP⁺-treated cells with morphological changes of apoptosis (condensed nucleus). Scale bar 15 μm. The graph shows the percentage of cells exhibiting AIF translocation at either 12 or 24 h after exposure to 30 μM MPP⁺. Data are expressed as mean ± SEM from three independent experiments, involving at least 1200 cell counts per data point. **p < 0.05 versus vehicle treatment.

AIF siRNA studies demonstrate a significant role for AIF in mediating MPP⁺ toxicity

In order to determine the functional significance of AIF release and nuclear translocation in MPP⁺ toxicity, we designed a short hairpin siRNA sequence subcloned into a pcDNA backbone (Fig. 5a). Using this, we produced several permanently transfected lines of MN9D, which showed reduced expression of AIF, but not of mitochondrial outer membrane, inner membrane or other intermembrane space proteins (Fig. 5b). Cells transfected with control siRNA (containing a scrambled AIF sequence) showed a normal distribution of mitochondrial AIF, whereas AIF siRNA effectively inhibited AIF without disrupting the normal staining pattern of the mitochondrial marker MitoTracker Red (Fig. 5c). There were no effects of AIF siRNA on basal viability, basal mitochondrial complex I activity, or the ability of MPP ⁺ and rotenone to inhibit complex I activity in AIF-deficient MN9D cells compared with control siRNA and parental MN9D cells (data not shown).

Fig. 5.

siRNA-mediated AIF knockdown in MN9D cells. (a) Construction of a siRNA expression vector for mouse AIF. (b) Knockdown of AIF expression in MN9D cells stably expressing AIF-siRNA. Multiple clones were screened to identify cells that contained reduced levels of AIF expression. AIF knockdown had no effect on the expression of other mitochondrial proteins (Cyto c, cytochrome c; COX, cytochrome c oxidase; VDAC, voltage-dependent anion channel). All blots are representative of three independent experiments with similar results. (c) Confocal microscopic images showing AIF immunofluorescence in control siRNA- and AIF siRNA-treated cells. Scale bar 15 μm.

MN9D cells with reduced expression of AIF were significantly protected from MPP⁺ toxicity (Fig. 6a) compared with control siRNA transfectants, which showed equivalent dose–response curves to untransfected cells. Moreover, AIF knockdown resulted in reduced chromatin condensation and TUNEL staining (Figs 6b and e), as well as inhibiting large-scale DNA fragmentation characteristic of caspase-independent AIF-dependent cell death (Fig. 6f). Equivalent results were observed using three independent MN9D AIF siRNA clones (clones 1, 3 and 6 in Fig. 5b). To determine whether the protective effect from AIF knockdown manifested as a delay in cell death, cell viability was also measured at 48 and 72 h after induction of MPP⁺ toxicity. At each time point tested, AIF knockdown cells showed significantly increased viability and decreased nuclear condensation compared with control siRNA transfectants (Figs 6c and d).

Fig. 6.

AIF release contributes to MPP⁺ toxicity. (a, b) AIF knockdown attenuates MPP⁺-induced cell death. Cell viability and percentage of condensed nuclei were quantified at 24 h after exposure to MPP⁺ at difference concentrations. (c, d) Cell viability and percentage of condensed nuclei were quantified at different time points (24, 48 and 72 h) after exposure to 30 μM MPP⁺. All values in panels (a–d) are mean ± SEM from three independent experiments, involving at least nine replicates per data point for cell viability and 1200 cell counts for condensed nuclei. *p < 0.05, **p < 0.01 versus cells transfected with control siRNA. (e) Hoechst 33258 and TUNEL staining in MN9D cells treated with 30 μM MPP⁺ for 24 h. Left panel shows Hoechst staining, with increases in condensed nuclei in cells transfected with control siRNA; condensed nuclei were present less frequently in cells transfected with AIF siRNA. Right panel shows dual label of TUNEL and Cyto11 green fluorescent nuclear staining. TUNEL-positive nuclei (yellow–red) were present less frequently in cells transfected with AIF siRNA than in those transfected with control siRNA after MPP⁺ treatment. (f) MPP⁺-induced nuclear large-scale DNA fragmentation (mainly at the size of 50-kbp) was diminished in AIF knockdown cells, as determined using pulse-field gel electrophoresis. This image is representative of two experiments with similar results.

Toxic effects of MPP⁺ on primary mouse midbrain dopaminergic neurons

To determine whether primary dopaminergic neurons also undergo caspase-independent cell death involving AIF release and nuclear translocation, primary ventral midbrain cultures were treated with different doses of MPP⁺ for 48 h. Double immunofluorescence staining for TH, the rate-limiting enzyme for dopamine synthesis, and for MAP2, a general neuronal marker, was performed. Significant decreases in the number of TH-immunoreactive neurons (TH neurons) were observed at MPP⁺ concentrations of 1 μM and above (Fig. 7a). At concentrations of 50 μM and above, significant injury to all MAP2-expressing neurons, irrespective of TH expression, was observed. Thus, selective injury to TH neurons occurred at doses of 1–10 μM. For all subsequent primary culture studies, a dose of 5 μM was used. Compared with control cultures, which displayed multiple long, smooth neuronal processes (Figs 7b and d), TH neurons treated for 48 h with 5 μM MPP⁺ show blunted or absent processes (Figs 7c and e).

Fig. 7.

Dose dependence of MPP⁺ toxicity in primary midbrain neurons. (a) Primary mouse embryonic midbrain cultures were treated with different concentrations of MPP⁺ for 48 h. The total number of neurons expressing TH and MAP2 were expressed as a percentage of numbers in control cultures that did not receive MPP⁺. Data are expressed as the mean ± SEM from quadruplicate wells. Selective toxicity to TH neurons occurred at 1, 5 and 10 μM MPP⁺. *p < 0.01 versus MAP2 positive. (b) Photomicrograph of a representative control culture stained for MAP2 (green) and TH (red). Note long, smooth processes in TH neurons (yellow owing to co-localization of MAP2 and TH). (c) Photomicrograph of a culture treated with 5 μM MPP⁺ for 48 h and stained with dual fluorescence as in (b). Note reduction in numbers of TH neurons (yellow). Higher-magnification view of the TH channel of control (d) and MPP⁺-treated (e) cultures highlighted retraction or blunting of processes in MPP⁺-injured TH neurons. Scale bars 50 μm.

Inhibition of caspases does not confer protection from MPP⁺ injury to primary TH neurons

In primary TH neurons, MPP⁺ elicited cytochrome c release (Fig. 8a) and increases in the percentage of TH neurons exhibiting cleaved caspase-3 immunoreactivity (Figs 8b and 9a), although the overall level of caspase activation remained low at 8, 24 and 48 h after treatment (Table 1). To determine whether caspase activation mediates neurotoxin-induced death, we used both a caspase-3 selective inhibitor (Fig. 9b) and a broad-spectrum caspase inhibitor (Fig. 9c). Neither inhibitor affected MPP⁺-elicited loss of TH neurons, supporting a role for caspase-independent death pathways.

Fig. 8.

MPP⁺ elicits AIF redistribution in TH neurons. (a) Primary TH neurons exhibited a diffuse distribution of cytochrome c (Cyt c) in contrast to the normal granular staining pattern observed in adjacent non-TH neurons. (b) Most MPP⁺-treated TH neurons showed no evidence of caspase activation (bottom left); a subset of TH neurons exhibited immunoreactivity for cleaved caspase-3 associated with nuclear fragmentation (top right, arrow). (c) Control cultures showed normal punctate cytoplasmic staining for AIF. (d, e) In contrast, TH neurons in MPP⁺-treated cultures showed either nuclear AIF (d) or a clumped perinuclear distribution (e). Scale bars 10 μm.

Fig. 9.

Quantitative analysis of cleaved caspase-3 expression and AIF nuclear translocation in MPP⁺-treated TH neurons and effects of caspase inhibitors. (a) MPP⁺ elicited increased cleaved caspase-3 immunoreactivity in TH neurons. Values are mean ± SEM for eight experiments. *p < 0.01 versus vehicle. (b) Primary midbrain cultures were treated with 5 μm MPP⁺ for 48 h in the presence or absence of the caspase-3 inhibitor Ac-DVED-cmk at the indicated concentrations. The number of TH neurons was expressed as percentage of that in control cultures that did not receive MPP⁺. Results are mean ± SEM of three independent experiments. The caspase-3 inhibitor had no effect on MPP⁺-elicited TH neuron loss. (c) Primary midbrain cultures were treated with MPP⁺, the broad-spectrum caspase inhibitor Boc (100 μM) or a combination of MPP⁺ and Boc. Values are mean ± SEM of five independent experiments. The broad-spectrum caspase inhibitor did not protect against MPP⁺ toxicity to TH neurons. (d) Quantitative analysis of AIF subcellular distribution in control (vehicle-treated) and MPP⁺-treated cells was performed independently by two persons blinded to the treatment condition. Values are mean ± SEM of three independent experiments. There was a significant increase in both nuclear and clumped patterns of AIF in MPP⁺-injured TH neurons (*p < 0.01 vs. control).

Table 1.

Time course for caspase-3 immunoreactive TH neurons (fold increase over control cultures)

	MPP+ (5 μM)
	8 h	24 h	48 h
Fold increase	1.2 ± 0.6	2.4 ± 1.6	6.3 ± 1.3*

Values are mean ± SEM.

^*p < 0.01 versus control cultures in eight independent experiments. (ANOVA) followed by post-hoc students’ t test with Bonferroni correction).

AIF nuclear translocation is elicited in MPP⁺-treated primary TH neurons

The normal distribution of AIF in both TH and non-TH neurons results in a granular or punctate cytoplasmic staining pattern consistent with its localization in mitochondria (Fig. 8c). In MPP⁺-treated cells, nuclear translocation of AIF was observed in TH neurons showing cell shrinkage, nuclear condensation and irregular beading of processes (Fig. 8d). Other TH neurons exhibited perinuclear clustering of AIF (Fig. 8e). Quantitative analysis indicated a significant decrease in the percentage of TH neurons showing a normal mitochondrial distribution of AIF, accompanied by nuclear translocation of AIF in MPP⁺-treated TH neurons (Fig. 9d). These data suggest that, like MN9D cells, primary dopaminergic neurons undergo caspase-independent cell death involving AIF release.

Discussion

The data presented here indicate that AIF redistribution plays an important role in caspase-independent cell death in dopaminergic neuronal cells. Although MPP⁺ activates pathways associated with both caspase-dependent and caspase-independent mechanisms, neither overexpression of AIP, a dominant negative apoptosome regulatory protein, nor use of broad-spectrum caspase inhibitors conferred protection. In contrast, siRNA-mediated knockdown of AIF expression effectively reduced large-scale DNA fragmentation, chromatin condensation and cell death in the neuronal differentiated MN9D cells.

The role of caspases in MPP⁺-elicited cell death has been inconsistent in the literature. Caspase inhibitors have been reported to protect against MPTP toxicity (Yang et al. 2004) and MPP⁺-elicited cell death in primary cultures, but there is no protection from neuritic dysfunction (Bilsland et al. 2002). However, other studies found little evidence to support either phosphatidylserine externalization or apoptotic DNA laddering (Lotharius et al. 1999; Han et al. 2003). These groups also found no protection using broad-spectrum caspase inhibitors, and others have reported that caspase inhibition potentiates necrotic cell death (Hartmann et al. 2001). Although studies using higher doses of MPP⁺ have suggested necrosis as an explanation for caspase-independent cell death (Choi et al. 1999), our data indicate that MPP⁺ elicits an active form of cell death in differentiated MN9D cells that requires transcription, translation and AIF. Expression profiling studies of MPP⁺-treated neuronal cell lines indicate induction of transcription factors such as CHOP/Gadd153 (Ryu et al. 2002; Holtz and O’Malley 2003). Interestingly, overexpression of CHOP promotes stress-related death, resulting in decreased cellular glutathione, increased production of reactive oxygen species and decreased levels of Bcl-2 (McCullough et al. 2001). Given the common role of mitochondrial oxidative stress in major models of parkinsonian injury (Betarbet et al. 2002; Dawson and Dawson 2003; Callio et al. 2005), and the ability of Bcl-2 overexpression to regulate AIF release (Cao et al. 2003), it is possible that similar mechanisms are linked to AIF-mediated death.

A multiplicity of cell death pathways is not surprising in pathological cell death associated with disease. The eventual decline in viability of MPP⁺-treated AIF-deficient cells at 72 h implies the existence of additional caspase-independent pathways. In addition to apoptosis and necrosis, caspase-independent forms of cell death involving increased autophagosomes have been reported in neurons (Zaidi et al. 2001; Tolkovsky et al. 2002; Gomez-Santos et al. 2003; Florez-McClure et al. 2004). We found that autophagy is induced during MPP⁺ toxicity (J.-h. Zhu and C. T. Chu, unpublished data), and that autophagocytosed mitochondria are present in human Parkinson’s/Lewy body disease neurons (Zhu et al. 2003), suggesting the possibility of additional pathways not prevented by either AIF siRNA or caspase inhibitors. Although cytochrome c release was observed in both MN9D and primary TH neurons, caspase-3 activation was not robust. This may be explained by expression of endogenous caspase inhibitors in neuronal cells (Eberhardt et al. 2000; Potts et al. 2003), and/or reduced ATP levels in MPP⁺-treated cells (Han et al. 2003). Cellular context and pre-existing ATP levels determine available pathways of cell death (Eguchi et al. 1997; Nicotera et al. 2000; Han et al. 2003). It is interesting to note that AIF nuclear translocation can be triggered by ATP depletion itself (Daugas et al. 2000), suggesting that AIF-mediated cell death may occur under conditions where caspase activation is impaired.

Given that multiple pathways can be initiated in a given cell injury paradigm, the siRNA results are especially important in demonstrating a direct role for AIF in mediating MPP⁺ toxicity. A role for AIF as a caspase-independent mediator was derived from observations that overexpression of AIF induces neuronal cell death in a caspase-independent manner (Cregan et al. 2002). Other studies supporting a role for AIF as a death mediator include observations that AIF nuclear translocation corresponds with early commitment to apoptosis (Bidere et al. 2003), precedes cytochrome c release (Wang et al. 2004) and correlates with large-scale DNA fragmentation (Daugas et al. 2000; Cao et al. 2003). However, a temporal correlation does not necessarily predict causality. Although cytochrome c release precedes AIF release in our system, caspase activation is not necessary for MPP⁺-mediated cell death. Intracellular delivery of neutralizing antibodies (Cregan et al. 2002; Wang et al. 2004), genetic inactivation in embryonic stem cells (Joza et al. 2001) and our current siRNA studies directly support an executionary role for AIF.

The siRNA studies also support a mechanism involving AIF-mediated large-scale DNA fragmentation, as suggested by studies using purified nuclei (Daugas et al. 2000). An alternative possibility for detrimental effects of AIF release may include loss of some normal protective mitochondrial function. This possibility is not substantiated by the siRNA studies as the MN9D siRNA lines with reduced or absent AIF expression did not exhibit problems with viability. Moreover, there were no significant effects of AIF knockdown on mitochondrial complex I activity. The significance of the abnormal clumped pattern of AIF staining observed in primary neurons is unknown, but may reflect injury-induced alterations in mitochondrial distribution.

The in vivo evidence for a role of apoptosis in MPTP toxicity include studies using transgenic mice overexpressing Bcl-2 (Offen et al. 1998; Yang et al. 1998) and knockout mice lacking pro-apoptotic bax (Vila et al. 2001). It is interesting to note that overexpression of Bcl-2 reduces AIF nuclear translocation in other neuronal cell systems (Cao et al. 2003), and that Bax is capable of promoting AIF release (Bidere et al. 2003). Moreover, in contrast to the expression pattern of a number of apoptosis regulatory gene products during brain development, the expression of AIF increases with brain maturation and peaks in adulthood (Cao et al. 2003). Thus, the involvement of AIF in caspase-independent MPP⁺ toxicity to MN9D and primary dopaminergic neurons in culture could reconcile conflicting interpretations derived from transgenic mouse studies that support a role for active, regulated cell death and primary midbrain culture systems interpreted as ‘necrotic’ owing to inability of caspase inhibitors to confer protection.

To summarize, MPP⁺ elicits an active form of cell death in dopaminergic neuronal cells that involves AIF nuclear translocation, but is caspase independent. Given the multiplicity in potential cell death pathways in dopaminergic neurons, combination therapies targeted at multiple death pathways or those regulating signaling mechanisms that act before commitment to death may be beneficial (Chu et al. 2004; Horbinski and Chu 2005).

Abbreviations used

AAV

adeno-associated virus

AIF

apoptosis inducing factor

AIP

apoptotic protease activating factor 1 interacting protein

Apaf-1

apoptotic protease activating factor 1

Boc

Boc-aspartyl(OMe)-fluoromethyl ketone

DAPI

4′,6-diamidino-2-phenylindole, dihydrochloride

HA

hemagglutinin

MAP2

microtubule-associated protein 2

MPP⁺

1-methyl-4-phenylpyridinium

MPTP

1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine

PBS

phosphate-buffered saline

siRNA

small interfering RNA

STS

staurosporine

TH

tyrosine hydroxylase

TUNEL

Terminal deoxynucleotidyl transferase biotin-dUTP Nick End Labelling

XTT

3,3′-{1-[(phenylamino)carbonyl]-3,4-tetra-zolium}bis(4-methyoxy-6-nitro)benzene sulfonic acid)

z-DEVD-fmk

Benzyloxycarbonyl-Asp(OMe)-Glu(OMe)-Val-Asp(OMe)fluoromethyl-ketone

z-VAD.fmk

Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone