Research Article

Mechanisms of AIF-Mediated Apoptotic DNA Degradation in Caenorhabditis elegans

See allHide authors and affiliations

Science  22 Nov 2002:
Vol. 298, Issue 5598, pp. 1587-1592
DOI: 10.1126/science.1076194


Apoptosis-inducing factor (AIF), a mitochondrial oxidoreductase, is released into the cytoplasm to induce cell death in response to apoptotic signals. However, the mechanisms underlying this process have not been resolved. We report that inactivation of theCaenorhabditis elegans AIF homolog wah-1 by RNA interference delayed the normal progression of apoptosis and caused a defect in apoptotic DNA degradation. WAH-1 localized in C. elegans mitochondria and was released into the cytosol and nucleus by the BH3-domain protein EGL-1 in a caspase (CED-3)–dependent manner. In addition, WAH-1 associated and cooperated with the mitochondrial endonuclease CPS-6/endonuclease G (EndoG) to promote DNA degradation and apoptosis. Thus, AIF and EndoG define a single, mitochondria-initiated apoptotic DNA degradation pathway that is conserved between C. elegans and mammals.

Programmed cell death (apoptosis) is a fundamental feature in the development and tissue homeostasis of metazoans (1, 2). During apoptosis, the cell activates a suicide machinery that executes orderly cell disassembly, including condensation and fragmentation of the chromosomal DNA (1). Many key components that initiate and execute apoptosis are conserved across species (3). For example, the onset of apoptosis is controlled by a regulatory pathway involving conserved cell death activators and inhibitors: EGL-1 and BH3-domain-only proteins, CED-9 and Bcl-2, CED-4 and Apaf-1, and CED-3 and caspases, in nematodes and mammals, respectively (3). In addition, the degradation of chromosomal DNA involves a mitochondrial endonuclease: endonuclease G (EndoG) in mammals and its ortholog CPS-6 in worms (4, 5).

EndoG is one of several mitochondrial proapoptotic factors, including cytochrome c, apoptosis-inducing factor (AIF), Smac/DIABLO, and Omi/HtrA2, which are released into the cytosol to mediate various aspects of apoptosis in mammals (6–11). These factors have been proposed to act through caspase-dependent pathways (cytochrome c and Smac/DIABLO), caspase-independent pathways (AIF and EndoG), or both (Omi/HtrA2). CPS-6 is the only mitochondrial protein that has been implicated in invertebrate apoptosis (5). It is unclear whether additional mitochondrial factors regulate apoptosis in invertebrates, and if so, whether the mitochondrial cell death pathways are conserved.

AIF has been suggested to mediate caspase-independent apoptosis (12, 13). In response to apoptotic stimuli, AIF is released from mitochondria and translocated to both the cytosol and the nucleus, where it induces condensation and large-scale fragmentation of chromatin (7, 13,14). AIF is a flavin-adenine dinucleotide (FAD)–binding oxidoreductase, but neither its FAD-binding ability nor its oxidoreductase activity is required for its apoptogenic activity (7, 15). Therefore, it is unclear how AIF induces DNA fragmentation or apoptosis. Furthermore, as an evolutionarily conserved protein (12, 15), it is unclear whether the pro-apoptotic function of AIF is conserved in other species. To address these issues, we cloned the C. elegans AIF homologwah-1 (worm AIF homolog; GenBank accession number AY147006) and performed systematic studies to understand its biological functions, biochemical activities, and the mechanisms by which it mediates apoptosis.

WAH- 1 is important for apoptosis in C. elegans.

The sequence of the human AIF protein was used to search the C. elegans genome database. One putative open reading frame on linkage group III, C51G7.5, was identified, and a full-length cDNA clone corresponding to C51G7.5 was obtained with a reverse transcription–polymerase chain reaction strategy (16). The predicted 700–amino acid protein shows 37% sequence identity and 54% sequence similarity to the human AIF (fig. S1). C51G7.5 was renamed wah-1. Sequence similarity is particularly strong between residues 214 and 700 of WAH-1 and residues 107 and 613 of the human AIF, which correspond to the mature human AIF protein (7). The NH2-terminal portions of the two proteins do not have sequence similarity except for the first 30 amino acids, which encode a putative mitochondrial localization sequence (fig. S1) (17). WAH-1 also shares sequence similarity to the Dictyostelium discoideumAIF protein (18) and a homologous protein inDrosophila (fig. S1).

To understand the functions of wah-1 in C. elegans programmed cell death, we used RNA interference (RNAi) (19) to inhibit the expression of the wah-1 gene in C. elegans (16) [these are referred to as wah-1(RNAi) animals]. Unlike the situation in mice, where AIF-deficient embryonic stem cells fail to form a viable chimeric embryo (13), wah-1(RNAi) animals were viable but exhibited slower growth rate and a smaller brood size than wild-type animals or animals treated with an RNAi control (20). These features allowed us to examine the functions of WAH-1 in various aspects of C. elegans apoptosis during development.

A time-course analysis of embryonic cell corpses (5, 21) revealed that wah-1(RNAi)treatment delayed the appearance of embryonic cell corpses during development (Fig. 1A). The peak of cell corpses was shifted from the comma embryonic stage in wild-type animals to the twofold embryonic stage in wah-1(RNAi) animals (P < 0.0001, unpaired t test) (Fig. 1A). The numbers of cell corpses in later embryonic stages (2.5-fold and 3-fold) also increased in wah-1(RNAi) embryos (P < 0.0001, unpaired t test). This delayed corpse appearance phenotype is similar to that displayed by thecps-6(sm116) mutant (5). In addition, like the cps-6(sm116) mutation, wah-1(RNAi) treatment enhanced the delay-of-cell-death phenotype of theced-8(n1891) mutant (21), increasing the numbers of cell corpses in fourfold stage embryos and L1 stage larvae (P < 0.0001, unpaired t test) (Fig. 1B). However, wah-1(RNAi) treatment did not affect the profile of corpse appearance in either the cps-6(sm116) mutant (Fig. 1C) or the cps-6(sm116); ced-8(n1891) mutant (Fig. 1D), suggesting that wah-1 and cps-6 may function in the same pathway to affect the progression of apoptosis.

Figure 1

wah-1(RNAi) affects the progression of apoptosis and apoptotic DNA degradation in C. elegans. L1 stage C. elegans larvae were treated withwah-1(RNAi) (filled bars) or control(RNAi) (open bars), and the progeny of the RNAi-treated animals were scored for cell corpses (A to D) or TUNEL-positive nuclei (E to H) (16). In the cell corpse assays, N2 (A), ced-8(n1891) (B),cps-6(sm116) (C), and cps-6(sm116); ced-8(n1891)(D) animals were scored at the following embryonic or larval stages: comma, 1.5-fold, 2-fold, 2.5-fold, 3-fold, 4-fold, and early L1 stage larvae. In the TUNEL assays, N2 (E),cps-6(sm116) (F), nuc-1(e1392) (G), andced-3(n2433) (H) animals were counted at three embryonic stages: comma, 1.5-fold, and 3- and 4-fold. The y axis represents the average number of cell corpses or TUNEL-positive nuclei scored. Error bars indicate SEM. At least 15 animals were scored for each developmental stage in each assay.

To determine whether wah-1(RNAi) treatment could prevent cell death, we examined cells in the anterior pharynx of treated animals (21). Treatment with wah-1(RNAi) alone had little effect on the deaths of the 16 cells in this region that normally undergo apoptosis (Table 1). However,wah-1(RNAi) treatment enhanced the cell survival phenotypes of mutants that are partially defective in apoptosis, such as animals that are homozygous for a partial loss-of-function mutation in theced-3 gene (n2438 or n2447) or the ced-4 gene (n2273) (Table 1). For instance, in ced-3(n2447) animals, an average of 1.2 extra cells was seen in the anterior pharynx, compared with a mean of 2.7 extra cells observed in ced-3(n2447); wah-1(RNAi)animals (P < 0.003, unpaired t test) (Table 1). Treatment with wah-1(RNAi) also enhanced the cell survival phenotypes of mutants that are strongly defective in cell death, such as ced-3(n2433) or ced-4(n1162)(P < 0.02, unpaired t test) (Table 1). These data suggest that wah-1, like the cps-6gene, not only is important for the normal progression of apoptosis but also can promote cell killing.

Table 1

wah-1(RNAi) enhances the cell death defects of the ced-3 and ced-4 mutants. RNAi experiments were carried out with a bacterial feeding protocol (16). “Control(RNAi)” indicates that the animals were fed bacteria containing an expression vector without an insert as a negative control. Extra cells were counted in the anterior pharynx of L4 hermaphrodites using Nomarski optics (16); the data are shown as means ± SEM.

View this table:

WAH-1 promotes apoptotic DNA degradation.

Because human AIF induces large-scale chromosome fragmentation (7), we examined apoptotic DNA degradation in nematodes treated with wah-1(RNAi) using TUNEL (terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling) assay (22). Wild-type C. elegans embryos treated withwah-1(RNAi) contained more TUNEL-positive nuclei than did control embryos at several embryonic stages examined (Fig. 1E), suggesting that WAH-1 is involved in resolving the TUNEL-positive DNA breaks. The numbers of TUNEL-positive nuclei observed inwah-1(RNAi) embryos were lower than those observed in thecps-6(sm116) mutant (5) or a mutant(e1392) defective in the nuc-1 gene, which encodes a deoxyribonuclease II homolog and has been implicated in mediating apoptotic DNA degradation in C. elegans(22) (Fig. 1, F and G). The TUNEL-positive nuclei found inwah-1(RNAi) embryos likely represent cells undergoing apoptosis, because the ced-3(n2433) mutation that blocks almost all cell death in nematodes abolished the TUNEL staining in wah-1(RNAi) embryos (Fig. 1H).

Similar numbers of TUNEL-positive cells were observed incps-6(sm116); wah-1(RNAi) embryos or controlcps-6(sm116) embryos; however, nuc-1(e1392); wah-1(RNAi) embryos had higher numbers of TUNEL-positive nuclei than did control nuc-1(e1392) embryos orwah-1(RNAi) embryos (P < 0.0001, unpaired t test) (Fig. 1, F and G). These results suggest that wah-1 and cps-6 likely function in the same DNA degradation pathway, consistent with observations thatwah-1(RNAi) does not enhance the cell death defects of thecps-6(sm116) mutant (Fig. 1, C and D). In contrast,nuc-1 appears to be dispensable for apoptosis (5, 22, 23) and likely functions in a different DNA degradation process.

WAH-1 localizes to mitochondria and can be released by EGL-1.

To examine the subcellular localization of WAH-1, we expressed a fusion protein composed of WAH-1 and green fluorescent protein (WAH-1::GFP) under the control of the wah-1promoter (Pwah-1 wah-1::gfp) (16). WAH-1::GFP was widely expressed in embryos and larvae (Fig. 2) (20), displaying a punctate, cytoplasmic staining pattern similar to the mitochondrial localization pattern of CPS-6 (5) (Fig. 2A). The staining pattern of WAH-1::GFP coincided with that of Mitotracker Red, a mitochondria-specific fluorescent dye (24), confirming that WAH-1 localized within mitochondria (Fig. 2A).

Figure 2

WAH-1 localizes to mitochondria and can be released by EGL-1. The Nomarski, GFP, and Mitotracker Red images and the merged image of GFP and Mitotracker Red (from left to right) of an early C. elegans embryo from the following strains are shown: N2 animal transgenic for Pwah-1 wah-1::gfp(A); ced-1(e1735); egl-1(n3082) animal transgenic for both Pwah-1 wah-1::gfpand Phsp egl-1 constructs without heat-shock treatment (B) or after heat-shock treatment at 33°C for 1 hour (C) (16); ced-1(e1735); ced-3(n2433); egl-1(n3082) animal transgenic for both Pwah-1 wah-1::gfp and Phsp egl-1 after heat-shock treatment at 33°C for 1 hour (D and E) (16). WAH-1::GFP was not released from mitochondria in the embryo shown in (D) but was partially released in the embryo shown in (E).

In mammals, the BH3-domain-only proteins such as Bid and Bim can cause release of mitochondrial apoptogenic factors in response to apoptotic signals (4, 25–28). However, it is unclear whether a similar mitochondrial cell death pathway exists in C. elegans. Because the C. elegans BH3-domain protein EGL-1 is thought to be the most upstream cell death activator that receives and integrates cell death stimuli (29, 30), we determined whether induction of global expression of EGL-1 under the control of the C. elegans heat-shock promoters (Phsp egl-1) could cause the release of WAH-1 from mitochondria. Upon heat-shock treatment ofegl-1(n3082) embryos transgenic for both Pwah-1 wah-1::gfp and Phsp egl-1, WAH-1::GFP lost its normally punctate, cytoplasmic staining pattern (Fig. 2B) and, instead, adopted a more uniform localization pattern with more intense GFP fluorescence seen in the nuclei (Fig. 2C). This result indicates that WAH-1 was released from mitochondria and translocated to nuclei by EGL-1 and that a mitochondrial cell death pathway similar to the one used in mammals likely exists in C. elegans.

In mammals, AIF appears to mediate a caspase-independent cell death pathway (7, 13). We thus examined whether the CED-3 caspase activity in nematodes is dispensable for the release of WAH-1 from the mitochondria. Expression of EGL-1 in ced-3(n2433); egl-1(n3082) embryos transgenic for both Pwah-1 wah-1::gfp and Phsp egl-1 failed to induce the release of WAH-1 in most of the embryos (Fig. 2D). In 17% of the embryos (n = 197), WAH-1::GFP was observed in ∼10% of the nuclei, indicating partial release of WAH-1 (Fig. 2E). In contrast, WAH-1::GFP was completely released from mitochondria by EGL-1 in 98% of egl-1(n3082) embryos (n = 50). Time-course analysis indicated that the reduction or block of the WAH-1 release in ced-3(n2433); egl-1(n3082) embryos was not due to a delay in the mitochondrial release of WAH-1 (20). These observations suggest that the CED-3 caspase activity is important for the proper release of WAH-1 during apoptosis and that WAH-1 acts in a caspase-dependent manner inC. elegans.

WAH-1 cooperates with CPS-6 to promote DNA degradation.

To better understand how WAH-1 affects apoptosis in C. elegans, we performed biochemical characterization of WAH-1. When human AIF is imported into mitochondria from the cytoplasm, it is cleaved after residue 101 to generate a mature protein (7). WAH-1 is likely to be processed similarly inC. elegans. We thus generated and purified a truncated form of WAH-1, WAH-1(196–700), which contains the region most homologous to the mature human AIF protein (fig. S1). Intriguingly, the recombinant WAH-1(196–700) protein was completely colorless (31), and absorption spectrum analysis showed that it lacked the critical cofactor FAD required for typical oxidoreductases (fig. S2) (16, 31).

Because wah-1 and cps-6 appear to function in the same apoptotic DNA degradation pathway (Fig. 1), we tested whether WAH-1 interacts with CPS-6, using a glutathioneS-transferase (GST) fusion protein pull-down assay. Purified, histidine-tagged WAH-1 protein [WAH-1(196–700)-His6] interacted with a GST–CPS-6 fusion protein that was immobilized on glutathione Sepharose beads, but not with the GST protein (Fig. 3A). In contrast, an unrelated human protein, CDC34-His6, did not associate with the GST–CPS-6 protein in the pull-down assay (Fig. 3A), indicating that WAH-1 interacted specifically with CPS-6.

Figure 3

WAH-1 cooperates with CPS-6 to promote DNA degradation. (A) WAH-1 associates with CPS-6 in vitro. Purified GST or GST-CPS-6 (1 μg each) immobilized on glutathione-Sepharose beads was incubated with similar amounts of WAH-1(196–700)-His6 or CDC34-His6 at room temperature for 2 hours (16). The beads were washed extensively. The bound proteins were resolved on a 10% SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and visualized by Western analysis with antibody to His6. (B) WAH-1 cooperates with CPS-6 to degrade DNA. Plasmid (1 μg) was incubated with buffer alone (lane 1), with indicated amounts of CPS-6 or WAH-1(196–700), or with both at 37°C for 1 hour before the reactions were resolved on a 1% agarose gel (16). Supercoiled (sc) and open circle (oc) forms of plasmid are indicated. (C) WAH-1 and CPS-6 need to act together to promote DNA degradation. Plasmid (1 μg) was incubated in buffer alone (lane 1), with 20 ng of His6CPS-6(22–308) (lanes 2 and 6), with 1 μg of WAH-1(196–700)-His6 (lanes 3 and 5), or with 20 ng of His6CPS-6(22–308) and 1 μg of WAH-1(196–700)-His6 (lane 4) at 37°C for 30 min. The reactions in lanes 1 to 4 were stopped by freezing. The reactions shown in lanes 5 and 6 were passed through mini Ni2+–nitrilotriacetic acid columns to remove the His6-tagged proteins. Twenty nanograms of His6CPS-6(22–308) (lane 5) or 1 μg of WAH-1(196–700)-His6 (lane 6) was then added into the flow-through, and the reactions were incubated for an additional 30 min at 37°C before they were resolved on a 1% agarose gel together with the products from lanes 1 to 4.

We tested whether WAH-1 interacts with CPS-6 to promote DNA degradation, using a plasmid cleavage assay (4). A low concentration (0.5 ng/μl) of CPS-6 caused nicking of a plasmid DNA substrate, converting the supercoiled DNA into open circle forms (Fig. 3B). At higher concentrations (>5 ng/μl), CPS-6 further cleaved the plasmid DNA into smaller DNA fragments, yielding a smear pattern of degradation products (5). In contrast, the recombinant WAH-1(196–700) protein does not degrade DNA, even at a high concentration (50 ng/μl) (Fig. 3B) (20). However, when both CPS-6 and WAH-1 were incubated with the plasmid DNA at concentrations where neither protein alone exhibited significant nuclease activity, WAH-1(196–700) cooperated with CPS-6 to efficiently degrade DNA in a concentration-dependent manner (Fig. 3B), generating progressively smaller DNA fragments. This result suggests that WAH-1 and CPS-6 cooperate to promote DNA degradation.

To rule out the possibility that WAH-1 and CPS-6 act independently but in a sequential manner to promote DNA degradation, we incubated WAH-1(196–700)-His6 with plasmid DNA for a fixed amount of time and then depleted it from the reaction before His6CPS-6(22–308) was added to the reaction (Fig. 3C). We performed a similar experiment in which the incubation order of these two proteins was reversed (Fig. 3C). In both cases, the extent of DNA degradation was far less severe than that observed when WAH-1 and CPS-6 were incubated together with the DNA substrate (Fig. 3C), indicating that WAH-1 and CPS-6 need to be present simultaneously to promote efficient DNA degradation, possibly through protein-protein interaction. Together, these results demonstrate that WAH-1 and CPS-6, two mitochondrial proteins, associate and cooperate to promote DNA degradation and likely function in the same apoptotic DNA degradation pathway as that observed in vivo.

WAH-1 synergizes with CPS-6 to induce cell killing.

Because WAH-1 and CPS-6 can interact in vitro to promote DNA degradation, we determined whether they could cooperate in vivo to promote cell killing. To assess the killing activity of WAH-1 and CPS-6, we ectopically expressed one or both proteins in six touch receptor neurons under the control of the C. elegans mec-7promoter (32) and then scored the percentage of PLM touch receptor neurons that were killed. When expressed individually in the touch cells, WAH-1(214–700), which lacks the mitochondrial targeting sequence, or full-length CPS-6 had a marginal killing activity, resulting in the death of ∼5 to 8% or ∼15 to 23% of the PLM neurons, respectively (Table 2). In contrast, when WAH-1(214–700) and CPS-6 were coexpressed in touch cells, ∼53 to 75% of PLM neurons were killed, suggesting that WAH-1 and CPS-6 can synergize to induce cell killing (Table 2). The ectopic cell killing induced by WAH-1/CPS-6 was significantly inhibited by the strong loss-of-function ced-3(n2433) mutation (Table 2), indicating that WAH-1/CPS-6 induced cell killing partially throughced-3. These results further support the conclusion thatwah-1 and cps-6 cooperate to promote apoptotic DNA degradation and apoptosis.

Table 2

WAH-1 and CPS-6 synergize to promote cell killing. In the Pmec-7WAH-1ΔN construct, WAH-1(214–700) was expressed under the control of the mec-7promoter. In Pmec-7CPS-6, the full-length CPS-6 protein was expressed. The indicated transgene construct (50 μg/ml each) was injected into bzIs8 or ced-3(n2433); bzIs8 animals with pRF4 (50 μg/ml), a dominant coinjection marker (34). bzIs8 is an integrated transgenic array containing Pmec- 4 gf p, which directs GFP expression in six touch receptor neurons and allows scoring of the PLM neurons. Each numbered array represents an independent transgenic line. The percentage of surviving PLM neurons was scored in the L4 larval stage animals with a fluorescent Nomarski microscope. Thirty Roller transgenic animals were counted for each transgenic line. In bzIs8 or ced-3(n2433); bzIs8animals, 100% of PLM neurons survived.

View this table:

AIF and EndoG define a single conserved DNA degradation pathway.

As a FAD-binding oxidoreductase, mammalian AIF induces apoptosis independently of its oxidoreductase activity (7,15). Our finding that the recombinant WAH-1 protein can promote DNA degradation in the absence of FAD further supports the generality of this conclusion. It has been suggested that AIF may exert its functions by interacting with downstream effectors (7,13). Our observations that WAH-1 associated and cooperated with CPS-6 to promote DNA degradation in vitro, that wah-1(RNAi)animals and cps-6 mutants display similar cell death and DNA degradation defects, and that both WAH-1 and CPS-6 localized to the mitochondria (5) and synergized to induce cell killing indicate that CPS-6/EndoG is an important nuclease target or effector that WAH-1/AIF interacts with to promote apoptotic DNA degradation. Thus, EndoG and AIF appear to define a single conserved DNA degradation pathway initiated from the mitochondria.

AIF and EndoG are two of the expanding list of apoptogenic proteins that are released from mitochondria to promote apoptosis in mammals (6–11, 28). In invertebrates, the evidence for an important role of mitochondria in apoptosis has not been strong (3, 5, 33). In our studies, we showed that WAH-1, a C. elegans mitochondrial apoptogenic factor, can be released by the BH3 domain-only protein EGL-1 into the cytosol and the nucleus in a manner similar to the release of cytochrome c or EndoG from mammalian mitochondria by the BH3-domain proteins (4, 25–27). This finding strongly suggests that the mitochondrial cell death pathway is evolutionarily conserved.

It is somewhat surprising that the release of WAH-1 from mitochondria in C. elegans partially depends on the activity of the CED-3 caspase, in contrast to the observation that the release of mammalian AIF is caspase independent (12, 13). It is likely that the initial, limited release of WAH-1 may be CED-3 independent, whereas the complete release of WAH-1 depends on the activation of the CED-3 caspase, which could amplify the damage to mitochondria by cleaving downstream CED-3 targets and promote further release of WAH-1. Our studies of AIF-mediated apoptosis in C. elegans indicate that different cell death pathways interact and likely cross-regulate one another in the process to activate the entire apoptotic program.

Supporting Online Material

Materials and Methods

References and Notes

Figs. S1 and S2

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: ding.xue{at}


View Abstract

Navigate This Article