Intranuclear localization of apoptosis-inducing factor (AIF) and large scale dna fragmentation after traumatic brain injury in rats and in neuronal cultures exposed to peroxynitrite

Authors

  • Xiaopeng Zhang,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Jun Chen,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Steven H. Graham,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Lina Du,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Patrick M. Kochanek,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Romesh Draviam,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Fengli Guo,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Paula D. Nathaniel,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Csaba Szabó,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Simon C. Watkins,

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author
  • Robert S. B. Clark

    1. Departments of *Anesthesiology and Critical Care Medicine, †Neurology, §Pediatrics and ¶Cell Biology and Physiology, the Safar Center for Resuscitation Research and the Brain Trauma Research Center, University of Pittsburgh, Pittsburgh, Pennsylvania ‡Geriatric Research Educational and Clinical Center, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania **Inotek Corporation, Beverly, Massachusetts
    Search for more papers by this author

and reprint requests to Robert S. B. Clark, Safar Center for Resuscitation Research, 3434 Fifth Avenue, Pittsburgh, PA 15260, USA. E-mail: clarkrs@ccm.upmc.edu

Abstract

Programmed cell death occurs after ischemic, excitotoxic, and traumatic brain injury (TBI). Recently, a caspase-independent pathway involving intranuclear translocation of mitochondrial apoptosis-inducing factor (AIF) has been reported in vitro; but whether this occurs after acute brain injury was unknown. To address this question adult rats were sacrificed at various times after TBI. Western blot analysis on subcellular protein fractions demonstrated intranuclear localization of AIF in ipsilateral cortex and hippocampus at 2–72 h. Immunocytochemical analysis showed AIF labeling in neuronal nuclei with DNA fragmentation in the ipsilateral cortex and hippocampus. Immunoelectronmicroscopy verified intranuclear localization of AIF in hippocampal neurons after TBI, primarily in regions of euchromatin. Large-scale DNA fragmentation (∼50 kbp), a signature event in AIF-mediated cell death, was detected in ipsilateral cortex and hippocampi by 6 h. Neuron-enriched cultures exposed to peroxynitrite also demonstrated intranuclear AIF and large-scale DNA fragmentation concurrent with impaired mitochondrial respiration and cell death, events that are inhibited by treatment with a peroxynitrite decomposition catalyst. Intranuclear localization of AIF and large-scale DNA fragmentation occurs after TBI and in neurons under conditions of oxidative/nitrosative stress, providing the first evidence of this alternative mechanism by which programmed cell death may proceed in neurons after brain injury.

Abbreviations used
AIF

apoptosis inducing factor

CAD

of caspase-activated deoxyribonuclease

kbp

kilo-basepair

PFGE

pulsed field gel electrophoresis

TBI

traumatic brain injury.

Programmed cell death is an essential mechanism for the selective elimination of cells during development, homeostasis of tissues with cell turnover and removal of aging and abnormal cells (Steller 1995). In general, the process of programmed cell death is strictly regulated, with dysregulation occurring in, and contributing to, many acute and chronic pathological conditions (Thompson 1995). In the CNS, programmed cell death contributes to neuronal death after ischemia (Graham and Chen 2001), epilepsy (Henshall et al. 2000) and traumatic brain injury (TBI; Raghupathi et al. 2000). It is clear that experimental TBI is capable of eliciting most of the currently known steps in the programmed cell death cascade, including expression of death receptors (Beer et al. 2000), alterations in bcl-2 family genes (Clark et al. 1997b), mitochondrial release of cytochrome c (Buki et al. 2000), activation of caspases (Yakovlev et al. 1997; Clark et al. 2000b), disinhibition of caspase-activated deoxyribonuclease (CAD; Zhang et al. 1999), cytoskeletal disintegration (Pike et al. 1998) and DNA fragmentation (Rink et al. 1995; Colicos and Dash 1996; Clark et al. 1997a,b; Yakovlev et al. 1997; Conti et al. 1998; Fox et al. 1998). Importantly, many of these features of programmed cell death also occur in humans after head injury (Clark et al. 1999, 2000a; Smith et al. 2000). However, controversy exists over whether or not programmed cell death after acute injury in adult mammalian brain can culminate in morphologic ‘apoptosis’, typical of that seen during CNS development (MacManus and Buchan 2000). Activation of CAD results in internucleosomal cleavage of DNA and nuclear phenotypes characteristic of apoptosis; however, in PC12 cells or primary cortical neurons expressing a mutant CAD that is caspase-resistant, apoptotic stimuli still result in cell death, but without internucleosomal DNA fragmentation or apoptotic nuclear changes (Chen et al. 2000; Cao et al. 2001). These results suggest that a caspase-independent mechanism executing programmed cell death also exists in neuronal cells.

Recently, an additional and/or alternative pathway for programmed cell death has come to light. This pathway involves loss of mitochondrial membrane potential, translocation of apoptosis-inducing factor (AIF) into the nucleus, large-scale 50–700 kilo-basepair (kbp) DNA fragmentation, and does not produce archeotypal apoptosis (Susin et al. 1996; Susin et al. 1999; Susin et al. 2000; Joza et al. 2001). Rather, cells exhibit peripheral chromatin condensation without internucleosomal DNA cleavage, and somal and nuclear shrinkage are not prominent characteristics. This type of cell death is regulated by bcl-2 family proteins, but is independent of caspase activation. AIF translocation to the nucleus appears to be a key step in this process. AIF is a flavoprotein located in the mitochondrial intermembrane space that produces peripheral chromatin condensation and large scale DNA fragmentation directly when applied to isolated cell nuclei (Susin et al. 1996; Daugas et al. 2000a; Daugas et al. 2000b). Although nuclear translocation of AIF had not been reported after brain injury in vivo, several studies suggest that AIF may mediate a portion of cell death after acute brain injuries such as TBI. Specifically, large scale DNA fragmentation precedes internucleosomal DNA cleavage after cerebral ischemia (Charriaut-Marlangue et al. 1995; MacManus et al. 1995; MacManus et al. 1997), oxidant injury to neural cell cultures produces cell death that is independent of caspase-3 (Hill et al. 2000), and previous studies in models of TBI have reported cells with DNA fragmentation that did not have apoptotic morphologies, previously attributed to neuronal necrosis (Rink et al. 1995; Clark et al. 1997b). Relevant to other CNS diseases, AIF has been shown to mediate microglial apoptosis in an in vitro model of meningitis, where similar to TBI both caspase-dependent and independent cell death is reported (Braun et al. 2001).

Accordingly, we hypothesized that AIF would translocate from mitochondria to cell nuclei in injured brain after experimental TBI and that this would occur concomitantly with large scale DNA fragmentation in these regions. To further clarify the role of AIF in neuronal cell death and to begin to determine if a cause and effect relationship exists between intranuclear localization of AIF and large-scale DNA fragmentation, in vitro studies were also performed using a model of nitrosative/oxidative stress in neuron-enriched cultures. This model was chosen because it has been shown to produce neuronal cell death with phenotypes consistent with AIF-mediated cell death (Hill et al. 2000), and in vivo studies suggest that nitrosative stress is an important contributor to the pathogenesis of traumatic CNS injury (Scott et al. 1999; Whalen et al. 1999b).

Materials and methods

Model of TBI

Studies were approved by the institutional Animal Care and Use Committee. Controlled cortical impact (CCI) with secondary hypoxemic insult was used (Clark et al. 1997a, b, 2000b Sinz et al. 1999). Adult male Sprague-Dawley rats were anesthetized with 4% isoflurane, intubated, and mechanically ventilated with 2.0% isoflurane/66% N2O/balance O2. A femoral arterial catheter was inserted for monitoring of blood pressure and blood sampling. A craniotomy was made over the left parietal cortex. A temperature probe (Physiotemp, Clifton, NJ, USA) was inserted through a burr hole into the left parietal cortex. Rats were maintained at a brain temperature of 37 ± 0.5°C and allowed to equilibrate under anesthesia (1.1% isoflurane/66% N2O/balance O2) for 30 min. After removal of the bone flap, injury was produced using the CCI device (Dixon et al. 1991). For all studies a 5.0-mm rounded impactor tip, depth of penetration of 2.5-mm, a velocity of 4.0 ± 0.2 m/s, and a duration of deformation of 50 ms were used. To produce moderate hypoxemia, air and oxygen were blended to achieve an FiO2 of 0.11 (1.1% isoflurane/74% N2O/19% air/6% O2) 1 min after CCI. This produces a PaO2 in the rats of 44 ± 1 mmHg and a 40% reduction in mean arterial blood pressure by 30 min, with pH and PaCO2 remaining constant. Hypoxemia was maintained for 30 min during which time the bone flap was replaced. After 30 min catheters and probes were removed, anesthesia was discontinued, and rats were allowed to awaken. Rats were extubated, placed in supplemental O2 for 30 min, then returned to their cages. For all studies, naive rats were used as controls. To maximize the contrast between an injury producing cell death and the control condition, naive rats were used as controls in this study, rather than sham-injured rats. A craniotomy can produce some degree of local cortical damage, and variable changes in a variety of gene products have been observed in the superficial cortex; however, these are insufficient to produce alterations in functional outcome or cell death (Clark et al. 2000b).

Pulsed field gel electrophoresis (PFGE)

Naive rats and rats at 2, 6 and 24 h after TBI (n = 4/group) were anesthetized with isoflurane and perfused with ice-cold heparinized saline. Brains were removed and the ipsilateral hippocampus and cortex including and adjacent to the contusion were dissected on ice. To avoid non-specific DNA breaks caused by phenol/chloroform extraction, chromosomal DNA samples were prepared in agarose plugs (Walker et al. 1993; Charriaut-Marlangue et al. 1995) using a CHEF Mammalian Genomic DNA Plug Kit (Bio-Rad, Hercules, CA, USA). Tissue samples (15–20 mg) were homogenized in cell suspension buffer, mixed with preheated (50°C) 2% low melting point agarose, and transferred into agarose plug molds. After solidification at room temperature the plugs were incubated in 1 mg/mL proteinase K overnight at 50°C without agitation. Deproteinized DNA-containing agarose plugs were washed in buffer, added to wells and sealed with low melting point agarose. PFGE was carried out using a Field Inverse Gel Electrophoretic system (Bio-Rad). Fragments were separated on a 1.2% agarose gel at 14°C for 20 h. Field strengths were 180 V forward and 120 V reverse, initial and final switching time was set at 5–60 s with a linear ramp. The gel was stained with ethidium bromide and visualized under UV light.

Subfractionization of cellular proteins

Separation of cytosolic, nuclear, and mitochondrial proteins was performed as previously described (Clark et al. 2000b) with modification. Naive rats and rats at 2, 6, 24 and 72 h after TBI (n = 7/group) were anesthetized and perfused as described above. Brains were removed and the ipsilateral hippocampus and cortex including and adjacent to the contusion were dissected on ice. Each sample was homogenized in lysis buffer then centrifuged at 1025 g for 15 min at 4°C, with the supernatants containing mitochondrial and cytosolic proteins and the pellets containing nuclei. Nuclear pellets were resuspended in buffer, incubated on ice for 15 min, and centrifuged at 16 000 g at 4°C for 20 min. These supernatants were transferred and stored with 50 µL/mL glycerol at −80°C. The mitochondrial/cytosolic fractions were centrifuged at 735 g at 4°C for 10 min, these supernatants were further centrifuged at 10 000 g at 4°C for 15 min to pellet mitochondria. The supernatants from mitochondrial preparations were centrifuged at 100 000 g at 4°C for 1 h, and the final supernatants containing cytosolic proteins were stored at −80°C in 10% glycerol. The mitochondrial pellets were resuspended in 3% Ficoll solution containing 120 mm mannitol, 30 mm sucrose, and 25 µm EDTA, and were layered on top of a 6% Ficoll solution containing 240 mm mannitol, 60 mm sucrose, 50 µm EDTA. Ficoll gradients were centrifuged at 16 000 g at 4°C for 25 min and the pellets containing mitochondria were resuspended with cell lysis buffer. Protein concentration of lysates was determined with a Bradford-based protein assay (Bio-Rad).

To assess efficiency of separation of cellular compartments, immunoblotting for cytochrome c oxidase (Clark et al. 2000b) was performed pre hoc. Nuclear or cytosolic samples with gross mitochondrial contamination were not used for analysis. The suspensions were stored at −80°C until batch analysis.

Western blot analysis

Western blotting was performed using standard methods (Clark et al. 2000b). Briefly, electrophoretically separated proteins were incubated in primary antibody against the carboxy-terminus of AIF (D20, goat polyclonal, Santa Cruz, Santa Cruz, CA, USA) at a dilution of 1 : 200–1 : 1000 for 1 h at room temperature (22°C). The optimal dilution varied depending upon the batch of antibody and was determined for each batch used. Washed membranes were incubated in horseradish peroxidase-conjugated secondary anti-goat antibody (Santa Cruz) and immunoblotted proteins were detected using chemiluminicence (NEN, Boston, MA, USA). Protein abundance was determined by measuring the relative optical density of the respective protein bands using a Kodak Image Station 440CF (Kodak, Rochester, NY, USA).

Immunocytochemistry

Immunocytochemistry was performed using standard methods (Clark et al. 2000b). Briefly, naive rats and rats at 2, 6, 24 and 72 h after TBI (n = 3/group) were anesthetized as described above and perfused with 200 mL ice-cold heparinized saline followed by 500 mL 2% paraformaldehyde. Brains were removed, immersion-fixed for 30 min, cryoprotected in 30% sucrose, then frozen in precooled isopentane suspended in liquid nitrogen. Coronal brain sections (5-µm) were incubated at room temperature for 1 h in a 1 : 200 dilution of anti-AIF antibody, then incubated in a 1 : 200 dilution of secondary antibody Alexa 488 (Molecular Probes, Eugene, OR, USA) for 1 h. To assess nuclear morphology bis-benzimide (Sigma, St Louis, MO, USA) was applied prior to coverslipping. Fluorescence was visualized using excitation/emission wavelengths of 488/520 (green) and 346/460 (blue) λ for AIF and bis-benzimide, respectively. In sections from each specimen, the primary antibody was omitted to assess for non-specific binding of the secondary antibody.

To determine cell subtype, cellular localization, and presence of DNA fragmentation coronal brain sections from additional naive rats and rats 24 h after CCI were processed for triple-color immunofluoresence and examined using single-plane confocal microscopy. Brain sections were processed as above with the following exceptions. Sections were incubated in cold methanol for 30 min, washed with PBS, and then incubated at 37°C for 1 h in buffer containing terminal deoxynucleotidyl transferase and biotin-conjugated 16-dUTP (Boeringer Mannheim's, Indianapolis, IN) for detection of DNA fragmentation (TUNEL; Clark et al. 2000b). Sections were then incubated for 1 h in 1 : 200 dilution of strep-avidin conjugated Alexa 488 (Molecular Probes) at 4°C overnight. Slides were then incubated at room temperature for 1 h in a 1 : 200 dilution of anti-AIF antibody and a 1 : 500 dilution of the anti-neuronal antibody NeuN (Chemicon, Temecula, CA, USA). Sections were then incubated in 1 : 3000 dilutions of anti-goat Cy3.18 and anti-mouse Cy5 immunoconjugates (Jackson Immunochemicals, West Grove, PA, USA). Sections were examined using a Leica TCS NT confocal tri-laser scanning inverted microscope (Wetzlar, Germany). Pinhole size, photomultiplier tube settings, laser intensities, and magnifications (40 ×) were kept constant for all imaging analyses conducted.

Immunoelectronmicroscopy

Naive rats or rats at 24 h after CCI (n = 2/group) were perfused with 2% paraformaldehyde/0.01% glutaraldehyde. Brains were cut into 1-mm thick sections then incubated in polyvinyl pyrrolidone at 4°C overnight. The CA3 hippocampal region bilateraly was grossly dissected under an operating microscope, mounted on the head of a small metal screw, then snap frozen in liquid nitrogen. Ultrathin sections (80 nm) were mounted on grids, incubated in a 1 : 50 dilution of anti-AIF antibody for 1 h, then incubated in immunogold (12 nm) conjugated donkey anti-goat secondary antibody. Sections were fixed with 2.5% glutaraldehyde, counterstained with 2% neutral uranyl acetate, embedded in methyl cellulose, then observed using a JEOL JEM-100CX II electron microscope.

In vitro model of nitrosative/oxidative stress

Primary cultures of cortical neurons were prepared from 16 to 17-d-old Sprague-Dawley rat embryos as previously described with modification (Cao et al. 2001). Briefly, dissociated cell suspensions were placed in 96-well plates at a density of 5 × 104 cells/cm2 or in plastic dishes coated with poly D-lysine at a density of 1.3 × 107 cells/cm2 containing Neurobasal medium supplemented with antioxidant-rich B27 (Life Technologies, Gaithersburg, MD). Neurobasal medium provides neuron-enriched cultures with the near-complete absence of glia (Brewer 1995). Neuron survival was optimized by replacing glutamine with GlutaMaxl (Sigma). Cells were incubated at 37°C in a humidified chamber containing 5% CO2. On the third and sixth day in vitro (DIV) the culture media was replaced with fresh media. All experiments were done at 10 DIV, when cultures consisted primarily of neurons (> 95% MAP2-immunoreactive cells, < 5% GFAP-immunoreactive cells; n = 6).

On the tenth DIV cell medium was removed, cells were washed once with 37°C buffered Hanks' solution for 5 min at 37°C. Neuron-enriched cultures were exposed to 100 µm or 2 mm peroxynitrite (Cayman Chemical, Ann Arbor, MI) for 30 min. Cell viability at 30 min, 1 h, 2 h and 22 h after exposure was assessed by measuring lactate dehydrogenase (LDH) release and cellular respiration using a 3-[4,5-dimethylthiazol]-2,5-diphenyltetrazolium bromide (MTT) assay. Duplicate 20 and 150 µL supernatant samples from 96-well plates were used for the LDH and MTT assays, respectively. Values are expressed as the percentage of control (untreated) neurons and represent 3 separate experiments.

For analysis of AIF in subcellular compartments neurons at 0, 0.5, 1, 2, and 22 h after exposure to 100 µm or 2 mm peroxynitrite were lysed in buffer. Samples were centrifuged at 1000 g for 15 min at 4°C. The supernatants were centrifuged at 16 000 g for 20 min at 4°C. These supernatants were used for assessment of cytoslic proteins. The pellets containing mitochondria were lysed in buffer and sonicated until frothy. The initial pellet was re-suspended in lysis buffer then centrifuged at 16 000 g for 25 min at 4°C with the supernatant used for assessment of nuclear proteins. To determine the efficiency of separation of cellular compartments, immunoblotting for cytochrome c oxidase was performed pre hoc. Western blot analysis of AIF was performed as described above using a 1 : 500 dilution of anti-AIF antibody.

To determine whether large-scale DNA fragmentation occurred in this in vitro model, PFGE was performed in neuron-enriched cultures at 0, 0.5, 1, 2, and 22 h after exposure to 100 µm peroxynitrite as described above.

Pharmacological studies

Neuron-enriched cultures exposed to 100 µm peroxynitrite were pre- or post-treated with the potent metalloporphyrin-based peroxynitrite decomposition catalyst FP15 (Szabo et al. 2002) (Inotek, Beverly, MA) or PBS vehicle, or pretreated with the caspase inhibitors N-benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone (VAD; 100 µm; Enzyme Systems Products, Livermore, CA, USA) or boc-Asp(OMe)-fluoromethylketone (BAF; 100 µm; Enzyme Systems Products) or vehicle (DMSO). Dose–response curves for FP15 were first established using the MTT assay. Cells exposed to 100 µm peroxynitrite were then pretreated with the maximally effective dose of FP15 (100 µm) or vehicle and either western blot analysis for AIF or PFGE were performed as described above.

Statistical analysis

Descriptive data of DNA electrophoresis and immunocytochemistry were not statistically analyzed. Comparisons of the relative optical densities of AIF between groups, and LDH and MTT assays, were made using Kruskal–Wallis with post-hoc Dunn's test for non-parametric data or analysis of variance (mnova) with Student-Newman Keuls test if data were normally distributed. A p < 0.05 was considered significant.

Results

Intranuclear localization of AIF after TBI

Western blot analysis of subfractionated cellular proteins is shown in Fig. 1. A predominant ∼67 kDa protein band corresponding to the molecular weight of AIF was detected in mitochondrial fractions from ipsilateral cortex and hippocampus at all time points after TBI and controls (Fig. 1a). A truncated ∼57 kDa AIF peptide was infrequently detected in mitochondrial fractions from both control and injured rats. In nuclear fractions from ipsilateral cortex, the relative protein abundance of a truncated ∼57 kDa AIF peptide was increased at 2, 6, 24, and 72 h compared with control (Figs 1a and b; p < 0.05). Other, less prominent peptides < 57 kDa were also detected. The relative abundance of ∼67 kDa AIF in nuclear fractions from ipsilateral cortex were different over time ( p < 0.01); however, where this difference occurred could not be determined using multiple comparisons. In nuclear fractions from ipsilateral hippocampus, the relative protein abundance of the ∼67 kDa AIF peptide was increased at 2, 6, 24 and 72 h compared with control (Figs 2c and d; p < 0.05). No difference was detected in the relative protein abundance of the ∼57 kDa AIF band after TBI compared with control (p = 0.20) and other, less prominent peptides < 57 kDa were also detected, although there was considerable variability. AIF was not detected in the cytosolic fractions in either control or injured rats at any time point (data not shown). The anti-AIF antibody used detected a recombinant rat AIF protein (provided by JC), verifying the specificity of the antibody, and demonstrating the relative amounts of AIF in mitochondrial vs. nuclear samples (Fig. 1f ).

Figure 1.

Large-scale DNA fragmentation and intranuclear localization of AIF after TBI. A–D. Western blot analysis of AIF in mitochondrial and nuclear protein samples from ipsilateral cortex (a and b) and hippocampus (c and d). AIF was detected as an ∼67 kDa protein in mitochondrial samples from control and injured rats. In nuclear samples AIF was detected as both ∼67 kDa and ∼57 kDa (tAIF) peptides, with the ∼57 kDa peptide increased in injured cortex versus control and the ∼67 kDa peptide increased in injured hippocampus versus control (*p < 0.05). Relative AIF protein levels from nuclear fractions from cortex (b) and hippocampus (d) are shown in the box plots (median, 5th, 25th, 75th and 95th percentiles; n = 7/group; ROD = relative optical density). (e) PFGE after TBI in ipsilateral cortex and hippocampus. M = DNA markers. (f) Verification of anti-AIF antibody specificity using 10 µg recombinant AIF protein (amino acids 1–300, AIF300R), 15 µg/well of mitochondrial proteins, and 50 µg/well nuclear proteins. (g) Schematic representation of the intact AIF protein showing mitochondrial and nuclear localization sequences (MLS and NLS, respectively; Loeffler et al. 2001).

Figure 2.

Immunolocalization of AIF in ipsilateral parietal cortex (a, c, e, g, i and k) and CA3 hippocampus (b, d, f, h, j and l), from naive rats (a and b) and rats at 2 h (c and d), 6 h (e and f), 24 h (g and h) and 72 h (i and j). AIF immunoreactivity (green) increases with time within cell nuclei (labeled blue with bis-benzimide) in cortex and hippocampus versus naive controls. Bar ∼ 30 µm. (k and l) Single-plane, scanning confocal laser microscopic images of AIF in ipsilateral parietal cortex (k) and dentate gyrus (l) from rats at 24 h (Note fluorochromes are different than in panels a–j: AIF = red, TUNEL = green, NeuN = blue, AIF + NeuN = magenta, AIF + TUNEL + NeuN = white. Bar ∼50 µm.

In nuclear fractions, the molecular weight of the predominant protein band was smaller than that seen in mitochondrial fractions, and was ∼57 kDa. This likely represents post-translational modification of AIF, which contains two amino-terminal mitochondrial localization sequences (MLS; Fig. 1g; Loeffler et al. 2001). The antibody used was generated against the carboxy-terminus, and thus should detect both intact AIF, and AIF that has been truncated after the MLS, likely via proteolysis. While statistically it appears that regional differences in the relative amounts of nuclear AIF peptides exist between cortex and hippocampus after TBI (the 57 kDa band was increased in cortex whereas the 67 kDa band was increased in hippocampus vs. control), the temporal pattern and magnitude of nuclear AIF peptide expression were similar (Figs 1b and d).

Immunocytochemical analysis also demonstrated intranuclear localization of AIF after TBI (Fig. 2). As expected, AIF was detected throughout the cell bodies in each examined section. In cells from control brain, intense bis-benzimide staining of nuclear material with minimal AIF immunoreactivity over cell nuclei was seen. In contrast, many cells in the ipsilateral parietal cortex, dentate gyrus, and CA3 hippocampus demonstrated increasing AIF immunoreactivity within cell nuclei, with a concurrent reduction in bis-benzimide intensity, suggesting nuclear translocation of AIF. This labeling pattern was patchy, but was more consistent with increasing time after injury, to where ∼50% of cells in injured cortex and hippocampus demonstrated nuclear AIF immunoreactivity at 24 and 72 h. AIF immunoreactivity was not seen in sections incubated without primary antibody. Triple-label confocal microscopy demonstrated that the majority of cells with nuclear AIF labeling were NeuN-positive (Figs 2k–l). NeuN is specific for neurons and labels both nuclei and cell bodies (Mullen et al. 1992). Most neurons within the injured cortex and many neurons within the ipsilateral hippocampus with nuclear AIF also had evidence of DNA fragmentation (TUNEL-positive). The ipsilateral dentate gyrus is shown in Fig. 2(l). Similar labeling patterns were seen in CA3 hippocampus.

Intranuclear localization of AIF after TBI was further verified using immunoelectronmicroscopy (Fig. 3). In CA3 hippocampus from naive rats immunogold labeled AIF was rarely seen in cell nuclei (Fig. 3a), but was readily detectable in mitochondria within the cristae and within the inner membrane (Fig. 3b; > 30 neurons examined). In contrast, in CA3 hippocampus 24 h after TBI, immunogold labeled AIF was prominent in nearly all cell nuclei (> 30 neurons examined), primarily in regions of euchromatin and the euchromatin-heterochromatin boundries (Figs 3d–f). AIF labeling was also detected in mitochondria 24 h after TBI, but was reduced compared with naive. Sparse immunogold particles were also detected in nuclei of the contralateral CA3 hippocampus (Fig. 3c).

Figure 3.

Subcellular immunolocalization of AIF in CA3 hippocampus using immunogold labeling (12 nm gold particles) and electronmicroscopy. (a and b). Naive rat. (c) Contralateral CA3 hippocampus at 24 h. (d–f) Ipsilateral CA3 hippocampus at 24 h. Note prominence of immunogold particles in regions of euchromatin and at the euchromatin-heterochromatin boundaries (arrows), and the localization of AIF to cristae and inner membrane of mitochondrion. c = cytosol, m = mitochondria, n = nucleus. Bars ∼250 nm.

Large scale DNA fragmentation after TBI

PFGE was used because conventional DNA gel electrophoresis is incapable of separating large (>10 kbp) DNA fragments. DNA fragments of ∼50 kbp in size were detected in the ipsilateral cortex in all samples at 2, 6, and 24 h after TBI (Fig. 1e). In ipsilateral hippocampus, ∼50 kbp fragments were detected in 2/4 samples at both 2 and 6 h, and were seen in 3/4 samples at 24 h (a representative sample from each time point is shown in Fig. 1e). In both cortex and hippocampus, ∼50 kbp fragments were most prominent at 24 h. In all cases, ∼50 kbp DNA fragments were not detected in cortical or hippocampal samples from uninjured rats, and DNA extracted from control rats did not migrate through the gel suggesting that it remained intact.

Intranuclear localization of AIF and large-scale DNA fragmentation in neurons exposed to peroxynitrite

As oxidative and nitrosative stress can produce large-scale DNA fragmentation and caspase-independent neuronal cell death (Hill et al. 2000), and nitrosative stress occurs after experimental spinal cord injury (Scott et al. 1999) and TBI (Whalen et al. 1999a), neuron-enriched cultures were exposed to peroxynitrite to verify nuclear localization of AIF. Doses of peroxynitrite were used that produce both complete and incomplete neuronal death as determined by both LDH release (Fig. 4a) and mitochondrial respiration (Fig. 4b). Intranuclear localization of AIF in neurons was seen as early as 30 min after exposure to either peroxynitrite dose (Fig. 4c). Intranuclear AIF was detected as an ∼67 kDa band, in contrast to the in vivo studies where both ∼67 and 57 kDa bands were seen. Nuclear AIF was sustained for 22 h after 100 µmnova peroxynitrite, but was not detected beyond 2 h after exposure to 2 mmnova peroxynitrite. The latter represents complete cell death and is consistent with a lack of LDH, MTT, and mitochondrial AIF in cultures 22 h after 2 mmnova peroxynitrite exposure.

Figure 4.

Peroxynitrite-induced intranuclear localization of AIF, large-scale DNA fragmentation, and cell death in rat cortical neuron-enriched cultures. (a and b) LDH release and conversion of MTT to formazin in neurons 0.5–22 h after exposure to 100 µmnova and 2 mmnova peroxynitrite. *p < 0.05 versus control, n = 3 experiments/timepoint done in duplicate. Note lack of detection of both LDH and MTT at 22 h after 2 mmnova peroxynitrite exposure representing complete cell death prior to 22 h. (c) Intranuclear localization of AIF in neurons 0.5–22 h after exposure to 100 µmnova and 2 mmnova peroxynitrite (representative of 3 independent experiments/timepoint). (d) PFGE showing ∼ 200 and ∼50 kbp DNA fragments in neurons 0.5–22 h after exposure to 100 µmnova peroxynitrite (representative of 4 experiments/timepoint). (e) Treatment with the peroxynitrite decomposition catalyst FP15, but not the caspase inhibitors VAD or BAF (100 µmnova), preserves the ability of cells to convert MTT to formazin after exposure to 100 µmnova peroxynitrite. *p < 0.05 versus control, n = 3/timepoint done in duplicate. (f) Pre-treatment with 100 µmnova FP15 prevents intranuclear AIF expression at 0.5–22 h and large-scale DNA fragmentation at 24 h after exposure to 100 µmnova peroxynitrite (representative of 2 independent experiments/timepoint).

Exposure to 100 µmnova peroxynitrite also produced large-scale DNA fragments; however, that pattern differed from the in vivo studies. DNA fragments of ∼200 kbp were detected at 30 min to 22 h (Fig. 4d). Similar to the in vivo studies, 50 kbp fragments could be detected at 2 h and were prominent at 22 h. In the in vitro studies this corresponded to a reduction in the larger ∼200 kbp fragments. DNA smears were also seen at all times after exposure.

Treatment with the peroxynitrite decomposition catalyst FP15 preserved mitochondrial respiration in a dose dependent manner 2 h after exposure to 100 µmnova peroxynitrite (Fig. 4e). FP15 was effective when given before and 30 min after peroxynitrite exposure. Caspase inhibitors were ineffective in preserving mitochondrial respiration 2 h after exposure to 100 µmnova peroxynitrite (100 µmnova VAD or BAF) compared with vehicle. Pretreatment with 100 µmnova FP15 completely blocked the appearance of AIF in nuclear fractions and large-scale DNA fragmentation after peroxynitrite exposure (Fig. 4f ). In these sets of experiments, mitochondrial AIF was reduced or undetectable at 22 h, presumably related to either translocation of AIF from mitochondria or cell death.

Discussion

The novel findings of this study are: (1) intranuclear localization of AIF, an event sufficient to produce cell death in non-neuronal cells in vitro, occurs in ipsilateral cortical and hippocampal neurons within hours after TBI with secondary insult in rats; (2) large-scale DNA fragmentation, the signature event in AIF-mediated cell death, also occurs in injured cortex and hippocampus within hours after experimental TBI; (3) intranuclear AIF, large-scale DNA fragmentation, and cell death also occurs in primary cortical neurons exposed to peroxynitrite; and (4) treatment with a peroxynitrite decomposition catalyst but not caspase-inhibitors is protective in this in vitro cytotoxicity model. These are the first data implicating nuclear localization of AIF in this additional/alternate pathway of programmed cell death in neurons in vitro and after brain injury in vivo.

AIF plays a dominant role in programmed cell death in some in vitro models (Joza et al. 2001). Upon loss of mitochondrial membrane potential (ΔΨm) and opening of the mitochondrial permeability transition pore, AIF is released and rapidly translocates to the nucleus, preceding cytochrome c release from mitochondria (Susin et al. 2000). In the present study, intranuclear AIF was detected by 30 min in vitro and 2 h in vivo, although earlier time points were not examined. Once AIF has entered the nucleus, it is sufficient to produce ∼50 kbp DNA fragmentation that is not inhibited by the caspase inhibitor VAD, but is inhibited by the magnesium chelator EDTA (Susin et al. 1999). Cytosolic microinjection of AIF, including an AIF deletion mutant lacking the mitochondrial localization and FAD binding domain, produces loss of ΔΨm, nuclear translocation of AIF, ∼50 kbp DNA fragmentation, and cell death (Susin et al. 1999). Thus, AIF may function as an endonuclease itself, or as a cofactor for a magnesium-dependent endonuclease. The presence of AIF primarily in regions of euchromatin or euchromatin-heterochromatin borders after TBI (Fig. 3) are consistent with either possibility. Magnesium-dependent endonucleases have been shown to generate large-scale DNA fragmention, whereas calcium/magnesium-dependent endonucleases produce internucleosomal DNA fragmentation (Sun and Cohen 1994; Walker et al. 1994; Zhivotovsky et al. 1994). Both are likely to be activated after TBI (Pravdenkova et al. 1996).

Large-scale DNA fragmentation was detected using PFGE as early as 2 h, and was most prominent at 24 h, after TBI in ipsilateral cortex and hippocampus. After TBI, DNA fragmentation can be detected by PFGE earlier than it can be detected by TUNEL, where fragmentation is not seen until 6 h (Clark et al. 2001). Previous studies in various models of TBI have reported cells with DNA fragmentation lacking typical apoptotic morphologies, attributed to neuronal necrosis (Rink et al. 1995; Clark et al. 1997a,b). Large-scale DNA fragmentation also precedes, and may occur independently of, oligonucleosomal DNA fragmentation after ischemic brain injury (Charriaut-Marlangue et al. 1995; MacManus et al. 1997). While it remains clear that necrosis contributes to cell death after TBI, particularly in contused cortex, a more cybernetic mechanism for large-scale DNA fragmentation, likely facilitated by AIF is supported. Large scale DNA fragmentation may also occur in necrosis, as brains from decapitated rats incubated at high temperatures indicate that both apoptotic and necrotic cell death pathways involve DNA damage (MacManus et al. 1995). Finally, it is important to bear in mind that the model of CCI with secondary hypoxemic insult was chosen for this study because it produces more DNA damage in the ipsilateral hippocampus than CCI alone (Clark et al. 1997a). Verification that nuclear AIF translocation does or does not occur in other TBI models is needed.

Intranuclear AIF and large-scale DNA fragmentation represents a previously undefined pathway of programmed cell death after TBI. Similar to the temporal appearance of large-scale DNA fragments, intranuclear AIF was detected by 2 h in injured cortex and granular layers of the ipsilateral hippocampus and dentate gyrus. Neuronal death that is only partially reduced by the relatively selective caspase-3 inhibitor N-benzyloxycarbonyl-Asp(Ome)-Glu(Ome)-Val-Asp(Ome)-fluoromethylketone occurs in these regions (Clark et al. 2000b). In PC12 cells or primary cortical neurons expressing a dominant-negative caspase-activated deoxyribonuclease, apoptotic stimuli still result in cell death, but without internucleosomal DNA fragmentation or apoptotic nuclear changes (Chen et al. 2000; Cao et al. 2001). Similar phenotypes were seen with the cytoxicity model used in the present study, where the caspase inhibitors VAD and BAF were found to be ineffective (Figs 4e and f ). Since AIF-mediated cell death in vitro is caspase-independent (Susin et al. 2000), it is possible that some degree of programmed cell death after TBI is also caspase-independent.

Truncated AIF was detected in nuclear fractions in vivo (Fig. 1), but not in vitro (Fig. 4). This phenomenon has also been reported in lymphocyte apoptosis (Dumont et al. 2000). The amino-terminal 1–120 polypeptide sequence of AIF contains two MLS (Loeffler et al. 2001). AIF deletion mutants Δ1-120 lack oxidoreductase activity, but retain the capacity to produce DNA fragmentation and cell death (Susin et al. 1999). We speculate that this peptide domain may be cleaved within the mitochondria. Since AIF is a relatively large protein and is detected within the mitochondrial cristae as well as the inner membrane, proteolyis may be required for AIF release. Proteolysis of AIF may also lead to conformational changes necessary for mitochondrial release, exposure of nuclear localization sequences at amino acids 277–299 and 445–450 (Susin et al. 1999), and/or stimulating endonuclease activity. The absence of detectable truncated AIF fragments in the in vitro model suggests differences in AIF proteolysis in cultured neurons exposed to peroxynitrite vs. vulnerable neurons after acute injury in vivo. Further study to determine the relevance of AIF proteolysis both in vitro and in vivo, is warranted.

In addition to the pathway producing archeotypal apoptosis, a heretofore undefined pathway involving intranuclear localization of AIF and large-scale DNA fragmentation occurs early after TBI and in neurons exposed to nitrosative stress. Molecular/pharmacological inhibitors of AIF-mediated DNA damage, including peroxynitrite decomposition catalysts, in in vivo models of acute brain injury deserves investigation.

Acknowledgements

We thank John Melick for expert technical assistance. Supported by National Institutes of Health/NINDS Grants RO1 NS 38620 (RSBC and XZ) and P50 NS30318 (RSBC, PMK and SHG), and by the Department of Veterans Medical Affairs Merit Review Program (SHG).

Ancillary