Dear Editor,

Apoptosis inducing factor (AIF), which is encoded by the pcd8 gene on the X chromosome, is a protein that is normally present in mitochondria.1, 2 The AIF flavoprotein can participate in the scavenging of reactive oxygen species,3 and normal AIF levels are required for the assembly and/or maintenance of the complex I of the respiratory chain.4 Thus, AIF-deficient human and mouse cells manifest a reduced abundance and function of complex I that may results from a deficient antioxidant defense in mitochondria.5 Upon outer mitochondrial membrane permeabilization, a process that generally occurs during apoptosis,6 AIF translocates from mitochondria via the cytosol to the nucleus where it participates in apoptotic chromatin condensation, presumably through a direct interaction with DNA.4 AIF can also participate in apoptotic chromatinolysis, presumably by recruiting catabolic enzymes to the so-called degradosome.7

The AIF gene has been knocked out or knocked down in several species including in human cell lines,8 mice,9 Caenorhabditis elegans10 and Saccharomyces cerevisiae.11 In yeast, the knockout of AIF leads to a partial resistance to oxidative stress and influences replicative aging.11 In nematodes, the knockdown of AIF leads to reduction of developmental cell death, in particular, in animals that bear loss-of-function mutations of caspases.10 In mammalian cells, removal of AIF can reduce apoptosis in some particular settings, although there is no general apoptosis defect.2, 8, 9 Thus, knockdown of AIF protects differentiated PC12 cells against the neurotoxin 1-methyl-4-phenylpyridinium,12 Jurkat T lymphoma cells against a combination of γ-irradiation and phytosphingosin,13 erythroleukemic HEL cells against CD44 ligation,14 melanoma cells against the Raf inhibitor BAY 43-9006 (Sorafenib),15U937 promonocytic cells against cadmium,16 and Raji B lymphoma cells against UV irradiation.17 Microinjection of AIF-neutralizing antibodies can also reduce the neurotoxic effects of NMDA in primary murine cortical cultures,18 the lethal effects of PARP-1 activators in several cellular systems,19 as well as the proapoptotic effects of staurosporin on non-small cell lung carcinoma cells.20

While embryonic stem (ES) cells lacking AIF (AIF−/y) are relatively resistant against serum withdrawal-induced apoptosis, as compared to wild-type ES cells,9 they respond to most apoptosis inducers including DNA-damaging agents and tyrosine kinase inhibitors normally. According to one report, AIF−/y ES cells fail to undergo cavitation upon culture in the absence of leukemia-inhibitory factor (LIF),9 a process that usually induces aggregation of ES cells followed by apoptosis of the cells in the inner mass.21 Cavitation is regarded as the earliest wave of programmed cell death during embryogenesis, Two recent reports based on the use of another technique of AIF inactivation (the cross of mice expressing a β-actin cre transgene with mice having a floxed AIF locus) came to the conclusion that AIF was not required for cavitation in vivo22 or in vitro22 and that the observed in vitro phenotype, the absence or presence of cavitation, was influenced by the culture conditions and/or the genetic background.23

Driven by these considerations, we addressed the question whether the defect in cavitation observed in AIF−/y ES cells might be secondary to a defect in complex I. Hence, we cultured AIF+/y and AIF−/y ES cells in the presence of variable doses of the highly specific complex I inhibitor rotenone24 for 3 days (when ES cells aggregate upon removal of LIF) or 10 days (when cavitation has occurred in normal circumstances, in wild-type embryoid bodies). Rotenone caused a dose-dependent inhibition in the cellularity of 6-day-old embryoid bodies (Figure 1a, b). Of note, at a dose of 1 or 10 nM, rotenone was more toxic for AIF+/y than for AIF−/y ES (Figure 1b), in line with previous observations suggesting that AIF-deficient cells compensate their partial deficiency in respiration.4 Rotenone used at doses of 1 or 10 nM also strongly inhibited cavitation in 10-day-old AIF+/y embryoid bodies (Figure 1c, d). Rotenone-treated AIF+/y and AIF−/y embryoid bodies exhibited a similar phenotype, without cavitation. Thus, specific inhibition of the complex I of the respiratory chain is sufficient to perturb cellular metabolism and/or the signal transduction pathways leading to cavitation-associated apoptosis. As a word of caution, it should be mentioned that embryonic bodies cultured in the presence of rotenone were smaller than those lacking AIF, suggesting subtle differences between the acute complex I defect induced by rotenone (that might affect homeostatic systems in the cell other than respiration) and the chronic defect induced by the AIF knockout.

Figure 1
figure 1

Response of ES cells to complex inhibition by rotenone. (a, b). Reduced vulnerability of AIF−/y cells to complex I inhibition. Embryonic bodies (EBs) derived from AIF−/y or AIF+/y ES cells were cultured for three days in the presence of the indicated concentration of rotenone (100 nM in (a). Representative microphotographs are shown in (a). Note that rotenone-treated AIF+/y EBs form much more debris than control EBs). The frequency of viable cells was determined after dissociation of the EBs using a cytofluorometric method in B. What does B actually show. To generate embryoid bodies (EBs), ES cells were cultured in Iscove's modified Dulbecco's medium (IMDM, Invitrogen) supplemented with 15% FCS (Hyclone, Perbio), 1% L-Glutamin (Invitrogen), 1% Penicillin/Streptomycin (Invitrogen), 50 μg/ml ascorbic acid (Sigma), 450 μ M monothioglycerol (Sigma) and 200 μg/ml iron-saturated transferrin (Sigma). Undifferentiated ES cells (4.5 × 104 cells/ml) were seeded in bacterial grade dishes (Nunc) and incubated at 37°C under a low O2 concentration (7% O2). By 3 days of differentiation, EBs were treated with various concentration of rotenone (Sigma) or with vehicle alone. Embryoid bodies were then allowed to grow for further 3 days before analysis by flow cytometry. EBs were collected, dissociated in non-enzymatic Cell Dissociation Solution (Sigma), and subjected to the quantitation of viable PI-negative, DIOC6(3)-positive cells on a FACS Sort cytofluorometer (Becton and Dickinson). (c, d). Inhibition of cavitation by rotenone. ES cells with the indicated genotypes were cultured as in (a), with the difference that cultures were allowed to proceed until day 10. Representative pictures showing one cystic EB (formed from AIF+/y ES cells) and compact EBs obtained after inhibition with rotenone (10 nM) in AIF+/y ES cells or AIF−/y ES cells are shown in c and the percentage of cystic embryonic bodies obtained in the presence of different concentrations of rotenone is quantified in (d). As reported previously, the AIF−/Y ES cells did not cavitate. All the experiments have been reproduced at least three times, with similar results

Full size image

These results illustrate the difficulty to separate the two functions of AIF, first as a mitochondrial redox enzyme required for normal respiratory function and, second, as a factor that can participate in the apoptotic execution phase, after its translocation to the nucleus. The observation that AIF participates in cavitation-associated apoptosis9 implicitly suggested that cavitation would depend on the lethal action of AIF. Based on recent insights on AIF biology, however, it is possible that at least some of the lethal processes that were interrupted by the removal of AIF from the experimental system actually were suppressed due to defective bioenergetic and/or redox metabolism. As a result, we recommend the use of rotenone as an internal control in experiments, in which the apoptosis-modulatory effects of human or mouse AIF are assessed. Only when rotenone fails to modify the apoptotic response, the interpretation that the lethal (rather than the vital) action of AIF is involved in the process can be maintained.

Knock-in mutations that affect only the lethal function of AIF yet do not interfere with its metabolic activity in mitochondria are being prepared in several laboratories,2 and the use of such mutants will yield important insights into the contribution of AIF to lethal signal transduction processes. Such knock-in mutations will also help to decipher the contribution of AIF, alone or in combination with mutations of the apoptosome components, to programmed cell death in physiological and pathological cell death.