DRONC Coordinates Cell Death and Compensatory Proliferation

Accidental cell death often leads to compensatory proliferation.In Drosophila imaginal discs, for example, ${gamma}$ -irradiation inducesextensive cell death, which is rapidly compensated by elevatedproliferation. Excessive compensatory proliferation can be artificiallyinduced by "undead cells" that are kept alive by inhibitionof effector caspases in the presence of apoptotic stimuli. Thissuggests that compensatory proliferation is induced by dyingcells as part of the apoptosis program. Here, we provide geneticevidence that the Drosophila initiator caspase DRONC governsboth apoptosis execution and subsequent compensatory proliferation.We examined mutants of five Drosophila caspases and identifiedthe initiator caspase DRONC and the effector caspase DRICE ascrucial executioners of apoptosis. Artificial compensatory proliferationinduced by coexpression of Reaper and p35 was completely suppressedin dronc mutants. Moreover, compensatory proliferation after ${gamma}$ -irradiation was enhanced in drice mutants, in which DRONC isactivated but the cells remain alive. These results show thatthe apoptotic pathway bifurcates at DRONC and that DRONC coordinatesthe execution of cell death and compensatory proliferation.

INTRODUCTION

Top

Introduction

Developing animal tissues have striking regenerative capacity.If a developing tissue is accidentally injured and a large numberof cells undergo cell death, the final shape and size of thetissue are often invariant, owing to additional cell proliferationthat compensates for the cell loss. The developing Drosophilalarval imaginal disc is a tissue with extremely high regenerativecapacity. Even when more than half of the cells are lost, theremaining cells undergo extra cell divisions and compensatefor the loss (22). This "compensatory proliferation" is, atleast in part, a local event that does not depend on the sizecontrol mechanism that governs the whole disc. For example,when apoptosis is locally induced by ectopic toxin expression,elevated cell proliferation is observed only around the siteof apoptosis (40), suggesting that cells can perceive apoptosisoccurring in the vicinity and undergo extra cell divisions untilthe original cell number is restored. Currently, little is knownabout how compensatory proliferation is regulated.

One attractive hypothesis is that dying cells actively induceproliferation of the neighboring cells by secreting mitogensbefore they disappear. This hypothesis assumes that the non-cell-autonomousinduction of compensatory proliferation is a process that isunder the control of the intracellular signaling pathway ofapoptosis. According to this model, if cells could be stimulatedto undergo apoptosis but artificially kept alive, the intracellularapoptosis signaling pathway would be activated and the "undeadcells" produce mitogens indefinitely as long as the apoptoticstimulus persists. As a consequence, the neighboring cells wouldbe misguided into conducting unrestrained compensatory proliferation.Several recent studies investigated this hypothesis using theDrosophila wing disc (25, 45, 51). The trick they used is tostimulate cells to undergo apoptosis and simultaneously blockcell death by overexpressing p35, a potent inhibitor of effectorcaspases. Effector caspases are proteases that lie at the finalstep in the apoptosis signaling cascade. Thus, in such "undeadcells," the signaling cascade up to effector caspases is activated,but the cells fail to undergo apoptosis. Indeed, they foundthat excessive non-cell-autonomous proliferation is inducedby p35-expressing cells if they are stimulated to undergo apoptosis(25, 45, 51). "Undead cells" ectopically express Wingless (WG)and Decapentaplegic (DPP), two major morphogens that shape thewing disc, of which DPP was found to be required for the abnormalovergrowth in a subsequent study (46).

What remains unanswered is which molecule of the apoptosis signalingpathway triggers compensatory proliferation. Two previous studiesaddressed this issue, with different conclusions (25, 51). Apoptosisin Drosophila is typically induced by transcriptional activationof the proapoptotic genes named reaper, hid, and grim (RHG)(1, 36, 37). Before apoptotic induction, caspases are constantlyinhibited by the ubiquitin ligase DIAP1. The RHG proteins triggerapoptosis by blocking the function of DIAP1, which allows autocatalyticactivation of the initiator caspase DRONC (Nedd2-like caspase-FlyBase).Interestingly, DRONC is insensitive to p35 inhibition (21, 39),which suggests that DRONC is activated in p35-expressing "undeadcells." Huh et al. suggested that the activation of DRONC isrequired for compensatory proliferation (25). They found thatunrestrained compensatory proliferation induced by "undead"cells is suppressed by dominant-negative DRONC. On the otherhand, Ryoo et al. suggested that DIAP1 downregulation activatesJNK signaling and mitogen production independently of DRONC(51). They showed that unrestrained compensatory proliferationis accompanied by activation of JNK signaling and ectopic WGexpression. The ectopic WG expression is suppressed by overexpressionof the JNK-inhibiting phosphatase Puckered, but not by dominant-negativeDRONC. The sources of these discrepancies are not clear. Thetwo studies used different dominant-negative forms of DRONC,different GAL4 drivers, and different UAS reporters, makingit difficult to directly compare the results.

The idea that DRONC, the most crucial component of the apoptosissignaling pathway, also regulates compensatory proliferationis particularly interesting. However, the prior studies areat odds with each other over the role of DRONC in compensatoryproliferation. Unfortunately, the demonstration by Huh et al.is not conclusive because it is conceivable that the dominant-negativeDRONC, which is a catalytically inactive form that can stillbind to DIAP1, blocks compensatory proliferation not by inhibitingDRONC but by antagonizing DIAP1. Ideally, the requirement ofDRONC and other caspases in compensatory proliferation wouldbe unambiguously determined by testing whether the massive overgrowthinduced by coexpression of p35 and RHG can be completely suppressedin null mutants of caspases. In the present study, we addresswhether caspases are required for compensatory proliferationby mutant analysis. There are seven caspases in the Drosophilagenome: three initiator caspases (Dredd, DRONC, and STRICA [Dream-FlyBase])and four effector caspases (DCP-1, DRICE [Ice-FlyBase], DECAY,and DAMM) (7, 11, 12, 13, 14, 19, 52). Biochemical and geneticstudies have shown that DRONC is a major initiator caspase requiredfor the execution of apoptosis (8, 10, 11, 20, 39, 48, 55, 60).Although another initiator caspase DREDD has been reported tobe specifically expressed in apoptotic cells (7), null mutantsof DREDD have no defect in apoptosis but show defective innateimmunity (34). Biochemical studies have suggested that DRICEis a major effector death caspase (14, 15), but its role invivo has been elusive owing to the lack of drice mutants. Herewe report on the isolation of null mutations in dronc, drice,and decay and show that execution of apoptosis predominantlydepends on DRONC and DRICE. Using these mutants, we show thatthe initiator caspase DRONC is also a crucial regulator of compensatoryproliferation.

MATERIALS AND METHODS

Top

Materials and Methods

Fly stocks and genetics. The following fly stocks were used: EY08574 (2); GE284897 andGE2731 (from EP collection established by GenExel, Inc., Korea),GMR-Dmp53 (43), GMR-reaper (57), GMR-hid (17), GMR-grim (6),dredd^B118 (34), hs-FLP (18), ey-FLP (41), dcp-1^prev1 (31), yw;ey-GAL4 UAS-FLP/CyO y⁺; GMR-hid l(3)CL3L FRT2A/TM6 y⁺ (53),w; P[ovo^D1-18]^3L FRT2A/st betaTub85D^D ss e/TM3 Sb (9), UAS-reaper(17), UAS-p35 (62), and hh-GAL4 (kindly provided by TetsuyaTabata).

Deletion alleles of dronc, drice, and decay were obtained byremobilization of EY08574, GE284897, and GE2731, respectively.For each caspase locus, up to 500 candidate excision eventswere screened by genomic PCR that amplifies the whole locus.Four deletion lines—dronc^A8, dronc^B3, dronc^D8, and dronc^E10—wereobtained at the dronc locus. A precise excision line, dronc^A1was used as a wild-type control in compensatory proliferationafter ${gamma}$ -irradiation. Two deletion lines, drice^1E4 and drice^2C8were obtained at the drice locus. Both of them delete the entireopen reading frame and are null alleles. We could not obtaina precise excision line. drice^1A1 is an imprecise excision thatretains the drice open reading frame. drice^1A1 animals are viableand fertile, showing no defect in apoptosis. It was also usedas wild-type control in compensatory proliferation after ${gamma}$ -irradiation.Two deletion lines, decay^H10 and decay^K2, were obtained at thedecay locus. The regions deleted in decay^H10 and decay^K2 includethe transcription start site and part of the catalytic subunit;thus, we believe both of them are null alleles. A precise excisionline decayS3 was also obtained.

For eye-specific mosaic analysis of dronc, yw ey-FLP; droncA8FRT2A/TM6B was crossed to GMR-X/CyO; GMR-hid l(3)CL3L FRT2A/TM6B(see Fig. 3). Maternal/zygotic dronc embryos were obtained bythe FLP-DFS system (see Fig. 2B).

View larger version (111K):
[in this window]
[in a new window]
FIG. 3. Requirement of caspases in RHG-induced apoptosis. (A to L) Eyes of adult flies. GMR-hid induces dramatic reduction in eye size (B). These phenotypes are suppressed in dronc (C) and drice (E) backgrounds but not in dredd (D), dcp-1 (F), and decay (G) backgrounds. The exact genotypes are as follows: A, wild type; B, yw/Y; GMR-hid/+; C, yw ey-FLP/Y; GMR-hid/+; GMR-hid l(3)* FRT2A/dronc^A8 FRT2A; D, yw dredd^B118/Y; GMR-hid/+; E, yw/Y; GMR-hid/+; drice^2C8/drice^2C8; F, yw/Y; dcp-1^Prev1/dcp-1^Prev1; GMR-hid l(3)* FRT 2A/+; G, yw/Y; GMR-hid/+; decay^K2/decay^K2. ey-FLP and GMR-hid l(3)* FRT2A were used to create whole-eye clones of dronc^A8 (C and L). The same GMR-hid l(3)* FRT2A chromosome was used to examine the requirement of Dcp-1 in Hid-induced apoptosis (F). (H to J) Eye imaginal discs of wandering third-instar larvae stained with acridine orange. (H) GMR-hid induces extensive apoptosis posterior to the morphogenetic furrow. (I) dronc mutants completely suppress Hid-induced apoptosis. (J) In drice mutants, suppression of Hid-induced apoptosis is almost complete, but sporadic apoptosis persists, suggesting that another effector caspase may play a minor role in Hid-induced apoptosis. Genotypes: H, GMR-hid/+; I, GMR-hid/+; dronc^A8/dronc^A8; J, GMR-hid/+; drice^2C8/drice^2C8. (K and L) Eyes of adult flies; (M and N) eye imaginal discs of wandering third-instar larvae stained with acridine orange. Genotypes: K and M, yw/Y; GMR-Dmp53/+; L and N, yw ey-FLP/Y; GMR-Dmp53/+; GMR-hid l(3)* FRT2A/droncA8 FRT2A. GMR-Dmp53 induces apoptosis in the developing eye disc (M), which results in dramatic reduction in eye size (K). Apoptosis in the developing eye disc is suppressed in a dronc background (N), but the eye size reduction is only moderately suppressed (L). (O and P) Eye discs stained with anti-ELAV (green) and anti-Bar (magenta). Genotypes: O, wild type; P, yw ey-FLP/Y; GMR-Dmp53/+; GMR-hid l(3)* FRT2A/dronc^A8 FRT2A. In the wild-type eye disc, ELAV is expressed in all photoreceptor neurons, whereas Bar is expressed in R1/R6 photoreceptor (O). If Dmp53 is overexpressed in a dronc background, photoreceptor organization is disturbed and few cells express Bar (P).

View larger version (42K):
[in this window]
[in a new window]
FIG. 2. Requirement of caspases in developmental apoptosis. (A to F) Embryos were stained by using TUNEL. (A) Wild type; (B to F) maternal/zygotic caspase mutants. Genotypes: B, dronc^A8/dronc^A8; C, dredd^B118/dredd^B118; D, drice^2C8/drice^2C8; E, dcp-1^Prev1/dcp-1^Prev1; F, decay^K2/decay^K2. Maternal/zygotic dronc^A8 embryos were obtained from germ line clones. Apoptosis is completely absent in dronc mutants and greatly reduced in drice mutants. (G to M) Eye imaginal discs of wandering third-instar larvae stained with acridine orange. Genotypes: G, wild type; H, drice^2C8/drice^2C8; I, dcp-1^Prev1/dcp-1^Prev1; J, decay^K2/decay^K2; K, droncA8/dronc^A8; L, dredd^B118/dredd^B118; M, dcp-1^Prev1/dcp-1^Prev1; drice^2C8/drice^2C8. Apoptosis is completely absent in dronc (K). Apoptosis is greatly reduced in drice (H) and dcp-1 (I) mutants and completely absent in drice/dcp-1 double mutants (M).

Immunohistochemistry. Immunohistochemistry of imaginal discs was done as describedpreviously (49). The following antibodies were used: rabbitanti-phospho-histone H3 (1:1,000; Upstate Technologies), mouseanti-WG 4D4 (1:20; Hybridoma Bank), rat anti-ELAV 7E8A10 (1:20;Hybridoma Bank), rabbit anti-Bar (1:200) (24), rabbit anti-p35(1:100) (54), and rabbit anti-DIAP1 (1:1,000) (30). Fluoresceinisothiocyanate-, Alexa 488-, Cy3-, and Cy5-conjugated secondaryantibodies were obtained from Jackson Immunoresearch and MolecularProbes and used at 1:100 to 101:500. Pictures were taken usinga confocal microscope. Acridine orange staining of imaginaldiscs was done as described previously (27). Pictures were takenon an epifluorescence microscope. TUNEL (terminal deoxynucleotidyltransferase-mediateddUTP-biotin nick end labeling) staining of embryos was doneas described previously (28). Cy3-conjugated streptavidin wasused for detection. Pictures were taken on a confocal microscope.Either a Zeiss LSM5 Pa or Zeiss LSM510 META microscope was usedfor confocal microscopy.

${gamma}$ -irradiation. Wandering third-instar larvae were irradiated with 20 Gy or40 Gy of ${gamma}$ rays at 1 Gy/min using a ¹³⁷Cs source. They were subsequentlyallowed to recover at 25°C.

RESULTS

Top

Results

Phenotypic analysis of caspase mutants. Previous studies have shown that compensatory proliferationis induced by the apoptosis signaling pathway. In order to understandthe underlying molecular mechanisms, we need to know which componentof the apoptosis signaling pathway triggers compensatory proliferation.We chose to examine the requirement of caspases for compensatoryproliferation because these proteases may be able to activatea variety of substrates through their cleavage. There are sevencaspases in the Drosophila genome. At the time we started thepresent study, mutants were available only for dredd and dcp-1(31, 34). We created null mutants of three other caspases, decay,drice, and dronc by imprecise excision of P-element insertionsand analyzed their phenotypes. Figure 1 summarizes the extentof deletions in the mutant alleles.

View larger version (23K):
[in this window]
[in a new window]
FIG. 1. Generation of caspase mutants. The genomic structures of caspases are schematized. A P-element was remobilized to create deletions in each caspase gene. The translated and untranslated regions are depicted as empty and filled rectangles, respectively. The insertion sites of the original P-elements are indicated by the inverted triangles. The breakpoints of each deletion are indicated by the parentheses. (A) dronc locus. EY08574 is a P element inserted in the 5' untranslated region of dronc. The 5' breakpoints of the deletions in dronc^A8, dronc^B3, and dronc^D8 are identical to the insertion site of EY08574. The breakpoint of dronc^E10 is upstream of the transcription start site of dronc. (B) drice locus. The entire open reading frame is deleted in drice^1E4 and drice^2C8. (C) decay locus. The first exon, which contains part of the catalytic domain, is deleted in decay^K2 and drice^H10.

First, we examined apoptosis in initiator caspase mutants. Fourdeletions were recovered at the dronc locus. All alleles exceptfor dronc^E10 showed indistinguishable phenotypes. Homozygousdronc animals died as early pupae. A very low fraction (<0.1%)of dronc homozygous animals emerged as adult flies. Their externalmorphology was apparently normal, except that their wings wereopaque due to the failure of the wing cells to undergo apoptosis.Maternal/zygotic dronc mutant animals were embryonic lethal,with almost complete absence of apoptosis (Fig. 2B). Developmentalapoptosis was also abrogated in the mutant larval imaginal discs(Fig. 2K). Such developmental apoptosis was normal in dreddmutants (Fig. 2C and L). These results indicate that DRONC isa major initiator caspase required for apoptosis, although theother putative initiator caspase, STRICA, which was excludedfrom our current analysis, may also have minor contributionto developmental apoptosis (33). One dronc allele, dronc^E10,behaved differently from the other alleles. Homozygous dronc^E10animals showed late larval lethality with a prolonged larvalstage and occasional formation of melanotic tumors, phenotypessimilar to those of dronc⁵¹ (8). Both dronc^E10 and dronc⁵¹ removethe putative promoter region of the adjacent gene CG6685 alongwith the first exon of dronc. Their phenotypes likely resultfrom the combined effect of inactivating both dronc and CG6685.

We next examined effector caspase mutants. decay mutants werecompletely viable and fertile without any obvious morphologicalabnormalities. Developmental apoptosis was normal in decay mutants(Fig. 2F and J). drice mutants were mostly pupal lethal, andtheir lethal phase was later than that of dronc mutants (datanot shown). Many of the drice mutant animals developed intopharate adults, with occasional adult escapers emerging at afrequency of <10%. In maternal/zygotic drice mutant embryosand mutant imaginal discs, apoptosis was greatly reduced butwas not completely abolished (Fig. 2D and H). Another effectorcaspase, dcp-1 is a closely related homolog of drice and isnot essential for viability (31, 52). Biochemical studies haveshown that both drice and dcp-1 are substrates of dronc. Althoughdcp-1 mutants are completely viable, we found that developmentalapoptosis in the larval eye disc was significantly reduced (Fig.2I). Therefore, we speculated that the milder phenotype of dricewould be due to its functional redundancy with dcp-1. Indeed,developmental apoptosis in the eye disc was completely suppressedin drice/dcp-1 double mutants (Fig. 2M), and the lethal phaseof drice/dcp-1 double mutants was now indistinguishable fromdronc mutants (data not shown), suggesting that drice and dcp-1cooperatively act downstream of dronc, with drice playing amajor role.

In Drosophila, transcriptional upregulation of RHG proteinsis an essential step in the execution of all apoptosis. Accordingly,apoptosis requires specific transcription factors that regulatethe expression of RHG genes. For example, Dmp53, the Drosophilahomolog of the tumor suppressor p53, binds to the enhancer ofreaper and activates its transcription during stress-inducedapoptosis (5, 43, 44). Ectopic overexpression of RHG proteinsor Dmp53 in the developing eye imaginal disc using the GMR promoterinduces extensive apoptosis and causes dramatic reduction inadult eye size (6, 17, 43, 57; Fig. 3B, H, K, and M). To testthe requirement of caspases in RHG-induced apoptosis, we overexpressedRHG proteins in each caspase mutant background. The eye sizereduction caused by GMR-reaper, GMR-hid, and GMR-grim was allsuppressed in dronc and drice backgrounds (Fig. 3C, E, I, and J;also see Fig. S1 in the supplemental material). In contrast,we could not observe any phenotypic suppression in other caspasemutants (Fig. 3D, F, and G). Mutations in dronc and drice alsosuppressed apoptosis induced by overexpression of RHG proteins(Fig. 3H to J). Thus, apoptosis triggered by the RHG proteinsrequires the activity of caspases DRONC and DRICE. Althoughthe dronc mutation also suppressed apoptosis induced by overexpressionof Dmp53, it did not suppress the eye size reduction (Fig. 3K, L, M, and N).Indeed, the eye discs of GMR-Dmp53 dronc^–/– larvaeshowed a severe defect in neuronal differentiation (Fig. 3O and P).Targets of Dmp53 aside from RHG proteins likely affect neuronaldifferentiation, even when apoptosis is completely blocked.

DRONC is required for excessive compensatory proliferation. When RHG proteins and the baculoviral caspase inhibitor p35are locally overexpressed, enormous tissue overgrowth is observedin and around the area of ectopic expression. It has been proposedthat apoptosis signaling is constitutively activated in the"undead cells" overexpressing RHG proteins and p35, which inducesunrestrained compensatory proliferation (25, 51). We chose touse this phenotype as an indicator for the level of compensatoryproliferation. We overexpressed Reaper and p35 in the posteriorcompartment of the wing disc using hh-GAL4. The ectopic expressiondriven by hh-GAL4 is much higher than en-GAL4, another posterior-specificGAL4 driver, and the resultant overgrowth phenotype was indeedseverer than the previously reported phenotypes that were inducedby en-GAL4 (25, 51). As previously reported, WG was ectopicallyexpressed in the posterior compartment (Fig. 4C). Interestingly,overexpression of p35 alone often induced local overgrowths(Fig. 4B). During larval development, sporadic apoptosis occursin the wing disc (40). We assume that suppression of this naturallyoccurring apoptosis led to the observed local overgrowths.

View larger version (74K):
[in this window]
[in a new window]
FIG. 4. DRONC is required for artificially induced compensatory proliferation. (A to E) Single confocal sections of wing discs stained with anti-WG, anti-p35, and anti-DIAP1. Anti-p35 was used to visualize the posterior compartment of the wing disc (B to E). Genotypes: A, UAS-mGFP/+; hh-GAL4/+; B, UAS-p35/+; hh-GAL4/+; C, UAS-reaper UAS-p35/+; hh-GAL4/+; D, UAS-reaper UAS-p35/+; hh-GAL4 dronc^A8/dronc^A8; E, UAS-reaper UAS-p35/+; hh-GAL4 drice^2C8/drice^2C8. (A) Wild-type wing disc. The posterior compartment is marked by green fluorescent protein (GFP). DIAP1 protein is uniformly distributed throughout the wing disc. (B) The overall morphology and WG expression pattern of wing discs overexpressing p35 in the posterior compartment is identical to that of wild-type wing discs except for occasional local tissue overgrowth and ectopic WG expression (arrow). (C) Coexpression of Reaper and p35 in the posterior compartment induces dramatic tissue overgrowth, which is accompanied by ectopic WG expression. Reaper binds to DIAP1, thereby causing its destabilization. As a result, DIAP1 protein is reduced in the posterior compartment. (D and E) The overgrowth is suppressed in dronc mutants but not in drice mutants. Reduction of DIAP1 protein in the posterior compartment is unchanged in both dronc and drice mutants. All pictures were taken at the same magnification. Scale bar, 100 µm.

p35 is a broad-specificity caspase inhibitor, but it does notinhibit DRONC (39). Thus, we assume that the following eventstake place in the "undead" cells expressing Reaper and p35.First, Reaper downregulates DIAP1, which leads to activationof DRONC. Subsequently, DRONC transforms DRICE into an activeform by internal cleavage, but the cleaved DRICE is quicklyinactivated by p35. Therefore, there are four candidate eventsthat lead to the production of a compensatory proliferationsignal: expression of Reaper, downregulation of DIAP1, activationof DRONC, and cleavage of DRICE. To test these possibilities,we overexpressed Reaper and p35 in dronc and drice backgrounds.The overgrowth of the posterior compartment and the ectopicWG expression were not suppressed in a drice background (Fig.4E), indicating that neither the caspase activity of DRICE northe cleaved DRICE protein is required for compensatory proliferation.In contrast, they were completely suppressed in a dronc background,clearly demonstrating that compensatory proliferation requiresDRONC (Fig. 4D).

In agreement with our results, Huh et al. showed that the abnormalovergrowth and ectopic WG expression induced by "undead cells"is suppressed by a dominant-negative form of DRONC (25). However,under a different experimental regime, cells overexpressingHID, p35, and dominant-negative DRONC were still able to induceectopic WG expression (51). These contrasting results likelyarose from the use of dominant-negative forms of DRONC. Becausedominant-negative proteins act by competing with normal proteins,inactivation of endogenous protein can be incomplete (39).

Overexpression of RHG proteins has been shown to induce reductionof DIAP1 protein (23, 29, 50, 58, 59, 61). Indeed, DIAP1 proteinwas reduced in "undead cells" of the posterior compartment ofthe wing disc, which overexpressed Reaper and p35 (Fig. 4C).The reduction of DIAP1 protein in "undead cells" was unaffectedin a dronc mutant background, despite the fact that the tissueovergrowth was completely suppressed (Fig. 4E). Therefore, DIAP1downregulation alone is not sufficient to induce compensatoryproliferation. In contrast, overexpression of DRONC induceshyperproliferation without reducing the DIAP1 protein levels(25; see Fig. S2 in the supplemental material). Taken together,these observations suggest that activation of DRONC is sufficientto induce compensatory proliferation and that the major roleof DIAP1 downregulation in compensatory proliferation is toactivate DRONC.

Endogenous compensatory proliferation requires DRONC. Apoptotic cells are quickly removed by macrophages, usuallywithin a few hours. Accordingly, compensatory proliferationshould also occur immediately after apoptosis, since the mitogensthat induce compensatory proliferation are unlikely to be producedafter clearance of the apoptotic cells. The overgrowth of thewing disc caused by RHG proteins and p35 is a result of theiroverexpression throughout the larval stages, which amounts tomore than 100 h. In the "undead cells" that are artificiallykept alive for such a long period of time, many intracellularchanges are likely to take place. Therefore, we cannot ruleout the possibility that the overgrowth phenotype is a secondaryconsequence of apoptosis inhibition that occurs independentlyof the immediate compensatory proliferation.

To test the role of DRONC in naturally occurring compensatoryproliferation, we chose to examine the response of the wingimaginal disc upon damage caused by ${gamma}$ -irradiation (4, 26). High-dose ${gamma}$ -irradiation triggers immediate cell cycle arrest of the wingdisc cells within 1 h (4) (Fig. 6). During the cell cycle arrest,the DNA repair machinery attempts to repair double-strand breaksin the genomic DNA inflicted by the ${gamma}$ rays. If the damage isso extensive that it cannot be repaired, cells choose to undergoapoptosis. The cell loss is quickly restored by compensatoryproliferation, which manifests as elevated cell proliferationduring the recovery period (26).

View larger version (42K):
[in this window]
[in a new window]
FIG. 6. Compensatory proliferation is accelerated in drice mutants. Third-instar larvae were irradiated with 40 Gy of ${gamma}$ -irradiation and allowed to recover at 25°C. At the indicated times after irradiation, wing discs were dissected and stained with anti-phophohistone H3 to visualize mitotic cells. Wing discs stained without irradiation are indicated as "0 h." The genotypes of the animals shown in each column are wild type, dronc^A8/dronc^A8, drice^2C8/drice^2C8, and dronc^A8 drice^2C8/dronc^A8 drice^2C8 (from left to right). The wild-type strain shown in this figure is dronc^A1, a revertant line derived from precise excision of EY08574. We also examined mitosis after irradiation in the drice revertant line drice^1A1 and found that it was indistinguishable from that of dronc^A1. After irradiation, wing discs of all genotypes undergo immediate G₂ arrest, which is still evident at 4 h after irradiation. In drice mutants, recovery from G₂ arrest, which is likely to reflect compensatory proliferation, starts at 6 h after irradiation, much earlier than for wild-type animals. In dronc/drice double mutants, this rapid recovery is suppressed, suggesting that DRONC accelerates the onset of mitosis by inducing compensatory proliferation.

It has been shown that ${gamma}$ -irradiation induces reaper mRNA expression(42), and RHG deletion mutants are resistant to irradiation-inducedapoptosis (56). Thus, one would expect that ${gamma}$ -irradiation bringsabout apoptosis by upregulating RHG proteins, which promotedegradation of DIAP1 and subsequent activation DRONC and DRICE.To test whether ${gamma}$ -ray-induced apoptosis depends on DRONC andDRICE, we irradiated wild-type and caspase mutant third-instarlarvae with 40 Gy of ${gamma}$ -irradiation. Between 4 and 7 h after irradiation,extensive apoptosis is induced in wild-type wing discs (Fig.5A,D). ${gamma}$ -ray-induced apoptosis was completely suppressed in droncand drice mutants (Fig. 5B, C, E, and F), whereas apoptosiswas normally induced in dredd, dcp-1, and decay mutants (Fig.5G to I). These results establish that DRONC and DRICE are necessaryfor the induction of apoptosis induced by DNA damage.

View larger version (71K):
[in this window]
[in a new window]
FIG. 5. DRONC and DRICE are required for ${gamma}$ -ray-induced apoptosis. (A to I) Wing imaginal disc stained with acridine orange before and after ${gamma}$ irradiation. Wandering third-instar larvae were irradiated with 40 Gy of ${gamma}$ rays and allowed to recover for 4 to 7 h. Before irradiation, little apoptosis is observed in the wild type and in dronc and drice mutants (A to C). After irradiation, apoptosis is induced throughout the wild-type wing disc (D), which is completely suppressed in dronc and drice mutants (E and F). ${gamma}$ -ray-induced apoptosis is not suppressed in dredd, decay, and dcp-1 mutants (G to I). Genotype: A and D, wild type; B and E, dronc^A8/dronc^A8; C and F, drice^2C8/drice^2C8; G, dredd^B118/dredd^B118; H, decay^K2/decay^K2; I, dcp-1^Prev1/dcp-1^Prev1.

If compensatory proliferation is indeed triggered by DRONC,we can predict that, in drice mutants, cell proliferation after ${gamma}$ - irradiation will be enhanced compared to irradiated wild-typeanimals. Irradiated drice cells do not undergo apoptosis, butinduction of RHG proteins, downregulation of DIAP1, and activationof DRONC should normally occur, since these processes lie upstreamof DRICE in the apoptosis signaling cascade. Thus, in the "undeadcells," activated DRONC would induce continuous production ofmitogens, which enhances compensatory proliferation. To testthis possibility, we compared the time course of cell proliferationafter irradiation between the wild-type and caspase mutant animals.We irradiated wandering third-instar larvae with 40 Gy of ${gamma}$ -irradiationand allowed them to recover at room temperature. Both in thewild-type and in the caspase mutants, cell cycle arrest occurrednormally when examined at 1 h after irradiation. At this dose,mitotic cells appeared around 8 h after irradiation in the wild-typewing disc (data not shown). In drice mutants, mitotic cellswere already visible at 6 h after irradiation (Fig. 6), suggestingthat the resumption of mitosis was accelerated by the accumulatedmitogens produced by the "undead cells." In contrast, droncmutants were indistinguishable from the wild type (Fig. 6).To test whether the accelerated resumption of mitosis dependson DRONC, we examined dronc/drice double mutants. The accelerationin recovery was suppressed in dronc/drice double mutants (Fig.6), indicating that DRONC is responsible for the acceleratedresumption of mitosis that occurs in drice mutants. These resultsdemonstrate that DRONC is required for endogenous compensatoryproliferation that is induced by ${gamma}$ -irradiation.

DISCUSSION

Top

Discussion

In Drosophila imaginal discs, accidental cell loss due to damage-inducedapoptosis is rapidly recovered by a process known as compensatoryproliferation (40). Previous studies have shown that apoptoticcells actively induce proliferation of the neighboring cells,which suggests that induction of compensatory proliferationis part of the apoptosis signaling pathway (25, 45, 51). Thereare seven caspases in the Drosophila genome. We created mutationsin three of them, dronc, drice and decay and showed that droncand drice play major roles in apoptosis. Using these mutants,we tested the requirement of DRONC for compensatory proliferation.We found that the ability of "undead cells" to induce excessivecompensatory proliferation depends on DRONC. We then showedthat DRONC is required for immediate compensatory proliferationthat occurs after extensive apoptosis induced by high-dose ${gamma}$ -irradiation.Our results provide unambiguous genetic evidence that the apoptosissignaling pathway bifurcates at DRONC and that DRONC is requiredfor both DRICE activation and compensatory proliferation (Fig.7).

View larger version (14K):
[in this window]
[in a new window]
FIG. 7. Schematic diagram showing the proposed signaling cascade that regulates apoptosis and compensatory proliferation. First, apoptotic stimuli given to the cell activates RHG proteins. RHG proteins activate DRONC through inhibition of DIAP1 function. The signaling cascade bifurcates at DRONC, one pathway leading to apoptosis through cleavage of DRICE by DRONC. The other pathway leads to compensatory proliferation through cleavage of an as-yet-unknown factor (denoted as "X") by DRONC. Cleavage of factor "X" activates JNK signaling, which induces compensatory proliferation through transcriptional upregulation of mitogens including DPP.

Downstream events of DRONC activation. Activation of the apoptosis signaling pathway leads to transcriptionalupregulation of secreted factors WG and DPP, of which DPP isbelieved to induce proliferation of the neighboring cells (46).Since DRONC is a protease, it is unlikely that it activatesthe transcription directly. A candidate event responsible forthe transcription of dpp and wg is activation of JNK signaling.Ryoo et al. found that JNK signaling is activated in "undeadcells" and that it is required for the ectopic WG expressionand the disc size enlargement (51). Indeed, our preliminarydata also show that the massive tissue overgrowth induced by"undead cells" can be suppressed by blocking JNK signaling (ourunpublished observation).

Then, how is JNK signaling activated? It has been previouslyshown that a molecular pathway exists whereby downregulationof DIAP1 induces JNK activation through DTRAF1 (29, 47). However,downregulation of DIAP1 in "undead cells" is not sufficientto induce ectopic WG expression and massive overgrowth if DRONCactivity is blocked (25) (Fig. 4C). Therefore, we propose thatactivation of JNK signaling during compensatory proliferationis a downstream event of DRONC activation. Ryoo et al. showedthat JNK signaling is required for the ectopic WG expressioninduced by overexpression of HID and p35 (51). On the otherhand, Huh et al. showed that overexpression of DRONC is sufficientto induce ectopic WG expression (25). The simplest interpretationof these observations is that DRONC and JNK act in a singlelinear pathway to induce WG expression during compensatory proliferation.Indeed, coexpression of DRONC and p35 in the wing disc inducesJNK activation and abnormal overgrowth (see Fig. S2 in the supplementalmaterial). In mammals, Mst1, an upstream kinase of the JNK pathway,is activated through cleavage by caspase (16, 32). A similarupstream kinase may exist in Drosophila, which is specificallycleaved by DRONC and activates JNK signaling. Recently, it wasreported that activation of JNK signaling accompanies compensatoryproliferation of the liver (35). Thus, the mechanism uncoveredin Drosophila may also apply to compensatory proliferation inmammals.

Compensatory proliferation during Drosophila development. During normal development, a large number of cells undergo developmentallyprogrammed apoptosis, whereby specific cells are invariablyremoved according to the developmental program of the organism.We have shown that compensatory proliferation is induced by"unprogrammed" apoptosis, such as accidental stress-inducedapoptosis and RHG-induced artificial apoptosis of cells thatnormally do not undergo developmentally programmed cell death.On the other hand, there is little evidence that compensatoryproliferation is induced by developmentally programmed apoptosis.In fact, there are obvious cases in which no proliferation isinduced by apoptosis. For example, although a large number ofcells that are produced in excess undergo developmentally programmedapoptosis in the pupal retina of Drosophila, there is no accompanyingcell proliferation (3). This may be because there are no surroundingcells that respond to the mitogens released from the dying cell.It is also possible that developmentally programmed apoptosisoccurs only in a context that does not induce compensatory proliferation.

Interestingly, however, our result that inhibition of apoptosisby overexpression of p35 in an otherwise normal genetic backgroundinduces ectopic WG expression and overgrowth (Fig. 4A) indicatesthat there are apoptotic cells that induce compensatory proliferationeven during normal development. The sites of the ectopic overgrowthwere not consistent, which suggests that the inhibited apoptosiswas not developmentally programmed apoptosis but accidentalstress-induced apoptosis. This notion is further supported bythe fact that sensitizing cells to stress-induced apoptosissimply by half-dose reduction of the puckered gene induces massivetissue overgrowth upon inhibition of apoptosis by p35 (38).

In nature, animals are constantly exposed to environmental stressorssuch as UV radiation, dynamic temperature change, and low nutritionalconditions. Our finding suggests that, even under laboratoryconditions, stress-induced apoptosis occurs during normal developmentto induce compensatory proliferation. Since there are more environmentalstressors in nature, induction of compensatory proliferationduring normal development is probably a frequent event and isnot limited to extremely stressful conditions. Although thesignaling pathway downstream of DRONC in compensatory proliferationis not yet clear, it likely involves secreted molecules, suchas WG and DPP, which are also used for developmental patterningand proliferation. Thus, the developing tissue must interprettwo signals: spatiotemporal information within the tissue formorphogenesis and the signal from apoptotic cells. The Drosophilaimaginal disc is likely equipped with a mechanism that integratesthese two signals, by which a tissue with the normal patternis restored after injury.

ACKNOWLEDGMENTS

We thank the Bloomington Drosophila Stock Center; the DevelopmentalStudies Hybridoma Bank; GenExel, Inc.; Kimberly McCall; TetsuyaTabata; and Takaomi Saido for reagents. We thank Hoshio Eguchifor assistance with ${gamma}$ -irradiation. We thank members of the Miuraand Hiromi laboratories for helpful discussions.

This study was supported by grants from the Japanese Ministryof Education, Science, Sports, Culture, and Technology (to M.M.and Y.H.); the CREST program of the Japan Science and TechnologyAgency (to Y.H.); and the Cell Science Research Foundation (toM.M.).

FOOTNOTES

* Corresponding author. Mailing address for Masayuki Miura: Department of Genetics, Graduate School of Pharmaceutical Sciences, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Phone: 81-3-5841-4860. Fax: 81-3-5841-4867. E-mail: miura{at}mol.f.u-tokyo.ac.jp. Mailing address for Yasushi Hiromi: Division of Developmental Genetics, National Institute of Genetics, 1111 Yata, Mishima, Shizuoka 411-8540, Japan. Phone: 81-55-981-6809. Fax: 81-55-981-6868. E-mail: yhiromi{at}lab.nig.ac.jp.

Supplemental material for this article may be found at http://mcb.asm.org/.

Present address: Department of Genetics and Development, Collegeof Physicians and Surgeons, Columbia University, 701 West 168thSt., New York, NY 10032.

REFERENCES

Top