|
|
||||||||




* Centre National de la Recherche Scientifique, UPR420, F-94801 Villejuif, France;
Assistance Publique-Hôpitaux de Paris, Service de Néphrologie B, Hôpital Tenon, F-75020, France;
Unité dOncologie Virale, Institut Pasteur, F-75724 Paris cedex 15, France;
§ McMaster University Medical Centre and
|| Department of Biochemistry, McMaster University, Hamilton, Ontario, L8N 3Z5, Canada;
** Laboratoire dAnatomopathologie et INSERM U430, Hôpital Broussais, F-75014 Paris, France; and

The Amgen Institute and Ontario Cancer Institute, Department of Medical Biophysics and Immunology, University of Toronto, Toronto, Ontario M5G 2C1, Canada
1Correspondence: 19 rue Guy Môquet, B.P. 8, F-94801 Villejuif, France. E-mail: kroemer{at}infobiogen.fr
| ABSTRACT |
|---|
|
|
|---|
Key Words: antioncogene mitochondrial transmembrane potential oncogene permeability transition programmed cell death
| INTRODUCTION |
|---|
|
|
|---|
Opening of the PT pore or translocation of Bax-like proteins from the
cytosol to the mitochondrion cause the release of soluble intermembrane
proteins from the mitochondrion, via the outer mitochondrial membrane
(1
2
3
, 6
7
8
, 12
13
14)
. The first apoptogenic
intermembrane protein to be molecularly identified has been cytochrome
c. (2)
. Cytochrome c can interact
with the WD40 domains of the dATP/ATP binding protein Apaf-1 in a
reaction that finally leads to the autoactivation of pro-caspase-9 and
initiation of the caspase activation cascade (15)
. Many
investigators have tacitly or overtly assumed that cytochrome
c would be a major rate-limiting factor of apoptotic cell
death (2
, 4
, 6
, 7)
. However, this notion does not
withstand experimental verification because 1) in most if
not all models of stress-induced mammalian apoptosis, inhibition of
caspases does not prevent cell death (16
17
18
19
20)
and
2) mitochondria release factors other than cytochrome
c participate in the apoptotic process. This applies in
particular to apoptosis inducing factor (AIF), a mitochondrial
intermembrane protein that has the unique property to induce
apoptosis-like changes in purified nuclei in vitro in a
caspase-independent fashion (1
, 3)
.
AIF has recently been characterized at the molecular level and
discovered to be a novel mitochondrial intermembrane flavoprotein with
significant homology to bacterial and plant oxidoreductases
(21)
. This protein is synthesized as a nonapoptogenic
precursor in the cytoplasm and efficiently imported into the
mitochondrial intermembrane space. Import through the general impart
pathway culminates into the proteolytic removal of the amino-terminal
mitochondrial targeting sequence, attachment of a flavin adenine
dinucletide group, and refolding of the protein, which becomes
potentially apoptogenic (21)
. On induction of apoptosis,
AIF redistributes to the nucleus. Moreover, when added to purified
nuclei, AIF induces partial chromatin condensation as well as
large-scale (~50 kbp) DNA fragmentation in a caspase-independent
fashion. These nuclear changes resemble to some extent those observed
in intact cells in which apoptosis is induced in conditions of caspase
inhibition (16
17
18
19)
, suggesting that AIF may be
responsible for at least some of the caspase-independent features of
apoptosis.
Recent data indicate that the paradigmatic opposition between apoptosis
and necrosis is less clear than this has been generally thought. Thus,
necrosis can be prevented in some cases by Bcl-2 (22)
and
may involve an early mitochondrial permeabilization event
(22
23
24
25)
. Moreover, classical protocols of apoptosis
induction give rise to a more necrotic phenotype when cytosolic ATP
levels are low (26
, 27)
. These finding have been
interpreted to mean that the early phase of apoptosis and of some types
of necrosis may involve common mitochondrial events. Only at the
postmitochondrial level, depending on the ATP concentration and caspase
activation, would the decision between apoptotic and necrotic cell
death be made (24
, 26
, 27)
.
Intrigued by these possibilities, we have used confocal immunofluorescence microscopy as well as immune electron microscopy to determine the subcellular localization of AIF in a variety of different apoptosis-inducing conditions, in cells that overexpress Bcl-2, as well as in conditions in which caspase activation is prevented or ATP levels are low. As shown here, AIF translocation can occur in experimental conditions in which advanced apoptosis is prevented by caspase inhibition. Moreover, AIF translocation can occur in conditions of ATP depletion that lead to nonapoptotic cytolysis (necrosis).
| MATERIALS AND METHODS |
|---|
|
|
|---|

m, 20 nM, Molecular Probes, Eugene, Oreg.)
(30)
Immunostaining protocols
A rabbit antiserum was generated against a mixture of three
peptides derived from the mouse AIF aa sequence (aa 151170, 166185,
181200, coupled to keyhole limpet hemocyanine, generated by Syntem,
Nîmes, France). This antiserum (ELISA titer ~10.000) was used
(diluted 1/1000) on paraformaldehyde (4% w:v) and picric acid-fixed
(0.19% v:v) cells (cultured on 100 µm coverslips; 18 mm Ø;
Superior, Germany), and revealed with a goat anti-rabbit IgG
conjugated to phycoerythrine (PE) (Southern Biotechnology, Birmingham,
Ala.). Control experiments performed with the pre-immune antiserum or
in the presence of an excess (100 µM) of the three immunogenic
peptides confirmed that all detectable fluorescence was specific (not
shown). Cells were counterstained for the detection of cytochrome
c (mAb 6H2.B4 from Pharmingen, revealed by a goat anti-mouse
IgG fluorescein isothiocyanate (FITC) conjugate; Southern
Biotechnology), hsp60 (mAb H4149 from Sigma, revealed by the same
anti-mouse IgG FITC conjugate), or DNA (Sytox Green from Molecular
Probes, 25 nM, 15 min of incubation at room temperature).
Conventional, confocal, and immune electron microscopy
Conventional examination of samples was performed in a
Leitz Labolux S microscope equipped with standard filters for FITC and
PE, as well as a Leica camera. CMXRos and Sytox Green fluorescence were
monitored using the PE and FITC filters, respectively. Confocal
microscopy was performed on a Leica TC-SP (Leica Microsystems,
Heidelberg, Germany) equipped with an ArKr laser mounted on an inverted
Leica DM IFBE microscope with an 63 x 1,32 NA oil objective. To
avoid cross talk between different fluorochromes, images were acquired
in a sequential fashion. For immunoelectron microscopy, cells were
fixed for 1 h at 4°C with 4% paraformaldehyde in 0.1 M
phosphate-buffered saline (PBS), pH 7.4, washed three times in PBS, and
embedded in 10% gelatin at 37°C, then in 1.7 M sucrose + PVP15%
overnight. Ultrathin sections (60 nm) were floated on drops of anti-AIF
antiserum (1/100 dilution in PBS + 1% bovine serum albumin + 1%
gelatin, pH 7.4), then revealed by an immunogold (10 nm) anti-rabbit Ig
conjugate, Amersham, Arlington Heights, Ill.), counterstained with
uralylacetate-oxalate, and embedded in methylcellulose, following
standard protocols (31)
. Cells that demonstrate a clear
AIF staining of the nucleus were scored as positive for AIF
translocation. Cells having lost the mitochondrial staining pattern for
cytochrome c or CMXRos in favor of a diffuse cyosolic
staining were scored positive for cytochrome translocation or as

mlow, respectively. Using
these criteria, the frequency of intermediate cases (cytochrome
c release or loss of CMXRos staining only in a part of the
cell was low (<2%). Such intermediate cases were scored as negative
events. A minimum of 200 cells were monitored for these parameters for
each data point.
DNA gel electrophoresis
For the detection of oligonucleosomal DNA fragmentation, nuclear
DNA from lysed cells (treated with protease K and RNase according to
standard protocols) was subjected to conventional horizontal agarose
gel electrophoresis (1%) followed by ethidium bromide staining
(32
, 33)
. For pulse field gel electrophoresis (PAGE), DNA
was prepared from agarose plugs (1x106 cells)
(34)
digested twice with proteinase K (1 mg/ml; 50°C;
12 h) in NDS buffer (0.5 M EDTA, 10 mg/ml lauroyl sarcosine),
washed in TBE x 0.5, followed by electrophoresis in a
Bio-RadCHEF-DR II (Richmond, Calif.) equipment (1% agarose; TBE x 0.5; 200 V; 24 h; pulse wave 60 s; 120° angle).
Molecular weight standards were from Bio-Rad (Yeast chromosomes) and
Appligene (Raoul; Illkirch, France).
| RESULTS AND DISCUSSION |
|---|
|
|
|---|
|
In normal untreated cells, AIF (revealed by PE in red) spares the
nucleus (stained by Sytox-green, a DNA intercalating dye) (Fig. 2A
). However, after induction of apoptosis, a large portion of
AIF is found within the nucleus, as determined by confocal analyses
(Fig. 2B
). This interpretation was confirmed by
immunoelectron microscopy, showing that before induction of apoptosis
mitochondria stain for AIF (Fig. 3A
), yet lose most of the specific signal after STS treatment
(not shown). Nuclei from control cells lack AIF (Fig. 3B
).
In contrast, in nuclei from STS-treated cells, immunogold particles
reveal the presence of AIF (Fig. 3C
). Collectively,
these data confirm that after induction of apoptosis, AIF redistributes
from mitochondria to the nucleus.
|
|
Temporal relationship between AIF and cytochrome c
release
Staurosporin is known to induce the release of cytochrome
c from mitochondria to the cytosol (2
, 35)
.
Since the redistribution of cytochrome c has been reported
to be an early event of the apoptotic process (2
, 35)
, we
determined the temporal and spatial relationship between cytochrome
c and AIF release. As revealed by two-color
immunofluorescence detection of cytochrome c and AIF, both
proteins normally colocalize in mitochondria (Fig. 4A
. After 4 h of STS treatment (2 µM), most cells
(>80%) manifest a phenotype where the mitochondrial pattern of
cytochrome c distribution is lost and cytochrome
c relocalizes to the cytosol; AIF partially redistributes to
the nucleus, giving rise to a spatial separation of the two
mitochondrial intermembrane proteins. This same phenotype is found in
about half of the cells 3 h after initiation of the treatment
(example 2 in Fig. 4B
). Moreover, in a significant portion
of cells (36%), which probably represent a short transitional step of
the apoptotic cascade, AIF-specific fluorescence (red) is translocated
to the nucleus, whereas mitochondria still brightly stain for
cytochrome c (green, example 1 in Fig. 4B
). The
percentage of cells having translocated AIF but not cytochrome drops to
<2% when the incubation with STS is prolonged to 8 h. Within the
limits of detection by quantitative confocal immunofluorescence, these
data suggest that AIF is released shortly before cytochrome
c.
|
Temporal relationship between mitochondrial change and nuclear
chromatin condensation
Chromatin condensation, as monitored by staining with the
intercalating dye Hoechst 33342 or Sytox-green, advances gradually
during the apoptotic process. Several stages of nuclear apoptosis can
be distinguished: stage I with rippled nuclear contours and a rather
partial chromatin condensation; stage IIa with marked peripheral
chromatin condensation; and stage IIb with formation of nuclear bodies
(Fig. 5A
). Kinetic studies reveal that these stages actually reflect
successive step in the apoptotic process (Fig. 5B
). We have
correlated the translocation of cytochrome c and AIF with
the advancement of nuclear apoptosis. It appears that AIF is
translocated in nearly all cells in which morphological changes in the
nucleus have become apparent, including stage I of the process. In
contrast, cytochrome c translocation is only found in
approximately half of cells in stage 1 of the process and is only
complete (>90%) at stage IIa (Fig. 5C
). These data are
compatible with the two-color immunofluorescence determinations,
suggesting that AIF release precedes full cytochrome c
release (Fig. 4
; see above). Whenever nuclear apoptosis becomes
apparent (including stage I), the 
m is
reduced in most cells, as indicated by staining with the
potential-sensitive dye CMXRos (Fig. 5A, C
). However,
a significant fraction of cells with nuclear apoptosis (between 10 and
20%) retained a mitochondrial staining pattern of CMXRos, suggesting
that at least a few mitochondria can maintain the

m during apoptosis, as previously suggested
(35)
.
|
Regulation of AIF translocation by multiple effectors: Bcl-2,
c-Myc, DNA damage, and ceramide
Overexpression of Bcl-2 has been suggested to inhibit apoptosis
mainly by preventing the release of cytochrome c (6
, 7)
. Using a Bcl-2/ActA fusion protein specifically targeted to
the mitochondrion (28)
, we have confirmed that Bcl-2
prevents the release of cytochrome c in response to STS (not
shown). Mitochondrion-targeted Bcl-2 also prevents the release of AIF.
Thus, in conditions in which vector-only transfected cells readily
undergo a mitochondrio-nuclear translocation of AIF in response to STS
(Fig. 1B
, Fig. 2B
), Rat-1 cells overexpressing
Bcl-2 withhold AIF in their mitochondria (Fig. 1C
, Fig. 2C
). Release of AIF is not restricted to apoptosis induction
by STS-mediated tyrosine kinase inhibition. Rather, we have observed
that multiple different death triggers can provoke the relocalization
of AIF. This applies to apoptosis triggered by serum withdrawal and
simultaneous activation of an estrogen receptor c-Myc fusion protein,
which translocates to the nucleus after addition of estradiol
(Fig. 6A
). Moreover, several additional apoptosis inducers including
the topoisomerase II inhibitor etoposide (Fig. 6B
) and the
proapoptotic second messenger ceramide (Fig. 6C
), as well as
adriamycin and cis-platin (not shown), induce the
mitochondrio-nuclear translocation of AIF, suggesting that the
redistribution of AIF is a general feature of the apoptotic process.
|
AIF translocation is a caspase-independent phenomenon
Inhibition of caspases using Z-VAD.fmk does not prevent cell death
induced by several proapoptotic stimuli including STS
(19)
. However, Z-VAD.fmk prevents or retards advanced
nuclear apoptosis. Four hours after addition of STS, the preponderant
phenotype is stage I (56±5%), whereas 6 h later most cells are
in stage IIa+b (72±3%). In the presence of Z-VAD.fmk, cells mostly
remain in stage I (Fig. 7E
). Concomitantly, Z-VAD.fmk prevents the oligonucleosomal
DNA fragmentation induced by STS (Fig. 7D
). These internal
controls indicate that Z-VAD.fmk did prevent caspase-dependent features
of apoptosis. However, Z-VAD.fmk did not prevent the translocation of
AIF to the nucleus (Fig. 7A, B
) or loss of the mitochondrial
transmembrane potential (Fig. 7C
). Moreover, Z-VAD.fmk
failed to prevent large-scale DNA fragmentation to fragments of ~ 50 kbp, as determined by pulse field gel electrophoresis (Fig. 7D
). As a result, it appears that the mitochondrio-nuclear
translocation of AIF and concomitant ~ 50 kbp DNA fragmentation
is caspase-independent.
|
AIF translocation is ATP-independent
One strategy of preventing the acquisition of apoptotic
morphology consists in the depletion of ATP (27)
. It is
generally assumed that ATP depletion results in the functional
inactivation of the ATP binding caspase activator Apaf-1, thereby
preventing caspase activation (15
, 27)
. ATP depletion is
of pathophysiological interest because it mimics situations of ischemia
in which cells die from necrosis. As shown in Fig. 8
, cells kept in a glucose-free, oligomycin A-supplemented medium (which
abolishes both glycolytic ATP and respiratory ATP generation; ref
27
) die within 6 to 8 h. The aspect of nuclei
obtained in these conditions is clearly nonapoptotic, different from
all stages of apoptotic nuclear chromatin redistribution described
above. Cells manifest a homogeneous chromatin condensation (note that
the scale in Fig. 8A-C
differs from that in Figs. 1
, 2
and 4
5
6
7
), which we judged to be necrotic (27)
. At this point,
AIF accumulates in the nucleus (Fig. 8A, B
). Addition of STS
to ATP-depleted cells accelerates cellular demise (Fig. 8E
), and again cells with chromatin condensation
demonstrate mitochondrio-nuclear AIF translocation (Fig. 8A, B
). Thus, the translocation of AIF does not depend on ATP, similar
to what has been reported for the translocation of cytochrome
c (ref 36
) and data not shown). In the context
of ATP depletion, cells treated with STS only manifest a reduction in
nuclear size, without any detectable changes in chromatin distribution
(Fig. 8B, C
). ATP depletion does prevent STS-induced
oligonucleosomal DNA fragmentation, yet has no effect on ~50 kbp DNA
fragmentation (Fig. 8D
). Mitochondrion-targeted Bcl-2
prevents the necrotic cell death induced by ATP depletion (Fig. 8E
).
|
| CONCLUDING REMARKS |
|---|
|
|
|---|
When added to purified nuclei in a cell-free system, recombinant AIF
has two major effects. It causes peripheral chromatin condensation and
large-scale DNA fragmentation, yet fails to generate nuclear bodies or
oligonucleosomal DNA fragments (21)
. These latter changes
have been attributed to the caspase-dependent activation of
caspase-activated DNase (CAD), also called DNA fragmentation factor
(DFF40) (39
40
41
42)
. In conditions in which caspases are
inhibited, yet mitochondrial factors such as AIF and cytochrome
c are released, cells manifest only partial chromatin
condensation (stage I) and large-scale DNA fragmentation (Fig. 7)
.
These results are in complete agreement with those obtained in the
cell-free system, as well as with the staging of apoptosis (Fig. 5)
induced in the absence of caspase inhibitors. They suggest that AIF is
indeed the caspase-independent factor responsible for initial chromatin
condensation (stage I) and large-scale DNA fragmentation (Fig. 9
). In contrast, the data obtained in conditions of ATP depletion are
more difficult to be accommodated in this scheme. In intact cells,
shortage of ATP abolishes all signs of inhomogeneous (apoptotic)
chromatin condensation yet does not affect translocation of AIF into
the nucleus or prevent large-scale DNA fragmentation (Fig. 8)
. However,
in the cell free system, ATP is dispensable for AIF-induced chromatin
condensation (21)
. So far we have no explanation for this
apparent paradox. In any case, the data obtained in intact cells are
incompatible with the simple assumption that ATP depletion affects the
phenotypic manifestations of cell death solely by preventing caspase
activation. Rather, additional effects of ATP shortage must be
postulated.
|
Irrespective of these details, the data presented in this study suggest
an overall scheme of cell death regulation in which Bcl-2-inhibitible
mitochondrial alterations are rate-limiting for both apoptosis and
necrosis. Increase in mitochondrial membrane permeability results in
the release of apoptogenic factors, including AIF and cytochrome
c. AIF then accounts for initial apoptotic changes in the
nucleus, whereas cytochrome c triggers caspase activation-
and CAD-dependent downstream events leading to advanced nuclear
apoptosis (Fig. 9)
. At present, it is not clear whether AIF truly
participates in the regulation of cell death (e.g., by determining
thresholds of death induction) or whether its contribution is limited
to the post mortem manifestations of cell death, beyond the
Bcl-2-regulated checkpoint. Since some AIF distributes to the cytosol
and since AIF may affect mitochondrial function, both hypotheses appear
plausible (3)
. Genetic studies involving homologous
recombination of the AIF gene will resolve this issue in the future.
Irrespective of these considerations, it appears that the translocation
of AIF from a mitochondrial to an extramitochondrial localization
occurs both in apoptosis and necrosis and thus may participate in the
common death pathway.
| ACKNOWLEDGMENTS |
|---|
| FOOTNOTES |
|---|
| REFERENCES |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
S. Renolleau, S. Fau, and C. Charriaut-Marlangue Gender-Related Differences in Apoptotic Pathways After Neonatal Cerebral Ischemia Neuroscientist, February 1, 2008; 14(1): 46 - 52. [Abstract] [PDF] |
||||
![]() |
A. N. Kavazis, J. M. McClung, D. A. Hood, and S. K. Powers Exercise induces a cardiac mitochondrial phenotype that resists apoptotic stimuli Am J Physiol Heart Circ Physiol, February 1, 2008; 294(2): H928 - H935. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. S. Collingwood, E. V. Smirnova, M. Bogush, N. Carpino, R. S. Annan, and A. Y. Tsygankov T-cell Ubiquitin Ligand Affects Cell Death through a Functional Interaction with Apoptosis-inducing Factor, a Key Factor of Caspase-independent Apoptosis J. Biol. Chem., October 19, 2007; 282(42): 30920 - 30928. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. Chassin, M. Bens, J. de Barry, R. Courjaret, J. L. Bossu, F. Cluzeaud, S. Ben Mkaddem, M. Gibert, B. Poulain, M. R. Popoff, et al. Pore-forming epsilon toxin causes membrane permeabilization and rapid ATP depletion-mediated cell death in renal collecting duct cells Am J Physiol Renal Physiol, September 1, 2007; 293(3): F927 - F937. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Srivastava, H. Banerjee, A. Chaudhry, A. Khare, A. Sarin, A. George, V. Bal, J. M. Durdik, and S. Rath Apoptosis-Inducing Factor Regulates Death in Peripheral T Cells J. Immunol., July 15, 2007; 179(2): 797 - 803. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. Wang, L. Zhang, X. Han, J. Yang, J. Qian, S. Hong, F. Samaniego, J. Romaguera, and Q. Yi Atiprimod inhibits the growth of mantle cell lymphoma in vitro and in vivo and induces apoptosis via activating the mitochondrial pathways Blood, June 15, 2007; 109(12): 5455 - 5462. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Lubitz, S. Glaser, J. Schaft, A. F. Stewart, and K. Anastassiadis Increased Apoptosis and Skewed Differentiation in Mouse Embryonic Stem Cells Lacking the Histone Methyltransferase Mll2 Mol. Biol. Cell, June 1, 2007; 18(6): 2356 - 2366. [Abstract] [Full Text] [PDF] |
||||
![]() |
P. J. Adhihetty, V. Ljubicic, and D. A. Hood Effect of chronic contractile activity on SS and IMF mitochondrial apoptotic susceptibility in skeletal muscle Am J Physiol Endocrinol Metab, March 1, 2007; 292(3): E748 - E755. [Abstract] [Full Text] [PDF] |
||||
![]() |
G. Kroemer, L. Galluzzi, and C. Brenner Mitochondrial Membrane Permeabilization in Cell Death Physiol Rev, January 1, 2007; 87(1): 99 - 163. [Abstract] [Full Text] [PDF] |
||||
![]() |
E. P. Jane, D. R. Premkumar, and I. F. Pollack Coadministration of Sorafenib with Rottlerin Potently Inhibits Cell Proliferation and Migration in Human Malignant Glioma Cells J. Pharmacol. Exp. Ther., December 1, 2006; 319(3): 1070 - 1080. [Abstract] [Full Text] [PDF] |
||||
![]() |