Loss of putzig Activity Results in Apoptosis during Wing Imaginal Development in Drosophila

The Drosophila gene putzig (pzg) encodes a nuclear protein that is an integral component of the Trf2/Dref complex involved in the transcription of proliferation-related genes. Moreover, Pzg is found in a complex together with the nucleosome remodeling factor NURF, where it promotes Notch target gene activation. Here we show that downregulation of pzg activity in the developing wing imaginal discs induces an apoptotic response, accompanied by the induction of the pro-apoptotic gene reaper, repression of Drosophila inhibitor of apoptosis protein accumulation and the activation of the caspases Drice, Caspase3 and Dcp1. As a further consequence ‘Apoptosis induced Proliferation’ (AiP) and ‘Apoptosis induced Apoptosis’ (AiA) are triggered. As expected, the activity of the stress kinase Jun N-terminal kinase (JNK), proposed to mediate both processes, is ectopically induced in response to pzg loss. In addition, the expression of the mitogen wingless (wg) but not of decapentaplegic (dpp) is observed. We present evidence that downregulation of Notch activates Dcp1 caspase and JNK signaling, however, neither induces ectopic wg nor dpp expression. In contrast, the consequences of Dref-RNAi were largely indistinguishable from pzg-RNAi with regard to apoptosis induction. Moreover, overexpression of Dref ameliorated the downregulation of pzg compatible with the notion that the two are required together to maintain cell and tissue homeostasis in Drosophila.


Introduction
Cellular and tissue homeostasis describes a complex process ensuring the survival and correct development of an organism. Apoptosis, the major form of programmed cell death, contributes to tissue homeostasis by eliminating aberrant, surplus or malignant cells during 'normal' development and in response to stress induced conditions. This safeguarding system demands a fine-tuned control as an unregulated apoptosis has been connected to various human diseases including cancer (reviewed in [1][2][3][4]). Research in several model organisms including Drosophila has expanded our knowledge on the molecular mechanisms underlying the well conserved apoptotic execution program in metazoans: Under normal conditions, cell survival is guaranteed by the Inhibitors of Apoptosis Proteins (IAP, DIAP1 in Drosophila), which bind and inhibit caspases, the key executing enzymes of apoptosis (reviewed in [4][5][6]). In contrast, if cell death is triggered, e.g. under cellular stress conditions, pro-apoptotic gene activity is induced in Drosophila: Pro-apoptotic gene products include the DIAP1-antagonists Hid, Rpr and Grim, which themselves mediate the ubiquitin dependent degradation of DIAP1 thereby enabling caspases to provoke the death of the cell [7][8][9][10][11][12][13].
Although diverse stress signals provoke a strong apoptotic answer in the respective tissue, organisms can often compensate this cell loss allowing them to survive with no or only minor consequences on the final tissue or body size. This intriguing fact has been designated Apoptosis-induced Proliferation (AiP) and describes the striking property of dying cells to stimulate proliferation of adjacent surviving cells, and therefore tissue regeneration (reviewed in [14][15][16][17]). The model system Drosophila with its sophisticated genetic methods offers the opportunity to study the mechanisms underlying the communication between damaged cells and the surrounding tissues. Here, cell death was experimentally induced in cells of larval imaginal discs and, as a consequence, the ectopic induction of mitogens like wingless (wg), decapentaplegic (dpp) or hedgehog (hh) was often observed in the apoptotic cells [18][19][20][21]. These mitogens, the key players in the regulation of morphogenesis and growth in the course of Drosophila development [22,23], provide a well-founded explanation for the compensatory proliferation of neighboring cells. However, since the genuine dying cells are often rapidly eliminated, much of our current knowledge on AiP was deduced from a special type of apoptotic cells referred to as 'undead' cells. These cells are experimentally obtained by provoking cell death while expressing the caspase inhibitor p35 at the same time. p35 specifically blocks the function of the effector caspases Drice and Dcp-1 without affecting the activity of the initiator caspase Dronc [19,24,25]. Therefore, these cells are trapped in the execution of cell death, but fail to complete the process due to the blocked function of effector caspases [18,19,26]. This experimental approach led to the conclusion that the Jun N-terminal kinase (JNK) acts as a central player of AiP in the dying cells. A robust activity of this stress kinase is associated with diverse aspects of tissue regeneration, including the expression of the aforementioned mitogens and the delay of larval development that keeps the animal in the growth phase ( [27][28]; reviewed in [29]). Moreover, as dying cells lose their epithelial integrity, JNK-signaling enforces the restitution of an intact epithelium including the formation of actin cables and filopodia in accordance with its well-defined role in the healing of epidermal wounds [30][31][32][33].
Intriguingly, JNK-signaling activity is not only associated with emanating proliferative signals from the apoptotic cells, but also with the non-autonomous induction of secondary cell death at a considerable distance from the primary cell death source. This phenomenon was recently termed Apoptosis induced Apoptosis (AiA) and might be the mechanistic explanation for the systemic cell death occurring both in normal development of metazoans as well as in some human pathologies like e.g. neurodegenerative disorders ( [34]; reviewed in [35]). Obviously, tissue homeostasis is coordinated by a cross-regulatory relationship of widespread signaling molecules keeping proliferation and apoptosis in a balanced ratio.
The DNA replication-related element-binding factor (Dref)-complex as well as the Notch (N) signaling pathway are both suitable candidates for being part of such a regulatory network, as they are known to govern numerous developmental processes including cell proliferation and apoptosis (reviewed in [36][37][38]). Dref acts as a transcription factor in Drosophila and is proposed to regulate the expression of a multitude of genes required for cell proliferation such as cell cycle regulators, growth factors or DNA replication factors (reviewed in [37]). A similar pleiotropic influence is mediated by the highly conserved N signaling pathway, being reiteratively used during the development of a variety of tissues in higher eumetazoa. Depending on the cellular context, N can either promote or inhibit growth processes emphasizing the importance of a tight and fine-tuned regulation of the signaling cascade (reviewed in [38,39]). In Drosophila, both signaling networks depend on putzig (pzg), an essential positive regulator of the Dref and N signaling cascades [40,41]. Pzg is an integral component of the Trf2/Dref protein complex that regulates proliferation-related genes [40]. Moreover, Pzg acts Dref independently and promotes N target gene activation via the Nucleosome remodeling factor (Nurf), implying a Pzg-mediated epigenetic influence on N target gene activation [41].
Here, we show that reduction of pzg activity during larval wing development results in the induction of genuine dying cells which initiate AiP mechanisms and enable AiA. As expected, ectopic JNK-signaling activity is induced autonomously and non-autonomously, likely to mediate the systemic response. This spectrum of consequences is not mimicked by a downregulation of N receptor activity: Though apoptosis and proliferation are induced, the latter is not mediated by the ectopic induction of wg in the genuine dying cells, unlike in the pzg depleted cells. In contrast, a downregulation of Dref activity does not only provoke apoptosis but also AiP, mediated by the induction of wg, similar to the effects observed after pzg knockdown. We conclude that Pzg is fundamental for the fine-tuned homeostasis of cell survival and proliferation via its influence on important signaling networks during the development of Drosophila.
For labeling cells undergoing DNA synthesis, the Click-iT EdU Alexa Fluor 488 Imaging Kit (Invitrogen, Eugene, Orgeon, USA) was used. After dissection in ice-cold PBS, larval tissues were incubated for two hours at room temperature in M3 insect medium including 10 mM EdU stock solution. A 4% para-formaldehyde fixative was added for 25 minutes, removed, and washed at least four times with PBS, followed by a 20 minutes incubation with PBX (PBS with 0.3% Triton X-100) and two washing steps for 10 minutes each. The Click-iT reaction cocktail was added containing 443.5 μl H 2 O, 21.5 μl 10X Click-iT reaction buffer, 10 μl CuSO 4 , 0.5 μl Alexa Fluor Azide and 25 μl reaction buffer additive, and the discs were incubated for 30 minutes in the dark. After removal of the reaction cocktail solution, the tissues were washed with PBS two times, prepared and embedded in Vectashield (Vector Laboratories, USA).
Confocal images were acquired with a Zeiss Axioskop linked to a Bio-Rad MRC1024 scanhead by using Bio-Rad Laser Sharp 3.1 software.

Documentation of adult eyes and statistical quantification of eye size
Adult eyes of females were documented with an ES120 camera (Optronics, Goleta CA, USA) using Pixera viewfinder software version 2.0. For quantification, eye size of five females each was measured with Image J and eye area was calculated. Statistical significance of probes was determined according to Student's T-test (http://www.physics.csbsju.edu/stats/t-test.html) and p-value was scaled accordingly: p>0.05 (not significant, n.s.); p<0.05 (weakly significant; Ã ); p<0.01 (significant; ÃÃ ); p<0.001 (highly significant; ÃÃÃ ).

Results
Pzg genetically interacts with pro-apoptotic genes and is required for cell survival Knock down of pzg gene activity by pzg-RNAi induction during larval development results in tissue size reduction (Fig 1A-1C). Previously, we have shown that pzg-RNAi effects are negatively correlated with cell cycle progression, thereby influencing cell proliferation and cell growth [40]. In order to test whether tissue loss and size reduction might also be a consequence of the induction of programmed cell death, we performed genetic interaction assays with the well defined apoptosis-inducing factors hid, rpr and grim, whose transcriptional activation is an essential step in the execution of most apoptotic events in the development of Drosophila [54][55][56][57] (reviewed in [58]). It is well established that ectopic expression of pro-apoptotic genes in the developing eye imaginal disc using the Gmr promoter causes cell killing and therefore conspicuously small eyes in the adults (Fig 1D-1F). Eye size reduction was considerably suppressed by additional expression of Pzg ( To address this phenomenon in more detail, we examined the expression of several components of the cell death machinery in wing imaginal discs upon pzg depletion. We have chosen the wing disc tissue for our analyses, as there the endogenous level of apoptosis is comparatively low [18,59]. Thus higher levels of apoptosis can be attributed to a reduced pzg activity. Pzg activity was downregulated with the UAS-pzg-RNAi shown before to efficiently reduce Pzg protein levels [40]. Pzg-RNAi induction either in the posterior compartment (using en-Gal4) or in a more central area (using omb-Gal4) of the wing disc triggered the execution of the apoptotic program: Pro-apoptotic gene activity, visualized by the activation of a rpr-lacZ reporter ( To exclude, however, off-target effects, a second independent RNAi-line (VDRC v25542) was included in the analysis of apoptosis induction and gave the same overall results. These data show that an impaired pzg activity leads to an increase in apoptosis.

pzg silencing triggers JNK-mediated stress responses
A key factor known to convey apoptosis in Drosophila and mammals alike is the c-Jun N-terminal kinase (JNK) pathway (reviewed in [60,61]). To measure JNK signaling activity in pzg mutant cells, we made use of the JNK downstream effector puckered (puc) and monitored the expression of a puc-lacZ enhancer trap line [62]. In wild type third instar wing discs, puc-lacZ expression is detected in the small stalk region attaching the disc to the larval epidermis [62]. Knock down of pzg robustly activated the puc-lacZ reporter gene within the depleted area and weakly in adjacent regions notably in older wing discs (   with the induction of JNK-mediated cell death. Great experimental insights were gained in the recent years demonstrating that JNK-mediated cell death in Drosophila is crucial for eliminating aberrant cells, thereby ensuring further development and morphogenesis (reviewed in [16,29,61]). This includes also the induction of a developmental delay and inhibition of metamorphosis, allowing the larva to compensate growth deficits and repair injured tissue. The gene encoding Dilp8, a member of the insulin-relaxin peptide family, was found to be upregulated in response to JNK signaling and is thought to delay metamorphosis by inhibiting ecdysone biosynthesis [53,63]. Tracing Dilp8 in wing discs, where pzg-RNAi was induced in the posterior half, revealed no obvious accumulation in younger discs (approximately 96 h AEL; Interestingly, the older pzg mutant wing discs appeared more crumpled and folded than the younger ones (compare Fig 3A and 3C with 3B and 3D). This phenomenon is reminiscent of regenerative growth during larval development experimentally induced with non-surgical tissue damage, e.g. pro-apoptotic gene activity ( [64][65]; reviewed in [66]). Moreover it has been shown that cell-adhesion is reduced by the caspase dependent cleavage of Armadillo (Arm) [67,68]. Indeed, we observed a strong reduction of membrane-associated Armadillo (Arm, beta-Catenin) in the pzg silenced area of the wing disc ( Fig 3E-E''). In a wild type wing disc, the antero-posterior compartment border is straight and defined. In contrast we observed 'fingerlike' structures notably in older pzg-silenced discs (Fig 3F-F''). Apparently, pzg-depletion enabled the mutant cells to invade the anterior compartment while still retaining posterior identity ( Fig 3F-F'').

Loss of pzg activity results in 'genuine' dying cells and consequently AiP
Genetic studies in Drosophila have shown that different mechanisms are triggered in apoptotic cells leading to an increase in proliferation and cell division rates of adjacent cells, a process which is referred to as Apoptosis induced Proliferation (AiP) (reviewed in [14,17,69,70]). In order to investigate the consequences of pzg depletion with regard to AiP, we monitored the proliferation rates in wing discs, where pzg-RNAi was induced in the posterior compartment. As pzg was shown to be required for the activation of cell cycle related genes [40], the autonomous decrease of actively dividing cells upon pzg depletion compared with controls was expected ( Fig 4A-B'''). In addition we noted an elevated number of cells undergoing DNA synthesis (EdU labeled) as well of mitotic cells (PH3 labeled) abutting the pzg-RNAi mutant territory at the anterior, implying the induction of a non-autonomous cell division response (Fig 4A-B''').
Apoptotic cells ectopically activate morphogenetic genes like wingless (wg) or decapentaplegic (dpp), responsible for the proliferation stimulus in directly adjacent cells [18,19]. To test if this is also a response to pzg loss, we firstly analyzed the distribution of Wg protein in wing discs, where pzg-RNAi was induced in the posterior compartment. At later stages (approximately 120 h AEL) we observed an ectopic expression of Wg in areas of the posterior compartment (Fig 4C', arrows), whereas the wild type expression along the dorso-ventral boundary in the posterior compartment of the disc was interrupted (Fig 4C', arrowhead). As wg is a well known target of N signaling, a reduction of wild type Wg accumulation along the dorso-ventral boundary was expected [40]. Intriguingly, no ectopic induction of dpp was observed in pzg silenced cells, inferred from the normal expression levels of a dpp-lacZ reporter along the putzig Depletion Causes Apoptosis antero-posterior compartment boundary (Fig 4D-D''). This demonstrates that pzg mutant cells behave like 'genuine' apoptotic cells with respect to the induction of ectopic wg, and that they induce AiP without the involvement of ectopic dpp activity.
Loss of pzg activity is accompanied by Apoptosis induced Apoptosis (AiA) Interestingly, cells doomed to die not only trigger AiP but also induce non-autonomous secondary apoptosis, abbreviated AiA (Apoptosis induced Apoptosis) [34]. To examine if such an effect is also observed in pzg-RNAi mutant cells we followed the activation of Caspase3 and Dcp1 in late third instar larval wing discs. Under these conditions, a strong caspase activity was observed in the anterior half of the disc, either visualized by staining for cleaved Caspase-3 or Dcp-1 (Fig 5A-A'' and 5C-C''). The co-expression of the baculovirus caspase inhibitor p35 in the pzg-RNAi mutant cells strongly enhanced this effect and triggered tumorous overgrowth of the wing discs and non autonomous AiA (Fig 5B-B'' and 5D-D''). This can be explained by the induction of the apoptotic machinery but the prevention of execution through inhibition of effector caspases by p35, resulting in so-called 'undead' cells [18,19,26,71]. Altogether, these data indicate that loss of pzg activity results in apoptosis followed by AiP and AiA.

Apoptotic consequences of N depletion in wing discs
As Pzg was shown to be required for efficient N signaling to occur, we asked whether the observed apoptotic outcome after pzg-RNAi induction might be the consequence of an impaired N signaling activity [40,41]. It is well established that a reduction in N signaling activity during imaginal development is correlated with tissue loss and apoptosis [51,72]. Therefore, we induced N-RNAi in the posterior compartment of the wing disc to compare the effects with those obtained after pzg depletion. To this end, we used the same N-RNAi line that has been shown by others to provide RNAi-mediated knock-down of N activity in wing imaginal discs [38,73]. We observed an accumulation of activated Dcp-1 caspase autonomously in N mutant cells and also non-autonomously in the anterior compartment, indicating that a downregulation of N signaling contributes to AiA (Fig 6A-B''). Moreover, a robust induction of puc-lacZ was detected in both compartments, indicating that JNK-mediated activity was induced as well ( Fig 6C-C''). In contrast, cell proliferation, visualized with EdU labeling, was different from pzg-RNAi depleted cells: Cells within the N-RNAi depleted compartment were still able to cycle through the cell cycle concluded from EdU incorporation (Fig 6D-D''). Based on the modest reduction of EdU signals, less cells however, appeared to enter the S-phase. Moreover, AiP was only weakly observed as EdU labeled cells abutting the N-deficient area appeared only more densely spaced (Fig 6D-D''), and no ectopic induction of the mitogens Wg (Fig 6E-E'') or dpp-lacZ (Fig 6F-F'') was detected.  In order to investigate, whether an upregulation of N signaling activity could ameliorate the apoptotic consequences of pzg loss, we initially sought to overexpress the activated form of the N receptor in pzg-RNAi depleted cells. However probably due to the immensely hyper-activated N response, we failed to obtain any viable larvae of this genotype. A more gentle method to increase N signaling activity is to reduce the activity of the general antagonist Hairless (H), e.g. by inducing a H-RNAi construct proven to reduce H activity in the fly [47]. Reduction of H

Apoptotic consequences of Dref depletion in wing discs
Apart from its role in N target gene activation, Pzg is important for cell proliferation as a member of the Trf2/Dref complex in Drosophila [40,74]. The transcription factor Dref regulates the expression of many proliferation-related genes, and has been described as a master key factor for cell proliferation (reviewed in [37]). We hence examined the consequences of Dref-RNAi depletion on apoptosis, AiP and AiA. To this end the UAS-Dref-RNAi line was used that has been shown before to specifically target Dref activity [75]. Inducing Dref-RNAi in the posterior half of the wing disc indeed triggered a strong apoptotic response, including robust Dcp1 caspase activation in the anterior half of late third instar wing discs (appr. 120 h AEL), indicative of AiA (Fig 7A-B''). Moreover, JNK signaling, i.e. puc-lacZ, was induced autonomously in the affected compartment, and to a lesser degree also in the anterior compartment of the wing disc (Fig 7C-C''). As expected, reduction of Dref activity was correlated with impaired cell proliferation as demonstrated by reduced EdU labeling in these areas (Fig 7D-D''). Proliferation rates were, however, significantly increased in the tissue abutting the Dref mutant cells (Fig 7D-D''). Moreover, similar to the effects of pzg-RNAi depletion, ectopic Wg induction was detected in the Dref-RNAi mutant tissue (Fig 7E-E''), whereas dpp activity appeared unaffected ( Fig  7F-F''). As expected from our earlier work, Wg expression along the dorso-ventral boundary was not affected by the depletion of Dref, in support of the notion that the effects of Pzg on N regulation are Dref independent [40,41]. Overall, induction of apoptosis with all its consequences is very similar upon the depletion of either Dref or Pzg during larval wing development. We therefore asked whether an increase of Dref activity might ameliorate the apoptotic consequences of pzg depletion. To this end we overexpressed UAS-pzg-RNAi together with UAS-Dref in the posterior wing disc compartment. To account for possible titration effects of Gal4 by the extra UAS-dose, we compared the effects of Dref overexpression with those obtained in a pzg-RNAi plus UAS-lacZ background. We observed a weaker apoptotic response in the UAS-lacZ background, (Fig 8). However, a concomitant increase of Dref activity ameliorated the pzg-RNAi proliferation defects much more robustly and strongly attenuated the induction of activated Dcp-1 caspase (Fig 8). Therefore, although Pzg and Dref are part of a large multi-subunit complex, increasing the amount of Dref protein helps to counterbalance the apoptotic effects resulting from pzg loss.

Discussion
In this work we show that loss of pzg induces apoptosis, including the autonomous activation of rpr and of various Caspases, of JNK and of Dilp-8 expression. Moreover, we observed nonautonomous apoptosis induced proliferation (AiP) including the induction of Wingless as well as apoptosis induced apoptosis (AiA). Apparently, pzg is required for the survival of the cell which in its absence undergoes apoptosis with all its consequences.
Pzg is a nuclear protein found in different multimeric complexes including the Trf2/Dref complex and the NURF complex. The Pzg/NURF complex has been implied in the epigenetic regulation of N and EcR target genes [41,76], whereas Dref is involved in the regulation of replication and proliferation related genes (reviewed in [37]). Both, downregulation of N as well as of Dref induced apoptosis. Whereas the former was expected [51,72], the latter was not since overexpression rather than downregulation of Dref was associated with apoptosis so far [77,78], presumably because Dref overexpression is sufficient to drive terminally differentiated cells into a new cell cycle [78]. Very similar to pzg RNAi, downregulation of Dref not only resulted in autonomous JNK-and Caspase activation, but also induced ectopic expression of Wingless, which was not observed as a consequence of N depletion. Moreover, unlike the activation of N signaling activity, overexpression of Dref was sufficient to ameliorate the effects of pzg induced apoptosis. Overall we conclude that the apoptosis induced by a depletion of pzg is primarily triggered by a disturbance of the Pzg/Trf2/Dref complex. In this case, at least part of the apoptotic consequences may result from a dysregulatin of Pzg/Dref target genes involved in cell survival or cell death. For example, the failure to activate anti-apoptotic factors in the pzgand Dref-RNAi mutant background might directly induce apoptosis. Amongst the plethora of Dref target genes is the Drosophila proto-oncogene raf [79]. Indeed it was shown that overexpression of Dref stimulates MAPK signaling activity [46]. MAPK, however, is an important negative regulator of pro-apoptotic gene activity in Drosophila [55,80], easily explaining the pro-apoptotic effects of a Dref loss. Moreover, inspection of the promoter region of the Drosophila inhibitor of apoptosis (DIAP1) gene reveals the presence of several potential DRE sites (S4 Fig). Hence, Dref may support cell survival by activating DIAP1 under normal circumstances. Future will show whether Dref acts together with Pzg, i.e. whether Pzg binds to the promoters of cell death regulators as well. Dref, however, not only stimulates the expression of anti-apoptotic or survival factors, but it also promotes the transcriptional activation of the apoptosis inducer dmp53 [81]. Similar to the apparent disparate role of Dref in the regulation of apoptosis, Dref not only promotes cell proliferation by the transcriptional activation of proliferation related genes like Pol alpha or PCNA, but Dref also inhibits proliferation as different members of the Hippo signaling pathway are Dref target genes as well [82,83].
Alternatively to a direct role of Dref and Pzg in the transcriptional regulation of genes involved in cell survival, cell death may arise as a consequence of conflicting signals that result from a collective dysregulation of the numerous Dref target genes. Depletion of pzg or Dref presumably affects a wide range of Dref-target genes. Dref target sequences have been found in more than 200 Drosophila genes, including replication and proliferation related genes, genes involved in growth, development and differentiation as well as components of protein biosynthesis or RNA binding proteins (reviewed in [37]; [84]). Therefore, it appears likely that crippling pzg activity, impairs the Dref-mediated cellular homeostasis, and hence the balance between survival and death decisions.
Recent molecular studies in Drosophila demonstrated that Pzg protein can be specifically detected at the telomeres of the chromosomes [85]. In Drosophila, the maintenance of telomeres is realized by a repeated transposition of retrotransposons instead of the telomerase-dependent extension of other eukaryotes (reviewed in [86,87]). Despite this difference, mutations that cause dysfunctions of the telomeres give rise to chromosome-and DNA-damage in all eukaryotes, and consequently result in apoptosis and in an increased lethality ( [88,89]; reviewed in [90]). Pzg mutant animals were shown to suffer from moderate telomere instability, inferred from a significant increase in the in vivo incidence of telomere fusions in anaphase neuroblasts. Telomere fusions were attributed to major chromatin changes causing altered transcriptional activity of the retrotransposon Het-A due to the loss of pzg activity [91]. In accordance with a functional relationship between Pzg and Dref in this developmental context, Dref mutants show similar alterations in retrotransposon expression [92]. Moreover, specific Dref target sequences were identified in the promoters of several retrotransposons in Drosophila, implying a direct regulatory function of Dref on the retrotransposon activity and telomere elongation [92]. Therefore, apoptosis induction observed in pzg and Dref mutant tissues might involve excessive retrotransposon activity, destroying the fine-tuned genomic stability.