Mitogenic signaling from apoptotic cells in Drosophila


Author to whom all correspondence should be addressed.


Apoptotic cells of Drosophila not only activate caspases, but also are able to secrete developmental signals like Hedgehog (Hh), Decapentaplegic (Dpp) and Wingless (Wg) before dying. Since Dpp and Wg are secreted in growing tissues and behave as growth factors, it was proposed that they play a role in compensatory proliferation, the process by which a growing blastema can restore normal size after massive apoptosis. We discuss recent results showing that there is normal compensatory proliferation in the absence of Dpp/Wg signaling, thus indicating it has no significant role in the process. Furthermore, we argue that Dpp/Wg signaling is not a resident feature of apoptotic cells, but a side effect of the necessary activation of the JNK pathway. Nevertheless, the ectopic JNK/Dpp/Wg signaling may have an important role in tissue regeneration. Recent work in other organisms suggests that paracrine signaling from apoptotic cells may be of general significance in wound healing and tissue regeneration in metazoans.


Apoptosis is a form of cellular suicide in which cells activate the machinery for their own destruction. It is a conserved phenomenon that plays several major biological roles in all animal cells. There is a developmentally regulated apoptosis, necessary for the normal morphogenesis of the Drosophila larval head and of the leg joints of the adult (Lohmann et al. 2002; Manjon et al. 2007), and also for the control of cell number (Bergmann et al. 2002). In addition, it plays a safeguard role aimed to eliminate supernumerary, abnormal or malignant cells that may appear in development or during the life of an individual (Adachi-Yamada et al. 1999; Moreno et al. 2002; Igaki et al. 2009; Menendez et al. 2010). Because of the important functions it performs, the regulation of apoptosis is critical for the wellbeing of the organisms, for excess or defect of apoptosis has grave consequences (Jacobson et al. 1997; Hanahan & Weinberg 2000; Danial & Korsmeyer 2004).

The molecular and genetic mechanisms implicated in Drosophila apoptosis are well known (see a recent review in Steller, 2008): the executive role is performed by some cystein proteases, called caspases, which destroy the cell substrates. The function of caspases is normally kept in check by the product of the diap1 gene, which is active in all cells and is therefore essential for cell survival. The triggers of apoptosis are a group of genes (reaper, head involution defective, grim and sickle, termed pro-apoptotic genes), whose products bind to and inactivate the Diap1 protein (Wang et al. 1999; Goyal et al. 2000; Ryoo et al. 2002; Yoo et al. 2002), thus allowing caspases to function. In turn, the pro-apoptotic genes can be activated by various developmental or stress stimuli (Brodsky et al. 2000; Milan et al. 2002; Perez-Garijo et al. 2004, 2009).

One unexpected property of the apoptotic cells in Drosophila is that they activate ectopically the signaling genes decapentaplegic (dpp), wingless (wg) and hedgehog (hh) (Ryoo et al. 2004; Fan & Bergmann 2008). These genes are known to play major morphogenetic and growth control roles in Drosophila development (Lawrence & Struhl 1996; Tabata & Takei 2004), suggesting that this property may have an important biological significance. In this review we will focus on the phenomenon of ectopic Dpp/Wg signaling by apoptotic cells, describe the principal observations and discuss the implications.

Ectopic Dpp/Wg and Hh signaling in apoptotic cells

The discovery of Dpp/Wg signaling was originally made in a special type of apoptotic cells, the “undead” cells. These are artificially generated apoptotic cells in which the entire apoptotic pathway is active except that the effector caspase activity is blocked. A convenient property of undead cells is that they remain alive indefinitely, while showing many of the genetic features of apoptotic cells. It permits us to study aspects of the apoptotic pathway that are hard to examine in normal apoptotic cells, as these die soon after the initial stimulus. However, the features described for undead cells also apply to normal apoptotic cells.

Experimentally, undead cells are generated by inducing apoptosis in cells in which the baculovirus protein P35 blocks the function of the effector caspase (Hay et al. 1994). Apoptosis may be induced by removing the diap1 gene in clones of cells (Ryoo et al. 2004), by expressing pro-apoptotic genes (Wells et al. 2006), or by stress stimuli, like X-rays or heat shock (Perez-Garijo et al. 2004, 2009). The presence of P35 in cells homozygous for mutations causing cell lethality, for example lgl, also results in undead cells (Fig. 1A).

Figure 1.

 Induction of non-autonomous proliferation and activation of dpp and wg by undead cells. (A–A′′) Overexpression of p35 in lethal giant larvae (lgl) mutant clones (green) induces active proliferation in non-mutant cells close to the clones, as indicated by their high level of BrdU incorporation (white). (B–B′′) Disc of genotype hh-Gal4 > UAS-p35 UAS-GFP fixed 3 days after irradiation (1500R). Note the ectopic expression of both dpp (red) and wg (blue) in areas of the posterior compartment (labelled green by GFP). Note also in the inset that wg and dpp are co-expressed.

One significant property of undead cells is that they induce extra proliferation in the normal cells close to them (Perez-Garijo et al. 2004; Ryoo et al. 2004; Huh et al. 2004) (Fig. 1A). The non-autonomous effect on proliferation is linked to another feature: their acquisition of ectopic activity of the signaling genes dpp and wg (Fig. 1B). This was an unexpected finding that was reported almost simultaneously by three different groups (Huh et al. 2004; Perez-Garijo et al. 2004; Ryoo et al. 2004). Although more easily observed in undead cells, it was also shown that dpp and wg are also active in “genuine” apoptotic cells (Perez-Garijo et al. 2004; Ryoo et al. 2004).

Apart from the functional significance of this phenomenon, that we discuss below, the ectopic expression of wg and dpp in individual isolated cells points to a special activation mechanism. This is particularly intriguing in the case of dpp, which becomes expressed in apoptotic cells of the posterior wing compartment. Normally dpp is activated only in the anterior cells close to the A/P border. In the posterior compartment cells the presence of the Engrailed protein prevents dpp activation by the Hedgehog protein (reviewed in Lawrence & Struhl 1996 and Tabata & Takei 2004). However, the ectopic activation of dpp in the posterior compartment cells occurs in the presence of the Engrailed product (Martin et al. 2009) indicating that it is activated by a novel mechanism. In the case of wg it was found that its activation requires activity of the JNK pathway (Ryoo et al. 2004).

Dpp and Wg are not the only signals that can be released by the apoptotic cells. Induction of apoptosis in the differentiating cells posterior to the morphogenetic furrow in the eye disc causes the emission of the Hh signal (Fan & Bergmann 2008). Interestingly, unlike the proliferating cells anterior to the furrow, these cells secrete neither Wg nor Dpp. Moreover, the activation mechanism of hh appears to be different from that of dpp or wg. The latter requires normal activity of the apical caspase Dronc and occurs in the absence of function of the effector caspase Drice (Huh et al. 2004; Perez-Garijo et al. 2004; Ryoo et al. 2004). In contrast, hh induction requires normal Drice activity (Fan & Bergmann 2008).

All together the previous findings establish the ability of apoptotic cells to secrete developmental signals and also the versatility of the process, for they are able to generate different signals depending on the cellular context.

The secretion of paracrine signals by apoptotic cells is not a peculiarity of Drosophila. Recent work indicates that it also occurs in evolutionary distant organisms as Hydra and the mouse (Chera et al. 2009, Li et al. 2010). The functional significance of these findings is discussed below.

Persistent apoptotic status of undead cells

One particularly interesting feature of undead cells is that they maintain the apoptotic status indefinitely. This is especially clear in experiments in which the initial apoptotic stimulus was a brief irradiation or a heat shock, lasting from a few minutes to 3 h, given to late first or early second instar larvae (Perez-Garijo et al. 2004, 2009). A large proportion of imaginal cells from those larvae still show all the apoptotic markers several days after the end of the stimulus (Fig. 2). These markers include the Hid protein, the caspases Dronc and Drice, the Dpp and Wg signals and the function of the JNK apoptotic pathway.

Figure 2.

 Sustained pro-apoptotic activity in undead cells. Wing discs from irradiated larvae of genotype hh-Gal4 > UAS-p35 UAS-GFP were fixed 96 h after the end of the irradiation. (A–A″) Cells in the posterior compartment (marked in green) still express the pro-apoptotic markers Drice (blue) and Hid (red). (B–B″) wing disc of the same experiment showing cells expressing wg (blue) and dronc (red). (C–C″) Another disc from the same experiment showing high levels of ectopic wg (blue) and JNK expression (red). The latter was monitored by the activity of the puc-LacZ insert.

The long persistence of the apoptotic status suggests the existence of a maintenance mechanism. This is also supported by the observations of Wells et al. 2006; demonstrating a feed back process in the apoptotic pathway of undead cells. These authors showed that the p53 gene not only functions inducing rpr activity (Brodsky et al. 2000), but it is itself activated downstream rpr by a non-apoptotic role of dronc. This results in further activation of other pro-apoptotic genes. Thus, p53 appears to be a part of the feed back loop that maintains the activity of the apoptotic program and the associated Dpp/Wg signaling. Wells et al. 2006 argued that this phenomenon could be related with the additional cell divisions necessary for compensatory proliferation.

Some results concerning the pro-apoptotic activity of the JNK pathway also suggest a maintenance loop in undead cells. While JNK is able to induce the pro-apoptotic genes (Mcewen & Peifer 2005; Luo et al. 2007), it can also be activated downstream dronc (Kondo et al. 2006) and is required to induce dpp and wg. Thus, JNK induction downstream dronc possibly contributes to maintain the persistent pro-apoptotic status of undead cells. Whether this loop is a general feature of apoptosis or it is a just a characteristic of the undead cells remain to be elucidated.

Developmental consequences of the persistent Dpp/Wg signaling in undead cells. Role of the JNK pathway

The continuous emission of Dpp and Wg by the undead cells has developmental consequences, as these signals affect the behaviour of the neighbor non-apoptotic cells inducing in them additional proliferation. The result is an excess of growth, which is particularly clear when an entire compartment is affected (Fig. 2C). It is worth emphasizing that these hyperplastic overgrowths can be produced just by a brief, usually inconsequential, apoptotic stimulus (Perez-Garijo et al. 2004; Martin et al. 2009). It illustrates the importance of the mechanism of removal of abnormal cells to achieve normal development.

The role of the Dpp and Wg signals in these overgrowths has been assayed recently (Perez-Garijo et al. 2009). In those experiments an apoptotic stimulus (an X-radiation of 1500 R) was given to wing discs in which the posterior compartments in which all the cells contain P35. This resulted in the generation of a hyperplastic overgrowth only if the posterior compartment cells contain normal dpp and wg function. If either dpp or wg function was defective the overgrowth was not produced (see Fig. 2 in Perez-Garijo et al. 2009). Furthermore, and quite unexpectedly, in the absence of either gene the other is not ectopically activated. While these results demonstrate the implication of the two genes in the hyperplastic overgrowths, they also indicate a mutual requirement for their ectopic expression (see below). This again indicates that the mechanism of dpp and wg induction in the apoptotic cells is different from the normal one.

A key factor involved in dpp and wg activation by apoptotic cells appears to be the JNK pathway, known to be implicated in stress-induced apoptosis in Drosophila and also in mammals (Adachi-Yamada and O’connor 2002; Dhanasekaran & Reddy 2008; Igaki 2009). In Drosophila JNK function can be monitored by the expression of the puckered target gene – using a puc-lacZ insert (Martín-Blanco et al. 1998). In the wing disc there is normally no JNK function, except in a small proximal region implicated in thorax closure. However, it is activated in much of the disc in response to irradiation (Mcewen & Peifer 2005; Perez-Garijo et al. 2009). It plays a key role in inducing apoptosis because in the absence of JNK function there is a much reduced apoptotic response.

The observations that: (i) JNK is persistently active in undead cells (Fig. 2); (ii) it is co-expressed with wg (Fig. 2); and (iii) in absence of JNK there in no Dpp/Wg signaling (Ryoo et al. 2004), suggest a role of JNK in the ectopic activation of wg and dpp.

This idea was analyzed by forcing JNK expression in the wing disc and examining the dpp/wg response. With a Gal4 line (salEPv) directing expression in a relatively small portion of the wing pouch (Cruz et al. 2009), it is possible to drive constitutive expression of the JNK pathway using the hepACT construct (Adachi-Yamada et al. 1999). salEPv- Gal4 > UAS- hepACT wing discs show many cells of the sal domain with high levels of caspase activity but also showing dpp and wg expression (Fig. 3). These are normal (not undead) apoptotic cells, giving further support to the observation that the emission of the Dpp and Wg signals also occurs during normal apoptosis.

Figure 3.

 Role of the JNK pathway in the activation of wg and dpp. (A–A′) Forcing the JNK pathway in the wing pouch by overexpression of hepACT under the control of salEpv (the pattern of expression of which is shown in the inset) leads to extensive apoptosis, revealed by anti-Drice staining (A) and concomitant activation of wg (A′). (B–B′) Apoptosis in not required for wg induction because in dronc null mutants wg is still active upon hepACT overexpression. (C, D) Wildtype and salEpv-Gal4 > UAS- hepACTdronc discs showing normal dpp expression (C) and ectopic dpp expression in the sal domain (D).

However, the JNK pathway has other roles in addition to its apoptotic function. For example, it is known that dpp expression in the leading edge cells of the lateral epithelium during dorsal closure requires JNK activity (Riesgo-Escobar & Hafen 1997). It raised the possibility that the ectopic activation of dpp and wg may be caused by a non-apoptotic role of the JNK. This possibility was confirmed because in salEPv- Gal4 > UAS- hepACTdronc discs both genes are expressed (Perez-Garijo et al. 2009 and Fig. 3). The conclusion from this result is that the ectopic Dpp/Wg signaling observed in apoptotic cells is not a consequence of the apoptosis per se, but a side effect of the JNK pathway (see Fig. 4).

Figure 4.

 Signaling from Drosophila apoptotic cells. This figure illustrates how the activation of the apoptotic program leads to the emission of the Hh, Dpp and Wg signals. A stimulus leads to the activation of the pro-apoptotic genes reaper, hid and grim whose products bind to and inhibit Diap1. This results in the stability of Dronc, which in turn activates the effector caspase Drice, and importantly, the JNK pathway. The activation mechanism of JNK by Dronc is not known. Also by an unknown mechanism JNK induces dpp and wg, which may promote proliferation in neighbor cells. This Dpp/Wg activation is a side effect of JNK function and in normal circumstances has little or no effect, but their persistent emission by undead cells produces hyperplastic overgrowths. The induction of the Hh signal is originated by a different mechanism, as it requires Drice activity. Also, unlike Dpp and Wg, it is emitted by non-proliferating cells.

In this line of thought, if the ectopic Dpp/Wg signaling is solely due to JNK activity, and the hyperplastic overgrowths are caused by Dpp/Wg signaling, it follows that JNK should be able to generate overgrowths in the absence of apoptosis. We have recently shown that this is the case (Perez-Garijo et al. 2009): mis-expression of JNK in the wing pouch of dronc wing discs produces an excess of growth in that region. It has also been reported (Igaki et al. 2006) that the generation of tumours by polarity deficient cells is dependent on JNK activity.

The induction of dpp and wg by the JNK pathway under these circumstances is a novel phenomenon. With the present data it is not possible to say whether this induction is direct or mediated by other factors. Most of the JNK functions are mediated by AP-1 (Riesgo-Escobar & Hafen 1997; Zeitlinger & Bohmann 1999), so it could be of interest to look for possible AP-1 binding sites in the control regions of dpp and wg. An additional point of interest is that the induction mechanism requires normal function of dpp and wg, because it does not occur in the absence of either gene. It may suggest the need of a feed back from dpp and wg to maintain JNK activity.

Our results and those of others (Ryoo et al. 2004; Kondo et al. 2006; Wells et al. 2006) showing hyperplastic overgrowths caused by interfering with caspase activity during apoptosis, suggest a possible mechanism that may generate tumours in mammals. Mammalian cells often initiate apoptosis when infected by a virus as a defence to impede virus proliferation. If for whatever reason the effector caspase is not sufficiently effective (for example, the virus may encode a P35-like protein), it may result in the generation of undead cells with persistent activity of the JNK pathway and Dpp/Wg signaling. There are numerous reports reviewed in (Wagner & Nebreda 2009) linking JNK activity and cancer in humans.

Function of Dpp and Wg signaling by apoptotic cells – compensatory proliferation?

Since the ectopic Dpp/Wg signaling was reported in undead cells, it has been associated with compensatory proliferation. The latter phenomenon describes the ability of imaginal tissues of Drosophila to restore normal size after massive cell death, caused by stress events like irradiation. The removal of more that 50% of the cells in an imaginal disc provokes a response in the remaining cells that adjust their proliferation to compensate for the loss. It may be interpreted as the result of some interaction between the dying and the surviving cells leading to additional proliferation of the latter.

Two key features of undead cells, (i) their ability to induce additional proliferation of neighbor cells, and (ii) that they secrete growth signals like Dpp and Wg, immediately suggested a model for compensatory proliferation: the Dpp/Wg signals released by the dying cells would induce local proliferation in their neighbors that would compensate for the cell loss.

One problem with the model was that although it had been shown that normal apoptotic cells emit Dpp and Wg, there was no evidence that they were able to induce additional proliferation in their neighbors. Although the duration of the apoptotic process in Drosophila, from the initial stimulus until the physical elimination of the cells, has not been measured, it is supposed to take a short time. Whether under these circumstances cells in apoptosis are able secrete enough Dpp/Wg to induce additional proliferation was unclear. Besides, although Dpp functions as a growth inducer in the wing disc (Burke & Basler 1996), the Wg signal appears to arrest growth, especially during late development (Johnston & Sanders 2003).

We carried out an assay to test the role of the ectopic Dpp/Wg signaling in compensatory proliferation (Perez-Garijo et al. 2009). The experiments consisted of inducing massive cell death in compartments that were defective in dpp, in wg, or in wg and dpp. The conclusion is that there is compensatory proliferation in the absence of both genes: neither dpp nor wg play a significant role.

This result poses the problem about the mechanism behind the phenomenon of compensatory proliferation. We have argued (Perez-Garijo et al. 2009) that it may reside in the system that governs size control and that arrests growth once a compartment has reached final size (Martin & Morata 2006). The loss of even a large fraction of the cells of a compartment can be easily restored if all (or the majority) of the surviving cells perform just one or two additional divisions. The system would interpret the massive cell loss as the compartment becoming smaller, and then would allow the additional divisions until the final size is achieved. In this view compensatory proliferation would be intrinsic to the control system.

In our view the ectopic Dpp/Wg signaling in apoptotic cells is a consequence of the activation of the JNK pathway, which is necessary for the induction of the pro-apoptotic genes. The JNK pathway performs other functions promoting cell movements (Riesgo-Escovar et al. 1996; Pastor-Pareja et al. 2004; Llense & Martin-Blanco 2008) and also is required for dpp expression in the leading edge cells of the lateral epithelium during dorsal closure in late embryos (Riesgo-Escobar & Hafen 1997). In normal circumstances these other functions have negligible consequences because apoptotic cells die soon after the apoptotic stimulus. However, in the case of undead cells the situation is different because as they remain alive they manifest several features associated with JNK activity. Their persistent Dpp/Wg signaling causes hyperplastic overgrowths but they also often migrate. In compartments in which the undead cells belong to the posterior compartment they often alter the A/P compartment border, as they can move and push the boundary (Fig. 5). This phenomenon is not a genuine crossing of the A/P border, because the undead cells do not mix with anterior compartment cells (they also maintain their engrailed expression). More likely, it reflects their ability to migrate, conferred by the persistent activity of the JNK pathway.

Figure 5.

 Invasion of anterior compartment by JNK-expressing undead cells from the posterior compartment. (A–A′) Undead cells generated by irradiation of discs of genotype en-Gal4 > UAS-p35 UAS-GFP frequently penetrate into the anterior compartment. Note the three protrusions from the posterior compartment (arrow and inset). (B–B′) Magnification of the inset in A. Note the high levels of JNK activity (red) as monitored by the puc-LacZ insert.

Apoptosis, regeneration and growth factors release

Although we believe that Dpp/Wg signaling is largely irrelevant in compensatory proliferation in the imaginal discs of Drosophila, the ability of apoptotic cells to induce mitogenic signals may be of use in other situations and in other organisms. It has been shown recently that apoptotic cells in the mouse (Li et al. 2010) and in Hydra (Chera et al. 2009) can also release mitogenic signals that induce proliferation in nearby cells.

The phenomenon of mitogenic signaling may play an important role in tissue regeneration. In experiments inducing massive apoptosis in the wing disc of Drosophila (Smith-Bolton et al. 2009), it has been found that wg becomes upregulated and plays a role in the regeneration process, although only a fraction of wg expression is associated with dying cells. In similar experiments, (Bergantinos et al. 2010) it has been shown that there is ectopic activation of the JNK pathway, and also that this activity is necessary to regenerate the ablated tissue. The present evidence, however, suggests that the dying cells do not originate the adventitious JNK activity. Although there are still questions to be clarified, these experiments suggest a role of the JNK/Dpp/Wg signaling from apoptotic cells in regeneration.

In other organisms there is also evidence for a functional connection between apoptosis and regeneration. In the planarian Dugesia (Hwang et al. 2004) there is overall induction of the caspase-like gene DjClg3 during tail regeneration, and similar observations have been made in other planarian species (Pellettieri et al. 2010). Also, tail regeneration in the Xenopus tadpole is associated with the appearance of apoptotic cells close to the wound 12 h after the amputation. Moreover, this apoptotic activity is essential for regeneration, as caspase-inhibited embryos completely fail to regenerate (Tseng et al. 2007). Although the genetic/molecular factors involved in these processes are not known, it is reasonable to assume that the additional growth necessary for regeneration is mediated by mitogenic signals originating from the apoptotic cells.

There are two recent reports implicating signaling from apoptotic cells and regeneration. Studying regeneration in Hydra, Chera et al. 2009 have shown that amputation of the upper half causes a rapid apoptotic response in the region close to the cut site and that this apoptosis is necessary to regenerate the head. Moreover, they showed that the apoptotic cells express transiently the Wnt3 gene, reflected later by nuclear translocation of ß-catenin, which is associated with increased cell proliferation. Inhibition of Wnt3 activity by RNA interference results in suppression of regeneration.

Li et al. 2010 reported that lethally irradiated mouse fibroblasts can induce in co-culture proliferation of several types of stem cells and that this induction is dependent on caspase function. Furthermore these authors showed that mice lacking caspase activity are deficient in wound healing and liver regeneration, probably due to attenuation in cell proliferation at the wound site. The data suggest that the caspase-induced cell proliferation is mediated by secretion of Prostaglandin E2, a potent stimulator of stem cell proliferation.

Although the field is still at the beginning, the present evidence suggests that the ability to secrete growth factors is a universal feature of metazoan cells in apoptosis. It may be a fundamental process to respond to tissue damage.