DIFFERENTIAL REQUIREMENT FOR THE MITOCHONDRIAL APOPTOSIS-INDUCING FACTOR IN APOPTOTIC PATHWAYS

Nicholas Joza

A thesis submitted in conformity with the requirements for the degree of Master of Science Graduate Department of Immunology University of Toronto

O Copyright by Nicholas Joza (2001)

I+B

d Canada

National Library

Bibliothèque nationale du Canada

Acquisitions and Bibliographie Services

Acquisitions et services bibliographiques

395 Wellington Street OnawaON KlAON4 Canada

395, nre WeHington OUawaOhl K1AON4 Canada

The author has granted a nonexclusive licence allowing the National Library of Canada to reproduce, loan, distribute or seIl copies of this thesis in microform, paper or electronic formats.

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/nlm, de reproduction sur papier ou sur format électronique.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantid extracts fkom it may be printed or otheMrise reproduced without the author's permission.

L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.

Differential Requirement for the Mitochondrial Apoptosis-Inducing Factor in Apoptotic Pathways degree of Master of Science 2001 Nicholas Joza Department of Immunology University of Toronto

ABSTRACT

A family of cysteine proteases (caspases) form the core machinery of

programmed cell death (PCD) in mammals. However, in certain contexts, cell death proceeds normally in the absence of caspase function, suggesting that there exist caspase-independent pathways of PCD. Recently, a novel death effector called apoptosis-inducing factor (AIF) was cloned and shown in cell lines to induce hallmarks of PCD independent of caspases. To explore the physiological role of AIF, I disrupted the mouse aif gene in embryonic stem cells using gene targeting. I show that AIF is required for cell death in response to serum withdrawal, but not to DNA damage or inhibition of kinases. Moreover, in response to oxidative stress, AIF and caspases function redundantly to induce cell death. These results provide the first genetic evidence for a caspaseindependent pathway of PCD. Furthermore, they support a model whereby distinct death stimuli trigger distinct effector pathways of cell death.

TABLE OF CONTENTS Abst ract

1.

Table of Contents

ii.

List of Abbreviations

iii.

List of Figures

iv.

1.

Introduction Caspases and the cell death program Caspase-independent pathways of programmed cell death Apoptosis-lnducing Factor: a caspase-independent cell death effector

II.

Materialaethods Screening of rnouse genomic library using mAlF cDNA as probe Mapping the mAlF genomic clones using restriction enzymes Constructing the targeting vector Screening of ES cell colonies for homologous recombination Southern, Northem and Western blotting Cell death assays

111.

Results Screening of mouse genomic library using mAlF cONA as probe Mapping the mAlF genomic clones using restriction enzymes Constructing the targeting vector Screening of ES cell colonies for homologous recombination aif -IY ES cellç are sensitive to various death-inducing drügs aif

ES cells are resistant to growth factor deprivation

IV.

Discussion

v.

Conclusions

VI.

Acknowledgments

VII.

References

LIST OF ABBREVIATIONS

ES cell: embryonic stem cell

LA: long a m mAIF: mouse apoptosis-inducing factor MCS: multicloning site Neo: neomycin resistance cassette pBS: Bluescript KS vector

pEF: primary embryonic fibroblast PI: propidium iodide

s.e.m.: standard error of the mean ShA: short a m DiOCJ3): 3,3'-dihexyloxacarbocyanine iodide

LIST OF FIGURES Figure 1. Figure 2. Figure 3. Figure 4. Figure 5.

The essential role of caspases in programmed cell death. Screening of mouse genomic library using mAlF cDNA as probe. Mapping of genomic clone AIF5.1. Mapping of genomic clone AIF7.1. Construction of targeting vector and strategy for screening for gene-targeted clones.

ES cells.

Figure 6.

Generation of of aif

Figure 7.

aif

Figure 8. Figure 9. Figure 10.

Effects of caspase inhibition on ES cell death. AIF is important for cell death induced by serum withdrawal. A model for the stimulus-specific role of AIF in PCD.

ES cellç are sensitive to various death stimuli.

1.

Introduction Programmed cell death (PCD) is a physiological process whereby a cell actively

destroys itself. It is a fundamental property of al1 rnulticellular organisms and crucial to plant and animal development, organ morphogenesis, tissue homeostasis, the removal of infected or damaged cells, and repertoire selection of T and B cells (Jacobson et al., 1997; Vaux and Korsmeyer, 1999; Penninger and Kroemer, 1998). lmpaired cell death

can lead to cancer and autoimmunity, while dysregulated, excessive cell death may precipitate neurodegenerative diseases or immunodeficiencies (Thompson, 1995). A variety of intracellular and extracellular signals can trigger cell death, including

engagement of TNF or Fas receptors, ultraviolet (UV) radiation, anticancer drugs, growth factor deprivation, and overexpression of certain tumour suppressor genes. Despite the diversity of these death triggers, the machinery responsible for executing PCD is remarkably uniform. PCD involves the activation of an evolutionarily conserved family of cysteine proteases called caspases, which are able to carry out the ordered dismantling of the cell (see below) (Nicholson and Thomberry, 1997). PCD most often proceeds through a series of distinctive changes in the morphology

and

biochemistry

of

the

cel!,

collectively

termed

apoptosis.

Morphologically, the dying cell's cytoplasm and nucleus shrink, the nuclear chromatin condenses, the plasma membrane "blebsn, while the organellar ultrastructure is presenred and the plasma membrane remains intact (Kerr et al., 1972). These features are preceded by collapse of the mitochondrial membrane potential and the consequent release of apoptotic effector molecules, the cleavage of DNA into oligomers of

-

200

bp, the exposure of phosphatidyl serine residues on the outer leaflet of the plasma

membrane, and the cleavage of structural proteins (Kroemer and Reed, 2000). While the significance of al1 these changes is not altogether clear, it is believed that they partly facilitate phagocytic removal of dying cells before they can spill their cellular contents into the intercellular space (Green and Reed, 1998). Apoptosis thus contrasts with necrosis which arises from acute injury and leads to an inflammatory response due to premature rupture of the plasma, nuclear and organellar membranes. Rather, apoptosis amounts to an ordered, rapid and relatively inconspicuous destruction of the cefl.

Caspases and the cell death program The existence of a genetically-controlled suicida1 program that underlies the apoptotic morphology was first elucidated through genetic studies in Caenorhabditis elegans. In these studies, Horvitz and colleagues demonstrated that the physiological death of al1 131 cells of the 1090 somatic cells generated during the nernatodes's development was controlled by several genes, which they termed ced-3, ced-4 and ced-9 (Ellis and Horvitz, 1986; Horvitz et al., 1983). Orthologs of these genes in

vertebrates were subsequently discovered. Ced-3, which is reguired for apoptosis to occur in the worm, is homologous to the proapoptotic mammalian cysteine proteases known as the caspases. At least 13 members of the caspase family have now been

identified in mammals, and many of thern have been shown to play important roles in cell death (Li and Yuan, 1999). Synthesized as inactive zymogens, when activated through trans- or auto-proteolytic processing, they cleave select intracellular substrates, thereby disabling important homeostatic and repair processes (eg.

inactivation of DNA repair enzyme poly(ADP)ribose polymerase (PARP), dismantling structural components (nuclear lamins, the actin-regulatory protein gelsolin, the cytoskeletal protein fodrin) and activating latent death effecton (PAK 2, PKC

8),

culminating in the morphological changes of apoptosis (Nicholson and Thombeny, 1997; Thomberry, 1998). Caspases also serve to eliminate death antagonists: cleavage of ICAD (inhibitor of caspase activated DNAse), an inhibitor of a latent endonuclease, results in its dissociation from the DNAse CAD, allowing CAD to mediate fragmentation of nuclear DNA into oligomers of = 200 bp (Enari et al., 1998). Other components of the cell death machinery are conserved between nematodes and mammals. CED4, which functions in activating CED-3, is homologous to mammalian apaf-1, a key caspase activator in the mitochondrial PCD pathway (see below)(Li et al., 1997; Zou et al., 1997). The anti-apoptotic molecule CED-9 functions by inhibiting CED-4-induced activation of CED-3(Hengartner and Hoivitz, 1994). It is homologous to the rnammalian Bcl-2 gene family, of which some mernbers (Bcl-2, BelXL) inhibit apoptosis and others (Bax, Bid, Bad) stimulate cell death (reviewed in Kroemer, 1997). In vertebrates, two principal pathways of cell death have been delineated (Figure 1) (Green and Reed, 1998). In the "extrinsicnpathway, Fas ligand, TNF-a and other extracellular "death factorswbind to their cognate cell surface receptors, resulting in the recruitrnent, oligomerization, and activation of proximal caspases -8 and -10. This event triggen a caspase activation cascade culminating in the activation of the effectors caspase-3, -6, and -7 which can directly mediate the cell's destruction.

Figure 1. The essential role of caspases in programmed cell death. Apoptosis is initiated within a cell by either of two ways. In the "extrinsicnpathway, Fas ligand, TNF-a and other extracellular "death factorsn bind to their cognate cell surface receptors, resulting in the recruitment, oligomerization, and activation of proximal caspases -8 or -10. This event triggen a caspase activation cascade culminating in the activation of the effectors caspase-3, -6, and -7. These caspases directly mediate cellular destruction by disabling homeostatic and repair processes, dismantling structural components and activating latent death effectors. In the "intrinsic" or mitochondrial-dependentpathway, death signals induce opening of the permeability transition pores (PTPs), resulting in dissipation of the mitochondrial membrane potential and release of death effectors such as cytochrome c and AIF into the cytosol. Cytochrome c induces a caspase activation cascade, which culminates in activation of caspase effecton. In sorne cell types, crosstalk exists between the two pathways.

Another pathway, the "intrinsic" or mitochondrial pathway, responds to a distinct set of death signals, among them DNA damage and metabolic disturbances (Rathmell et al., 2000; Vander Heiden et al., 1999). The irreversible commitment of the cell to the death program is coincident with the dissipation of the inner mitochondrial membrane potential (Kroemer and Reed, 2000). This collapse of the mitochondrial potential reflects the opening of specialized "permeability transition pores" (PTPs), large multicomponent conductance channels spanning the inner and outer membranes. Composed of the ADPIATP transporter (ANT), the voltage-gated anion channel (VDAC), cyclophilin D and other molecules, PTPs respond to oxidative stress, calcium elevations, and ATP depletion (Kroemer and Reed, 2000). PTPs are also the major site of action of the BcWBax family molecules: Bax can cooperate with either VDAC (Shimizu et al., 1999) or the ANT (Marzo et al., 1998) to induce pore formation. PTP opening triggers pemeabilization of the outer mitochondrial membrane and release into the cytosol of pro-apoptotic factors such as cytochrome c (Kluck et al., 1997; Yang et al., 1997) and apoptosis-inducing factor (AIF) (Susin et al., 1999) which

normally reside within the intermembrane space of mitochondria. Cytochrorne c subsequently induces a conformational change in apaf-1 to f o m the "apoptosome", a multimeric caspase activation complex composed of apaf-1 and caspase-9 (Li et al., 1997; Liu et al., 1996). The apoptosome triggers a caspase activation cascade culminating in the activation of effector caspases. Thus, in both the extrinsic and the rnitochondrion-dependent, intrinsic pathways, the caspase family plays prominent roles both in initiating and executing the cell's demise.

The crucial role of some of these death molecules has been established largely through analysis of knock-out mice. Caspase-9 " and caspased" mice display ectopic masses of post-rnitotic neurons on the head due to impaired cell death of neural progenitors, and these mice die perinatally due to mechanical disruption and hemorrhaging of their large brain masses during birth (Hakem et al., 1998; Kuida et al., 1998; Kuida et al., 1996; Woo et al., 1998). Apaf-1"

mice have a more severely

affected central nervous systern, and exhibit retinal hyperplasia and craniofacial abnormalities due to impaired apoptosis of the retinal and palatal shelves during shelve fusion, respectively (Cecconi et al., 1998; Yoshida et al., 1998). The more severe phenotype of apaf-1-deficient mice suggest that apaf-1 may activate apical caspases other than caspase-9. Various cell types such as embryonic fibroblasts (EFs) and embryonic stem (ES) cells derived from these knockout mice are resistant to a range of death stimuli including the DNA-damaging agents etoposide and UV radiation and the protein kinase inhibitor staurosporine. However, these cells remain sensitive to death following ligation of the Fas and TNF receptors, consistent with the existence of independent death signalling pathways converging on common downstream caspases. In contrast, EF cells from caspase-8-deficient mice are resistant to apoptosis mediated by TNF-a receptors and Fas, but are sensitive to genotoxic dnigs (Varfolorneev et al.. 1998). Caspase-8 -'- mice display abnormal heart development, but this defect does not seem to be due to impaired cell death, indicating that caspase-8 may also have a non-apoptotic developmental function (Li and Yuan, 1999). Caspase-2 " mice have excess oocytes at birth due to impaired cell death, and apoptosis mediated by

granzyme €3 and peiforin are defective in caspase-zc 5 lymphoblasts (Bergeron et al., 1998).

Caspase-independent pathways of programmed cell death

While these knockout mice cleariy established that some caspases and their direct activators were crucial to some cell death pathways, it was orginally surprising that the knockout phenotypes were restricted merely to the CNS. Moreover, even the severe CNS phenotypes were variable, as evidenced by the fact that in some genetic backgrounds 5% of apaf-1

"

rnice survive to adulthood and exhibit no overt

abnormalities (Honarpour et al., 2000). Furthermore, cell lines derived from these mice are not unifomly resistant to death stimuli given, but rather undergo PCD in a manner specific to both cell type- and death signal (Amarante-Mendes et al., 1998; Cecconi et al., 1998; Hakem et al., 1998; Hirsch et al., 1997; Kuida et al., 1998; Li et al., 2000; Woo et al., 1998; Xiang et al., 1996; Yoshida et al., 1998). For example, c a s p a ~ e - 9 ~ -

ES cells are resistant to y and UV radiation, whereas thyrnocytes from the same knockout mice are resistant to y radiation but sensitive to UV radiation (Hakem et al., 1998; Kuida et al., 1998). In addition, cell lines treated with the broad-spectrum caspase inhibitor z-VAD.fmk remain sensitive to cell death in various instances. For example, enforced Bax expression in Jurkat cells treated with z-VAD.fmk only blocked DNA fragmentation, while chromatin condensation, membrane blebbing, and collapse of the mitochondrial potential were unaffected (Xiang et al., 1996). Finally, apaf-1"

ES

cells die when exposed to DNA-damaging dmgs or senim withdrawal, albeit with delayed kinetics (Haraguchi et al., 2000). Overexpression of Bcl-2, however,

completely blocks this cell death, suggesting that Bcl-2 controls another mitochondrial cell death pathway distinct from that requiring apaf-1 (Haraguchi et al., 2000). To explain these observations, it has been suggested that caspases are functionally redundant, capable of compensating for each other, andor caspaseindependent PCD pathways exist. The former possibilities, however, seems less likely since no functional apaf-1 homologues have been discovered nor are broad-spectrum caspase inhibiton able to block al1 cell death. In light of this evidence, there is intense interest in detennining the molecules responsible for triggering caspase-independent cell death.

Apoptosis-lnducing Factor: a caspase-independent cell death effector Recently, a novel, caspase-independent cell death effector was discovered, tenned apoptosis-inducing factor (AIF) (Susin et al., 1999). AIF is a 56 kD ubiquitouslyexpressed protein confined, in healthy cells, to the intermembrane space of mitochondria. It possesses oxidoreductase function, yet this activity can be uncoupled from its role in cell death, as shown in experiments which genetically rnutate the FADbinding sites (Loeffler et al., 2001). In response to various death stimuli, AIF translocates from mitochondria to nuclei, where it induces nuclear features of apoptosis, such as chromatin condensation and large-scale DNA cfeavage into 50 kb fragments (Daugas et al., 2000; Susin et al., 1999). Overexpressed or microinjected AIF is sufficient for other hallmarks of apoptosis, including permeabilization of mitochondrial membranes and exposure of phosphatidyl serine residues on the surface of the plasma membrane (Susin et al., 1999). These events cannot be

prevented in the presence of z-VAD.frnk, suggesting that AIF is a caspaseindependent death effector. In addition, its ability to stimulate mitochondrial membrane permeabilization leading to the release of cytochrome c and formation of the apoptosome, indicates that there exists potential cross-talk between the AIF and caspase death pathways. Recent evidence argues that, depending on the cell systern and death stimulus, AIF functions either upstream of and/or in parallel to the cytochrome c/apaf-lkaspase9 apoptosome. Thus, apoptosis of syncytia formed by the fusion of HeLa cells co-

expressing the HIV env surface protein and CD4lCXCR4 receptors is AIF-dependent, since neutralization of AIF function using AIF-specific antibodies prevents cytochrome c release and cell death (Ferri et al., 2000). In contrast, in mouse embryonic fibroblasts, AIF neutralization abolishes nuclear apoptosis only when the apoptosome pathway is simultaneously blocked by either genetic inactivation of apaf-1 or by addition of the pan-caspase inhibitor z-VAD.fmk (Susin et al.. 2000). These data indicate that AIF is required for cell death in certain systems but plays a functionally redundant role with caspases in others. Despite the accumulated in vitro evidence for a role of AIF in cell death, the physiological function of AIF remains elusive. Is AIF essential to cell death in response to certain stimuli, or does it play a functionally redundant role with caspases? Altematively, does AIF determine certain aspects of a dying cell's rnorphology, such as the pattern of nuclear chrornatin condensation? To explore these questions, I disrupted the mouse aif gene in mouse embryonic stem (ES) cells using gene targeting technology. Disruption of the aif gene resulted in nuIl AIF expression. aif -IY ES cells

exhibited resistance to apoptosis which depended on the type of death stimulus and whether the caspase pathway was simultaneously blocked. AIF is required for cell death induced by growth factor deprivation, plays a functionally redundant role with caspases in response to oxidative stress, but is not essential for other death stimuli including DNA damage. These data identify a second, physiologically relevant mitochondrial pathway of cell death distinct from caspases.

II.

Materials/Methods

Screening of mouse genomic library using mAlF cDNA as probe

A 50 ml confluent culture of LE392 cells was resuspended in 20 ml 0.01 M MgSO,.

22 tubes, each containing 300 pL of this bacterial resuspension, were each 5

infected with 7.5 x 10 pfu of phage library (giR of J Rossant, Toronto), followed by incubation at 37OC for 30 min (Figure 2). The mixtures were each mixed with 7 ml top agarose and poured ont0 22 x 15 cm LB bacterial dishes. After ovemight incubation at 37OC, nitrocellulose filters (Millipore) were applied to the plates, removed and floated for 5 min each in denaturation solution (1.5 M NaCI, 0.5 N NaOH), neutralization solution (1.5 M NaCI, 1 M Tris, pH 7.4), and finally briefly washed in 2X SSC. DNA was UV cross-linked (Stratalinker 2400, Stratagene) to the filten. Next, the filters were

incubated at 60°C ovemight with hybridization solution (5x Denhardt's solution, 6x SSC, 0.5% SDS) containing 5x10~c p d m l of denatured rnAlF probe (boiled 5 min, quick cool on ice) and 100 pg/ml of denatured salmon sperm DNA. The probe was

randorn hexamer labeled with

(32~1-CX-~CTP(Arnenharn). After 3 x 30 min washes in

0.15X SSC, 0.5% SDS, the filters were exposed to X-ray film for 5 days at 4 O C . "Positive" signals were picked with a pasteur pipette, dissolved in 0.5 ml SM buffer (50

mM Tris HCI pH8, 0.58% NaCI, 0.2% MgSO,

-

7H,O,

0.01% gelatin) and 25 pl

chlorofom, and centrifuged at 10,000g for 5 min. Serial dilutions of phage supematant were used to infect 100 pL of LE392 ovemight bacterial culture (prepared as above), and the mixture was plated on 10 cm LB bacterial dishes. The entire phage supematant from one "positive" plaque was then mixed with 300 pl LE392 cells and grown on 15 cm d i s h ~ sovemight. Phage were collected by applying 12 ml SM buffer and 1 ml chlorofom ont0 the LB and shaking on a rotating platform at RT for 3 h. The supematant was collected, and the genomic DNA recovered using the Qiagen Lambda MaxiPrep (Qiagen) according to the manufacturer's instructions.

Mapping the mAlF genomic clones using restriction enzymes Knowledge of the intron-exon boundaries of the mAlF locus was obtained from Loeffler et al, 2001. The oligonucleotides used for mapping were 5'

- CTG GCG GGT

GCT ï T C AAG CAG AAA C - 3' (nucleotides 19-43 of mAlF coding region, exon l), 5'

- CGT GlT CCT CTA GAA CTC CAG ATG - 3' (124-147, exon 2), 5' - TGG GAT TAG GAC TOT CCC CAG AAG AG - 3 (296-321, exon 3), 5' - CTC AGT TCC TCA exon 4). Oligonucleotides were 5' labeled with GAT GAG GGC ACC AA - 3' (360-385, [ 3 2 ~ ] --dATP y using T4 polynuclaotide kinase (Amersham), and hybridized to southern

blots of electrophoresed DNA digests. Hybridization was performed as described for library screening (above), except that hybridization was done at 55OC, and a phosphorimager (Molecular Probes) was used ta expose the blots. A 1 kb DNA ladder was used (Roche).

Constructing the targeting vector The PGK-neo vector (gift of T. Mak, Amgen) was constructed by inserting the PGKneomycin resistance cassette into the EcoRV and Sma I sites of pBS KS and destroying these sites.

Screening of ES cell colonies for homologous recombination E14K ES cells (isogenic with the 12915 phage genomic library used) were maintained on a layer of mitomycin C-treated primary embryonic fibroblasts (pEFs) (treated for 2 h with 10 pg/ml mitomycin C). ES cell media consisted of Dulbecco's modified Eagle culture medium (DMEM), supplemented with antibiotics, leukemia inhibitory factor (LIF), 15% fetal calf serum (FCS), L-glutamine, sodium pyruvate, and P-mercaptoethanol. 6

20 pg of Xho 1-linearized targeting vector was electroporated into 5x10 ES cells

in a volume of 0.8 ml, using the Biorad Gene Pulser (0.34 kV, 0.25 mF). Cetls were 6

plated on gelatin-coated 10 cm plates at 2.5 x 10 cells per plate. Gelatin-coated plates were prepared by coating with an autoclaved 1% gelatin solution dissolved in water, aspirating off excess gelatin, and allowing the plates to dry for at least 30 min. 24 h

af&ertransfection, 300 pg/ml of 0418 sulfate (Gibco) was added to the cells. Media was changed daily. After 9 days of selection, ES cell colonies were picked and screened for homologous integrants using PCR and southem blotting on ES cell genomic DNA. The primers used for screening for homologous recombination were: neornycin resistance casette-specific sense primer: 5'- GGG ATT AGA TAA ATG CCT GCT CTT

-3'and flanking region-specific antisense primer: 5'- CCC CCA AAC TTA TAT CAG CCT ACC TTC -3'.

Southern, Northern and Western blotting For Southem blotting, 15 pg of Hindlll-digested ES cell genomic DNA was electrophoresed at 30 V overnight on a 1% agarose gel, followed by capillary transfer to a nylon membrane (Hybond) (Amersham). The membrane was UV cross-linked, and hybridized with a random hexarner [ 3 2 ~ ] - a - d ~ ~ ~ - f a b e(Amersharn) lled probe corresponding to the region flanking the short a m (indicated in figure 5B). The 500 bp probe was generated by PCR using the sense primer 5' TTT GG

- GTA GGC TGA TAT AAG

- 3'and antisense primer 5' - AAG TTC AGT TGC ACC TT - 3' (indicated in

figure 5B). For Northem blotting, total RNA from ES cells was extracted using Tnzol (Gibco-BRL). 20 pg of RNA was separated on a 1% agarose gel and transferred to nylon membranes. The membranes were hybridized to a [ 3 2 ~ ] - a - d ~ ~ ~ - l a bfullelled

length mAlF cDNA probe. A human P-actin probe (Clontech) was used as loading control. For Western blotting, ES cells were lysed in 50 mM Tris-HCI (pH8), 1% TritonX100, 20 mM EDTA supplemented with protease inhibitor cocktail (Boehringer Mannheim, 183 6170). The lysate was centrifuged at 10,000g for 20 min. 25 pg of supernatant was separated by SDSIPAGE and transferred to poly(viny1idene difluoride) membranes (Roche, 3 010 040). The membranes were blocked with 2.5%

milk in PBS, followed by 2 hour incubation with anti-AIF polyclonal rabit senim (gift of

G. Kroemer, CNRS, France) at 1:2000 dilution in 2.5 % BSA. The anti-semm was generated against a mixture of 3 peptides corresponding to amino acids 151-170, 166185, 181-200 of mouse AIF, coupled to keyhole limpet haemocyanin. A horseradish peroxidise-conjugated anti-rabbit IgG (Amersham, NA 934) was used as secondary antibody at 1:5000 dilution in 1% BSA, and immunoblots were visualized by ECL (Amersham, RPN 2106). Anti-actin (Sigma, A-2066) was used as loading control.

Cell death assays To assay PCD of ES cells, 1x1 o5cells were plated in each well of gelatin-coated 24-well plates for one day prior to treatment. The cells were left untreated or treated with anisornycin (50 PM), staurosporine (2, 5 or 10 PM), menadione (150 PM) (al1 from Sigma), or ultraviolet radiation (120 m~lcrn* ) (Stratalinker 2400, Stratagene). In some experiments, cells were pre-treated with 100 pM of the broad-spectnim caspase inhibitor z-VAD.fmk (Sigma, V-116) for 1 h pnor to drug treatment. At various time points, cell viability was determined by collecting adherent cells by trypsinization and

pooling with floating cells, followed by staining with Annexin V and propidium iodide (PI) using the Apoptosis Detection Kit (R&D systems, TA 4638) according to manufacturer's instructions. Altematively, cells were incubated with DiOC6(3) (20 nM in PBS final concentration) (Molecular Probes) and 5 p g h l PI for 20 min at at 37°C. Staining was analyzed immediately by flow cytometry using a FACScaliburGB (Becton Dickinson). Annexin V and DiOC,(3) staining was detected on the FL1 channel, and PI staining on the FL3 channel. Data were analyzed using CeIlQuest software (Becton Dickinson).

III.

Results

Screening of mouse genomic library using mAlF cDNA as probe The mouse AIF (rnAIF) cDNA was previously cloned by B. Snow at the Amgen Institute, Toronto from a retinal cell cDNA library into pCR2.1 vector (Stratagene) (Susin et al., 1999). Nucleotides 1-512 of the coding region, generated by digestion of cDNA with EcoRl and BamHI, were used as probe to screen a 129J mouse genomic h phage library (gift of J Rossant, Toronto). This library was constructed by cloning 129J mouse DNA, partially digested with SauM, into the Barn HI sites of the Lambda DASH IIvector (Stratagene) (figure 3 A).

Screening of the mouse genomic library to obtain mAlF genomic clones is schematically illustrated in figure 2. LE392 bacterial cultures were infected with the  phage genornic library and plated on LB bacterial dishes. After ovemight incubation at

Figure 2. Screening of mouse genomic library using mAlF cDNA as probe LE392 bactena were infected with a mouse genomic phage library, and plated ont0 LB plates. After incubation ovemight at 37OC, the plates were confluent with phage plaques. Filters were applied to the plates, phage DNA was denatured and irrevenibly bound to the filten by UV cross-linking, followed by hybridization of filters to a radiolabelled mAlF cDNA probe, and exposure of filters to X-ray film. "Positive" phage plaques were picked with a pasteur pipette, dissolved in buffer, and serial dilutions were used to infect fresh bacteria. The process was repeated until a homogenous population of well-spaced "positiven plaques was obtained. Phage DNA was then purified.

Phage genomic llbrary

LE392 bacteria

Apply filters -Denature and crosslink DNA -Hybridim with 32Plabelled mAlF cDNA probe -Visualize signal on X-ray film

Colncubation at 3P C

Giow phage ovemight at 3P C

serial dilutions

r Purify phage DNA

37OC, the plates were confluent with phage plaques. Nitrocellulose filters were applied

to the plates, followed by sequential soakings in denaturation solution, neutralization solution, and 2X SSC. These steps resulted in the recovery of single-stranded phagederived genomic DNA altached to the filters. Next, the filters were hybridized with denatured,

32

P-labelled mAlF probe, washed, and exposed to X-ray film. The presence

of positive signals on the film indicated the presence of mAlF-specific genomic clones. "Positive" phage plaques were picked, and serial dilutions of phage were used to infect LE392 bacteria and plated. The process of making filter duplicates of LB plates,

probing the filters with radioactively-labelled probe, and isolating phage plaques was repeated until a homogenous population of well-separated "positive" plaques were obtained on a plate. Next, the phage DNA from one plaque was isolated using the Lambda Maxi Kit (Qiagen) according to the manufacturer's instructions. Two distinct genomic clones comprking exons 1 and 3 of the mAlF genomic locus were obtained (see below).

Mapping the mAlF genomic clones using restriction enzymes The genomic clones were digested with Not 1, and the DNA was purified and cloned into Not 1-digested pBluescript KS II vector (pBS) (figure 3 and 4, A and 6). Using

32

P-labelled oligonucleotides corresponding to exons 1, 2, 3 and 4, it was

detennined that two distinct genomic clones had been obtained. These clones were denoted "AIF5.1" and "AlF7.1". Only the exon 1-specific probe hybridized to AIF5.1 (figure 3C), while only an exon 3-specific probe hybridized with AIF7.1 (figure 4C),

Figure 3. Mapping of genomic clone AIF5.1 (A) Diagram of AIF5.1 phage DNA, comprised of 20 kb and 9 kb phage amis and a 17 kb rnouse genomic ONA fragment. The mouse DNA fragment was excised with Not 1,

and cloned into the Not I site of pBS. (B) Partial restriction map of AIF5.1. Exon 1 is indicated in red, pBS vector in blue. (C) Restriction digests and corresponding southem blots of AIF5.1. Abbreviations: El: EcoRI, EV: EcoRV, P: Pst 1, C: Cla 1, H: Hindlll, N: Not 1, MCS: pBS multicloning site.

Z W

20 kb

W Z

AIF5.1

11

I

= 17 kb senomic fragment

11

gkb

1 MCS o P 2 s o

Probed with exon 1

Probed with pBS

El: 5 kb, 4.5 kb, 2.9 kb(pBS), 2.8 kb, 2.1 kb(ex1)

EV: 15 kb(exl)(p8S),5.1 kb P: 11 kb(ex 1), 6 kb(pBS), 1.4 kb, 1.1 kb

EV+P: 7 kb(ex1), 6 kb(pBS), 2.8 kb, 1.4 kb, 1.1 kb C: 14 kb(ex1), 6 kb(pBS) H: 12 kb(ex1), 5 kb(pBS), 1.6 kb, 1.5 kb C+H: 12 kb (exl), 5 kb (pBS), 1.6 kb, 1 kb, 0.4 kb N:17 kb(ex1), 2.9 kb(pBS)

suggesting that AIF5.1 harboured exon 1, and AIF7.1 harboured exon 3. 60th clones were digested with various restriction enzymes and southem blotting and hybridization was perforrned using

32

P-labelled pBS or exon-specific oligos as probes in order to

obtain restriction enzyme maps of the two clones. The methodology is described below. For AfF5.1 (figure 38 and C), a Not I digest produces the 17 kb genomic fragment and the 2.9 kb pBS vector. The genomic DNA fragment contains exon 1 since it hybridizes to an exon 1-specific probe, while the pBS fragment hybridizes to the pBS probe. An EcoRV digest generates 2 fragments, a 15 kb fragment consisting of the 2.9 kb pBS and 12 kb of exon 1-containing genomic DNA (2.9

+ 12 = 15 kb),

and a 5.1 kb fragment consisting of genomic DNA only. Hence, the genomic clone contains an intemal EcoRV site which is 12 kb from the left pBS-genomic DNA boundary and to the right relative to exon 1 (figure 38).Digestion with two restriction enzymes aided in the positioning of restriction sites relative to each other. In addition, partial sequencing was used to both confirm restriction sites and to determine the relative 5'-

3' orientation of the genomic clone.

For AIF7.1, a Not I digest liberated a 14 kb genomic fragment containing exon 3 from the 2.9 kb pBS vector (figure 4C). Digestion of the AlF7.l clone with BamHl generated 4 fragments: an 8 kb genomic DNA fragment, a 4.3 lcb exon 3-containing fragment, a 4.2 kb fragment consisting of the 2.9 kb pBS and 1.3 kb of genomic DNA (2.9

+ 1.3 = 4.2

kb), and a 0.65 kb genomic DNA fragment. Hence, AlF7.l contains 3

intemal BamHl sites, as shown.

Figure 4. Mapping of genomic clone AIR.1 (A) Diagram of AIF7.1 phage DNA, comprised of 20 kb and 9 kb phage amis and

a 14 kb mouse genomic DNA fragment. The mouse DNA fragment was excised with Not 1, and cloned into the Not I site of pBS. (B) Partial restriction map of AIF7.1. Exon 3 is indicated in red, pBS vector in blue. (C) Restriction digests and corresponding southem blots of AIF7.1. Abbreviations: B: Barn HI, El: EcoRI, H: Hindlll, N: Not 1, P: Pst 1, MCS: pBS multicloning site

mapping information, not al1 Pst Isites are shown.

* Due to lack of

20 kb

z w

11

1

-

wz

AlFi.1

11

14 kb genomlc fragment

9kb

J

MCS

3' x

I

Probed with exon 3

mam

11 1

r

I

III Er

m xizz

II

m l3rev

+

Probed with pBS

B: 8 kb, 4.3 kb(ex3),4.2 kb(pBS), 0.65 kb El: 11.5 kb, 2.9 kb(pBS), 2.5 kb(ex3),0.45 kb B+El: 7 kb, 3.4 kb, 2.9 kb(pBS), 1.3 kb, 1 kb(ex3),0.65 kb, 0.45 kb H: 6.5 kb, 3 kb, 2.9 kb(pBS), 2.6 kb, 2.1 kb(ex3) N: 14 kb(ex3), 3 kb(pBS)

* P: 8.5 kb(ex3)(pBS),3 kb, 1.9 kb, 1.2 kb, 0.8 kb

Constructing the targeting vector AIF7.1 was used to construct a targeting vector designed to replace exon 3 with the PGK-neomycin resistance cassette (neo). Deletion of exon 3 should result in no functional protein being generated since exon 3 is the first exon of the mature AIF protein; that is, it is the first exon after removal of the amino-terminal mitochondrial import sequence.

Two areas of homology, designated as the "long ami" (LA) and the "short amn (ShA), were cloned into either side of neo in the PGK-neo vector (figure 5A). The LA

was generated by first cloning the 6.5 kb Hindlll fragment of AlF7.l into pBS KS, such that the fragment was flanked by Not f on either side, followed by ligation of the 6.5 kb fragment into the Not I site of the PGK-neo vector (figure 5A). The 0.6 kb ShA was generated by PCR using AIF7.1 as template and the primers 5' GCA CAA CAG GGA GCA

- GGA TCG ATT ACC

- 3'and 5' - AAC TCG AGC TAG CAT GGC TGG CCT

CTG A - 3'. containing Cla I and Xho I site overhangs, respectively. The PCR product

was directionally cloned into the Cla I and Xho I sites of the PGK-neo vector (figure 5A). In addition, the targeting vector is designed to introduce a Hindlll site (astenx in

figure 5A) into the mAlF locus such that gene-targeted clones can be confirrned using southem blotting on Hindlll-digested DNA (see below).

Screening of ES cell colonies for homologous recombination E14K ES cells (isogenic with the 129J mouse strain) were transfected with targeting vector, and clones beanng neo were selected for using G418-containing medium. After 9 days of selection, ES cefl colonies were screened for potential

Figure 5. Construction of targeting vector and strategy for screening for genetargeted clones. (A) Construction of a targeting vector designed to delete exon 3. A 6.5 kb LA was

cloned into the Not I site, and a 0.6 kb ShA was cloned into Cla I and Xho I sites, of a

pBS vector containing a PGK-neomycin resistance cassette. (B) The areas of homology between the targeting vector and the endogenous locus are shown. Homologous recombination results in the introduction of an Hindlll site and the predicted appearance of a 1.8 kb band on a southem blot of Hindlll-digested ES cell DNA hybridized with the indicated flanking probe. A 2.1 kb band appears in parental ES cell DNA. In blue is shown the 1.5 kb template used for optimization of the PCR

screening conditions. In red is shown the PCR primers used for screening. a sense primer within neo and an antisense primer outside the targeted region.

homologous integrants using PCR. The screening strategy is based on the notion that homologous recombination results in the juxtaposition of neo with the mAlF endogenous locus; only in this situation, and not in a random integration event, will

PCR amplification occur when primers specific for neo and the endogenous locus are used.

During screening, PCR is performed on pools of 16 colonies to identify "positive" pools, followed by PCR on each colony to identify the "positive" colony within a "positive" pool. This means that the amount of "positiven DNA is substantially underrepresented in the first pooling step, thus requinng that PCR be optimized. For this purpose, a 1.5 kb DNA ternplate which reproduces the genomic organization of a homologous recombination event was generated. The template consists of the 3' end of neo, the ShA, and 500 bp of the region flanking the ShA, and was generated using the primers indicated in figure 58 and a mixture of the targeting vector and AIF7.1 DNA as templates. Using this 1.5 kb template, a series of primers was tested for the ability to arnptify 100 copies of ternplate in the presence of =1 ug of (nonspecific) E14K ES cell DNA. A set of primers (indicated in figure 56) was found to satisfy these conditions, and was used in the PCR screening strategy.

ES cell clones positive by PCR were confimed by southem blotting on Hindllldigested ES cell DNA, followed by hybridization to a 3' probe flanking the targeted region. A Hindlll digest was performed because homologous recombination results in introduction of an Hindlll site from the targeting vector into the mAlF locus (figure 5B). Targeting of the mAlF locus was confirmed in 3 of 4500 G418-resistant colonies

Figure 6. Generation of sH -IY ES cells. (A) Southem blot on Hindlll-digested ES cell DNA, hybridized with a 3' flanking probe

(indicated in figure 58).The expected 1.8 kb mutant band in aif

ES cells is show".

(6) Northern blot on total ES cell RNA, hybridized with full-length mAlF cDNA. No

transcript is detectable in aif

ES cellç. The blot was repmbed with p-actin as a

loading coritrol. (C) Western blot on ES cell lysate, probed with anti-AIF antibodies. No AIF protein is detectable in aif -IY ES cells. The blots were stripped and reprobed with anti-actin antibodies as a loading control. The same blots were stripped and reprobed with anti-Apaf1 antibody. Equivalent levels of Apafl protein are detected in aif aif

-jY

ES ce~~s.

+IY and

screened, as determined by the presence of a 1.8 kb Hindlll fragment (figure 5 8 and 6A). A 2.1 kb Hindlll band is seen in parental ES cell DNA. To control for ES cell changes during selection, several colonies with randomlyintegrated neo (denoted aif

) were obtained. Since aif is located on the X

chromosome in mice (Susin et al., 1999), targeting of aif in XY ES cells results in nuIl AIF expression, as determined by northern and western blotting on total ES cell RNA and protein lysates, respectively (figure 6 B and C).

aif

ES cells are sensitive to various death-inducing drugs Since targeting of aif results in nuIl AIF expression, the obtained aif

ES cell

clones could be analyzed for their response to death inducen. AIF is reported to be released from mitochondria in response to a wide variety of death stimuli including the kinase inhibitor staurosporine, DNA-damage induced by ultraviolet (UV) radiation, and inhibition of the translational machinery by anisomycin (Susin et al.. 1999). To test whether AIF is crucial to cell death in response to these death inducers, the suivival of aif -PI ES cellç after treatment with these drugs was rnonitored. Several parameters of cell death were assessed. Cells committed to the apoptotic program simultaneouly undergo collapse of the mitochondrial inner membrane potential, consequently releasing apoptotic effectors such as cytochrome c and AIF into the cytosol. This mitochondrial depolarization can be analyzed by incubating cells with the cationic lipophilic fluorochrome 3,3'-dihexyloxacarbocyanine iodide (DiOC6(3)). DiOC, (3) only

enters cells and distributes to the inner side of the inner mitochondrial membrane in mitochondria with intact membrane potentials (Penninger and Kroemer, 1998). Subsequently to permeabilization of mitochondnal membranes is a translocation of phosphatidyl senne residues from the inner to the outer leaflet of the plasma membrane. Exposed phosphatidyl serines can be detected by staining with annexin V conjugated to FITC. A relatively late event in apoptosis is the permeabilization of the plasma membrane, detectable by uptake of propidium iodide (PI) into the cell (Penninger and Kroemer, 1998). At various times after treatrnent, aif +/Y and aif

ES cells were harvested

and stained with FlTC-conjugated annexin V and propidium iodide (PI) (figure 7 and 8A). Altematively, cells were stained with the fluorochrome DiOC,(3)

and PI (figures

8B and 9). Staining was followed by flow cytornetric analysis. Untreated aif +IY and aif

ES cells were largely Annexin V- PIanisornycin or staurosporine, aif

(figure 7). In response to UV radiation,

and aif

ES cells displayed similar

+ +

percentages of Annexin V PI cells (figure 7A), indicating that AIF is not essential for death in response to these stimuli. The sensitivity of aif -lY ES cells to theçe death stimuli rnay indicate functionally redundant roles of AIF and caspases in cell death. To investigate this possibility, cells were incubated with the pan-caspase inhibitor z-VAD.fmk, followed by addition of the death stimulus. z-VAD.fmk was able to substantially block cell death in staurosponnetreated aif +/Y and aif -IY cells (61% vs 19% surviving aif +/"cells, figure 8A. right),

Figure 7. aif ajf

+N

-IY ES cells are sensitive to various death stimuli.

and aif

-/Y

ES cells were left untreated (UT) or treated with anisomycin

(50 PM), staurosporine (1 PM) or ultraviolet radiation (120 m~/cm*).36 h later,

cells were collected and stained with FITC-Annexin V and propidiurn iodide (PI), followed by flow cytometric analysis. Shown are dot plots of 10,000 events, ungated, representative of two independent experirnents. All three aif lines gave equivalent results. Control aif

neoN

cells subject to the same death

stimuli exhibited similar sensitivity profiles as aif percentage of cells in each quadrant.

ES cell

+/Y

ES cells. Numbers indicate

anisomycin FL 1Height

staurosporine

UV

FL 1Height

Annexin V

yet no preferential survival of aif -/Y cells cornpared to aif +/Y cells was obsewed (61% vs 67% surviving aif

and aif -lY

calls, figure 8A, "ght). These rasults suggest

that, at least in response to staurosporine, AIF cannot functionally substitute for the cell death function of caspases, and may indicate that in this setting, AIF functions downstream of caspase activation. Intriguingly, z-VAD.fmk failed to block cell death of aif +/Y cells treated with

rnenadione, a potent generator of reactive oxygen species (82 + 20% ~i0~,(3f* PI+ dead aif +IY

cellç, rnean values + s.e.rn., figure 8 B) and ait -IY

cellç treated with t-

VAD.fmk were largely protected from death due to menadione treatment (19 5 10% D~oc,(~)"

PI+ dead aif -IY cells, figure 88). These results indicate that AIF can

function in parallel with caspases to induce cell death. Moreover, this functionally redundant role of AIF is stimulus-specific.

aif

ES cells are partially resistant to growth factor deprivation It has been proposed that virtually alf cells, during development and tissue

homeostasis, continually require suwival factors from neighbouring cells andor the extracellular rnatrix to avoid PCD (Jacobson et al., 1997; Raff, 1992). Consequently, withdrawal of serum from cells in culture induces apoptosis. To determine the role of AIF in death induced by çerurn deprivation, aif

and aif

cells were depnved of

serurn for 72 h followed by staining with DiOC6(3) and PI to assess viability.

Figure 8. Effects of caspase inhibition on ES cell death. (A) aif

+N

and aif

-/Y

ES cells were left untreated (CO., left), or treated with

staurosporine (5 PM) without (middle) or with z-VAD.frnk (100 PM, added 1 h before staurosporine treatment, right). After 24 h, cells were collected and stained with FITC-Annexin V and propidium iodide (PI), followed by flow cytometric analysis. Shown are dot plots of 10,000 events, ungated. Numbers indicate percentage of cells in each quadrant.

Representative of two

experiments.

( B ) aif

+N

and aif

-/Y

ES cells were left untreated (CO., left), or treated with

menadione (150 PM) without (middle) or with z-VAD.fmk (same as in A, nght). After 24 h, cells were collected and stained with DiOC,(3)

and propidium iodide

(PI), followed by flow cytometric analysis. Numbers are mean percentage s.e.m. for two control and three aif

-/Y

+

ES cell lines corresponding to the

indicated quadrants, from three independent experiments.

Co.

Menadione

Menadione

Figure 9. AIF is important for cell death induced by serum withdrawal. aif

+N

and aif

-/Y

ES cells were cultured in ES cell media (15% FCS) (CO.),or

cultured in the absence of serurn (0% FCS) for 72 h, followed by staining with

DiOCJ3) and propidium iodide (PI), and flow cytometric analysis. Shown are dot plots of 10,000 events, ungated. Numbers are mean percentage 2 s.0.m. for two control and three aif

-/Y

ES cell lines corresponding to the indicated quadrants,

from three independent experiments.

Semm withdrawal

Interestingly, while aif +/Y cells succumbed to death, ait -IY

were largely protected

(61 + 19% vs 22 2 10% D~oc,(~)" PI+ dead aif +I" and aif

cells, respectively,

mean values + s.e.m., figure 9). These data identify an important role for AIF in death indüced by growth factor deprivation, and indicate that AIF is required for dissipation of the mitochondnal potential, and hence commitment to apoptosis, in serum-deprivationinduced cell death. Future studies will determine whether aif

cellç retain their

clonogenetic potential after growth factor deprivation.

IV.

Discussion PCD is a physiological mechanisrn to remove aged, infected, or othennrise

damaged cells. A family of cysteine proteases known as caspases are crucial effector molecules in cell death. However, not al1 cell death is blocked by genetic ablation or biochemical inhibition of caspases. Indeed, the phenotypes of ca~pase-3~, caspase9"

, and apaf-1"

mice are restricted to the CNS, whereas most other body parts

undergo normal PCD-dependent morphogenesis (Amarante-Mendes et al., 1998; Cecconi et al., 1998; Hakem et al., 1998; Kuida et al., f 998; Li et al., 2000; Woo et al., 1998; Yoshida et al., 1998). Caspase-8-'- and caspase-2"

mice also exhibit tissue-

restricted phenotypes (Bergeron et al., 1998; Varfolomeev et al., 1998). Moreover, cell lines derived from these knockout mice are not uniformly resistant to death stimuli, but instead undergo PCD in a manner specific to both cell type and death signal. These findings point to the probable existence of caspase-independent cell death pathways.

Based on overexpression analyses in cell lines, AIF was detemined to be a novel cell death effector, able to induce hallmarks of cell death independent of caspases. However, its role in vivo remained elusive. Moreover, AIF has never been finked to PCD at the genetic level. To determine the physiological role of AIF in cell death, ES cells lacking AIF were generated through gene targeting. aif -/Y ES cells exhibited selective resistance to certain death stimuli but not others, indicating the existence of stimulus-specific and both functionally redundant and nonredundant roles for AIF in cell death (figure 10).

ES cells were cornparably sensitive as aif

aif

cells to staurosporine, UV

radiation, and anisomycin (figure 7). indicating a nonessential role for AIF in cell death induced by these agents. Caspases are required for cell death in response to these inducers, as evidenced by the significant resistance of apaf-1-'-

and casPase-9-'-

ES

cells to cell death (Hakem et al., 1998; Yoshida et al., 1998). In light of this, I detemined whether addition of the pan-caspase inhibitor z-VAD.fmk would uncover an AIF-specific contribution to cell death. However, even in the presence of z-VAD.fmk, aif +/"and

aif

cells succumbed to staurosporine-induced death to comparable

extents (figure 8A). These data may suggest that AIF translocation from mitochondna is downstream of caspase activation. Indeed, it has been shown that in response to UV radiation, mitochondrial membrane depolarization (and presumably AIF release from mitochondria) requires cytochrome c translocation from mitochondna and subsequent caspase activation (Bossy-Wetzel et al., 1998: Goldstein et al., 2000).

Figure 10. A model foi the stimulus-specific role of AIF in PCD. Apoptosis in response to DNA damage, inhibition of kinases, or inhibition of the protein translational machinery does not require AIF. However, AIF is essential for cell death due to growth factor deprivation. In response to oxidative stress, AIF and caspases function in parallel and redundant pathways to induce apoptosis.

aif

and aif

cells treated with menadione, an inducer of oxidative stress,

were comparably sensitive to death (figure 8B, middle). However, in the presence of zVAD.fmk, aif -IY cells were largely protected whereas aif +lY cells remained sensitive (figure 8B, right). These data indicate that AIF and caspases participate in parallel pathways of cell death and each can functionally substitute for the other in response to menadione. Consistent with these results is the notion that

mitochondrial

depolarization is the commitment step to cell death, triggenng pemieabilization of the outer mitochondrial membranes and release of both cytochrome c and AIF into the cytosol. Mechanistically, this would occur either through the opening of the PTPs to create a conductance channel to the cytosol for apoptotic effectors, or through osmotic matnx swelling and rupture of the outer mitochondrial membrane secondary to PTP opening (Kroemer and Reed, 2000). Intriguingly, AIF was found to be important for death induced by serum withdrawal. It haç been reported that apaf-1-1-

ES cells and cytochrome c-'-

embryonic cells are partially resistant to serum starvation (Haraguchi et al., 2000; Li et al., 2000). Whether AIF, cytochrome and apaf-1 participate in the same linear pathway, or whether AIF cooperates with caspases needs to be resolved. In light of evidence suggesting that extramitochondrial AIF can induce pemieabilization of mitochondrial membranes (Susin et al., 1999), it is tempting to speculate that AIF may induce cell death by controlling cytochrome c release and activation of the caspase cascade. There is substantial evidence for the notion that the survival of individual cells depends on exogenous growth factors generated by thernselves, neighbouring cells or

the extracellular matrix (ECM) (Jacobson et al., 1997; Raff, 1992). Withdrawal of growth factors or removal from contact with ECM causes a cell to undergo PCD. The dependence of cells on exogenous survival signals provides a mechanism to regulate cell number and prevent cells accurnulating in ectopic areas. The mechanistic basis for this growth factor-induced survival has begun to be elucidated in several physiological systems. For example. IL-3 has been shown to tngger phosphorylation of the Bcl-2 family member BAD, resulting in the binding and sequestration of BAD by 14-3-3 in the cytosol. Nonphosphorylated BAD translocates to mitochondrial membranes where it binds and inactivates Bcl-XL to promote cell death (Zha et al., 1996). Moreover, the paucity of mature T cells in developing IL-7R-deficient mice can be rescued in a Bcl-2 transgenic background, suggesting that Bcl-2 can control growth factor-induced survival of developing T cells (Akashi et al., 1997). In addition, mice with double disruptions of the pro-apoptotic Bcl-2 members Bax and Bak exhibit elevated numbers of granulocytes and lymphocytes and enlarged spleen and lymph nodes due to a defect in homeostatic control of cell death (Lindsten et al., 2000). It is thus tempting to speculate that AIF may play a role in these and other physiological systems of cell death in response to growth factor deprivation.

Future Directions Since mutation of aif results in early lethality in the mouse (Joza et al., 2001). 1 will generate floxed aif mice (reviewed in Marth, 1996) to explore the role of AIF in various physiological compartments.

PCD is crucial to immunological tolerance (Penninger and Mak, 1994). In the thymus, immature T cells with potentially autoreactive antigen receptors undergo negative selection through PCD. In addition, thymocytes with antigen receptors unable to recognize peptide antigens in the context of self MHC undergo PCD by default. PCD also ensures homeostasis after an immune response. The death effector mechanisms responsible for immunological tolerance and homeostasis, however, are unclear. Peripheral T cells from caspase 3" mice are susceptible to CD3-, Fas- and activationinduced cell death (AICD) (Woo et al., 1998). ~paf-1" and caspase-9-'- thyrnocytes are resistant to certain death stimuli but sensitive to others, such as fas (Hakem et al., 1998; Yoshida et al., 1998). Furthemore, no caspase knockout mouse exhibits a

defect in 6 or T ceIl maturation. It thus remains probable that caspase-independent pathways of PCD, perhaps mediated by AIF, play a role in PCD in the immune system. To investigate the role of AIF in negative selection, knock-in mice carrying the floxed A!F allele will be crossed with H-Y TCR transgenic mice in order to generate a monoclonal T cell repertoire. These mice will then be crossed with transgenic mice expressing cre under the control of the T-cell specific promoter Ick. Mice bearing the genotype H-Y TCR; Ick-cre; floxed AIF will have specific deletion of mAlF in their (monoclonal) T cell population. Expression of the negatively-selecting peptide H-Y (in male mice) will allow me to detemine whether disruption of AIF rescues immature ~

0

PCD to allow developrnental progression to the 8 double-positive ~ ~ ~ T cells 4 from ~

CD8/CD4 single positive lineages.

In addition to T cell ontogeny, PCD plays a critical role in many aspects of mammalian embryogenesis. For example, apoptosis is crucial to morphogenic

processes such as the sculpting of digits and the hollowing out of the ectoderm to form the proamniotic cavity (Jacobson et al., 1997). Apaf-1"

mice exhibit normal, albeit

delayed, removal of interdigital mesenchyme, whereas adult mice doubly deficient in the pro-apoptotic Bcl-2 members Bax and Bak fully retain their interdigital webs (Yoshida et al., 1998; Lindsten et al., 2000). These results suggest that other apoptotic pathways downstrearn of the Bcl-2 cell death checkpoint are crucial to these processes. To address the in vivo importance of AIF in these developrnental processes, knock-in mice carrying the floxed AIF allele will be crossed to msx-2-cre transgenic mice to direct AIF disruption to the apical ectodermal ridge (AER) of the developing limb bud (Reginelli et al., 1995). In many developing organs, cells are overproduced and subsequentîy culled by

PCD to adjust their numbers. For exarnple, cells in the developing nervous system compete for limiting amounts of neurotrophic factors secreted by the target cells they innervate. Only those neurons able to make functional connections to their targets survive, while the remaining cells undergo cell death (Oppenheim, 1991; Oppenheirn et al., 1991). The consequences of dysregulated neuronal cell death is illustrated in the apaf-1-"-, caspase-9-'-, and caspase-3-'- mice, which exhibit an excess of cells in the

CNS due to irnpaired PCD of neural progenitors (Yoshida et al., 1998; Cecconi et al.. 1998; Hakem et al., 1998; Kuida et al., 1998; Kuida et al., 1996). To explore the

potential role of AIF in the developing CNS, I will cross floxed AIF knock-in mice to nestin-cre transgenic mice to direct AIF ablation to the neuronal lineage (Isaka et al., 1999).

Conclusions In summary, my data suggest that the role of AIF in ES cell death is stimulusspecific (figure 10). Apoptosis in response to DNA damage, inhibition of kinases, or inhibition of the protein translational rnachinery does not require AIF. However, AIF is essential for cell death due to growth factor deprivation. Furthemore, in response to oxidative stress, AIF and caspases function in parallel and redundant pathways to induce apoptosis. These results support a model whereby distinct death signals trigger distinct responses from mitochondria. Certain death stimuli (DNA damage) may preferentially release cytochrome c and induce caspase activation, while others (withdrawal of growth factors) may selectively trigger AIF release.

V.

Acknowledgments

I thank Drs. Guido Kroemer, Eric Daugas and Santos A. Susin for providing data on

the cell death assays, specifically the menadione and serum withdrawal experiments.

VI.

References

Akashi, K., Kondo, M., von Freeden-Jeffry, U., Murray, R., and Weissman, 1. L. (1997). Bcl-2 rescues T lymphopoiesis in interleukin-7 receptor-deficient mice. Cell 89, 1033-41.

Amarante-Mendes, G. P., Finucane, D. M., Martin, S. J., Cotter, T. G., Salvesen, G. S., and Green, D. R. (1998). Anti-apoptotic oncogenes prevent caspasedependent and independent cornmitment for cell death. CeII Death Differ 5,298306. Bergeron, L., Perez, G. I., Macdonald, G., Shi, L., Sun, Y., Jurisicova, A., Vamuza, S., Latham, K. E., Flaws, J. A., Salter, J. C., Hara, H., Moskowitz, M.

A., Li, E., Greenberg, A., Tilly, J. L., and Yuan, J. (1998). Defects in regulation of apoptosis in caspase-2-deficient mice. Genes Dev 12, 1304-14. Bossy-Wetzel, E., Newmeyer, D. D., and Green, D. R. (1998). Mitochondrial cytochrome c release in apoptosis occurs upstream of DEVD-specific caspase activation and independently of mitochondrial transrnembrane depolarization. Embo J 17, 37-49. Cecconi, F., Alvarez-Bolado, G., Meyer, B. I., Roth, K. A., and Gniss, P. (1998). Apafl (CED-4 homolog) regulates programmed cell death in mammalian development. Cell 94, 727-37. Daugas, E., Susin, S. A., Zamzami, N., Ferri, K. F., Irinopoulou, T., Larochette, N., Prevost, M. C., Leber, B., Andrews, D., Penninger, J., and Kroemer, G. (2000). Mitochondrio-nuclear translocation of AIF in apoptosis and necrosis. Faseb J 14,729-39. Ellis, H. M., and Horvitz, H. R. (1986). Genetic control of prograrnmed cell death in the nematode C. elegans. Cell 44,817-29. Enari, M., Sakahira, H., Yokoyama, H., Okawa, K., Iwamatsu, A., and Nagata, S. (1998). A caspase-activated DNase that degrades DNA during apoptosis, and its inhibitor ICAD. Nature 391, 43-50. Ferri, K. F., Jacotot, E., Blanco, J., Este, J. A., Zamzami, N., Susin, S. A., Xie, Z., Brothers, G., Reed, J. C., Penninger, J. M., and Kroemer, G. (2000). Apoptosis Control in Syncytia lnduced by the HIV Type 1-Envelope Glycoprotein Complex. Role of mitochondria and caspases. J Exp Med 192.1O81-1092. Goldstein, J. C., Waterhouse, N. J., Juin, P., Evan, G. l., and Green, O. R. (2000). The coordinate release of cytochrome c during apoptosis is rapid, complete and kinetically invariant. Nat Cell Biol 2,156-62. Green, D. R., and Reed, J. C. (1998). Mitochondria and apoptosis. Science 281, 1309-12. Hakem, R., Hakem, A., Duncan, G. S., Hendenon, J. T., Woo, M., Soengas, M. S., Elia, A., de la Pompa, J. L., Kagi, D., Khoo, W., Potter, J., Yoshida, R., Kaufman, S. A., Lowe, S. W., Penninger, J. M., and Mak, T. W. (1998). Differential requirement for caspase 9 in apoptotic pathways in vivo. Cell 94, 339-52. Haraguchi, M., Torii, S., Matsuzawa, S., Xie, Z., Kitada, S., Krajewski, S., Yoshida, H., Mak, T. W., and Reed, J. C. (2000). Apoptotic protease activating factor 1 (Apaf-1)-independent cell death suppression by Bcl-2. J Exp Med 191, 1709-20.

Hengartner, M. O., and Horvitz, H. R. (1994). C. elegans cell suivival gene ced-9 encodes a functional homolog of the mammalian proto-oncogene bcl-2. Cell 76, 665-76. Hinch, T., Marchetti, P., Susin, S. A., Dallaporta, B., Zamzami, N., Marzo. I., Geuskens, M., and Kroemer, G. (1997). The apoptosis-necrosis paradox. Apoptogenic proteases activated after mitochondrial perrneability transition determine the mode of cell death. Oncogene 15, 1573-81. Honarpour, N., Du, C., Richardson, J. A., Hammer, R. E., Wang, X., and Herz, J. (2000). Adult Apaf-1-deficient mice exhibit male infertility. Dev Biol218, 248-58. Horvitz, H. R.. Sternberg, P. W., Greenwald, 1. S., Fixsen, W., and Ellis, H. M. (1983). Mutations that affect neural cell lineages and cell fates during the development of the nematode Caenorhabditis elegans. Cold Spring Harb Symp Quant Biol48,453-63. Isaka, F., lshibashi M, Taki W. Hashimoto N, Nakanishi S, and Kageyama R. (1999) Ectopic expression of the bHLH gene Math1 distubs neural development. Eur 3 Neurosci 11,2582-8. Jacobson, M. D., Weil, M., and Raff, M. C. (1997). Programmed cell death in animal development. Cell 88,347-54. Joza, N., Susin, S. A., Daugas, E.,Stanford, W. L., Cho, S. K., Li, C. Y., Sasaki, T., Elia, A. J., Cheng, H. Y.. Ravagnan, L., Ferri, K. F., Zamzami, N., Wakeham, A., Hakem, R., Yoshida, H., Kong, Y. Y., Mak, T. W., Zuniga-Pflucker, J. C., Kroemer, G., and Penninger, J. M. (2001). Essential role of the mitochondrial apoptosis-inducing factor in programmed cell death. Nature 4 70, 549-54. Kerr, J. F., Wyllie, A. H., and Currie, A. R. (1972). Apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics. Br J Cancer 26, 239-57. Kluck, R. M., Bossy-Wetzel, E., Green, D. R., and Newmeyer, D. D. (1997). The release of cytochrome c from mitochondria: a pnmary site for Bcl-2 regulation of apoptosis [see cornments]. Science 275, 1132-6. Kroemer, G. (1997). The proto-oncogene Bcl-2 and its role in regulating apoptosis. Nat Med 3, 614-620. Kroemer, G., and Reed, J. C. (2000). Mitochondrial control of cell death. Nat Med 6, 513-9.

Kuida, K., Haydar, T. F., Kuan, C. Y., Gu, Y., Taya, C., Karasuyama, H., Su, M. S., Rakic, P., and Flavell, R. A. (1998). Reduced apoptosis and cytochrome cmediated caspase activation in mice lacking caspase 9. Cell 94,325-37. Kuida, K., Zheng, T. S., Na, S., Kuan, C., Yang, D., Karasuyama, H., Rakic, P., and Flavell, R. A. (1996). Decreased apoptosis in the brain and premature lethality in CPP32-deficient mice. Nature 384, 368-72. Li, H., and Yuan, J. (1999). Deciphering the pathways of life and death. Curr Opin Cell Biol 11, 261-6. Li, K., Li, Y., Shelton, J. M., Richardson, J. A., Spencer, E., Chen, 2. J., Wang, X., and Williams, R. S. (2000). Cytochrome c deficiency causes embryonic lethality and attenuates stress-induced apoptosis. Cell 101, 389-99. Li, P., Nijhawan, D., Budihardjo, I., Snnivasula, S. M., Ahmad, M., Alnemri, E. S., and Wang, X. (1997). Cytochrome c and dATP-dependent formation of ApafVcaspase-9 complex initiates an apoptotic protease cascade. Cell 91, 479-89. Lindsten, T., Ross, A. J., King, A., Zong, W.-X., Rathmell, J. C., Shiels, H. A., Ulrich, E., Waymire, K. G., Mahar, P., Frauwirth, K., Chen, Y., Wei, M., Eng, V. M., Evan, G., Konmeyer, S. J., MacGregor, G. R., and Thompson, C. B. (2000). The combined functions of proapoptotic Bcl-2 family mernben bak and bax are essential for normal development of multiple tissues. Mol Cell 6, 1389-99. Liu, X., Kim, C. N., Yang, J., Jemmerson, R., and Wang, X. (1996). Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome c. Cell 86, 147-57. Loeffler, M., Daugas, E., A., S. S., Zamzami, N., Metivier, D., Nieminen, A. L., Brothers, O., Penninger, J. M., and Kroemer, G. (2001). Dominant cell death induction by extramitochondrially targeted apoptosis-inducing factor. FASEB 15, 758-67. Marth, J. D. (1996). Recent advances in gene mutagenesis by site-directed recombination. J Clin lnvest 97, 1999-2002. Marzo, I., Brenner, C., Zamzami, N., Jurgensmeier, J. M., Susin, S. A., Vieira, H. L., Prevost, M. C., Xie, Z., Matsuyama, S., Reed, J. C., and Kroemer, G. (1998). Bax and adenine nucleotide translocator cooperate in the mitochondrial control of apoptosis. Science 281,2027-31. Nicholson, D. W., and Thomberry, N. A. (1997). Caspases: killer proteases. Trends Biochem Sci 22,299-306.

Oppenheim, R. W. (1991) Ce11death during development of the nentous system. Annu Rev Neurosci 14,453-501. Oppenheim, R. W., Prevette, D., Yin, O. W., Collins, F., and MacDonald, J. (1991) Control of embryonic motoneuron survival in vivo by ciliary neurotrophic factor. Science 251, 1616-8. Penninger, J. M., and Kroemer, G. (1998). Molecular and cellular mechanisms of T lymphocyte apoptosis. Adv lrnmunol 68, 51-144. Penninger, J. M., and Mak, T. W. (1994). Signal transduction, mitotic catastrophes, and death in T-cell development. lmrnunol Rev 142, 231-72. Raff, M. C. (1992). Social controls on cell survival and cell death. Nature 356, 397-400. Rathmell, J. C., Vander Heiden, M. G., Harris, M. H., Frauwirth, K. A., and Thompson, C. B. (2000). In the absence of extrinsic signals, nutrient utilization by lymphocytes is insufficient to maintain either cell sire or viability [In Process Citation]. Mol Cell 6, 683-92. Reginelli, A. D., Wang, Y. Q., Sassoon, D., and Muneoka, K. (1995). Digit tip regeneration correlates with regions of Msxl (Hox 7) expression in fetal and newborn mice. Development 121, 1065-76. Shimizu, S., Narita, M., and Tsujimoto, Y. (1999). Bcl-2 family proteins regulate the release of apoptogenic cytochrome c by the mitochondrial channel VDAC. Nature 399, 483-7. Susin, S. A., Daugas, E., Ravagnan, L., Samejima, K., Zamzami, N., Loeffler, M., Costantini, P., Ferri, K. F., Irinopoulou, T., Prevost, M. C., Brothers, G., Mak, T. W., Penninger, J., Eamshaw, W. C., and Kroemer, G. (2000). Two Distinct Pathways Leading to Nuclear Apoptosis. J Exp Med 192, 571-580. Susin, S. A., Lorenzo, H. K., Zamzami, N., Marzo, I., Snow, 8.E., Brothers, G. M., Mangion, J., Jacotot, E., Costantini, P., Loeffler, M., Larochette, N., Goodlett, D. R., Aebersold, R., Siderovski, D. P., Penninger, J. M., and Kroemer, G. (1999). Molecular characterization of mitochondrial apoptosis-inducing factor [see comments]. Nature 397, 441-6. Thompson, C. B. (1995). Apoptosis in the pathogenesis and treatment of disease. Science 267, 1456-62. Thomberry, N. A. (1998). Caspases: key mediators of apoptosis. Chem Biol5, R97-103.

Vander Heiden, M. G., Chandel, N. S., Schumacker, P. T., and Thornpson, C. B. (1999). Bcl-xL prevents cell death following growth factor withdrawal by facilitating rnitochondrial ATP/ADP exchange. Mol Cell 3, 159-67. Varfolomeev, E. E., Schuchmann, M., Luria, V., Chiannilkulchai, N., Beckmann, J. S., Mett, 1. L., Rebrikov, D., Brodianski, V. M., Kemper, O. C., Kollet, O., Lapidot, T., Soffer, D., Sobe, T., Avraham, K. B., Goncharov, T., Holtmann, H., Lonai, P., and Wallach, D. (1998). Targeted disruption of the mouse Caspase 8 gene ablates cell death induction by the TNF receptors, FadApol, and DR3 and is lethal prenatally. lmmunity 9, 267-76. Vaux, D. L., and Konmeyer, S. J. (1999). Cell death in development. Cell 96, 245-54. Woo, M., Hakem, R., Soengas, M. S., Duncan, G. S., Shahinian, A., Kagi, D., Hakem, A., McCurrach, M., Khoo, W., Kaufman, S. A., Senaldi, G., Howard, T., Lowe, S. W., and Mak, T. W. (1998). Essential contribution of caspase 3/CPP32 to apoptosis and its associated nuclear changes. Genes Dev 12, 806-19. Xiang, J., Chao, D. T., and Konmeyer, S. J. (1996). BAX-induced cell death may not require interleukin 1 beta-converting enzyme-fike proteases. Proc Natl Acad Sci U S A 93, 14559-63. Xiang, J., Chao, D. T., and Korsmeyer, S. J. (1996). BAX-induced cell death may not require interleukin 1 betatonverting enzyme-like proteases. Proc Natl Acad Sci U S A 93, 14559-63. Yang, J., Liu, X., Bhalla, K., Kim, C. N., Ibrado, A. M., Cai, J., Peng, T. 1.. Jones, D. P., and Wang, X. (1997). Prevention of apoptosis by Bcl-2: release of cytochrome c from mitochondria blocked [see comments]. Science 275, 1129-32. Yoshida, H., Kong, Y. Y., Yoshida, R., Elia, A. J., Hakem, A., Hakem, R., Penninger, J. M., and Mak, T. W. (1998). Apafl is required for mitochondrial pathways of apoptosis and brain development. Cell 94, 739-50. Zha, J., Harada, H., Yang, E., Jockel, J., and Korsmeyer, S. J. (1996). Serine Phosphorylation of Death Agonist BAD in Response to Survival Factor Results in Binding to 14-3-3 Not BCL-XL. Cell 87, 619-628. Zou, H., Henzel, W. J., Liu, X., Lutschg, A., and Wang, X. (1997). Apaf-1, a human protein homologous to C. elegans CED-4, participates in cytochrome cdependent activation of caspase-3 [see comments]. Cell 90, 405-13.