BIOLOGICAL OVERVIEW

Apaf-1-related-killer (Ark) encodes a Drosophila homolog of mammalian Apaf-1 and Caenorhabditis elegans CED-4, cell-death proteins. Like Apaf-1, but in contrast to CED-4, Ark contains a carboxy-terminal WD-repeat domain necessary for interactions with the mitochondrial protein cytochrome c (see Drosophila Cytochrome c proximal and Cytochrome c distal). Ark selectively associates with another protein involved in apoptosis, the fly apical caspase, Dredd. Ark-induced cell killing is suppressed by caspase-inhibitory peptides and by a dominant-negative mutant Dredd protein, and enhanced by removal of the WD domain. Loss-of-function mutations in Ark attenuates programmed cell death during development, causing hyperplasia of the central nervous system, and other abnormalities, including ectopic melanotic tumors and defective wings. Moreover, ectopic cell killing by the Drosophila cell-death activators, Reaper, Grim and Hid, is substantially suppressed in Ark mutants. These findings establish Ark as an important apoptosis effector in Drosophila and raise evolutionary considerations concerning the relationship between mitochondrial components and the apoptosis-promoting machinery (Rodriguez, 1999).

To determine whether Ark expression is sufficient to trigger cell death, conditional expression of the protein was directed in cultured fly cells. Epitope-tagged versions of Ark were transiently transfected into Drosophila Schneider L2 (SL2) cells. Robust killing was induced by expression of the apoptosis activator Grim. In parallel tests, moderate cell killing was associated with expression of full-length Ark, whereas a C-terminal truncation of the WD-repeat region, Ark(1-411), showed markedly enhanced killing activity. In both cases, cell killing was completely suppressed by the caspase inhibitor peptide Z-VAD and moderately attenuated by the Z-DEVD peptide. Thus Ark-mediated cell death requires caspase activity (Rodriguez, 1999).

Coexpression of an active-site C408A mutant of the fly apical caspase, Dredd [producing Dredd(C/A)], substantially attenuates cell killing triggered by Ark. In contrast, a comparable C211A mutation in the putative effector caspase drICE [producing drICE(C/A)] did not have similar effects even though it was prominently expressed. Therefore, Ark-mediated cell killing is generally not suppressed by the coexpression of mutant caspases, and the effect of Dredd(C/A) is specific. These data indicated that the Dredd mutant might exert a dominant-negative effect through a physical interaction with Ark. Whether Ark associates with Dredd was tested. A strong interaction between these proteins was detected when using either Ark(1-411) or the full-length protein. Similar tests with a comparable mutant form of drICE showed no evidence for an interaction between this caspase and Ark. These results do not address the question of whether a cofactor is necessary to regulate the Ark-Dredd interaction, since apoptotic SL2 cells may contain other proteins needed for their association. Nevertheless, Ark specifically interacts with the apical caspase Dredd but not with the effector caspase drICE. These data raise parallels to the binding observed between counterparts in the worm (CED-4 and CED-3) and in mammals (Apaf-1 and caspase-9) (Rodriguez, 1999).

Ark interacts with cytochrome c. Release of cytochrome c from the mitochondrial compartment and its association with Apaf-1 is a common feature of apoptosis in mammalian cells. In Drosophila, changes in cytochrome c also occur but the protein remains tethered to mitochondrial membranes, which are sufficient to trigger cytosolic caspase activation if isolated from apoptotic cells. Tests were performed to see whether Ark, like Apaf-1, might similarly associate with fly cytochrome c. Considerable levels of cytochrome c co-precipitate from Ark-expressing cells. To determine whether Ark's C-terminal WD domain might be important for this association, an identically tagged (3 x Myc) C-terminal truncation version of Ark(1-411) was tested. After transient transfection, substantial expression of Myc3-Ark(1-411) occurred but no co-immunoprecipitation with cytochrome c was observed. Therefore, a specific association of Ark with cytochrome c requires residues mapping between residues 412 and 1,440 in the Ark protein. These results indicate that Ark and Apaf-1 share homologous functions engaged by cytochrome c, and that the apoptosis-inducing activity of cytochrome c, through Apaf-1/Ark-like molecules, may be broadly conserved. Consistent with this idea, purified Drosophila cytochrome c was able to substitute for human cytochrome c in activating caspases via Apaf-1 in vitro. Future studies will determine the subcellular localization of Ark and the functional role of its interaction with cytochrome c (Rodriguez, 1999).

Ark loss-of-function mutations cause pleiotropic defects. To determine the function of Ark, mutants defective at this locus were isolated. Using a nearby P-element (P1041), a genetic screen was initiated to obtain loss-of-function Ark alleles. From ~700 transposition events, three were identified bearing a P-element insertion within the Ark locus. Genomic polymerase chain reaction (PCR) analysis reveals that these alleles [dark^CD4 (CD4), dark^CD8 (CD8) and dark^DD1 (DD1)] contain an insertion in the first intron and retain the original P1041 insertion upstream of the Ark promoter. Consistent with the differential severity of these alleles, CD4 and CD8 map several hundred base pairs downstream of DD1 (Rodriguez, 1999).

To characterize the nature of these mutations further, RNA was studied from CD4, CD8, DD1 and wild-type animals for Ark expression by Northern blot analysis. In the wild type, a single Ark transcript, migrating at ~5 kilobases (kb), was detected at all stages examined. Increased levels of Ark occur at a time coincident with the histolysis of larval tissues (during the third-instar and early pupation stages), whereas lowest expression occurs in embryos and adults. In CD4 animals, Ark messenger RNA is not detected even after long exposures of the blot, so this mutation represents either a null allele or a strong loss-of-function hypomorphic allele. By this criterion, CD8 is a hypomorphic allele because a reduced amount of transcript was found. A slightly larger-sized transcript occurs in DD1 animals, but no significant changes in the levels of Ark expression are seen, consistent with the idea that this insertion represents a weak Ark allele. Most animals homozygous for CD4 and CD8 survive to the adult stage, albeit at lower frequencies relative to heterozygous siblings. Many of these CD4/CD4 and CD8/CD8 animals derived from heterozygous parents exhibit melanotic tumours and abnormalities affecting the wings and/or bristles. Although these flies also have impaired viability and fecundity, progeny derived from homozygous parents suffers from notably increased penetrance and expression of similar defects. Substantial numbers of these F1 homozygotes are either sterile and/or die prematurely within several days. CD4 and CD8 animals show the same classes of phenotypes, albeit at different frequencies. Correlating with this molecular analysis, CD4 and CD8 show the strongest penetrance for all classes of abnormalities whereas DD1 animals (which express Ark RNA at wild-type levels) show milder defects that are limited to extra bristles (no defective wings or melanotic tissue are seen in these mutants). Transheterozygous combinations of the stronger CD4 mutation with the CD8 allele fail to complement, showing that lesions at Ark are the cause of the phenotypes observed (Rodriguez, 1999).

Wing abnormalities in CD4 and CD8 Ark homozygotes fall into different classes. The most affected individuals exhibit severe wing defects similar to a 'gnarled' and or 'wrinkled' phenotype. Other afflicted flies had wing blisters or burnt 'notched' wings. At moderate frequencies, melanotic tumors are also observed protruding from the body of the animal next to the haltere. Melanotic tumors in Drosophila are thought to arise from abnormalities in hematopoietic blood cells during larval growth or autoimmune defects. Among Ark mutants with extra bristles, most have one extra anterior scutellar macrochaetae while a minority have two extra such macrochaete and/or an extra posterior scutellar macrochaetae (Rodriguez, 1999).

Ark mutants are defective in programmed cell death. Neuronal PCD is important in the patterning of a functionally mature nervous system. For instance, Drosophila Df(3L)H99 embryos, which lack cell death, suffer from a greatly enlarged central nervous system (CNS), and mouse strains lacking Apaf-1 exhibit cell-death defects leading to hyperplasia of the nervous system. To determine whether the CD4 insertion also causes defects in the CNS, the brain lobes and ventral ganglion were dissected from wandering third-instar CD4/CD4 larvae. Most Ark mutants had a significantly overgrown CNS as compared with wild type. Occasionally, hyperplasia of the CNS was observed in the brain lobes only, while in other animals the ventral nerve cord (for example, the ganglion) was abnormally extended (Rodriguez, 1999).

Since the hyperplasia of the larval CNS might result from defective cell death earlier in development, the patterns of apoptosis were studied in CD4 embryos using the in situ TUNEL technique. Compared with wild-type embryos, CD4/CD4 homozygotes exhibit markedly reduced levels of apoptosis. Although there is a general decrease in TUNEL labelling throughout these embryos, the reduction in number of apoptotic cells is most noticeable within the CNS and the epidermis (Rodriguez, 1999).

Reduced apoptosis is most easily detected within the ventral nerve cord of Ark mutant embryos, but patterns of cell death are diminished in other tissues as well. For instance, whereas a large number of ventral epidermal cells normally undergo PCD shortly after germ-band shortening, Ark mutant embryos have a greatly reduced number of TUNEL-positive cells during this stage (Rodriguez, 1999).

Mutations in Ark suppress reaper-, grim- and hid-induced apoptosis. Since Ark embryos are defective in apoptosis, attempts were made to determine whether reaper, grim or hid signaling might require Ark function. Directed expression of transgenes expressing reaper (P[GMR-reaper]), grim (P[GMR-grim]) and hid (P[GMR-hid]) in the eye disc triggers ectopic apoptosis and, depending on expression levels, the effect is seen in adults as phenotypes that can range in severity from complete ablation of eye tissue to milder 'rough' eye phenotypes. Alterations in the magnitude of cell-death signaling resulting from such transgenes are an effective means of identifying genetic components that regulate apoptosis. Therefore cell-death phenotypes caused by P[GMR-reaper]-97A, P[GMR-grim]-1 and P[GMR-hid]-1M were examined in backgrounds that were either wild-type, heterozygous or homozygous for dark^CD4 (Rodriguez, 1999).

Overt suppression of cell killing was not evident in flies heterozygous for Ark. In contrast, substantial suppression was observed in homozygous individuals. The eyes of tester strains with two copies of P[GMR-grim]-1 were completely ablated, whereas homozygous Ark flies with two copies of P[GMR-grim]-1 retained substantial retinal tissue with many surviving ommatidia that were properly pigmented. Similarly, the P[GMR-hid]-1M phenotype was dramatically suppressed in Ark mutants and, compared with the wild-type tester strain, a large number of retinal cells that would have otherwise died persisted in the Ark background. Parallel tests with P[GMR-reaper]-97A uncovered a similar suppressive effect on reaper signalling. These results indicate that Ark functions as a pro-apoptotic effector of reaper-, grim-, and hid-induced cell killing (Rodriguez, 1999).

These data favor a shared evolutionary lineage for Ark, Apaf-1 and ced-4. Among these, the fly and mammalian genes share considerable homologies not found in ced-4. Therefore, Ark and Apaf-1 probably share the most recent common ancestry. Accordingly, the WD-repeat domain, which may represent the site of engagement by mitochondrial signals, could be a recent acquisition or, alternatively, this domain may have been lost from the worm protein. However, although these are the possibilities suggested at present, 'truer' orthologs of ced-4 in flies and mammals may yet be discovered. Functional studies also support a more recent lineage shared by Ark and Apaf-1. Ark, Apaf-1 and CED-4 each associate with apical caspases, but only the fly and mammalian counterparts are known to associate with cytochrome c, through a WD-repeat domain. Moreover, deletion of this region in Ark and Apaf-1 produces enhanced effects upon cell killing. Therefore, analogous to the scenario that has been proposed for Apaf-1 function, these results indicate that the association between cytochrome c and Ark may function to derepress an inhibitory effect imposed on Ark by the WD-repeat domain. One implication of these results is that an ancient mitochondrial circuit for propagating apoptotic signals (binding of cytochrome c to a WD-repeat domain) is preserved in insects and thus probably existed in a common ancestor of insects and mammals (Rodriguez, 1999).

The identification of Ark also raises important mechanistic questions. For instance, the C. elegans Bcl-2 homolog CED-9 can directly bind and repress the activity of CED-4, and analogous interactions can occur between Bcl2-family members and Apaf-1. However, it is not yet known whether Bcl-2 proteins represent an 'obligate' component of the apoptosome; future studies on Ark and its regulators may shed light on this issue. It will also be interesting to determine whether Ark can engage and/or function to activate other fly caspases (Rodriguez, 1999).

Some aspects of the Ark mutant phenotype suggest striking parallels to those reported for Apaf-1-deficient mice. In both cases, the CNS appears to be preferentially affected and a decrease in apoptosis leads to hyperplasia of this tissue. Although PCD is overtly compromised in darkCD4 embryos, clearly not all deaths are affected. Thus, certain apoptotic deaths may not use an Ark-dependent pathway; similar inferences have been drawn from study of the Apaf-1-deficient mouse embryos (where partial PCD suppression also occurs). Since it is formally possible that darkCD4 is not a null mutation, an absolute requirement for Ark activity in all PCDs cannot be excluded. However, the fact that no lethal complementation groups or stronger phenotypes have been mapped to Ark argues that other CED-4/Apaf-1-like molecules with partially redundant functions may be present in flies (Rodriguez, 1999).

Although the existence of an enlarged CNS can be easily reconciled with reduced apoptosis, the precise origins of other abnormalities are less obvious. Nevertheless, conspicuous parallels to phenotypes associated with mutations in caspases and other cell-death regulators in the fly are apparent. For example, larvae homozygous for loss-of-function mutations in the caspase gene DCP-1 also suffer from melanotic tumours, and dredd-deficient flies have wing and bristle abnormalities. Bristle defects are also associated with mutations at the Drosophila inhibitor of apoptosis-1 gene and wing defects are associated with mutation of a hid allele. Some of these defects (such as melanotic tumors) could arise from the incomplete histolysis of larval tissue during metamorphosis but, for the most part, many defects appear to be confined to tissues derived from the wing disc. It will be interesting to determine whether defects occur in tissues derived from other imaginal tissues (Rodriguez, 1999).

Other questions to be addressed relate to the role of Ark as an effector of signaling by the death activators reaper, grim and hid. Since these proteins can activate both pro-Dredd processing and the accumulation of dredd RNA in cells that are specified to die, it will be interesting to determine the precise role of Ark in these pathways. For example, it should be possible to test directly whether the apoptosis activators function through Ark to trigger Dredd processing. Ark functions as a pro-apoptotic effector of ectopic cell killing by reaper, grim and hid. However, the effect of these cell-death activators is not entirely abolished in Ark homozygotes and, therefore, Ark-independent apoptosis pathways, also triggered by reaper, grim and hid, are implicated. Further investigation into the function of Ark as it relates to signaling by reaper, grim and hid is now possible within the context of normal PCD and cell-type-specific fates (Rodriguez, 1999).

GENE STRUCTURE

cDNA clone length - 4676

Bases in 5' UTR - 77

Exons - 11

Bases in 3' UTR - 275

PROTEIN STRUCTURE

Amino Acids - 1440

Structural Domains

A TBLASTN search of the high-throughput genomic sequence (HTGS) database using the first 600 amino acids of human Apaf-1 protein has identified a segment of Drosophila genomic DNA that encodes a stretch of homology to the 'CED-4' domain. The predicted exon sequences of this DNA segment were used to isolate the full-length complementary DNA encoding Ark. Comparison of the cloned cDNA to the genomic sequence identified 11 exons spanning 6.5 kilobases. Ark, Apaf-1, and CED-4 share a homologous CED-4 domain of ~320 amino acids in length, but Apaf-1 and Ark also contain another region of homology of an extra 70 amino acids. The CED-4-domain homologies include the Walker A and B loops thought to mediate essential ATP-driven functions and an extra conserved six-amino-acid stretch that is conserved between Apaf-1 and CED-4. Analysis using the PSA sequence-analysis server also predicts the presence of at least eight complete WD repeats in Ark. A region of Apaf-1 containing its last WD repeat could be aligned with 31% identity and 49% similarity to a 74-amino-acid area of Ark that contains its last WD repeat. Ark contains two groups of WD repeats separated by more than 200 amino acids in an arrangement similar to that in Apaf-1. The presence of incomplete WD repeats, and tertiary-structure predictions (made using the PSA server) indicative of ß-propeller structures over incomplete WD repeats from residues 700 to 950, indicates that extra WD folds not detected by the computer algorithm may exist (Rodriguez, 1999).

Apaf-1 bears an amino-terminal CARD domain thought to be important for caspase recruitment and oligomerization, and similar profiles of CED-4 uncover 'subsignificant' similarities to this motif. The N terminus (amino acids 1-91) of Ark shows moderate similarities to both proteins, and identical and/or similar residues tend to cluster at important core hydrophobic sites predicted from the solution structure of a CARD. More important, like its counterparts in both worm and mammal, the minimal region of Ark necessary for caspase interaction (residues 1-411) includes this N-terminal 'CARD-like' domain (Rodriguez, 1999).

The N terminal CARD domain shares significant homology with APAF-1 (20% identity and 41% similarity for amino acids 65-486) and the plant disease-resistant proteins (e.g., 19% identity, 34% similarity for amino acids 147-462 in comparison to RPR1, GenBank accession number AB019186). Like APAF-1, the Drosophila protein also has nine WD repeat elements following the CED-4 homology region. The sequence homology between Ark and CED-4 is more limited and mainly restricted to the two nucleotide-binding motifs and the ARC domain (van der Biezen, 1998). Nevertheless, the biochemical and functional data indicate that this homology is significant (Zhou, 1999).

Using a cDNA as a probe, two clones, dapaf-1L and dapaf-1S, were isolated from a Drosophila S2 cell cDNA library. dapaf-1L encodes a protein of 1440 amino acids, and dapaf-1S produces a 531-amino acid protein by alternative splicing using the stop codon at the end of exon 5 in the Ark genomic region. Each form of Ark has one putative nucleotide-binding site (GXXGXGK) and several related motifs, NB-ARC motifs and 16 WDRs, found in the C-terminal portion of Dapaf-1L (Kanuka, 1999b).

In comparing Ark with Apaf-1 and CED-4 using a routine BLAST search, the amino-terminal region of Ark does not show any sequence homology with them. However, close inspection of the sequence reveals that the amino-terminal region of Ark is closely related to a CARD. The putative secondary structure and hydrophobicity pattern of Ark are similar to those of RAIDD and other CARD proteins, and the core residue, which is also critical for the function of CED-4 (L27), is also conserved as a similar residue (V32) in Ark. These similarities with Apaf-1/CED-4, especially the fact that Dapaf-1L contains CARD, NB-ARC, and WDR motifs, suggest that Ark may be a member of the Apaf-1/CED-4 family. Structural features of the two Ark isoforms suggest that Dapaf-1L corresponds to mammalian Apaf-1, and Dapaf-1S to C. elegans CED-4 (Kanuka, 1999b).

The Drosophila Apaf-1 related killer (Dark) forms an apoptosome that activates Dronc, an apical procaspase in the intrinsic cell death pathway. To study this process, a large Dark complex was assembled in the presence of dATP. Remarkably, cytochrome c was found not to be required for assembly and when added, cytochrome c does not bind to the Dark complex. A 3D structure of the Dark complex was assembled at 18.8A  resolution using electron cryo-microscopy and single particle methods. In the structure, eight Dark subunits form a wheel-like particle and two of these rings associate face-to-face. In contrast, Apaf-1 forms a single ring that is comprised of seven subunits and each Apaf-1 binds a molecule of cytochrome c. Relevant crystal structures were then used to model the Dark complex. This analysis shows that a single Dark ring and the Apaf-1 apoptosome share many key features. When taken together, the data suggest that a single ring in the Dark complex may represent the Drosophila apoptosome. Thus, this analysis provides a domain model of this complex and gives insights into its function (Yu, 2005).

The data lend support to the idea that cytochrome c may not be a physiological activator of Dark. Since cytochrome c does not bind to the regulatory region of Dark, it appears that an additional role must exist for the WD40 repeats. Previous work has implicated the WD40s in maintaining an inactive conformation of Apaf-114 and 48 and a similar observation has been made for Dark.20 Thus, the data suggest that the role of cytochrome c may be subsumed by an undiscovered co-factor, which activates the Dark monomer in vivo. The analysis also suggests that beta-propellers in Apaf-1 and Dark may have been added to a CED4-like molecule, to help regulate assembly before the split between chordates and arthropods. Hence, the ability to bind cytochrome c may have occurred in chordates at a later stage in evolution, or it may have been lost in arthropods (Yu, 2005).

Until recently, the role of nucleotide hydrolysis in apoptosome assembly was not fully understood. It now appears that assembly starts with the Apaf-1 monomer in a dATP bound state, and one cycle of dATP hydrolysis occurs during cytochrome c binding, to help trigger a conformational change from a compact monomer to a more open form. This Apaf-1 monomer may then exchange bound dADP with free dATP, to create a second dATP bound conformation that is competent for assembly. While Apaf-1 has a slightly higher affinity for dATP than ATP, it was found that ATP and a non-hydrolyzable analog (AMP-PCP) can substitute for dATP in promoting assembly of the human apoptosome (Yu, 2005).

The lack of a reliable assembly system for Dark has precluded similar studies on the nucleotide requirements for this system. This is due in part to the fact that a physiological activator has not been identified. Remarkably, the experiments indicate that Dark assembly has a strict requirement for dATP over ATP. However, a large excess of dATP is needed to drive Dark assembly in the absence of an activator. It is surmised that the high concentration of dATP may have promoted nucleotide exchange, thereby triggering assembly of the Dark monomer. These considerations suggest that differences in the nucleotide state of purified Dark might explain the variable assembly observed in different preparations when using a lower concentration of dATP (Yu, 2005).

Previous analysis of the Apaf-1 apoptosome revealed a central CARD ring and suggested that the procaspase-9 binding surface would be oriented towards the top of the central hub Thus, the central CARD ring of the Dark apoptosome was modeled in a similar way. This model predicts that binding sites for Dronc will be located between the Dark apoptosomes in the double ring. Hence, these binding sites would be inaccessible, rendering the double apoptosome inactive. However, it is suggested that Dark may bind Dronc during assembly and concentrate this zymogen on the CARD ring to activate it. Importantly, this process would likely block formation of the double ring. The structures of the Apaf-1 and Dark apoptosomes also suggest that the number of subunits in these platforms may not be a critical factor in their ability to bind and activate procaspases. Further studies are now required to investigate Dronc activation and to ascertain whether Dronc will promote the assembly of a single Dark apoptosome (Yu, 2005).

date revised: 10 October 2001

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.