The caspases, a family of cysteine proteases, function as central regulators of cell death. Recently, caspase activity and caspase substrates identified in the absence of cell death have sparked strong interest in caspase functions in nonapoptotic cellular responses; these functions suggest that caspases may be activated without inducing or before apoptosis, thus leading to the cleavage of a specific subset of substrates. This review focuses primarily on the caspase enzymatic activity. Detailed genetic analyses of caspase-deficient Caenorhabditis elegans, Drosophila, and mice have shown that caspases are essential, not only for controlling the number of cells involved in sculpting or deleting structures in developing animals, but also for dynamic, nonapoptotic cell processes, such as innate immune response, tissue regeneration, cell-fate determination, stem-cell differentiation and neural activation. Our understanding of the spatio-temporal caspase activation mechanisms has advanced, primarily through the study of Drosophila developmental processes. This review will discuss current findings regarding caspase functions in cytoskeletal modification, morphogenetic regulation of cell shape, cell migration and the production of mechanical force during embryogenesis.
The fundamental cellular response of apoptosis, or programmed cell death, plays a crucial role in shaping developing bodies and in regulating tissue homeostasis by eliminating unwanted cells. Dysregulation of the cell death mechanism has detrimental consequences, including tumorigenesis, autoimmunity, exencephaly, neurodegenerative disease, hematopoietic deficiencies and infertility (Vaux & Korsmeyer 1999; Ranger et al. 2001). However, precisely how cell death functions to the benefit or detriment of the organism is still largely unknown, and elucidating the mechanisms that regulate cell death is of obvious importance.
Cell death pathways are well conserved in all metazoans, from invertebrates to vertebrates. Caspases, a family of cysteine proteases that are highly conserved in multicellular organisms, function as central regulators of cell death. It has been suggested that caspases are involved in development not only through cell death in the process of sculpting and deleting structures, but also as dynamic regulatory molecules in immunity, cell differentiation and cell morphology (reviewed in Kuranaga & Miura 2007; Kuranaga 2011; Miura 2011).
Classical caspase signaling machinery
In Caenorhabditis elegans, apoptosis in germ cells is mediated by the same core cell death machinery that controls apoptosis in somatic cells (Conradt & Xue 2005), involving the activity of CED-3 caspase, the CED-3 activator CED-4 and the cell death inhibitor CED-9 (Ellis & Horvitz 1986; Yuan et al. 1993). Three additional C. elegans caspase-related genes, csp-1 (caspase homologue-1), csp-2 and csp-3, have also been identified (Shaham 1998). These proteins are similar in sequence to caspase proproteins. CSP-2 and CSP-3 function to block CED-3′s autoactivation via their sequences that are similar to both of CED-3′s catalytic subunits, thereby preventing apoptosis in somatic cells (Geng et al. 2009).
In mammals, caspase-1 reported as a CED-3 homologue was interleukin-1β-converting enzyme (ICE), a cysteine protease responsible for proIL-1β secretion and processing (Cerretti et al. 1992; Thornberry et al. 1992; Miura et al. 1993). Thus, the first mammalian CED-3 homologue was described with both apoptotic and nonapoptotic functions.
The regulatory functions of caspases after apoptotic stimulation have been studied in detail. The core machineries of the caspase-activating pathway are conserved throughout evolution (Fig. 1) (Schafer & Kornbluth 2006). Although CED-3 is the sole caspase required for programmed cell death in C. elegans, multiple caspases are required for the apoptosis in more complex organisms such as flies and mammals. This evolutionary expansion of the caspase family may have arisen around the dual purposes of executing apoptotic cell death and carrying out a number of other cellular processes not necessarily related to cell death. Caspases are synthesized as zymogens that must be cleaved at select aspartate residues for activation. Inactive caspase is initially processed by separating large (p20) and small (p10) subunits, after which the N-terminal domain is removed to form a catalytically active protease (Degterev et al. 2003).
Caspases can be classified into initiator and executioner caspases. Initiator caspases have a long N-terminal prodomain, which mediates the formation of protein complexes that provide the molecular platform for caspase activation and inhibition. Initiator caspases cleave a few specific substrates, including executioner caspase zymogens. This cleavage activates the executioner caspases, which in turn cleave their respective substrates, eliciting apoptotic cell death, with its characteristic morphological features of membrane blebbing, pyknotic nuclei, cell rounding and apoptotic vesicle formation (Clarke 1990).
Caspase activation can be regulated through an extrinsic or intrinsic signaling pathway (Schafer & Kornbluth 2006). The extrinsic pathway, which involves Fas and TNFR stimulation, activates caspase-8. The intrinsic pathway, which may be the primary means of activating apoptotic caspase in mammals, triggers the mitochondrial release of cytochrome c (cyt-c), which oligomerizes with Apaf-1 and procaspase-9 to form the apoptosome complex. Activated caspase-9 in this complex activates caspase-3 to execute apoptosis.
It has been suggested that the basic biochemical features of apoptosis are also present in Hydra and in sponges. Hydra, a member of the ancient metazoan phylum Cnidaria, has been used extensively as a model organism for investigating the evolutionary origins of metazoan processes. The molecular mechanisms leading to apoptosis in Hydra are surprisingly extensive and are comparable to those in mammals (David et al. 2005). Genome-wide sequence analysis has shown the presence of caspases, Bcl-2 family members, Apaf-1, IAPs and components of a putative death receptor pathway (Lasi et al. 2010). The presence of potential initiator caspases with CARD, DED and DD domains in Hydra suggests it has both extrinsic and intrinsic caspase activation mechanisms similar to those known in mammals. Wiens et al. (2003) identified caspase sequences in the marine sponge Geodia cydonium, indicating that caspases are also present in the Phylum Porifera. Because Hydra and sponges are the earliest multicellular animals, these findings suggest a possible correlation between the origin of multicellularity and the origin of apoptosis.
Some initiator caspase nonapoptotic functions, such as that of caspase-1 in inflammation, are well characterized. Although executioner caspases are considered apoptotic because of their strong enzymatic activity and relatively broad substrate specificity, recent studies have also pointed out the importance of their nonapoptotic functions. These findings indicate that although caspases are prominently associated with apoptosis, they are also dynamically active in the compensatory proliferation of neighboring cells, cell-fate determination, stem-cell maintenance, tissue regeneration, actin-cytoskeleton reorganization (which impacts cell shape and migration) and in the immune system (Kuranaga & Miura 2007; Yi & Yuan 2009). Although caspases can signal independently of their enzymatic activity through protein–protein interactions, as in NF-κB activation (Lamkanfi et al. 2007), this review will focus mainly on the enzymatic activity and functions of caspases.
Nonapoptotic caspase functions
Caspases in defense mechanisms
Caspase-1 (interleukin-1β-converting enzyme, ICE) was first identified as a unique cysteine protease that processes proIL-1β to mature IL-1β (Cerretti et al. 1992; Thornberry et al. 1992). Caspase-1 activation promotes IL-1β secretion by macrophages in response to various bacterial compounds, viral infections or endogenous molecules released from damaged macrophages (Nadiri et al. 2006). As caspase-1 also processes IL-18 and IL-33, it is thus responsible for both inflammatory and innate immune responses.
The caspase-1-activating complex, called the inflammasome (Nadiri et al. 2006; Schroder & Tschopp 2010), is composed of caspase-1 and the apoptosis-associated speck-like protein containing a caspase-recruitment domain (ASC), in addition to nucleotide-binding oligomerization domain (NOD)-like receptor family (also referred to as the Apaf-1-like protein family) members such as NLRP1 (Defcap/Nac/Card7), NLRP3 (Cryopyrin/Cias1/Pypaf1) and Ipaf/Card12, which can sense various bacteria, toxins or endogenous danger signals (e.g., monosodium urate crystals) released from damaged cells (Mariathasan et al. 2004, 2006; Martinon et al. 2006; Sutterwala et al. 2006). Thus, caspase-1 activation in macrophages may be an important cellular response to infection or cellular damage sensed by the Apaf-1-like components (Ogura et al. 2006). Inflammasome caspase-1 activation is required not only for inflammation, but also for protecting injured cells. For example, bacterial pore-forming toxins simultaneously induce K+ efflux and activate both NLRP3 and Ipaf inflammasomes (Gurcel et al. 2006); activated caspase-1 then promotes lipid metabolism by activating sterol regulatory element-binding proteins (SREBPs) to repair toxin-damaged membranes. Recent identification of a ROS-dependent NLRP3 ligand showed several molecular events that may direct inflammasome activation. NLRP3 agonists trigger the association of NLRP3 with thioredoxin-interacting protein (TXNIP) in human macrophages, and this association is suppressed by inhibiting ROS (Zhou et al. 2009). Furthermore, knocking down or deleting TXNIP suppresses the NLRP3-agonist-triggered caspase-1 activation and IL-1β secretion in human or mouse macrophages (Zhou et al. 2009), and knocking down the TXNIP inhibitor thioredoxin augments inflammasome activation in human macrophages (Dostert et al. 2008).
Caspase-1 regulates not only inflammatory cytokine maturation, but also the unconventional secretion of leaderless proteins such as fibroblast growth factor-2 (FGF-2) and thioredoxin-1 (Keller et al. 2008). Because many caspase-1-dependent secreted proteins are involved in inflammation, cyto-protection and tissue repair, caspase-1 may play a role in reconstructing stressed or damaged tissues.
Caspases in tissue regeneration
Caspases can integrate the processes of cell death and cell proliferation to shape regenerating tissue. Animal tissues such as the Drosophila larval imaginal disk have a striking regenerative capacity not only during development, but also in adulthood. Although large numbers of cells undergo cell death when these tissues are injured, the tissue often heals with the same size and shape as the uninjured tissue, owing to cell proliferation that compensates for the cell loss; this phenomenon is called compensatory proliferation. A full-sized mammalian liver, for example, can be regenerated after 75% of the organ has been removed (Lesurtel et al. 2006). Another example is found in the Drosophila wing imaginal disk, a larval monolayer epithelium that develops into the adult wing; in response to irradiation-induced cell death, cells adjacent to apoptotic cells undergo extra cell proliferation, resulting in adult wings of nearly normal size (Haynie & Bryant 1976; Brockes & Kumar 2008). Ectopic toxin expression results in locally induced apoptosis accompanied by increased cell proliferation around the apoptosis site, suggesting that cells can perceive apoptosis in their vicinity and undergo cell division until the original cell number is restored (Milan et al. 1997).
When the Drosophila cell-death-inducing gene reaper (rpr) or head involution defective (hid) is expressed in the imaginal disk along with p35, a potent inhibitor of effector caspases such as DrICE and Dcp-1, the signaling cascade is activated only up to the initiator caspase Dronc, and the cells fail to undergo apoptosis. This creates an excess of ‘undead’ cells that ectopically express Wingless (Wg) and Decapentaplegic (Dpp), resulting that neighbors of ‘undead’ cells undergo overproliferation (Huh et al. 2004a; Perez-Garijo et al. 2004, 2005; Ryoo et al. 2004). Dronc is required for compensatory proliferation in this experimental system, suggesting that it is involved in inducing mitogen expression (Huh et al. 2004a; Kondo et al. 2006; Wells et al. 2006). However, in differentiating eye tissues in Drosophila, undead apoptotic cells induce compensatory proliferation by up-regulating Hedgehog (Hh), not Wg or Dpp (Fan & Bergmann 2008). Therefore, caspases may coordinate both cell death and compensatory proliferation during development and regeneration.
During the development of the adult Drosophila abdomen, preexisting larval epidermal cells (LECs) are eliminated by apoptosis and are replaced by histoblasts, the adult precursor cells (Madhavan & Madhavan 1980). The replacement of larval abdominal epidermis with adult epithelium in Drosophila pupae is a simple model of tissue remodeling. Histoblasts undergo rapid proliferation and expansion, and LECs subsequently undergo a caspase-mediated apoptotic process controlled by an intriguing nonautonomous relationship with proliferation or cell cycle progression in histoblasts (Ninov et al. 2007; Nakajima et al. 2011). Inhibiting caspase in LECs severely delays LEC delamination and significantly retards the progression of histoblast nest expansion and the proliferation of histoblast cells, suggesting that caspase activation in LECs positively participates in tissue remodeling.
The link between the caspase signal and cellular proliferation in regeneration has also been illustrated in Hydra. During basal head regeneration after midgastric bisection, cells from the interstitial lineage immediately undergo cell death. These apoptotic cells provide an immediate but transient source of Wnt3 that activates the Wnt-β-catenin pathway in the surrounding cycling cells, which rapidly divide (Chera et al. 2009). In Drosophila and Hydra, both caspase activity and the extent of apoptosis play roles in tissue regeneration, but the caspase substrates that induce compensatory proliferation remain to be investigated.
Caspase-related apoptosis has also been implicated in increased stem-cell activity in the Drosophila midgut. Homeostatic turnover that replaces the cells of the entire midgut takes 2–3 weeks in healthy flies (Jiang et al. 2009). However, when epithelial cell loss is induced by bacterial infection, experimental injury or caspase signal activation, compensatory proliferation by intestinal stem cells (ISCs) regenerates the entire midgut within 2–3 days (Amcheslavsky et al. 2009; Buchon et al. 2009, 2010; Jiang et al. 2009, 2011). This increased ISC activity can be at least partially blocked by p35, which suggests that it can be induced by the executioner caspases DrICE and Dcp-1.
The mechanisms leading to increased compensatory ISC activity are less clear. It has been reported that dying enterocytes in the midgut produce the IL-6-like cytokine unpaired (Upd) in a JNK-dependent manner, which stimulates JAK-STAT signaling and leads to ISC proliferation, thereby driving gut epithelium renewal (Buchon et al. 2009, 2010; Jiang et al. 2011). Although the crosstalk mechanisms between JNK and caspase signaling are complicated, the upstream mechanism activating JNK also activates caspase under stress-inducing stimulation (Kuranaga et al. 2002). Therefore, it is possible that JNK and caspase signaling integrate under stressful conditions to promote compensatory ISC activity for gut regeneration. These processes may in fact reflect one of the primary functions of caspases, i.e., to restore homeostasis after tissue damage by linking apoptotic cell death with the induction of tissue repair.
Caspases in cell-fate determination and stem-cell differentiation
Caspase activation during apoptosis ultimately disrupts the nuclear structure. Some cells lose their nuclei during differentiation, which can be considered a specialized form of apoptosis. Such cells include keratinocytes, megakaryocytes, erythrocytes and lens cells, and it is not surprising that caspases are involved in the terminal differentiation of these cells (Lamkanfi et al. 2007). Caspases can mediate irreversible signal transduction through substrate cleavage, a type of signal regulation that may be suitable for determining cell fate (Kuranaga 2011). Caspase-8 cleaves receptor-interacting protein (RIP), a death-domain-containing kinase that regulates NF-κB, thus down-regulating NF-κB during macrophage differentiation (Rebe et al. 2007). Sterile twenty-like kinase (MST1), which is required for muscle differentiation, is activated when it is cleaved by caspase-3 (Fernando et al. 2002). In Drosophila, macrochaetes, external sensory organs typical of the Drosophila peripheral nervous system, are located on the notum. Four large macrochaetes are observed on the wild-type scutellum. However, one extra macrochaete often appears on each side of the scutellum in dark-mutant flies (Kanuka et al. 1999; Rodriguez et al. 1999). Both dark mutants and flies expressing dominant-negative Dronc have an extra sensory organ precursor (SOP) cell on each side, showing that caspase activation is involved in controlling SOP cell formation in the scutellum area (Kanuka et al. 2005). Furthermore, the increased number of SOP cells in caspase-inhibited wing disks is not a result of the artificial survival of cells that otherwise would have undergone apoptotic cell death, but is rather because of inhibited apoptosis-independent caspase activity (Kanuka et al. 2005). Genetic screening identified a novel substrate cleaved by the Dark-dependent caspase DrICE, the Shaggy46 protein. This isoform is encoded by the shaggy gene, a Drosophila ortholog of gsk-3β, and is essential for the negative regulation of Wg signaling. Cleaving Shaggy46 converts it to the active isoform Shaggy10, which contributes to SOP cell formation (Fig. 2) (Kanuka et al. 2005).
Caspase enzymatic activity is critical in stem-cell differentiation. Nonapoptotic caspase-3 activation has been observed in several stem-cell lineages, including embryonic stem cells (ESCs), hematopoietic stem cells (HSCs) and neural stem cells (Miura et al. 2004). Caspase-3-deficient mice show significant bone defects during early development; the alteration of the TGF-β/Smad2 pathway and cell cycle progression decreases the osteogenic differentiation of bone marrow stromal stem cells (BMSSCs), resulting in delayed ossification and decreased bone mineral density (Miura et al. 2004). Based on the premise that stem-cell differentiation corresponds with a loss of capacity for self-renewal, two recent studies explored the role of caspase-3 as a probable gatekeeper of stem-cell function (Fujita et al. 2008; Janzen et al. 2008). Stem cells lacking caspase-3 had marked defects in differentiation (Fujita et al. 2008). Fujita et al. (2008) reasoned that caspase-3 mediates its effects through the targeted cleavage of a pluripotent factor and showed Nanog to be a priority target substrate, because the expression of a caspase-3-resistant Nanog promoted ESC self-renewal while inhibiting differentiation. Janzen et al. (2008) reported that caspase-3 loss resulted in an accumulation of phenotypically defined long-term repopulating HSCs, with a corresponding reduction in circulating mature hematopoietic cells. Cytokine-mediated signals were also elevated in caspase-3-deficient HSCs. A recent report indicates that inhibiting caspase-3 or -8 in human fibroblast cells partially or completely prevents iPSC induction, and the transient activation of caspase-8 is sufficient to enhance the frequency of iPSC formation (Li et al. 2010a). These results indicate that caspases may be key nuclear reprogramming facilitators in iPSC induction.
Caspase in neural activation
Caspase-3 is also dynamically involved in neural activation. In the zebra finch auditory forebrain, birdsong exposure increased the concentration of activated caspase-3 in the caudomedial nidopallium (NCM), and inhibiting caspase-3 activity in the NCM during song training disrupted habituation memory development (Huesmann & Clayton 2006). In this case, activated caspase-3 was observed in both the NCM and the unstimulated regions of the brain, but was normally bound to the inhibitor of apoptosis protein (XIAP) to prevent unwanted cell death (Huesmann & Clayton 2006). These results support a mechanism in which active caspase-3 is always present in living cells, but is sequestered by its inhibitors and released only transiently for essential nonapoptotic functions. Consistent with the caspase-dependent memory consolidation in zebra finches, a recent study showed that synaptic depression and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor internalization in mouse hippocampal neurons requires caspase-9 and caspase-3/7 activity and can be blocked by over-expression of the anti-apoptotic proteins Bcl-xL and XIAP (Li et al. 2010b). Li et al. showed that stimulating NMDA receptors transiently activates caspase-3 without causing cell death, and long-term synaptic depression cannot be induced by stimulating NMDA receptors in hippocampal slices from caspase-3 knockout mice (Li et al. 2010b).
It was recently reported that a caspase-3-dependent mechanism drives synaptic failure and contributes to cognitive dysfunction in Alzheimer’s disease. Caspase-3 activates calcineurin in dendritic spines, in turn triggering the dephosphorylation and removal of the AMPA receptor GluR1 from postsynaptic sites (D’Amelio et al. 2011). These molecular modifications alter glutamatergic synaptic transmission and plasticity and correlate with spine degeneration and deficits in hippocampal-dependent memory.
Caspase activation mechanisms in nonapoptotic settings
Temporal regulation of caspase activation
The existing literature implicates caspases and their regulators in nonapoptotic functions as described above, raising compelling questions as to how caspase activity is controlled to prevent apoptosis. Evidence is for at least two mechanisms by which a cell can safely activate caspase temporarily for nonapoptotic functions.
Mechanisms to control caspase activation are essential for maintaining cell integrity, and much of this function is carried out by the inhibitor of apoptosis proteins (IAPs), originally found in baculoviruses (Fig. 2). DIAP1, which contains a carboxy-terminal RING finger domain and functions as an E3 ubiquitin ligase, suppresses caspase activation by binding directly to caspases and promoting their degradation (Wilson et al. 2002) or nondegradative inactivation (Ditzel et al. 2008). The proapoptotic proteins Rpr, Hid and Grim promote this DIAP1 degradation during the periods of programmed cell death (Hays et al. 2002; Ryoo et al. 2002; Wilson et al. 2002; Yoo et al. 2002). Programmed cell death is initiated when caspases are released from DIAP1 inhibition. Therefore, the balance between the DIAP1 protein level and caspase activation determines whether cells will survive or die by apoptosis.
A genetic modifier screen for genes that regulate caspase activation identified a novel caspase activation regulator that determines the threshold of caspase activation by regulating DIAP1 turnover (Kuranaga et al. 2006). Drosophila IKK-related kinase (DmIKKε) is a homologue of the noncanonical members of IκB kinase (IKKε/IKKι or NAK/T2K/TBK1), which regulate the activation of NF-κB or interferon regulatory factor (IRF)-3 and -7 in mammals (Kawai & Akira 2006). DmIKKε determines the level of DIAP1 by modifying its phosphorylation; ectopic DmIKKε expression causes DIAP1 phosphorylation and degradation. NAK/TBK1/T2K, the mammalian homologue of DmIKKε, phosphorylates mammalian XIAP, a potent caspase inhibitor. NAK/TBK1/T2K expression promotes XIAP phosphorylation and degradation under low-serum culture conditions, thus suggesting that the IKK-related kinase function in IAP phosphorylation and degradation is conserved in Drosophila and mammalian cells (Kuranaga et al. 2006). DmIKKε may maintain caspase activity at the threshold required for nonapoptotic functions as a determinant of the DIAP1 protein level. This is consistent with the observation that knocking down DmIKKε leads to an increase in SOP cells, which phenocopies the dark mutant (Kuranaga et al. 2006).
Another mechanism that restrains or shuts down caspase activity under such nonapoptotic conditions is the Drosophila inhibitor of apoptosis protein 2 (DIAP2), which controls the level of caspase activity in living cells (Fig. 2). Although diap2-deficient cells remain viable in vivo, they have increased DrICE activity and are sensitized to cell death after treatment with sublethal X-irradiation doses. DIAP2 forms a covalent adduct with the catalytic portion of DrICE, and it requires a functional RING finger domain to target DrICE for ubiquitination and block cell death. These data suggest that DIAP2 controls caspase in a nonapoptotic context by efficiently interacting with activated DrICE to prevent cell death (Ribeiro et al. 2007).
Spatial regulation of caspase activation
Some cells appear to have mechanisms that resist caspase-mediated cell death even in the face of high caspase activation levels. One mechanism involves sequestering caspase activity in specific subcellular regions (Arama et al. 2003; Huh et al. 2004b; Kuo et al. 2006; Williams et al. 2006). To shape neuronal architecture and neural circuits, excessive dendrites and axonal projections must be eliminated. During insect metamorphosis, larval neurons undergo massive loss and reconstruction of dendrites and axons without cell death. This dendritic pruning requires the ubiquitin-proteasome system and ubcD1, an E2 ubiquitin-conjugating enzyme involved in degrading the caspase-inhibitory molecule DIAP1 (Ryoo et al. 2002; Kuo et al. 2006). DIAP1 prevents activation of the initiator caspase Dronc, and UbcD1 activation is likely to lead to DIAP1 degradation and, subsequently, to Dronc activation. A loss-of-function dronc mutant prevents dendritic pruning, and its dendrites are labeled by antibodies that detect activated caspase-3-like activity (Kuo et al. 2006; Williams et al. 2006). Dendritic pruning resembles apoptosis in that it includes cytoskeletal disruption (dendritic), fragmentation and the cleanup of cellular fragments by phagocytic cells. The localized caspase activation in dendrites may induce apoptotic changes that affect only parts of cells, without executing cell death.
The mechanisms allowing localized caspase activation for dendritic pruning while preventing cell death have recently been elucidated. One study suggested that the molecular subcellular destruction mechanisms found in developmental neurite pruning and injury-induced neurite degeneration (Wallerian degeneration) are similar (Nikolaev et al. 2009; Schoenmann et al. 2010). Nikolaev et al. (2009) found that caspase-6, but not caspase-3, was required for axonal pruning in cultured dissociated mouse DRG neurons after trophic-factor deprivation. Schoenmann et al. (2010) showed that increasing NAD+ and inhibiting either caspase-3 or caspase-6 in axons efficiently blocked the expression of both caspase-3 and caspase-6 and prevented axonal degeneration. Moreover, Schoenmann et al. (2010) showed that the expression in Drosophila of mouse Wallerian degeneration slow (Wld) protein, which consists mainly of the full-length sequence for the NAD+ biosynthetic enzyme Nmnat1, suppresses dendritic pruning in sensory neurons, and that caspase is activated even in dendrites that escaped degeneration. Therefore, caspases and the NAD+-sensitive pathway, which operate in parallel to execute the degeneration process, may cooperate in evolutionarily conserved neurite destruction systems during development and after injury.
Spermatid individualization provides another example of localized caspase activation. During sperm differentiation in Drosophila, the 64 haploid spermatids of each cyst are connected by cytoplasmic bridges; these bridges are subsequently eliminated, and most of the cytoplasm is expelled to form the individual sperm. This process is termed individualization. Immunoreactivity for activated caspase-3 is detected in the individualization complex (IC), a cytoskeletal membrane complex that moves along the length of the cyst to the sperm tail (Arama et al. 2003; Huh et al. 2004b). Although dark and dronc mutants fail to complete individualization (Arama et al. 2006), mutants of the initiator caspase dredd and its adaptor molecule dfadd, which can activate Dredd, have partial defects in individualization (Huh et al. 2004b), as does a mutant of the executioner caspase drice (Muro et al. 2006). The Drosophila genome contains two cytochrome c genes, cyt-c-d and cyt-c-p. A cyt-c-d mutation also causes defects in IC caspase activation and in spermatid individualization (Arama et al. 2003, 2006). Arama et al. (2006) indicated that only cyt-c-d is required to activate caspase during spermatid differentiation, whereas cyt-c-p is required for respiration in the soma (Fig. 2). Whereas cytochrome c is crucial in apoptosome-mediated caspase activation in mammals, the existence of a comparable cytochrome c function in Drosophila is still controversial. Cytochrome c is not required to activate caspase in stress-induced apoptosis in Drosophila (Dorstyn et al. 2004); however, specifically cyt-c-d, and not the somatic cyt-c-p, activates caspase in retinal cell death (Arama et al. 2006; Mendes et al. 2006). It was also indicated that cyt-c-d functions in SOP development, because cyt-c-d loss leads to an extra-bristle phenotype mediated by imbalanced caspase activation (Arama et al. 2006).
Arama et al. (2007) found that a testis-specific Cullin-3-based E3 ubiquitin ligase complex is required for caspase activation in spermatids (Fig. 2). In a similar experiment, Kaplan et al. (2010) identified Soti, which inhibits the Cullin-3-based E3 ubiquitin ligase complex required for caspase activation during spermatid terminal differentiation; they found that Soti competes with an inhibitor of the apoptosis protein dBruce (a target of the E3 complex) to bind Klhl10, the E3 substrate recruitment subunit (Fig. 2). Interestingly, Soti is expressed in a subcellular gradient within spermatids, which in turn promotes the formation of a similar dBruce gradient, with the result that caspase is activated in an inverse gradient (Fig. 2) (Kaplan et al. 2010). These findings provide insight into how specific caspase regulation processes promote caspase-dependent differentiation while preventing cell death.
Cytoskeletal modification by caspase activation
Caspases in morphogenetic signaling
A subset of the morphological changes associated with cell death may be related to normal changes in cells during development, because cell death was originally described as a series of morphological changes. The unique morphological changes that define apoptosis include condensation of the cytoplasm, breakdown of nuclear integrity, cell rounding, membrane blebbing, and, in epithelial cells, loss of cell polarity and cell junction subsequent to caspase activation (Brancolini et al. 1997; Coleman et al. 2001; Sebbagh et al. 2001). Caspase-3 cleaves rho-associated kinase-1 (ROCK-1) to create an activated form of the kinase, which in turn regulates cytoskeletal rearrangement to create the apoptotic-specific cell shape (Chang et al. 2006).
The shape of the Drosophila antenna arista, a terminal segment of the antenna, is because of specific cell patterns and morphologies. Mutant alleles of either diap1 or hid cause an abnormal arista pattern. The thread1 mutant (loss-of-function diap1 mutation) lacks aristal branching and shows excessive cell death restricted to the antennal imaginal disk (Cullen & McCall 2004). Moreover, the hidWR+X1 mutant (loss-of-function hid mutation) has large, hairy aristae, suggesting the possibility that the level of apoptosis determines the number of branches in the arista (Cullen & McCall 2004); however, it has not been clarified how bristle branching might be regulated by apoptotic events. As Hid antagonizes DIAP1′s caspase-inhibitory function, another hypothesis is that the arista morphology is controlled by nonapoptotic caspase activity. DmIKKε, a DIAP1-degrading kinase, negatively regulates F-actin polymerization (Oshima et al. 2006). Dominant-negative DmIKKε expression causes an excess-branching phenotype that is suppressed by DIAP1 knockdown and enhanced by DIAP1 over-expression. A mild reduction in the expression of dronc and its activator dark by RNAi enhances the lateral branching phenotype caused by dominant-negative DmIKKε (Oshima et al. 2006). Although this phenotype was not observed when p35 was expressed, each single mutant for drice, dcp-1 and dronc showed excessive aristal branching, suggesting that caspase signaling contributes to the shaping of normal aristae in Drosophila (Fig. 2) (Muro et al. 2006; Baum et al. 2007).
DIAP1 protein metabolism is critical for temporal and quantitative caspase control, as it has a RING finger domain and functions as an E3 ubiquitin ligase. Koto et al. (2009) focused on precisely how the balance of caspase roles in cell death and nonapoptotic functions is maintained. They examined the protein turnover of the endogenous caspase inhibitor, DIAP1, which they monitored in the external SOP lineage of living Drosophila with a fluorescent probe, PRAP (PRe-Apoptosis signal detecting probe based on DIAP1 degradation). The SOP divides asymmetrically to make the shaft, socket and sheath cells, and then the neurons that innervate each sensory organ. They found that the DIAP1 quantity changes dramatically depending on the cell type and maturity. The physiological significance of DIAP1 dynamics during sensory organ development was then studied by manipulating DIAP1 levels in SOP linage cells. Knocking down DmIKKε delayed DIAP1 degradation and caused the shorter, thicker bristle phenotype in the adult notum. However, either DIAP1 knockdown or Rpr over-expression resulted in bristle loss. These results suggest that the temporal regulation of DIAP1 turnover determines whether caspases function nonapoptotically in cellular shaping or cause cell death (Koto et al. 2009).
Caspases in cell migration
During Drosophila oogenesis, a group of follicle cells known as border cells migrate to the center of the developing egg chamber; this phenomenon provides an excellent model for studying cell migration in vivo. Dominant-negative Rac expression inhibits border cell migration, and the over-expression of DIAP1, but not of p35, rescues the migration defect. Although p35 can suppress the executioner caspases DrICE and Dcp-1, it does not inhibit the initiator caspase Dronc. Dominant-negative Dronc expression suppresses the migration defects caused by dominant-negative Rac. These results suggest that Dronc inhibits border cell migration downstream of Rac (Geisbrecht & Montell 2004). DmIKKε, a kinase that promotes DIAP1 degradation, inhibits border cell migration (Oshima et al. 2006). This is consistent with the observation that DIAP1 positively regulates border cell migration by down-regulating Dronc.
In in vitro cell motility assays, caspase-8-null mouse embryonic fibroblasts (MEFs) are motility-deficient, whereas caspase-3-null MEFs are not. Cell motility requires calpain activity, which is lower in caspase-8-null cells than in wild-type cells. Calpain mediates Rac activation, and caspase-8-null cells activate Rac inefficiently (Helfer et al. 2006). Caspase-8-null MEFs are defective in lamellipodia generation, which is also initiated by Rac activation. Therefore, caspase-8 affects calpain-mediated cellular migration processes (Helfer et al. 2006). Although caspase-8 promotes motility in MEFs, a loss of caspase-8 expression occurs in metastatic neuroblastoma in mice, and the expression of caspase-8 suppresses metastases (Stupack et al. 2006). Therefore, the caspase-8 loss seen in certain tumors may facilitate tumor invasion. These observations warrant further investigation of the molecular mechanisms of caspase-mediated cell migration regulation.
Caspase contributes to mechanical force in tissue morphogenesis
Apoptotic epithelial cells quickly lose cell–cell contact and are actively extruded by neighboring cells. A ring of actin and myosin forms not only within the apoptotic cell, but also in the surrounding cells; contraction of the ring formed in neighboring cells is required to extrude the apoptotic cell, and injecting a Rho GTPase inhibitor into these cells completely blocks the extrusion in mammalian cells (Fig. 4A) (Rosenblatt et al. 2001). At the point at which apoptotic cells lose contact with neighboring cells, the progressive destruction of the adherens junction might be to coordinate the elimination of dying cells, thus restoring epithelial organization. Kessler & Muller (2009) showed that Armadillo/β-catenin, which is a major regulator of cadherin-mediated adhesion, is proteolytically cleaved by DrICE, so that DE-cadherin is removed from the plasma membrane during apoptosis in Drosophila (Fig. 3). These apoptosis-related morphological changes are correlated with tissue movement and tension.
Apoptotic force, an active mechanical function of apoptosis, has been showed during epithelial cell sheet fusion in Drosophila embryogenesis (Fig. 4B) (Toyama et al. 2008). Genetic manipulation and imaging analyses showed that proteins that suppress or enhance caspase activity in the amnioserosa, an extraembryonic tissue, cause the speed of dorsal closure to decrease or increase, respectively. It is possible that mechanical force produced by caspase activation during apoptosis is not only to force dying cells out of tissues, thus preserving tissue integrity, but also to change the morphology of neighboring cells to fill the space originally occupied by dying cells.
The possible role of caspase in dynamic organogenesis was recently shown in Drosophila. The organogenesis of Drosophila male terminalia requires apoptosis; apoptosis mutants have an orientation defect phenotype. The developing male terminalia normally rotates 360°, and the orientation defect is caused by incomplete rotation (Gleichauf 1936; Adam et al. 2003). As adult male terminalia orientation defects are observed in mutants of cell death pathway components, including hid, drice and dronc, caspase has been thought to have an important role in this organogenesis (Abbott & Lengyel 1991; Muro et al. 2006; Krieser et al. 2007). Time-lapse imaging showed that rotation of the genitalia accelerates as development progresses, finally completing a full 360° rotation in normal flies. However, this acceleration is impaired in caspase-inhibited flies (Suzanne et al. 2010; Kuranaga et al. 2011). Interestingly, the timing and extent of apoptosis correlates strongly with the start of rotational acceleration. Up-regulating the caspase signal increases the speed of genitalia rotation, suggesting that caspase might contribute to the production of a driving force to complete the male genitalia organogenesis within a limited developmental time window (Fig. 4C) (Kuranaga et al. 2011). The underlying mechanisms of caspase-mediated mechanical force during organogenesis require further investigation.
Caspases transduce irreversible or long-lasting signaling through substrate cleavage. Although strong caspase activation determines a cell fate of apoptosis, localized caspase activation, the caspase activation level and caspase substrate specificities appear to determine immunity and cellular behavior. Caspases must therefore have acquired multiple activation mechanisms in the process of animal evolution to control their unique signal-transducing roles in both apoptotic and nonapoptotic processes. Although the caspases’ nonapoptotic signaling mechanisms are more controversial, this aspect of their activity merits a careful examination and interpretation of published data. However, the involvement of prominently apoptotic caspases in cellular differentiation, immune and inflammatory responses and cytoskeletal modification shows that the therapeutic inhibition of caspase activity may have broader implications than initially conceived. A better understanding of caspase functions may lead to novel pharmaceutical strategies to prevent inflammation, oncogenesis or autoimmune diseases.
I apologize to colleagues whose work could not be cited because of space limitations. I especially thank M. Miura and H. Kanuka for the generous suggestion and helpful discussion and to all the members of Prof. Miura’s laboratory for valuable discussions. Studies by our group were supported in part by grants from the Takeda Science Foundation, the Uehara Memorial Foundation, the Kanae Foundation for the Promotion of Medical Science, and the Japanese Ministry of Education, Science, Sports, Culture and Technology.