Schnurri regulates hemocyte function to promote tissue recovery after DNA damage

To ensure tissue homeostasis, the repair, death, removal and replacement of damaged cells need to be tightly coordinated. Cells in which apoptosis is initiated secrete a variety of signaling molecules that can induce proliferation in neighboring cells and that recruit immune cells and macrophages to ensure removal of dead cells and limit local inflammation and tissue damage (Elliott and Ravichandran, 2010; Jiang et al., 2009; Babcock et al., 2008; Pastor-Pareja et al., 2008; Morata et al., 2011; Fan and Bergmann, 2008). Drosophila imaginal discs have served as a productive model with which to explore the communication of damaged cells with their immediate environment, allowing the identification of a number of growth factors and cytokines that are secreted from dying cells to induce compensatory proliferation or recruit hemocytes (Pastor-Pareja et al., 2008; Morata et al., 2011; Fan and Bergmann, 2008; Uhlirova et al., 2005; Huh et al., 2004; Ryoo et al., 2004; Pérez-Garijo et al., 2004). These interactions between apoptotic and viable cells are believed to be crucial for maintaining tissue homeostasis in developmental contexts.

In response to radiation-induced DNA damage, however, widespread tissue damage can overwhelm localized control and compensation mechanisms to significantly compromise tissue homeostasis. Accordingly, irradiation of Drosophila pupal retina with sublethal doses of UV-C light results in widespread tissue loss, which persists to adulthood. The extent of this loss is dose dependent and can be influenced by genetic conditions in which DNA repair is impaired or apoptosis is compromised (Jassim et al., 2003; Luo et al., 2007). Assessment of tissue loss in the eye after UV irradiation thus serves as a convenient model system with which to examine the regulation of DNA repair, apoptosis and tissue recovery in the context of a living organism. By comparing the extent of tissue loss in the irradiated eye with the untreated eye in the same animal, the extent of UV-induced apoptosis can be quantified precisely. In previous studies it has been established that p53 is a crucial regulator of DNA repair in this context, and that an antagonism between receptor tyrosine kinase (RTK) and Jun N-terminal kinase (JNK) signaling pathways regulates the decision between cell repair and cell death in this system (Jassim et al., 2003; Luo et al., 2007).

We have found that in photoreceptors and cone cells of the retina, the transcription factors Foxo and Fos (Kayak) integrate these signaling pathways and cooperate to induce the pro-apoptotic molecule Head involution defective (Hid, also known as Wrinkled), causing widespread cell death. This apoptotic program is activated by JNK through direct phosphorylation of Fos and is modulated by RTK signaling through Akt-mediated phosphorylation and 14-3-3-mediated cytoplasmic retention of Foxo (Luo et al., 2007; Nielsen et al., 2008).

To further characterize the mechanisms that regulate apoptosis and tissue maintenance in the Drosophila retina, we performed a dominant interaction screen for modulators of JNK-induced tissue loss. This screen was performed in flies overexpressing a constitutively active form of the JNK-activating kinase Hemipterous (Hep) under the control of the photoreceptor and cone cell driver Sep-Gal4 (Luo et al., 2007; Jasper et al., 2002), and tested a collection of P-element insertions from the Drosophila Gene Disruption Project (Bellen et al., 2004) as maintained by the Bloomington Drosophila Stock Center. Expression of Hep^act in photoreceptors is sufficient to induce apoptosis by transcriptional induction of the pro-apoptotic gene hid (Luo et al., 2007; Nielsen et al., 2008). One of the identified enhancers of Hep-induced apoptosis carried a P-element insertion in the schnurri locus [shn^K00401 (Rusten et al., 2002)], indicating that Shn might modulate the extent of tissue loss observed in conditions of sustained JNK activation. Accordingly, reducing the shn gene dose by introducing the shn¹ loss-of-function allele (Arora et al., 1995), or by overexpressing a double-stranded (ds) RNA against shn (shn^RNAi), significantly increased the loss of ommatidial structures observed in flies expressing Hep^act under the control of Sep-Gal4. Consistently, overexpressing Shn using the EP line shn^EP2359 (Rørth, 1996), which effectively increases shn expression in response to Gal4 (supplementary material Fig. S1C), was sufficient to rescue the loss of ommatidia, indicating that Shn activity is both required and sufficient to limit JNK-induced tissue loss in the developing retina (Fig. 1A and supplementary material Fig. S1D).

Shn is a large zinc-finger transcription factor best characterized as a component of the Decapentaplegic (Dpp)/TGFβ/BMP signaling pathway, where it acts in a heterotrimeric complex with the Smad proteins Mad and Medea (Arora et al., 1995; Torres-Vazquez et al., 2000). Interestingly, modulating the gene dose of other components of the Dpp signaling pathway did not influence JNK-induced tissue loss in the retina, indicating that Shn might act independently of Dpp signaling in this context (supplementary material Fig. S2B). Consistently, overexpression of shn cDNAs encoding fragments incapable of interacting with Mad and Medea (Udagawa et al., 2000; Pyrowolakis et al., 2004) significantly reduced JNK-induced tissue loss (supplementary material Fig. S2C). Interestingly, increased Shn expression did not reduce the number of apoptotic cells observed in the developing retina of Hep^act-expressing flies (supplementary material Fig. S1E). Furthermore, the shn gene dose did not affect tissue loss elicited by overexpression of the pro-apoptotic molecules Grim, Reaper or Hid, indicating that Shn does not directly influence cell death but might be associated with tissue repair or recovery (supplementary material Fig. S1F).

To assess whether Shn affects tissue recovery after a genotoxic challenge, we tested whether modulating shn gene dose would be sufficient to influence the extent of tissue loss observed after UV irradiation of the pupal retina. Irradiation of the pupal retina 24 hours after puparium formation results in adult flies that exhibit severe loss of tissue in the irradiated eye, a phenotype mediated by JNK-induced Hid expression (Fig. 1B) (Jassim et al., 2003; Luo et al., 2007). Overexpression of Shn, or downregulation of Shn by RNAi, significantly reversed or increased tissue loss, respectively, when compared with wild-type animals (Fig. 1C). Shn thus emerges as a crucial regulator of tissue maintenance and/or recovery in epithelial tissues challenged by genotoxic stress.

We tested whether Shn regulates JNK signaling in the retina, and found that Shn overexpression in retinal cells indeed resulted in increased expression of the JNK phosphatase puckered (puc) in dissected third instar eye imaginal discs (Fig. 2A). The puc gene is transcriptionally induced in response to JNK activation, acting in a negative-feedback loop to repress JNK activity (Martín-Blanco et al., 1998). Shn-mediated induction of puc could thus be interpreted either as Shn-mediated activation of JNK and subsequent activation of puc expression, or as direct activation of puc expression by Shn with subsequent inactivation of JNK (which would be consistent with the Shn-mediated reduction of the JNK-induced apoptotic phenotype in the retina). Strikingly, however, the induction of puc expression could not be explained by Shn-mediated cell-autonomous induction of puc in retinal cells, as detecting puc expression using a puc::lacZ reporter line [puc^E69 (Martín-Blanco et al., 1998)] showed no increase in lacZ expression in the retina, but expression in eye disc-associated hemocytes (Fig. 2B). We confirmed the identity of these lacZ-expressing cells using a hemocyte reporter expressing RFP under the control of the hemocyte-specific promoter of the Hemolectin (Hml) gene (Fig. 2C). A cell-autonomous effect of Shn on JNK activity in the retina could thus be excluded. Supporting this view, hid transcript expression in the eye disc was not affected by Shn overexpression (Fig. 2A).

Hemocytes recognize and engulf bacteria and are crucial for survival after pathogenic challenge (Vlisidou et al., 2009; Elrod-Erickson et al., 2000; Charroux and Royet, 2009). Besides mediating the innate immune response to microorganisms, macrophage recruitment to disrupted epithelia is important in both promoting wound repair and restricting tumor growth (Pastor-Pareja et al., 2008). Hemocytes are attracted to wound sites and apoptotic cells and are crucial for elimination of apoptotic bodies after wounding and during development (Stramer et al., 2005; Babcock et al., 2008; Zhou et al., 1995; Sears et al., 2003; Moreira et al., 2010; Tepass et al., 1994). These cells further mediate a systemic metabolic and growth response to localized DNA damage (Karpac et al., 2011). Interestingly, the hemocytes associated with the retina change morphology over the course of pupal development, acquiring a spread, macrophage-like morphology (Fig. 2D). These cells contain large numbers of phagocytic vesicles with DAPI-positive punctae, indicating that they actively engulf and remove other cells. This morphological transition is reminiscent of the acquisition of macrophage appearance by embryonic hemocytes in response to apoptotic cells (Babcock et al., 2008; Tepass et al., 1994; Cho et al., 2002; Lanot et al., 2001).

The increase in eye disc-associated puc-expressing hemocytes after retina-specific overexpression of Shn, along with the associated reduction in UV-induced tissue loss, indicated a potential link between hemocyte function and improved tissue maintenance in shn gain-of-function conditions. Importantly, it also suggested the existence of a Shn-induced humoral signal from retinal cells to hemocytes that induces their response to tissue damage. Supporting this view, when puc^E69 pupae were subjected to lateral UV irradiation, we observed puc induction in hemocytes distal to the site of damage (Fig. 2E; note that in this case, puc is also activated cell-autonomously in the retina, as JNK is activated by DNA damage in these cells) (see Luo et al., 2007). Consistent with our recent study (Karpac et al., 2011), localized UV irradiation of the retina thus results in a system-wide activation of hemocytes (Fig. 2E).

To test the importance of hemocyte function for tissue maintenance in the UV-irradiated retina, we compared tissue loss in wild-type animals with that in flies lacking hemocytes due to genetic ablation (by expression of Hid with the hemocyte-specific Gal4 driver Hml-Gal4). Strikingly, ablation of hemocytes significantly increased tissue loss in irradiated retinas, confirming that hemocytes are required to prevent excessive cell death and ensure tissue maintenance (Fig. 3A). A similar increase in sensitivity of the retina to UV irradiation was observed when the phagocytic and migratory activities of hemocytes were disrupted by expressing a dsRNA against myoblast city (mbc, also known as dock180) (Fig. 3B) (Wood et al., 2006; Duchek et al., 2001; Ishimaru et al., 2004; Paladi and Tepass, 2004; Ohsawa et al., 2011). The phagocytic activity of hemocytes is thus crucial to limit excessive tissue loss under genotoxic conditions.

Hemocytes are regulated by a number of cytokines and growth factors in Drosophila, including the TNF homolog Eiger (Egr), the IL-6-like Upd1-3 and the PDGF-like growth factors Pvf1-3. To determine whether Shn expression in the retina would activate hemocytes through one of these humoral interactions, we quantified the expression of these cytokines in third instar larval eye discs in which Shn was overexpressed under the control of GMR-Gal4 (Fig. 4A and supplementary material Fig. S3). Whereas Pvf2 and upd2/3 expression was unchanged compared with control animals, we observed a significant increase in egr and Pvf1 expression in response to Shn overexpression. Egr can activate the JNK pathway in target tissues (Igaki et al., 2002; Moreno et al., 2002) and is required for prophenol oxidase release from crystal cells that generate melanotic clots in response to injury (Bidla et al., 2007). It might thus be involved in activating JNK signaling in retina-associated hemocytes. However, inhibiting the Egr pathway systemically (in homozygotes for the egr¹ loss-of-function allele) or specifically in hemocytes (expressing dsRNA against the Egr receptor wengen using Hml-Gal4) did not significantly influence tissue loss in UV-treated pupal retinae, indicating that egr induction is not required for Shn-mediated modulation of tissue homeostasis (supplementary material Fig. S3C).

We therefore tested whether Pvf1 induction by Shn might influence tissue loss in UV-irradiated retina (Fig. 4). Pvf1 and Pvf2 signal redundantly through the receptor PVR and regulate embryonic hemocyte migratory patterns (Moreira et al., 2010; Cho et al., 2002; Wood et al., 2006) and border cell migration (McDonald et al., 2003; Janssens et al., 2010; Poukkula et al., 2011) and can promote epidermal cell migration in injured larvae (Wu et al., 2009). We find that Pvf/PVR signaling also regulates the activity of retina-associated hemocytes: overexpression of PVR in hemocytes was sufficient to promote early spreading of these cells in third instar (Fig. 4B,B′), and expression of a dsRNA against Pvr strongly inhibited acquisition of macrophage-like morphology in hemocytes of the early pupal stage (Fig. 4D,D′). Interestingly, this effect of PVR loss-of-function is reminiscent of the recently described function of PVR in imaginal disc cells surrounding oncogenic cells, in which activation of PVR is required to promote engulfment of oncogenic cells (Ohsawa et al., 2011). Furthermore, Mbc is a well-described effector of PVR signaling in Drosophila cells (Wood et al., 2006; Duchek et al., 2001; Ishimaru et al., 2004; Paladi and Tepass, 2004; Ohsawa et al., 2011). In both imaginal cells and hemocytes, PVR activation thus results in an increase in cell shape dynamics and macrophage-like activity, presumably by regulating cytoskeletal function and endocytosis through Mbc. A potential caveat to this interpretation is the described role of Pvf/PVR signaling in promoting survival of embryonic hemocytes (Brückner et al., 2004). However, we did not observe a reduction of hemocyte numbers in third instar larvae when Pvr RNAi was expressed using Hml-Gal4 or Pxn-Gal4 (Fig. 4C and supplementary material Fig. S3D), indicating that hemocytes at this stage do not require PVR signals for survival. This discrepancy might be explained by the distinct signaling status of embryonic and larval hemocytes. Although the larval hematopoietic system is founded by embryonic hemocytes (Tepass et al., 1994; Lanot et al., 2001; Holz et al., 2003; Makhijani et al. 2011), larval hemocyte behaviors are uniquely different, owing to inductive cues from the hematopoietic microenvironment, i.e. epidermal–muscular pockets that line the larval body wall (Makhijani et al. 2011). Furthermore, from the pupal stage onward, larval hemocytes are complemented by the developmentally distinct population of lymph gland hemocytes (Holz et al., 2003; Lanot et al. 2001), which do not depend on PVR signaling for their trophic survival (Mondal et al., 2011; Jung et al., 2005).

The reduction in hemocytes acquiring a macrophage-like morphology in PVR loss-of-function conditions corresponded with increased tissue loss after UV irradiation, indicating that a PVR-mediated increase in phagocytosing hemocytes in early pupal stages sets up a crucial surveillance mechanism to limit tissue loss in severely stressed tissues (Fig. 4E). Supporting this interpretation, we observed a significant rescue of UV-induced tissue loss in flies overexpressing Pvf1 in the retina (Fig. 4F). Importantly, Pvf1 overexpression also rescued the increased loss of tissue in retina-specific shn loss-of-function conditions, confirming the role of Pvf1–PVR signaling downstream of Shn in retinal homeostasis (Fig. 4G).

How does PVR regulate the activation of retina-associated hemocytes? PVR has been reported to activate JNK signaling (Ishimaru et al., 2004), and it is thus possible that the JNK activation we observe in hemocytes after UV irradiation is induced by PVR. JNK activation in hemocytes can cause filopodia formation (Williams et al., 2007), further suggesting that JNK activation contributes to the PVR-induced transition of retina-associated hemocytes into macrophage-like cells. Interestingly, a recent study has shown that PVR can also be activated by JNK signaling in Drosophila imaginal disc cells, in which JNK activation causes upregulation of PVR, which in turn promotes engulfment activity (Ohsawa et al., 2011). Eiger-mediated activation of JNK in hemocytes might thus prime them for responding to Pvf ligands, promoting phagocytic activity. However, we have not observed significant consequences for hemocyte morphology when activating JNK by overexpressing Hep in these cells, and have found a significant requirement for the JNK Basket in hemocytes in promoting tissue maintenance after UV irradiation only in males and not in females (supplementary material Fig. S4). Although JNK signaling may thus contribute to the activation of hemocytes in cooperation with PVR signaling, additional studies are needed to characterize the full downstream signaling mechanisms regulating hemocyte activity.

Our results indicate that Shn is crucial for tissue maintenance after DNA damage, and establish that the transcriptional induction of Pvf1 by Shn stimulates macrophage-like activity of retina-associated hemocytes. Strikingly, these cells are required for tissue maintenance, limiting the extent of tissue loss observed in irradiated retinas. Our findings thus support an emerging model that proposes that failure to properly engulf and eliminate apoptotic cells impairs tissue recovery after a challenge (Elliott and Ravichandran, 2010). Importantly, our findings also establish a novel function for Shn that is independent of the Dpp–TGFβ signaling pathway. This function is reminiscent of recent reports showing that SHN3 (HIVEP3) regulates NFκB-mediated responses to infection in human fibroblast and macrophage cell lines, and that Shn3 null mice exhibit reduced IL2 levels (Oukka et al., 2002; Oukka et al., 2004). Furthermore, murine Shn2 is required for T-cell development and maturation (Takagi et al., 2001; Staton et al., 2011). A role for Shn proteins in the regulation of the interaction between damaged tissue and blood cells, and in the control of tissue recovery after genotoxic damage as observed here, is thus likely to be evolutionarily conserved.

Flies were raised on standard cornmeal and molasses-based food. Unless noted otherwise, all experiments were performed at 25°C. Whenever possible, mutant lines were compared with wild-type siblings of the same sex. Both sexes gave the same result in all experiments. Fly lines shn¹, shn^K00401, dpp^d12, tkv⁷, med¹, GMR-grm, GMR-rpr, GMR-hid, Hml-Gal4 and shn^RNAi(TriP) were provided by the Bloomington Drosophila Stock Center. RNAi lines shn^RNAiv3226, wgn^RNAiv9152 and Pvr^RNAiv105353 were obtained from the Vienna Drosophila RNAi Center. The Szeged Stock Center provided shn^EP2359. We received the following lines as gifts: GMR-Gal4 and Sep-Gal4 (M. Mlodzik, Mt Sinai School of Medicine, New York); UAS-brk (S. Cohen, IMCB, Singapore); puc^E69 (E. Martin-Blanco, CSIC, Barcelona); UAS-hid (J. K. Billeter, University of Toronto, Toronto); UAS-Flag-shn1-909, UAS-Flag-shn909-1888, UAS-Flag-shn1888-2529, UAS-Flag-shn1888-2529ΔZF6-8 and UAS-Flag-shnΔZF6-8 (G. Pyrowolakis, Universität Freiburg, Freiburg, Germany); Pxn-Gal4, UAS-Pvr and UAS-Pvf1 (M. Galko, University of Texas MD Anderson Cancer Center, Houston); and egr¹ (M. Miura, University of Tokyo, Tokyo). The fly line Sep-Gal4, UAS-hep^ACT is as described (Jasper et al., 2002). Generation of the Hml::RFP construct and transgenes was as described (Makhijani et al., 2011).

Pupae raised at 25°C were collected 24 hours after puparium formation (pupae raised at 29°C were collected 16 hours after puparium formation) and the anterior pupal case was removed to expose the developing head and retinas. UV irradiation was administered at 17.5 mJ by a UV crosslinker (Stratalinker 1800) while shielding one eye, and treated animals were reared in the dark. Adult eye size (in pixels) was measured in Photoshop (Adobe).

Twenty third-instar imaginal eye discs were collected for each RNA extraction. Total RNA was isolated using Trizol (Invitrogen) and cDNA was synthesized with SuperScript II reverse transcriptase (Invitrogen). Quantitative real-time PCR was performed on a BioRad MyiQ Single-Color Real-Time PCR Detection System using SYBR Green (Roche). The following primer pairs (5′ to 3′) were used for RT-PCR: Actin 5C, ATTCAACACACCAGCGCTCTCCTT and ACCGCACGGTTTGAAAGGAATGAC; shn, GTGCAGCAACCGGATGTCAATGAA′ and TGATGTTGCCGCTGAAACGTCTTG; puc, AAATCCAAGTCGGATAGCGAGGCT and TGCTTGTGGAGCTGGGTGAATCTA; hid, TGCGAAATACACGGGTTCA and CCAATATCACCCAGTCCG; Pvf1 TGGAGCAGGCCGAGAACAAGTATT and CCTGGACAATGAAGCGTTTGCGAT; Pvf2, TCCTTTGCCATCATCAGTCGGTGT and AGTAATTCTTCAGGGCCGCAGTCT; dilp6 (Ilp6), GTTCTCAAAGTGCCGACGTCCAAA and CCGACTTGCAGCACAAATCGGTTA; upd2, TGCGGAACATCACGATGAGCGAAT and TCTTCTGCTGATCCTTGCGGAACT; upd3, ACAAGGCCAGGATCACCACCAAT and TGTACAGCAGGTTGGTCAGGTTGA; egr, GTGAATCTCTTTCTTTCAGCCA and TGTGAATGTCCTTCACATCCAT.

To maintain attachment of hemocytes to dissected tissue, discs were dissected in fixation buffer (4% paraformaldehyde in PBS, 0.1% Triton X-100) and fixation was performed for 1 hour with no agitation. X-gal staining for puc^E69 and immunostaining of imaginal discs were performed as previously described (Luo et al., 2007). In fluorescent images Hoechst was used to stain DNA. Antibodies were from Cell Signaling (anti-cleaved caspase 3 and anti-phospho-tyrosine), Developmental Studies Hybridoma Bank (anti-β-galactosidase) and Jackson Immunochemicals (fluorescently labeled secondary antibodies). Images were obtained using a Leica TCS SP5 confocal microscope. Morphological changes in size were determined in Photoshop on z-stack images spanning the epithelium and hemocytes. Larval hemocyte populations were quantified by bleeding four third-instar larvae into 15 μl PBS and counting using a hemocytometer.

We thank G. Pyrowolakis for the generous gift of fly lines, and members of the Jasper lab for discussions.