Interplay of cell proliferation and cell death in Drosophila tissue regeneration



Regeneration is a fascinating process that allows some organisms to reconstruct damaged tissues. In addition to the classical regeneration model of the Drosophila larval imaginal discs, the genetically induced tissue ablation model has promoted the understanding of molecular mechanisms underlying cell death, proliferation, and remodeling for tissue regeneration. Recent studies have also revealed that tissue injury responses occur not only locally but also systemically, even in the uninjured region. Genetic studies in Drosophila have demonstrated the dynamic role of the cell death-induced tissue response in the reconstruction of damaged tissues.


Since tissue regeneration is based on tissue repair and remodeling processes, understanding the tissue responses to injury or homeostatic cell renewal can provide mechanistic insights into tissue regeneration. Although several model organisms such as planaria and hydra have a high regenerative capacity, their genetic analysis is hampered due to the lack of abundant tools. In contrast, Drosophila larval imaginal discs that are the primordia of adult structures, exhibit remarkable potential to induce regenerative growth upon tissue damage. Thus, imaginal discs, combined with the Drosophila genetics, have provided a sophisticated system to facilitate the investigation of the molecular mechanisms underlying tissue regeneration (Smith-Bolton et al. 2009; Bergantinos et al. 2010b; Belacortu & Paricio 2011; Repiso et al. 2011; Worley et al. 2012). Recent genetic research has demonstrated the importance of the communication between primary damaged cells and the surrounding cells in the process of regeneration. Furthermore, tissue damage is known to activate not only a local tissue response, but also systemic responses that are beneficial for tissue regeneration and homeostasis.

Methods for the study of regeneration using Drosophila imaginal discs

The groundwork research for regeneration using imaginal discs dates back to a series of classic experiments performed in the 1940s to 1970s (Worley et al. 2012; Fig. 1). In these experiments, some regeneration was observed when imaginal disc fragments from larvae were transplanted into the abdomen of other larvae (Hadorn & Buck 1962). Similarly, regenerative growth was observed in genital disc cut in half, following the transplantation of the disc into an adult female host (Hadorn 1963). The implanted disc fragment was able to regenerate the rest of the disc following pattern regulatory rules (intercalation; Haynie & Bryant 1976). Intercalation of the disc regeneration was also genetically re-evaluated (Repiso et al. 2013). The Pinching method, that is, pressing the disc with a needle over the cuticle, was also developed to damage the disc in situ (Bryant 1971). Although this method can cause disc injury without puncturing the cuticle and allow regeneration to occur in its normal developmental context, it is unsuitable for large-scale experiments such as genetic screening because of the associated technical difficulties and limited reproducibility (Pastor-Pareja et al. 2008; Diaz-Garcia & Baonza 2013).

Figure 1.

Classical methods to induce tissue regeneration in Drosophila imaginal disc. (A) In vivo culture method: The imaginal disc can regenerate after surgical dissection and transfer to the adult abdomen or younger larvae. (B) The Pinching method: Pressing the disc with a pair of forceps or a needle over the cuticle can damage the disc without puncturing the cuticle. This allows regeneration to occur in its normal developmental context.

On the other hand, Drosophila genetics has enabled us to non-surgically remove fragments of a developing tissue. Hariharan and colleagues developed such an ablation method by combining the Gal4/UAS-based ectopic expression system with a temperature-sensitive Gal4 suppressor, Gal80ts, to induce cell death in a subdomain of the wing disc within a limited time period (Smith-Bolton et al. 2009; Fig. 2). In this study, the cell death-inducing gene eiger, which is the sole tumor necrosis factor (TNF) superfamily in Drosophila (Igaki et al. 2002; Moreno et al. 2002; Kauppila et al. 2003), was overexpressed for tissue ablation (Smith-Bolton et al. 2009). Similarly, proapoptotic genes like hid or reaper were also expressed to induce tissue ablation in a spatially and temporally controlled manner (Smith-Bolton et al. 2009; Bergantinos et al. 2010a; Herrera et al. 2013). Using this procedure, ablation was transiently induced in early- to mid-third instar larvae.

Figure 2.

Schematic view of the genetic ablation system for studying regeneration in the wing disc. (A) Example of the genetic ablation system: The cell death-inducing gene eiger is expressed under the control of rn-Gal4 driver (wing pouch specific Gal4 driver), and it ablates the wing pouch region, which is the primordium of the adult wing. A temperature-sensitive suppressor of Gal4, Gal80ts, is expressed under the control of an ubiquitous promoter. This enables the inhibition of the function of Gal4 at the permissive (18°C) but not the non-permissive temperature (30°C). The temperature shift bi-directionally switches Gal4 activity, thus temporally controlling the ectopic expression of eiger. (B) The time course for ablation and regeneration experiments: At the early-third instar of larval development, the temperature is raised to 30°C and the wing pouch is ablated followed by a developmental delay. Forty hours after the ablation, the temperature is restored to 18°C to repress cell death. If regeneration is sufficient, the adult flies eclose with intact wings.

Since pupariation limits the regeneration of the disc, larvae were allowed to recover within the third instar larval period after the induction of ablation (Smith-Bolton et al. 2009; Halme et al. 2010). Tissue injury at this larval stage causes a developmental delay and inhibits the initiation of metamorphosis (Hussey et al. 1927; Simpson et al. 1980; Smith-Bolton et al. 2009), suggesting that the animal body can sense the local tissue damage and systemically alter the developmental timing to repair the injured tissue. The transition from the larval to the pupal stage of development is primarily regulated by a neuroendocrine mechanism. Tissue damage delays the production of the neuropeptide prothoracicotropic hormone (PTTH), which promotes the production and release of the steroid hormone ecdysone (McBrayer et al. 2007). Dilp8 was secreted from the injured disc, and it inhibited pupariation in a non-tissue-autonomous manner (Colombani et al. 2012; Garelli et al. 2012). Retinoic acid was also suggested to play a role in the developmental delay (Halme et al. 2010). Thus, Dilp8 and retinoic acid may function upstream of PTTH production; however, the precise molecular mechanisms underlying Dilp8- and retinoic acid-dependent developmental delays need to be elucidated.

Although the imaginal disc possesses regenerative capacity, it is somewhat difficult to distinguish between developmental plasticity and regeneration, since a “regenerating” disc is also undergoing development. It may be possible to dissect and separate these two processes by ablating the disc after larval development has been completed. A method for prolonging the larval stage was developed using an erg2 mutant yeast strain as fly food, since it cannot produce substrates for biosynthesis of ecdysteroids, and larval development was arrested at the end of the third instar larvae (Katsuyama & Paro 2013). Using this method, tissue injury was induced after the larval development was completed to some extent, and then the subsequent disc regeneration was observed. Thus, regenerative growth can be studied separately from developmental growth, to facilitate the investigation of mechanisms that are specific to the regenerative process.

From death to birth: apoptosis and compensatory proliferation

Although exposure of larvae to X-ray irradiation induces massive cell death in the wing disc, normal wings can be generated even after the elimination of nearly half of the wing disc cells (Haynie & Bryant 1977). This is because the surviving cells proliferate to replace the lost parts of the disc and restore it to its original size. This phenomenon is known as “compensatory proliferation.” This regenerative cellular response after irradiation is mediated by Drosophila melanogaster p53 (Dmp53) and the c-Jun N-terminal kinase (JNK) pathway (Brodsky et al. 2000; Ollmann et al. 2000; Lee et al. 2003; Sogame et al. 2003; McEwen & Peifer 2005). Since Dmp53 and human p53 can stimulate JNK activity (Gowda et al. 2012), it is speculated that p53 and JNK may cooperate to regulate tissue regeneration. When the posterior compartment of the wing disc was ablated by Ricin, a toxin of plant origin that inhibits protein synthesis, cell proliferation was stimulated within the same posterior region, suggesting that mitogenic factors may be produced in the ablated region (Milan et al. 1997). However, since the dying cells are rapidly eliminated, it had been technically difficult to examine if and how the apoptotic cells contribute to the production of mitogens.

Expression of a caspase inhibitor p35 together with apoptotic stimulation, such as X-ray irradiation or expression of a caspase-activating gene (reaper or hid), could theoretically be able to generate the apoptotic “undead” cells, in which the initiator caspase Dronc was activated but the executioner caspases such as DrICE and Dcp-1 were not. The undead cells continuously expressed the morphogens, Wingless (Wg) and Decapentaplegic (Dpp), which promoted cell proliferation, resulting in hyperplastic overgrowth of the wing disc (Huh et al. 2004; Perez-Garijo et al. 2004, 2005; Ryoo et al. 2004; Kondo et al. 2006; Wells et al. 2006; Morata et al. 2011; Fig. 3A). In this system, Dronc enhanced the compensatory proliferation via the JNK pathway and Dmp53 (Ryoo et al. 2004; Kondo et al. 2006; Wells et al. 2006; Wells & Johnston 2012). Activation of the JNK pathway in damaged epithelia was required for the proliferation of the surrounding cells (Perez-Garijo et al. 2009; Bergantinos et al. 2010a; Warner & Longmore 2010; Suissa et al. 2011). It was reported that tissue damage disrupted the localization of Par polarity complex proteins, which resulted in JNK activation in damaged cells, and subsequently induced the secretion of morphogens for proliferation of the surrounding cells (Warner & Longmore 2010; Warner et al. 2010). Furthermore, activation of the JNK and Dmp53 signaling pathways regulated compensatory proliferation, both upstream and downstream of the apoptotic pathway, thus forming a feedback amplification loop for apoptosis signaling (Wells et al. 2006; Warner & Longmore 2010; Shlevkov & Morata 2012). However, the precise molecular mechanisms of Dronc-mediated activation of Dmp53 and JNK remain unclear.

Figure 3.

Interrelationship between cell death, morphogens, and compensatory proliferation in the imaginal disc. (A) Apoptotic undead cells trigger compensatory proliferation by secreting morphogens, Wg and Dpp. This is regulated by the initiator caspase Dronc. (B) Under UV irradiation, dying cells display ectopic induction of Wg and Dpp; however, these morphogens are not required for compensatory proliferation. (C) During eiger- or reaper-induced genetic ablation in the wing disc, proliferating but not dying cells express Wg. (D) During hid-induced genetic ablation, the expression pattern of both Dpp and Wg remains unaltered. (E) When apoptosis is induced by the ectopic expression of hid in the posterior part of the morphogenetic furrow of the eye disc, upregulation of Hh is observed in dying post-mitotic cells. This signal triggers the cell cycle re-entry of the neighboring post-mitotic but as yet undifferentiated cells. This compensatory proliferation requires the effector caspases, Drice and Dcp-1.

As described above, Wg and Dpp were required for the proliferation of apoptotic “undead” cells; however, they were dispensable for the compensatory proliferation of the surrounding cells after irradiation of the wing disc, despite being ectopically expressed in apoptotic cells (Perez-Garijo et al. 2009; Fig. 3B). Furthermore, when cell death was induced by ectopic expression of eiger or reaper in the wing disc, Wg was expressed in the proliferating but not the dying cells (Smith-Bolton et al. 2009; Fig. 3C). On the other hand, Morata and colleagues have shown that neither the expression pattern nor the protein levels of Wg and Dpp were altered when hid was ectopically expressed (Herrera et al. 2013; Fig. 3D). Thus, despite the accumulating experiments about the role of morphogens in compensatory proliferation, how the source and beneficiary of morphogens are determined is unknown.

Fan and Bergmann also reported the significance of another morphogen, Hedgehog (Hh) for compensatory proliferation. When apoptosis was induced by expressing hid in the differentiating region of the larval eye disc, Hh was expressed in the dying, post-mitotic cells. Hh triggered the cell cycle re-entry of the neighboring post-mitotic cells that were not yet differentiated into photoreceptor neurons. In this case, the effector caspases, Drice and Dcp-1, were required for Hh expression (Fan & Bergmann 2008; Fig. 3E).

The adult Drosophila gut is a well-known self-renewing epithelia. When gut enterocytes (ECs) were damaged chemically or by infection, the cells secreted the interleukin-6 (IL-6)-like cytokines, Upds (Unpaired 1–3 in Drosophila), and the epidermal growth factor receptor (EGFR) ligands, Keren (Krn) and Spitz (Spi), via JNK activation (Buchon et al. 2009, 2010; Jiang et al. 2009; Xu et al. 2011). Wg was also induced and secreted from the enteroblasts (EBs; Cordero et al. 2012). Thus, the Wg, EGF, and Janus kinase/Signal Transducer and Activator of Transduction (JAK/STAT) signaling pathways were activated in the intestinal stem cells (ISCs) to induce homeostatic proliferation (Buchon et al. 2010; Xu et al. 2011). Nrf2/Keap1, a master regulatory mechanism of the cellular redox state, was also found to regulate ISC proliferation under various stress conditions in the gut. In addition, reactive oxygen species (ROS) may function as regulators of ISC proliferation (Hochmuth et al. 2011). Thus, similar to the compensatory proliferation shown by the larval imaginal disc, adult gut ISCs also demonstrate homeostatic proliferation mechanisms following tissue injury.

Non-autonomous cell regulatory mechanisms other than compensatory proliferation for tissue homeostasis

Besides compensatory proliferation, there are other mechanisms that maintain tissue homeostasis in damaged regions. For instance, when major clusters of cells in the wing pouch region were eliminated by hid-induced apoptosis, cells outside the damaged area proliferated and migrated into the damaged area (Herrera et al. 2013). In case of postmitotic follicular epithelia, compensatory cellular hypertrophy occurred instead of compensatory proliferation (Tamori & Deng 2013). During cell competition in postmitotic follicular epithelia, winner cells underwent compensatory cellular hypertrophy implemented by acceleration of the endocycle, which was relied on insulin/IGF-like signaling (IIS), but not on JNK signaling, dMyc, Yki, and neoplastic tumor-suppressor genes (Tamori & Deng 2013). This mechanism may also be involved in maintaining the fine control of tissue integrity during regeneration.

Non-autonomous cell death induced after tissue ablation

Multiple studies have revealed that cells respond to tissue loss. Intriguingly, it has been demonstrated that cell death not only triggers compensatory proliferation, but also induces non-autonomous cell death. When Ricin was overexpressed in the posterior region, Dronc was activated not only in the Ricin-expressing cells but also in cells of the anterior compartment of the wing disc (Mesquita et al. 2010). The co-expression of p35 in the Ricin-expressing domain prevented cell death in that region, and partially prevented cell death in the anterior region as well. Dmp53 was activated in the Ricin-expressing region, which induced the activation of effector caspases. The co-expression of a dominant negative form of Dmp53 in the Ricin-expressing region also attenuated cell death in the anterior region (Mesquita et al. 2010). These results suggest that coordination of cell death occurs in the posterior and anterior parts of the wing disc.

Steller and colleagues also reported that when apoptotic undead cells were generated by the co-expression of hid and p35 in the posterior region of the wing disc, overgrowth of cells in the posterior region as well as the non-cell autonomous apoptosis of cells in the anterior region were observed (Perez-Garijo et al. 2013). This ectopic secondary cell death was termed as apoptosis-induced apoptosis (AiA; Fig. 4). JNK was strongly activated in the posterior region, and weakly activated in the dying cells of the anterior region. In addition, in the posterior region, Eiger was upregulated and its downregulation in this region prevented apoptosis in the anterior region, indicating that AiA was regulated by Eiger. TNF-α was found to be crucial for coordinated apoptosis and hair cycle progression in mice, suggesting that AiA plays an important physiological role in both flies and mammals (Perez-Garijo et al. 2013). In planarians, amputation triggered two waves of apoptosis. Initially, apoptosis was observed at the wound site (peaking at 1–4 h after injury), after which cell death spread to the entire body during the regeneration process (peaking 3 days after amputation) (Pellettieri et al. 2010). It is interesting to know whether the induction of systemic cell death has a beneficial role in regulating tissue remodeling and regeneration.

Figure 4.

Model of apoptosis-induced apoptosis (AiA). In the wing disc, apoptotic cells in the posterior region induce ectopic secondary apoptosis in the anterior region. This phenomenon is remotely controlled by the activity of JNK signaling in the posterior region.

Systemic response after tissue damage

As described above, AiA was an unexpected phenomenon in which ectopic secondary cell death was induced in a region distant from the primary lesion. In addition to this local systemic response, it was revealed that a more body-wide, inter-organ communication occurred in response to tissue damage (Takeishi et al. 2013). When the Drosophila cuticle was wounded by pricking, caspase-dependent cell death was observed in ECs in the gut within 30 min (Takeishi et al. 2013; Fig. 5). Caspase activation in ECs after cuticle wounding was mediated by NADPH oxidase 1 (Nox1)-dependent ROS in the ECs (Takeishi et al. 2013). Dual oxidase-1 (DUOX)- or Nox1-dependent ROS were also involved in controlling ISC proliferation during infection or homeostasis (Buchon et al. 2009; Jones et al. 2013; Lee et al. 2013). Thus, EC apoptosis following cuticle wounding may stimulate ISC proliferation via ROS. EC cell death followed by ISC proliferation forms the basis for gut epithelial turnover. In apoptosis-deficient mutants (dark/apaf-1 mutants), ISC proliferation was found to be severely impaired. On the other hand, when ISC proliferation was inhibited by PTEN expression, wound-induced caspase activation in ECs was inhibited, indicating the presence of mutual interactions between EC death and ISC proliferation for the maintenance of gut homeostasis, which is in turn essential for survival after wounding (Takeishi et al. 2013). This report also demonstrated that EC turnover is required for dampening the production of lethal factors after wounding.

Figure 5.

Local and systemic tissue responses initiated by cell death. (A) Injury of the imaginal disc activates a local tissue response to initiate tissue remodeling. Cell death induces compensatory proliferation. (B) Tissue damage causes systemic effects in tissues distant from the primary lesion. The secondary tissue activates the cell death program, which initiates tissue stem cell activation. For example, cuticle wounding induces gut enterocyte (EC) death, which is in turn required for intestinal stem cell (ISC) proliferation (Takeishi et al. 2013; See Text).

Participation of immune cells in tissue repair and regeneration

Following tissue ablation, various non-autonomous cellular responses function to maintain homeostasis. Homeostatic mechanisms are also mediated by blood cells at the systemic level. Around 24 h after tissue ablation by the pinching method, the wound site in the regenerating disc recruited a number of adherent hemocytes, the immune cells of Drosophila (Pastor-Pareja et al. 2008). Disruption of the basement membrane by wounding allowed hemocytes to access the damaged epithelia, facilitating a systemic tissue damage response (Pastor-Pareja et al. 2008). In the case of wound healing of larval epithelia, hemocytes were recruited to the damaged area to promote tissue repair and the removal of dead cells (Babcock et al. 2008).

UV irradiation of the pupal retina led to widespread apoptosis and tissue loss in the retina. Under these conditions, the genetic ablation of hemocytes enhanced tissue damage, suggesting a functional and protective role of hemocytes in the prevention of tissue damage (Kelsey et al. 2012). These findings imply that hemocytes provide an appropriate environment for regeneration and positively enhance the process. However, ablation of hemocytes and fat bodies by reaper-induced apoptosis did not cause defects in the regeneration of the wing disc (Katsuyama & Paro 2013). Katsuyama and Paro speculate that dead cells are probably engulfed by the surrounding epithelial cells rather than by hemocytes, in this case (Li & Baker 2007; Ohsawa et al. 2011).


Tissue remodeling during regeneration involves not only local events, but also cooperation with other tissues (Fig. 5). A study of tissue damage responses following epidermal DNA damage in mutant larvae lacking the DNA repair enzyme Mei-9 (XPF homologue), suggested the presence of inter-organ communications during tissue damage (Karpac et al. 2011). In mei-9 mutants, the cuticle is vulnerable to low doses of UV, which limits DNA damage to the exposed epidermis, thus creating the local injury model. Following epidermal damage by UV irradiation, hemocyte expansion and widespread melanization were observed. Upregulation of Upd3, a cytokine that positively regulates the JAK/STAT signaling, was also detected in hemocytes. In the early phase of response, JAK/STAT signaling was activated in diverse tissues. This signal is also known to depend on the hemocyte-mediated repression of secretion of Drosophila insulin-like peptides, produced by median neurosecretory cells (mNSC) in the brain. In the later phase, repression of insulin signaling led to Forkhead box O (FOXO)-dependent activation of nuclear factor (NF)-κB/Relish in the fat body, after which normal IIS levels recovered systemically. Thus, examining the cross-regulation across different tissues (hemocytes, fat body, and brain) is crucial in understanding the systemic regulation of regeneration.


This study was supported by the Japanese Ministry of Education, Science, Sports, Culture, and Technology (to M.M.).