Establishing and maintaining a morphogen gradient are important in the growth and patterning of developing organs. When a discontinuity in a morphogen signal gradient is created by somatic mutant clones with aberrant intensities of morphogen signals within the Drosophila wing disc, the clones can be removed by apoptosis to restore the morphogen signal gradient. This apoptosis is termed “morphogenetic apoptosis” and has been observed to occur in a cell autonomous or non-cell autonomous manner. This review discusses possible molecular mechanisms of both autonomous and non-cell autonomous apoptosis in addition to similar cellular events in reference to recent findings.
Morphogen is a secretory and diffusible molecule that is produced in a localized source and spreads across immature organs, forming a concentration gradient for morphogenesis (Tabata & Takei 2004; Affolter & Basler 2007). Within developing organs, the morphogen gradients create a primitive pattern of target genes expression in a concentration-dependent manner. The formation and function of morphogen gradients have been extensively studied in the Drosophila wing imaginal disc, an organ that is the precursor of the adult wing and composed of a sac-like epithelium. The simple structure of the wing disc and sophisticated genetic tools available in Drosophila have enabled us to better understand morphogens in the wing disc, including Decapentaplegic (Dpp, a transforming growth factor [TGF]-beta superfamily protein; Lecuit et al. 1996; Nellen et al. 1996), Wingless (Wg, a Wnt family protein; Zecca et al. 1996; Neumann & Cohen 1997), and Hedgehog (Hh, a Hh family protein; Heemskerk & Dinardo 1994). Establishing and maintaining proper morphogen signal gradients during development are essential for morphogenesis. If the morphogen signal gradients are disrupted so that they become discontinuous, several solutions are available for removing the abnormal cells. One safeguard mechanism that has been reported is morphogenetic apoptosis (Adachi-Yamada & O’Connor 2002, 2004).
In this review, morphogenetic apoptosis is discussed in comparison with similar phenomena by using the phenotypes in mutant for Dpp type I receptor Thick veins (Tkv) as an example, and possible regulatory mechanisms are illustrated.
Morphogenetic apoptosis occurs when relatively mild but prolonged discontinuities in the morphogen signal gradients are produced between neighboring cells. In the wing disc, a morphogen Dpp is expressed in a narrow belt of cells in the anterior compartment along the anteroposterior compartment boundary. Dpp spreads from the production source, forming a concentration gradient across the disc. Clones mutant for tkv with decreased Dpp signaling in the medial region of the wing disc (near the Dpp source) are removed by apoptosis, while clones expressing activated Tkv with higher Dpp signaling in the lateral region of the wing disc (far from the Dpp source) also undergo apoptosis (Adachi-Yamada & O’Connor 2002). Since both apoptotic responses seem to be reasonable for restoring the morphogen signal gradient to a normal shape, this process was dubbed morphogenetic apoptosis. It is interesting that the cells in the developing wing disc are not simply affected by the value of their own morphogen signal activity but rather their survival or apoptotic fate also depends on their neighbor’s morphogen signal activity. Thus, this response can be interpreted as a safeguard mechanism that ensures smooth morphogen gradient formation in the developing field, at least several cells in width.
The morphogenetic apoptosis is mediated by c-Jun N-terminal kinase (JNK), a member of the mitogen-activated protein kinase (MAPK) superfamily. In Drosophila, as in mammals, JNK signaling is involved in many biological processes such as cell morphogenesis (Jacinto et al. 2002), wound healing (Galko & Krasnow 2004; Martin & Parkhurst 2004), cell proliferation (Ryoo et al. 2004), and apoptosis (Adachi-Yamada & O’Connor 2004; Kanda & Miura 2004). JNK was proven to be necessary for all of these processes, which raises the possibility that JNK acts to enhance various cellular signaling processes. Furthermore, in Xenopus oocyte and mammalian culture cells, JNK molecules in each cell were all activated or not activated at all in response to various stressful stimuli. Thus, JNK has been suggested to support various cellular signaling processes as an all-or-none factor (Bagowski & Ferrell 2001; Bagowski et al. 2003). This will be discussed in a later section. Accordingly, JNK activation in the apoptosis described above due to abnormal morphogen levels should be restricted to within the area of mutant clones prior to apoptosis. However, in morphogenetic apoptosis, JNK activation as well as subsequent cell death is observed autonomously in the abnormal cell population and non-cell autonomously in the surrounding cells (Fig. 1). The following sections discuss the molecular induction mechanisms of both autonomous and nonautonomous JNK activation in the morphogenetic apoptosis in comparison with similar apoptotic or homeostatic phenomena.
Mechanisms of cell autonomous apoptosis
In the wing disc, Dpp signaling is thought to be required for cell survival because cells mutant for Dpp signaling components or cells with reduced Dpp signaling activities are removed by apoptosis (Burke & Basler 1996; Adachi-Yamada et al. 1999; Moreno et al. 2002). Additionally, experimental observations suggest that Dpp signaling is required for epithelial integrity and that clones with low levels of Dpp signaling are basally extruded from the columnar epithelium of the wing disc (Gibson & Perrimon 2005; Shen & Dahmann 2005). Similar morphological phenotypes can often be seen in mammalian skins as epidermoid cysts. These clones with low levels of Dpp signaling frequently undergo JNK-mediated apoptosis, suggesting that Dpp acts as a survival factor, at least in part, by maintaining epithelial integrity.
How does aberrant Dpp signaling lead to JNK activation? Recent studies suggest that a Rho family small GTPase, Rho1 (the RhoA homologue in Drosophila), provides a link between Dpp signaling and JNK activation. Dpp signaling promotes the columnar cell shape in the wing disc through the regulation of the subcellular distribution of Rho1 (Widmann & Dahmann 2009). In the medial region of the wing disc, where Dpp signaling is active, increased accumulation and activation of Rho1 are observed in the apicolateral subcellular region, which is responsible for elongating and maintaining the columnar epithelium. Furthermore, strong activation or inactivation of Rho1 elicits JNK signaling and induces apoptosis (Neisch et al. 2010). Rho1 is co-immunoprecipitated with upstream components of the JNK pathway, including Slipper, Tak1, Hep, and POSH, suggesting that Rho1 activates JNK signaling by forming a complex with them (Neisch et al. 2010). Therefore, autonomous apoptosis of clones with aberrant levels of Dpp signaling can be attributed to Rho1-mediated JNK activation. While the strongest activation of JNK and retraction in the height of the columnar cells are not observed in cells with weak Dpp signaling, they can be found in the cells around the discontinuities in the Dpp signal levels, which results in folding formation along the edge of the mutant clones, as discussed later.
Another possible explanation for autonomous JNK activation is the mechanism of cell competition. Cell competition is a similar apoptotic phenomenon by which a cell population with a reduced growth rate disappears in the background of a cell population with a higher growth rate (see review article by Morata in this issue). Moreno et al. (2002) reported that the cell competition observed in the cell population heterozygous for the Minute(2)60E (M(2)60E) mutation, which reduces ribosomal protein levels and therefore proliferates slowly, was caused by reduced Dpp signaling activity, which provides a survival ability to cells under normal conditions (Moreno et al. 2002). This is because cell death in M(2)60E heterozygous cells is suppressed when combined with a mutation in brinker, a transcriptional repressor on the Dpp signal-mediated transcription (Moreno et al. 2002). Interestingly, this cell death looks cell autonomous in contrast to morphogenetic apoptosis. Thus, only the slow-growing cells are removed when there is contact between two different kinds of cell populations.
Recently, two evolutionarily conserved proteins involved in the process of cell competition were identified. One is a transmembrane protein called Flower (Fwe) (Rhiner et al. 2010), and the other is a transcriptional corepressor known as dNAB (Ziv et al. 2009). Fwe has three isoforms that were named Fweubi, FweLose-A, and FweLose-B. Fweubi is ubiquitously expressed independently of cell death induction, while FweLose-A and FweLose-B are exclusively expressed in the outcompeted cells. When fwe was knocked down by RNAi, cell competition was no longer observed. In contrast, when FweLose-A or FweLose-B is artificially induced in mosaic clones, the cells undergo apoptosis. It is thought that the contact of Fweubi-expressing cells and FweLose-expressing cells is crucial for cell death induction in the latter cells.
dNAB is expressed in the prospective wing blade region in the normal wing disc. However, during cell competition, the level of dNAB is increased, which is required for apoptosis. dNAB is also thought to bind to Brinker to induce JNK activation for subsequent apoptosis.
Mechanisms of non-cell autonomous apoptosis
An intriguing characteristic of morphogenetic apoptosis is that, in addition to mutant clones, the surrounding wild-type cells undergo apoptosis. How is this non-autonomous apoptosis induced? One possible set of factors is the leucine-rich repeat (LRR) family of transmembrane proteins, Capricious (Caps) and Tartan (Trn), for which transcription is negatively regulated by Dpp signaling so that both genes are expressed in the lateral region of the wing disc due to low levels of Dpp signaling (Milan et al. 2002). Clones ectopically expressing spalt, which is a target gene of Dpp signaling, are removed by apoptosis in the lateral region. At that time, the expressions of caps and trn are thought to be repressed within the spalt-expressing clones, as is the case for clones with increased Dpp signaling activities. However, coexpression of cap or trn rescues the spalt-expressing clones in the lateral region, suggesting that the cells without LRR in the field of normal LRR-expressing cells result in apoptosis.
Another LRR family transmembrane protein, Fish-lips (Fili), is similarly involved in cell survival (Adachi-Yamada et al. 2005). Clones expressing spineless (the mammalian dioxin receptor homologue in Drosophila) induce a homeotic transformation of organ identity, from wing to leg or antenna, resulting in apoptosis through JNK activation. In this case, the cell autonomy of apoptosis and JNK activation show mirror-image symmetry in accordance with the distance from the dorsoventral (DV) boundary. This suggests that cell autonomy is affected by Wg, a morphogen that is expressed at the DV boundary and diffuses along the dorsoventral axis. Fili shows a typical expression pattern regulated by Wg signaling, and its loss in the spineless-expressing transformed clones induces non-cell autonomous JNK activation and apoptosis. Together with these findings, the loss of LRR transmembrane proteins between adjacent cells could account for the non-autonomous apoptosis observed in morphogenetic apoptosis.
Another proposed mechanism of the non-autonomous induction of apoptosis is mechanical stress in the surrounding cells. As mentioned above, clones with reduced Dpp signaling are basally extruded from the wing epithelium (Gibson & Perrimon 2005; Shen & Dahmann 2005). These extruded mutant clones might produce shear stress in the neighboring cells, which might activate JNK and subsequent apoptosis (Gibson & Perrimon 2005). However, the presence of such mechanical force around the mutant clones is still unproven, and cells may change their apicobasal positions without the occurrence of mechanical stress by relocalizing various cell junctional proteins on their cell surfaces.
Interestingly, Fwe, the above-mentioned transmembrane protein responsible for cell competition, might also be required for non-cell autonomous apoptosis. This is because when the clones artificially expressing FweLose are induced, apoptotic cells are observed frequently in the outermost part of the clones but not uniformly throughout the clones (Rhiner et al. 2010). Thus, the function of Fwe may involve regulation of non-cell autonomous apoptosis. Indeed, the authors observed an elevation of FweLose expression levels in the clones mutant for tkv. The detailed mechanism will be clarified by the authors.
One recent interesting observation about such nonautonomously induced apoptosis was the discovery of this type of cell death in normal development. During normal pupal leg development, nonautonomous apoptosis was found around the discontinuities in Dpp signaling gradients to sculpt the leg epithelium for multiple joint formation (Manjon et al. 2007). Thus, nonautonomous apoptosis is also useful in such particular cases of normal development without any aberrant or invasive stimuli.
JNK signaling as an all-or-none switch
Clones with reduced Dpp signaling activity can induce various responses such as cell competition (Moreno et al. 2002), morphogenetic apoptosis (Adachi-Yamada & O’Connor 2002), basal extrusion (Gibson & Perrimon 2005; Shen & Dahmann 2005), and folding formation (Umemori et al. 2007; Fig. 2). Folding formation in the imaginal disc epithelia occurs along the border where two types of cell populations with different levels of morphogenetic signaling activities artificially contact each other. However, similar folding can also be observed during normal development such as between the wing blade and hinge regions. Therefore, it is thought that the epithelial folding is generally created to separate a new or narrower secondary field for morphogenesis for later development, as observed in the leg disc (Kojima et al. 2000).
Two factors, JNK and Dpp signaling activities, are considered to be important in determining which response occurs when there is unusual contact of cell populations with different morphogen signals. The higher basal activity of JNK prior to occurrence of mutant cells means that it may prefer cell death as a consequence rather than a change in epithelial morphology (Adachi-Yamada & O’Connor 2002; Gibson & Perrimon 2005; Shen & Dahmann 2005). In contrast, the severity of the discontinuity in the Dpp signaling gradient may affect the area of the response. For example, milder tkv mutant clones grow more due to a loss in autonomous apoptosis or basal extrusion observed in more severe tkv mutant clones. This results in either nonautonomous apoptosis or folding formation. In fact, prolonged contact between mutant and wild-type cells is likely to induce nonautonomous apoptosis (Adachi-Yamada & O’Connor 2004). Furthermore, the response at an earlier stage seems to restrict the choice at a later stage. For example, if mutant cells do not disappear at an early stage, they continue to proliferate in an extruded cyst and no longer choose an apoptotic fate later (Gibson & Perrimon 2005). In this situation, although basal JNK activity throughout the wing disc seems to be lower, local JNK activity around the boundary between the cyst and the underlying epithelium can easily be observed (Gibson & Perrimon 2005), suggesting that JNK also plays a role in the induction of the cyst among normal epithelial cells.
JNK plays a similar role in the generation of the ectopic epithelial folding in the optomotor-blind mutant (Umemori et al. 2007). Folding formation and basal extrusion seem to be the two distinct final results derived from the same primary cause, apicobasal cell retraction. If the cell retraction occurs before clone growth, the entire clone undergoes basal extrusion. In the opposite situation, cell retraction is observed only along the clone boundary, as seen in clones with increased Dpp signaling activity (Shen et al. 2010).
Thus, JNK is apparently a key component common to all four of the above responses. How can a single kind of signal transducer, JNK, strongly govern different responses? This question may be common to various cell-signaling transduction pathways. One answer may be a particular trait of JNK: it shows an all-or-none activation, as seen in Xenopus oocyte (Bagowski & Ferrell 2001) and mammalian cells (Bagowski et al. 2003). Although the entire cell population displays a graded activation in response to stimuli, all or none of the JNK molecules are activated for a single cell. This characteristic of JNK signaling indicates that the role of JNK may be, depending on the cellular context, to ensure that a stimulus sufficiently elicits or does not elicit each output as an expression of bistability (Bagowski et al. 2003).
Generality in nonautonomously induced morphogenetic apoptosis
In morphogenetic apoptosis, cells on either side of the discontinuity in the morphogen signaling gradient undergo apoptosis. Clones of cells with artificially high or low levels of Dpp signaling can survive if they are in an appropriate region where the surrounding cells activate similar levels of Dpp signaling. Thus, cell survival is regulated by not only the condition of the cell itself but also the situation of surrounding cells. This can be interpreted to mean that each cell is monitored by its neighbor cells and that each cell’s survival depends on its neighbors. This would be an excellent mechanism for eliminating aberrant cells, thereby achieving robust patterning of the distribution of various types of cells in the developing field.
Even in cases where morphogens are not involved in such cell distribution processes, similar nonautonomous cell death induced by unusual contact between different types of cells could be a general mechanism in development and homeostasis. In fact, in the above-mentioned case of cells with homeotic transformation, this kind of nonautonomously induced apoptosis contributes to the removal of aberrant cells. As shown in Figure 3, undefined organ identities due to loss of expression of master genes such as Pax6 (eyeless (ey) in Drosophila) leads to massive apoptosis (Gehring 2004). Accordingly, the adult flies mutant for ey have a tiny head without the presence of undefined organs at the position of their compound eyes. Furthermore, conflicting organ identity specification due to artificially combined expressions of master genes also induces apoptosis. For example, coexpression of wing-directing vestigial and leg-directing spineless leads to a loss of appendage in a wide region (Adachi-Yamada et al. 2005). These two types of apoptosis occur autonomously. In contrast, mosaic formation by cell populations with distinct organ identities shows both autonomous and nonautonomous cell death similar to morphogenetic apoptosis, which seems to be the only way to remove abnormal organ primordia. This is because, throughout the development with organs of homeotic transformation, abnormalities can be detected only at the boundary between two types of organs. For example, the “existence” of wing and leg cells should not indicate an abnormal state since normal flies always possess them both. However, the “contact” between both types of cells should always be an alert signal because this state does not appear in normal flies. Since neither wing nor leg cells recognize which type of cells is abnormal, apoptosis induction on both sides of the organ boundary may be reasonable. Thus, nonautonomous induction of apoptosis is considered to be another important mechanism, as well as the well-characterized autonomous apoptosis, for detecting and repairing abnormal development and will be found in a wide variety of organ developments in multicellular organisms. A report has already been published referring to non-autonomous cell death after transplantation of liver tissue between different genotypes in rats (from wild types to mutants for dipeptidyl peptidase IV) (Oertel et al. 2006).