Tissue repair and regeneration in Drosophila imaginal discs


Author to whom all correspondence should be addressed.
Email: fserras@ub.edu


Exploring the mechanisms involved in tissue regeneration is one of the main challenges in biology and biomedicine. Multiple examples of tissue regeneration exist across the animal phyla, ranging from the recovery of the whole animal (e.g. flatworms) to the limited capability of the human liver. Studies performed in the 1960s showed that Drosophila imaginal discs are able to regenerate. This property, together with multiple genetic tools available, make fly an excellent model for the study of the regenerative process. Here we present an overview of the use of Drosophila for the study of regeneration and describe major recent advances in the understanding of this process. Current studies in Drosophila have unraveled some of the pathways and factors needed for a tissue to regenerate. Many observations point to the reuse of developmental programs and genetic reprogramming to drive regeneration. We discuss how this reprogramming could be orchestrated by the initial activity of the JNK pathway.


Regeneration is the ability of an organism to rebuild parts of the body that have been damaged or amputated. How some tissues or even some entire organisms can regenerate while others cannot is one of the most challenging issues in regenerative biology and medicine. There are plenty of examples of regeneration in nature, in almost all branches of the phylogenetic tree. The most prominent examples are flatworms and hydrozoans, which recover full size and organs by a variety of regeneration mechanisms. Tissue regeneration is also a property of more complex animals. The fins and heart of fishes, the limbs and spinal chord of amphibians, and the human liver are clear-cut examples of tissues capable of regeneration (Sanchez Alvarado & Tsonis 2006; Galliot et al. 2008). Regeneration also occurs in insects; for instance, cockroaches and crickets regenerate legs after amputation (French 1976; Bando et al. 2009). Adult Drosophila are not capable of regenerating appendages; however, the imaginal discs, are able to regenerate (Hadorn 1963; Bryant 1971; Schubiger 1971). Imaginal discs are sac-like invaginations of ectodermal cells, whose fates are determined during early embryonic development. These cells proliferate mitotically, following a program distinct from that of the larval tissues and are fated to become the adult legs, wings, halteres, eyes, antennas, head capsule and genital organs.

Two early discoveries were fundamental in establishing Drosophila imaginal discs as a model for regeneration. First, the discovery by Hadorn and colleagues (Hadorn & Buck 1962; Hadorn et al. 1968) of the capability of imaginal discs to regenerate after fragmentation. Second, the finding that imaginal disc tissue can recover and regenerate following massive cell death (Haynie & Bryant 1977).

Regenerative capabilities of imaginal discs

Hadorn’s Legacy

Hadorn et al. (1968) established that each disc and disc region is determined to give rise to a specific part of the adult body (Schubiger et al., 1969Schubiger 1971). They demonstrated that, when transplanted into host larvae, fragmented imaginal discs or even cell aggregates from disassociated discs retain the ability to differentiate and subsequently metamorphose. Fate maps, constructed by topographic correlation between the original site of the fragment and the differentiated disc (Schubiger 1968; Bryant and Schneiderman 1969; Bryant 1975) reveal that small regions within a disc or even individual cells remain determined for region-specific structures, even after dissociation and re-aggregation. Cells from dissociated imaginal discs, specifically recognize neighbor cells with similar properties and tend to re-aggregate to reconstruct the original pattern (Garcia-Bellido 1966). All of these studies imply that imaginal disc cell fates are determined during normal development, before experimental fragmentation and transplantation. This conclusion led to imaginal discs becoming a prototype of an inflexibly-determined developmental system. However, the fact that the fate of imaginal disc cells is determined implies that fragments must generate new tissue identities from previously committed cells. Therefore regeneration must entail a mechanism that confers a certain plasticity in order to reprogram cell fates.

This plasticity is revealed after long cultures of regenerating discs, where some determined cells switch fates to that of other imaginal discs, a phenomenon known as transdetermination (Hadorn 1978). This suggests the existence of zones with enormous cellular plasticity that can redirect the fate of cell populations (Steiner et al. 1981; McClure & Schubiger 2007; Salzer & Kumar 2010). In all these experiments, imaginal disc fragments were cultured in the abdomen of an adult female, permitting regenerative growth of the disc fragments but not differentiation (Hadorn 1963). Following this protocol, regenerated fragments recovered from the adult host were subsequently transplanted into larvae to undergo metamorphosis. A detailed analysis of the resulting differentiated adult structures demonstrated that regenerated fragments not only differentiate into the fate for which they are determined but also into other tissues. These observations indicate that although the patterned fate of disc cells is determined and mappable, it is not restricted in a cell-heritable manner until differentiation occurs, as evidenced by the new structures that arise from regeneration (Schubiger 1971; Hadorn 1978).

Compensatory proliferation as response to cell death

A second piece of evidence validating imaginal discs as a model for tissue regeneration was discovered when Haynie and Bryant eliminated about 50% of disc cells after X-ray irradiation, and yet a normal adult could develop (Haynie & Bryant 1977). They demonstrated that the remaining living tissue can recover, repopulate the disc, and reach normal size, thus compensating for the cells lost during irradiation. This so-called compensatory proliferation has been the subject of several recent studies based on interference with the apoptotic response after irradiation (Hay et al., 1994) or after induction of pro-apoptotic genes by coexpressing the baculovirus protein p35. This protein blocks the activity of caspases, thus cells remain in an “undead” state even though the apoptotic pathway has been induced. “Undead” cells induce non-autonomous proliferation in surrounding cells of the same compartment by continued release of mitogenic signals normally secreted by apoptotic cells (Huh et al. 2004; Perez-Garijo et al. 2004, 2009; Ryoo et al. 2004; Fan & Bergmann 2008). This proliferation in the absence of apoptosis results in an overgrowth phenotype (Huh et al. 2004; Perez-Garijo et al. 2004; Ryoo et al. 2004; Kondo et al. 2006; Wells et al. 2006; Fan & Bergmann 2008). It has also been suggested that organs re-establish appropriate size after cell death by re-using the normal growth processes that regulate size during development within the disc compartments (Perez-Garijo et al. 2009). Compensatory proliferation is emerging as a homeostatic cellular property activated by apoptosis or injury to trigger tissue repair and organ size control.

Regeneration and wound repair

The most striking advantage of using Drosophila as an experimental model is the ease of performing genetic analysis, including the use of a variety of sophisticated genetically-engineered tools. One of the most powerful is the yeast Gal4-UAS system (Brand & Perrimon 1993), which is a transgene-based method used to study gene expression and function. Using specific promoters to drive Gal4 expression, the activation of a transgene cloned 3′ to the upstream activating sequence (UAS) can be spatially and temporally controlled. Tight temporal control of Gal4 function can be achieved using a Gal80 thermo-sensitive construct expressed under the tubulin ubiquitous promoter (tubGal80ts; Zeidler et al. 2004). Active Gal80ts blocks Gal4 function at the permissive temperature. At the restrictive temperature, Gal80ts is inactive and Gal4 can induce transcription of the transgene under control of the UAS sequence. In regeneration studies in particular, this system allows the precise temporal activation of apoptotic genes in specific cells and tissues (Fig. 1).

Figure 1.

 Genetic ablation for regeneration studies in the wing imaginal disc. (A) Genetic design. Flies carrying a Gal4 transcription factor construct driven by a tissue specific enhancer/promoter (E/P-Gal4) will activate the pro-apoptotic gene reaper cloned 3′ of the Upstream Activation Sequence (UAS-rpr). At 29°C cells that activate reaper will enter apoptosis and will be extruded basally. However, at 17°C the temperature sensitive Gal80, driven under a tubulin (tub) ubiquitous promoter, is active and blocks Gal4. Therefore, when larvae are transferred from 17 to 29°C, cells of that specific tissue die, and when they are back to 17°C, the tissue will regenerate. In the experiments shown in this figure, the rpr activation was initiated at third instar larvae and kept for 16 h at 29°C. During the first 6 h the Gal80 protein is still active (grey box). (B) Wing imaginal disc in which a spalt promoter (dark region in the center of the disc) has been used to activate rpr for 10 h at 29°C. (C) At 17°C, cells have regenerated that central domain. Cyan, nuclei; Magenta, F-actin; Yellow, dorsoventral boundary labeled with anti-Wingless. (D) Sequence of images of a wing disc after genetic ablation, from basal (left) to apical (right). Note that the ablated region is open at basal (central dark zone) and healed at apical (right image). Cell debris is visible in some basal section (grey). Magenta, F-actin. Scale bar: 25 μm.

The application of this method has recently allowed the identification of several key features of regeneration after activation of apoptosis in specific domains of the wing disc. Ablated domains are able to regenerate by means of locally concentrated proliferation. Smith-Bolton et al. (2009) found that the apposition of dorsal and ventral tissue promotes the reappearance of the DV boundary, suggesting that not all commitments are maintained and re-specification can occur. The same authors found that wingless (wg) and dmyc are upregulated during regeneration. Wg appears to be expressed in a vast domain surrounding the ablated zone, in a pattern that is reminiscent of earlier stages of development. After 3 days of recovery from genetic ablation, the wg pattern returns to normal. It has been proposed that wg can promote growth by a double-repression mechanism: wg signalling inhibits Notch (N) activity, allowing the expression of the growth promoters dmyc and the microRNA bantam (Herranz et al. 2008). Wg was found in proliferating cells and N activity was absent or reduced during wing regeneration. As most of the wing pouch was ablated in these experiments, expression of wg was a requirement to re-specify the wing domain. Concomitantly, dmyc expression was elevated in the regenerating wing pouch and its overexpression enhanced regeneration. Thus, activation of cell death in particular domains can be used to test the role of key regulators of regeneration.

One of the aims of regeneration research is to discover the early signals that are triggered after cell death or injury and that are responsible for the re-epithelization. Several features of this process, which involves proliferation and cell shape changes to restore the cell sheet, have recently been discovered. Detailed analysis of imaginal disc epithelial reconstitution has revealed F-actin-rich projections emitted from the leading edge, proliferation concentrated near the wound and apical to basal wound healing with dead cells and debris extruded basally (Bergantiños et al. 2010a). This study also demonstrates that proliferation is predominantly associated with cells of the damaged-compartment lineage of the disc, suggesting that compartments regenerate their own lost portions. This is in agreement with the observation that expression patterns of selector genes such as engrailed (en, posterior compartment) and cubitus interruptus (ci, anterior compartment) are either intact or rapidly re-established (Smith-Bolton et al. 2009). Moreover, lineage analysis shows that the cells responsible for reconstructing the killed area originate primarily from cells near the wound of the damaged compartment (Bergantiños et al. 2010a). In addition to a rapid burst of local proliferation, a secondary peak of proliferation is found all over the compartment when small regions are killed. This is postulated to constitute a compartment-associated size re-adjusting event.

One signalling network dedicated to maintaining cell, tissue, and organism fidelity in response to trauma involves stress-activated protein kinases, the JUN NH2-terminal kinases (JNK). JNK activation can be monitored by expression of the phosphatase puckered (puc; Fig. 2; Bosch et al. 2005, 2008; Lee et al. 2005; Mattila et al. 2005; Bergantiños et al. 2010a), a transcriptional target of the JNK cascade (Martin-Blanco et al. 1998) and direct target of Fos (Lee et al. 2005). Those studies have shown localized JNK activity in the leading edges of the wound and zones of the regenerating domain. Using specific JNK cascade mutants, it has been demonstrated that JNK is required for wound healing and for early regenerative events after injury or genetic ablation. For example, alleles of hemipterous gene (hep), a Jun kinase kinase (JNKK) group mitogen-activated protein kinase kinase (MAPKK), can be used in heterozygosis to sensitize the tissue for analysis of functional requirements (Glise et al. 1995; Wu et al. 1997). The hypomorphic hepr75 allele in heterozygosis does not affect normal wing development, but severe alterations in F-actin cytoskeleton and in proliferation occur after activation of cell death, resulting in disruption of healing in this genetic background. Moreover, the sharp interface between the intact epithelium and the dead domain found in normal discs disappears as living and dying cells form an irregular edge in hep tissues, strongly altering or leading to complete failure in regeneration (Bergantiños et al. 2010a). Cytoskeleton rearrangements and JNK activity, in addition to cell shape changes, seem to be general features of healing and are comparable to other post-injury events in embryos (Jacinto et al. 2002; Campos et al. 2010), larvae (Galko & Krasnow 2004; Kwon et al. 2010), and adult cuticle (Ramet et al. 2002).

Figure 2.

 Activation of the phosphatase puckered (puc) in wing discs after genetic ablation. (A) Basal view of a disc showing the ablated area filled with cell debris (outlined in white). (B) Apical view of the same disc, where cells are repopulating the lost tissue. Puckered-lacZ (green) is highly expressed in that region. F-actin was labeled with rhodamine phalloidin to mark the cell contour (magenta). In order to see the puc labelled nuclei, the confocal image corresponds to a sub-apical optical section in the central region of the disc. Note that due to disc folding and morphology, the edges correspond more apically where F-actin accumulation is higher.

In contrast to microsurgical removal of a piece of tissue, genetic ablation requires elimination of dead cells from the epithelium to allow regeneration. As mentioned above, these cells and other debris appear to be eliminated from the basal cell sheet as wound healing occurs. One candidate for regulating this process is the p53 tumor suppressor gene, which has been proposed to be involved in maintaining tissue homeostasis and renewal by suppressing the accumulation of DNA-damaged cells in mammals (Yazinski et al. 2009). In the absence of p53, tissue deterioration and accumulation of damaged cells impede regeneration by undamaged cells, suggesting that p53-mediated removal of damaged cells is a pre-requisite for efficient progenitor driven renewal (Schoppy et al. 2010). Thus, p53 plays an important role in tissue renewal by limiting the accumulation of damaged cells. In Drosophila, p53 is required to stimulate growth in “undead cells” and appears that a non-apoptotic function of the initiator caspase Dronc is the regulation of p53 to stimulate compensatory proliferation (Huh et al. 2004; Wells et al. 2006).

Early regeneration responses

Apart from wound healing, another main issue is how regenerative growth and re-patterning is stimulated. In discs cut by microsurgery or after cell death induction, puc expression is observed after just a few hours (Fig. 2). This is one of the first signals activated during regeneration and points to the JNK pathway as one of the early responses. But how is this pathway activated? A network for JNK phosphorylation has been proposed following a screen to identify regulators of the JNK cascade (Bakal et al. 2008). Some of these regulators are genes encoding components of apico-basal polarity complexes that act as JNK suppressors. We speculate that epithelial perturbation may be associated with unrestrained JNK activation. A similar situation is found in fly tumors, where mutations in polarity genes activate JNK signalling and downregulate the E-cadherin/β-catenin adhesion complex, conditions necessary and sufficient to cause oncogenic RasV12-induced tumors (Igaki et al. 2006). Thus, it is possible that derepression of JNK signalling and alterations in cell polarity induced in the wound edge cells after injury or massive death, are related.

There is evidence that apoptosis and pro-apoptotic genes reaper and hid can also activate the JNK pathway (Adachi-Yamada et al. 1999; Ryoo et al. 2004), inducing mitogenic signals from apoptotic cells that in turn stimulate compensatory proliferation (Perez-Garijo et al. 2004, 2009; Ryoo et al. 2004). However, it has also been shown that JNK signalling switches its pro-apoptotic role to a pro-growth effect in the presence of oncogenic Ras (Uhlirova et al. 2005; Igaki et al. 2006; Uhlirova & Bohmann 2006), indicating that JNK signalling is context dependent. JNK is required for healing and early regeneration as puc-expressing cells do not die (Bosch et al. 2005, 2008; Lee et al. 2005; Mattila et al. 2005; Bergantiños et al. 2010a) and cell lineage studies of the puc-expressing domain show that these cells contribute to the regenerated tissue (Bosch et al. 2008; Bergantiños et al. 2010b). Thus, the JNK pathway responds rapidly to epithelial disc perturbations, possibly triggered by cell polarity, cytoskeleton variations, and/or stress.

There is a significant difference between the cellular response after irradiation and after injury. In the first, compensatory proliferation replaces the scattered dead cells. In the second, wherein a piece of a disc is microsurgically removed or ablated by directed cell death, proliferation and regenerative growth is preceded by cytoskeletal rearrangements resulting in cell shape and polarity changes (Bosch et al. 2005; Mattila et al. 2005; Bergantiños et al. 2010a). After imaginal disc injury, the columnar and squamous epithelia that characterize the disc establish transient heterotypic contacts (Reinhardt et al. 1977), the wound edges contract, and the surface becomes reduced (Reinhardt & Bryant 1981). Cell-shape changes occur that direct the epithelia toward the wound edges (Bosch et al. 2005). And finally, actin extensions develop at the wound edges to seal the wound.

Wound healing after epithelia injury also appears to involve additional signal transduction proteins. Grainy head (Grh) is a transcription factor involved in septate junction, epithelial sheet integrity, and wound healing in both flies and mice (Mace et al. 2005; Pearson et al. 2009). A transcriptional target of Grh, stitcher (stit), encodes a Ret-family receptor tyrosine kinase that activates Grh and extracellular signal-regulated kinase (ERK) signalling, and coordinates cytoskeletal rearrangements resulting in wound re-epithelization (Wang et al. 2009). Regulation of actin cytoskeleton through the small Rho GTPases is critical in Drosophila and vertebrate wound healing (Brock et al. 1996). The small GTPases also play a critical role in WNT/Frizzled planar cell polarity (PCP) signalling in Drosophila (Strutt et al. 1997; Yan et al. 2009), leading to the hypothesis that the PCP pathways could be important for epidermal repair. This has been demonstrated clearly in mammals where multiple PCP gene mutants exhibit defects in wound healing (Caddy et al. 2010).

Another question is what are the primary responses triggered by initial activation of signalling following injury. Matrix metalloproteinase-1 (Mmp-1), which is important for basement membrane degradation, is regulated by JNK and fos (Uhlirova & Bohmann 2006), and has been found localized near the wound in regenerating discs (McClure et al. 2008). Other effectors include the cell polarity factor Bazooka and the cytoskeleton adaptor D-Paxilin, through which JNK modulates cell–cell contacts and maintains sheet migration (Llense & Martin-Blanco 2008). Accordingly, it has been proposed that JNK controls epithelial organization not only in regeneration and tissue repair, but also in collective cell movements during morphogenesis (Llense & Martin-Blanco 2008). Additionally, a major class of genes whose expression increases during early regeneration have AP1 transcription factor binding sites in their regulatory regions (Blanco et al. 2010). These sequences are likely targeted by the Fos and Jun dimer that forms the AP-1 transcription factor. Many of these genes are transcription factors that somehow might be involved in tissue remodeling.

JAK/STAT signalling promotes cell proliferation in different cellular contexts (Zeidler et al. 2000). After wounding, unpaired (upd) genes, which encode the JAK/STAT-activating cytokines related to interleukin-6, are elevated in a JNK dependent manner (Pastor-Pareja et al. 2008). It has been found, using tumor models, that the JAK/STAT signalling pathway is involved in the growth of highly-active tumors generated after RasV12 overexpression in cells mutant for scribble (Wu et al. 2010). Moreover, JNK is upregulated in tumors generated by scribble mutations (Brumby & Richardson 2003; Pastor-Pareja et al. 2008; Igaki et al. 2009). In these flies, JAK/STAT signalling and overgrowth is reduced in hepr75 backgrounds (Wu et al. 2010). In addition, JNK activity propagates after wounding (Wu et al. 2010). Thus, it is possible that the JAK/STAT signal downstream of propagating JNK activation may function in regeneration.

Reprogramming model of disc regeneration

We have seen that regeneration entails healing and re-epithelization after genetically induced cell death or injury. Early responses to trauma include GTPase-, JNK-, and Grh-promoted cytoskeleton organization and cell-polarity reconstitution. As regeneration proceeds, JNK is found, not only in wound edge cells, but in broader domains (Bosch et al. 2005, 2008; Smith-Bolton et al. 2009; Bergantiños et al. 2010a; Wu et al. 2010). Interestingly, the phosphatase puc establishes a domain where tissue organization and regenerative growth interplay. This contrasts with the proposed role of JNK acting as a cell death-promoting signal to eliminate developmentally aberrant cells from a tissue (Igaki 2009). It has also been found that imaginal disc cells lacking puc die and ectopic expression of puc prevents apoptosis (McEwen & Peifer 2005). As puc inactivates the Drosophila JNK basket by dephosphorylation, the function of this phosphatase may be to maintain low JNK signalling levels to avoid cell death. This fits with a model of regeneration wherein the JNK is rapidly activated near the wound and a threshold of JNK activation is maintained by puc to protect the cell from death and to allow healing and subsequent regenerative growth (Fig. 3).

Figure 3.

 Model for disc regeneration. Microsurgical injury or removal of domains by cell death activation results in epithelial damage. The signal response includes transient JNK pathway activation. A first peak of JNK could be important to initiate the cytoskeleton reorganization and re-epithelization, including polarity. However, sustained high levels of JNK should be avoided to prevent JNK dependant cell death. Thus high levels of JNK will result in puckered (Puc domain) expression that will temper the signal by dephosphorylation of the JNK basket. This will result in the establishment of the puckered (puc) domain, wherein the disc epithelium will be protected from death. Regeneration requires growth and re-patterning, and also the recovery of the morphogen sources. Permissive epigenetic remodeling and developmental reprogramming of the existing tissue will allow re-epithelization.

Regeneration, in addition to healing and proliferation, requires repatterning of the lost tissue. This implies genetic reprogramming of the cells in order to switch fates. After apoptosis, newly formed tissue is derived from nearby cells and these cells must change their state of determination to contribute to the lost structure. Epigenetic marks ensure the maintenance of particular genetic programs and are responsible for the fixed-determination states that characterize late imaginal disc development. Some polycomb group (PcG) proteins are epigenetic modifiers that maintain cellular fates by controlling the expression patterns of developmental regulators (Ringrose & Paro 2004). They form chromatin complexes able to silence large numbers of genes by histone modification. PcG silence target genes by depositing repressive marks such as histone H3 lysine 27 trimethylation (H3K27me3). There is evidence that JNK activated during regeneration downregulates PcG genes and therefore allows the transcription of genes otherwise silenced (Lee et al. 2005). Wound healing and tissue repair in mammals is also associated with suppression of PcG genes, dramatic reduction of the repressive H3K27me3 mark, and upregulation of demethylases (Shaw & Martin 2009). The requirement of demethylases has also been reported in zebrafish fin regeneration (Stewart et al. 2009). This creates a cellular context in which transcription of growth-promoting genes are released from inhibition. For example, the dmyc proto-oncogene, which has been associated to disc growth (Johnston et al. 1999; de la Cova et al. 2004; Moreno & Basler 2004), is one of the potential targets of PcG and is derepressed in a PcG dependent manner in mammalian tissue repair (Shaw & Martin 2009). Interestingly, dmyc is derepressed during Drosophila wing disc regeneration (Smith-Bolton et al. 2009).

However, epigenetic changes will only be permissive. The increase of JNK activity, below the threshold controlled by puc, and the inhibition of gene silencing might be a prerequisite to release regenerative growth signals and switch to a new genetic program. Gradients, signalling molecules, and transcription factors are required to reconstitute the gene-regulatory network mimicking the developmental scenario. This is dependent on the capabilities of the damaged tissue to reuse signals required for its own development and therefore reprogram cells to reconstruct the lost part. Reprogramming exists in normal development; for example, Drosophila embryos use JNK signalling in the anterior compartment to promote anterior-to-posterior reprogramming of a few cells through de novo expression of engrailed, which is used during normal development as a posterior selector gene (Gettings et al. 2010). In summary, regeneration requires a field of JNK expression, elimination of epigenetic marks and activation of developmental and genetic instructions that will promote reprogramming. These requirements must be maintained at a level sufficient to reuse developmental instructions but not so much as to cause anomalous transformation.


This work was supported by the Consolider-Ingenio 2010 Program (CSD2007-00008) and BFU2009-09781 grants from the Spanish Ministerio de Ciencia e Innovación. We wish to thank Marina Ruiz for her support and valuable contribution. We also thank Cherie Byars for critically reading the manuscript.