Troponin-I mediates the localization of selected apico-basal cell polarity signaling proteins

Cell polarity results from the focal accumulation of proteins and organelles and is a key determinant of multiple features of cell biology from cell proliferation to organ morphogenesis (Harris and Peifer, 2005; Luxton and Gundersen, 2011; St Johnston and Ahringer, 2010). Out of the several cell polarity systems known, the apico-basal system (AB) is characteristic of epithelia. AB polarity is internal to the cell and some of its determinant protein complexes have been identified (Humbert et al., 2003; Rolls et al., 2003; Wodarz et al., 2000). AB polarity is unlikely to result from a single universal mechanism, even among epithelial tissues. For example, Par3 [also known as Bazooka (Baz) in flies] is a main requirement for AB polarity in many epithelial cells, while it plays an auxiliary role, if at all, in the follicular cells of Drosophila ovaries (Shahab et al., 2015). In addition to cell internal factors, mechanical interactions with neighboring cells influence the establishment of membrane domains in a polarized manner (Asnacios and Hamant, 2012; Goehring and Grill, 2013). As with all cellular features, cell polarity is dynamic and the corresponding factors move along mostly conserved actin, myosin II, keratin and tubulin cytoskeletal scaffolds (Flores-Benitez and Knust, 2016; Monier et al., 2015; Noordstra et al., 2016; Salas et al., 2016). Presumably, motor systems would provide the required force to execute these movements. However, how these cytoskeletal apparatuses regulate their activity and their directional traffic remains largely unknown.

Phosphoinositide 3-kinases (PI3Ks) are a family of lipid kinases of which class I members are most thoroughly characterized. While vertebrates have four isoforms of the class I PI3K, Drosophila has only one. All forms, however, signal through the generation of the phosphatidylinositol (3,4,5)-triphosphate [PtdIns(3,4,5)P3] lipid, which regulates AB polarity and epithelial cell morphogenesis, among other functions (Comer and Parent, 2007; Gibson and Perrimon, 2003; Krahn and Wodarz, 2012; Shewan et al., 2011). In addition, PI3K is often upregulated in several forms of cancer (Juvin et al., 2013; Tzenaki and Papakonstanti, 2013) suggesting a chain of causal links between an excess of PI3K activity, loss of AB cell polarity, aberrant mitosis, DNA damage and, finally, tumorous growth (McCaffrey et al., 2012). Whether these events are indeed causally linked or if, by contrast, they are mechanistically independent is still open.

We have shown previously that Drosophila TnI (also known as Wings up A) is expressed early in the syncytial embryo before muscles or any other cell types are specified. The TnI protein traffics between the nucleus and the cytoplasm as a function of cell cycle status using a sumoylation-dependent mechanism (Sahota et al., 2009). We have also shown that TnI is required for cell proliferation and enhances the oncogenic activity of the classical oncogenes Ras, Lgl and N for tumor outgrowth (Casas-Tintó et al., 2016). Here, we set out to characterize the mechanisms by which TnI depletion affects the AB polarity of epithelial cells in the wing imaginal discs. In addition, we address the context dependence of AB polarity mechanisms by comparing TnI-depleted epithelial cells and neuroblasts, a cell type that undergoes asymmetric cell divisions and is not epithelial in arrangement.

To visualize TnI we use the validated monoclonal antibody SC1 (Casas-Tintó et al., 2016). Wing disc cells show a strong TnI signal, which is localized apically with respect to baso-lateral Scribbled (Scrib) protein (Fig. 1A,B). In its apical position, TnI colocalizes with Actin (Fig. 1C–E). This is consistent with the canonical Actin-binding function of TnI. In neuroblasts, TnI is also accumulated in an apical position with respect to basal protein Prospero (Fig. 1F). Neuroblasts divide asymmetrically to yield another neuroblast and a ganglion mother cell (GMC). This later cell type undergoes symmetric divisions to yield neuron or glial cells. TnI is transiently found in the nucleus of GMC and their daughter cells (Fig. 1G). In salivary gland cells, whose large size facilitates visualization of protein localization, TnI is also accumulated in the apical domain with respect to the basolateral Dlg and the dorsal Crumbs (Fig. 1H). In these cells, we can also see that TnI accumulates in the apical domain of the cell nucleus (Fig. 1I). Thus, in all cell types analyzed TnI shows apical localization in the cytoplasm and, possibly, also in the nucleus.

The polar localization of TnI prompted an analysis of well-characterized polarity proteins under TnI-deficient conditions. To downregulate TnI we used a validated RNAi (Casas-Tintó et al., 2016), genetically driven by the UAS/Gal4 binary system (Brand and Perrimon, 1993). We generated FLP/FRT clones using the heat-shock inducible flipase (FLP) activity and actin-Gal4 cassettes that allow identification of the TnI^RNAi-expressing cells through the expression of GFP or RFP reporters (see Materials and Methods and Fig. S1). In wing-disc cells, Bazooka (Baz) is an apical polarity signal, whereas Discs large (Dlg1, hereafter Dlg) is the baso-lateral counterpart (Kaplan et al., 2009; Roberts et al., 2012; Rolls et al., 2003). We monitored Baz localization through two different procedures, a GFP-tagged construct or an anti-Baz antibody. The data show that the depletion of TnI results in the delocalization of Baz as determined by either of the two visualizing procedures (Fig. 2A,B; Fig. S2A,B). Likewise, the basal polarity signal Dlg fails to accumulate at the basal domain in TnI-depleted wing cells (Fig. 2C,D). This effect, however, is not bidirectional, because depleting Dlg does not alter TnI localization (Fig. 2E–G), nor does the downregulation of aPKC affect TnI localization (Fig. S2C,D). As a consequence of polarity loss after TnI depletion, cells undergo basal extrusion from the wing epithelium (see arrow in Fig. 2D).

Miranda (Mir, also known as Mira) is another basal polarity signal, although specific for neuroblasts (http://flybase.org/reports/FBgn0021776). We addressed its localization in larval type II neuroblasts (worniu-Gal4) (Fig. 2H,I). Under TnI depletion, Mir remains homogeneously distributed in the cytoplasm (Fig. 2J,K). Thus, neuroblasts also require TnI to localize another basal polarity signal, Mir in this case. Noticeably, the non-polar localization of Dlg in neuroblasts is not affected by TnI downregulation, suggesting that the TnI–Dlg interaction is cell type specific (Fig. S2E,F). In addition, Dlg is known to contribute to basal proteins targeting in neuroblasts through the interaction with Lethal giant larvae [Lgl; also known as L(2)gl] (Peng et al., 2000).

The contribution of TnI to the localization of AB polar factors, however, does not seem to be a general requirement for all polarity proteins. The basolateral localization of Scrib in epithelial and salivary gland cells (Fig. S2G–J), and the apical accumulation of Baz in neuroblasts are not affected by TnI depletion (Fig. 2L,M). These results reveal a degree of specificity in the mechanisms to localize polarity proteins that was previously unsuspected.

Co-immunoprecipitation (co-IP) assays determined that TnI physically interacts with Baz as well as with Dlg. The interaction was validated in both directions; either pulling down from Dlg or from TnI (Fig. 3A). Since Baz and Dlg show opposite polar localizations, we determined whether they form either a single or two independent complexes with TnI. To that end, we performed triple co-IP and antibody staining with anti-Baz, anti-Dlg and anti-TnI antibodies. While Baz precipitated with TnI, no signal for Dlg was detected (Fig. 3A). This result suggests that TnI forms independent complexes to position dorsal Baz and ventral Dlg. In addition to Baz, the apical complex includes atypical PKC (aPKC) (Harris and Peifer, 2005). We reproduced the Baz–aPKC binding in co-IP assays (Fig. 3B). Furthermore, pulling down with TnI also co-immunoprecipitated aPKC and Baz (Fig. 3B). Since TnI is an Actin-binding protein, these co-IP data between TnI and selected AB polarity proteins indicate that their localization is mediated through a mechanism using Actin filaments.

Apico-basal polarity complexes are linked to adherens junctions, which provide stability to the epithelium, among other functions (reviewed in Tepass, 2012). We monitored the integrity of these junctions, and the cytoskeleton in general, in TnI-deficient wing cells using β-Catenin [known as Armadillo (Arm) in flies], E-Cadherin and γ-Tubulin as reporters. In these experiments, we generated large mosaics by driving the corresponding constructs to the whole posterior wing compartment (en-Gal4). The data show the expected reduction of reporter signals when TnI is downregulated (Fig. 4). The weakening of these structures is expected to result in some form of cell unfitness, leading to the observed extrusion from the epithelium by the neighboring cells (Baffet et al., 2012; Marinari et al., 2012).

Searching for additional mechanisms involved in the localization of polarity factors, we noted that mammalian PI3K mediates AB polarity in epithelial cells and contributes to basal membrane formation (Peng et al., 2015). Thus, we analyzed the potential effects of PI3K depletion in wing disc cells and found no alterations in the apical localization of TnI or Baz, or in the basal localization of Dlg (Fig. S3). Nevertheless, like in mammals, fly PI3K (PI3K92E) must play a role in AB polarity because its upregulation rescues TnI-deficient cells from apoptosis and extrusion (Fig. S4). In addition, the changes in β-Catenin/Arm, γ-Tubulin and Dlg expression described above are not seen in TnI-depleted cells overexpressing fly PI3K (Fig. S5).

Correct AB polarity is required for normal cell proliferation in epithelia. As we have shown previously, TnI-depleted cells undergo cell competition leading to eventual extrusion or apoptosis after Caspase 3 activation (Casas-Tintó et al., 2016). That form of apoptosis includes the loss of dIAP1, requires Bsk/JNK activity and can be suppressed by the upregulation of Secreted protein, acidic, cysteine-rich (Sparc) or the downregulation of flower (fwe), two genes encoding cell survival and death signals, respectively (see fig. 3 in Casas-Tintó et al., 2016).

Here, we aimed to analyze cell structural features caused by TnI depletion in the absence of cell death. To achieve this, we monitored β-Catenin/Arm and γ-Tubulin in TnI-deficient cells under conditions in which cell competition is restrained by the upregulation of survival signals or by the downregulation of cell death signals. The death-regulator Need2-like caspase (Dronc) encodes an endopeptidase involved in apoptosis (Yang et al., 2010). Its joint downregulation with TnI rescues cells from apoptosis (Fig. S6) but the levels of β-Catenin/Arm are still reduced (Fig. 5A–C). The same effect was observed upon upregulating the cell survival factor gene Sparc (Portela et al., 2010) (Fig. 5D–F) or with the downregulation of the cell death signal gene fwe (Rhiner et al., 2010) (Fig. 5G–I). Concerning Tubulin, cells rescued from apoptosis through the overexpression of p35 (also known as Cdk5α) (Hay et al., 1994), still show the γ-Tubulin loss from the apical end that the TnI depletion causes by itself (Fig. 5J–L). In addition, the rescue of apoptosis seen upon p35 expression fails to prevent cell extrusion (Fig. S6). Further evidence of the differential mechanisms for cell polarity versus cytoskeletal defects in a TnI-depletion context was obtained by analyzing wing discs with concomitant downregulation of the three pro-apoptosis signals reaper (rpr), head involution defective (hid) and grim. The drastic reduction of β-Catenin/Arm persists in the TnI-depleted domain (Fig. 5M,N).

In summary, TnI loss causes defects in epithelial cell adhesion and the γ-Tubulin cytoskeleton that are maintained even if cells are rescued from apoptosis. This suboptimal status of imaginal wing cells is likely to cause the morphological defects observed in the resulting adult wings after metamorphosis (Fig. 5O–R).

E-Cadherin and γ-tubulin are also relevant for mitotic spindle orientation and chromosome segregation (Baffet et al., 2012; Wang et al., 2004). In that context, we monitored the cell centrosome using a GFP-tagged form of asterless (pEYFP.asl^FL) that binds γ-Tubulin in the pericentriolar material of the centrosome (Gopalakrishnan et al., 2011). TnI mutants show aberrant number of centrosomes around single nuclei in the syncytial embryo stage (Fig. 6A,B; Fig. S7). The feature is also reproduced in wing disc cells expressing a TnI^RNAi at 24–36 h post clone induction, a time prior to cell extrusion from the epithelium (Fig. 6C–G). Although the precise allocation of centrosomes per cell in the clones is somewhat uncertain, the abnormal number is evident. The similarity between mutant and RNAi phenotypes confirm that both tools to deplete TnI are similarly effective.

Chromosome mechanics is likely affected if centrosomes are abnormal. Previous reports indicate that microtubules and conserved PAR proteins are essential mediators of cell polarity, and mitotic spindle positioning depends on heterotrimeric G protein signaling and the microtubule motor protein Dynein (Ahringer, 2003). In particular, a Baz–centrosome positive-feedback loop contributes to the maintenance of the adherens junctions and cell integrity (Jiang et al., 2015). In our previous TnI study, we had already shown chromosome instability in TnI-null mutant syncytial embryos (Casas-Tintó et al., 2016). Here, to monitor DNA damage in wing cells, we relied on the γH2Av (known as γH2Ax in mammals) immunosignal. This histone H2 reporter is a standard indicator of both double-strand DNA breaks and replication stress, two events that may result in chromosome instability (Madigan et al., 2002). However, the γH2Av immunosignal may also reflect non-apoptotic events that do not imply fragmented nuclei (Khan et al., 2017) (see Discussion). As in the experiments above, apoptosis was prevented by upregulating p35. The data show a significant increase of the γH2Av signal in the TnI-deficient cells (Fig. 6H–J). As control, cells with upregulated p35 alone did not show changes in the γH2Av signal (Fig. 6K–M). Upon two additional methods to rescue TnI-depleted cells from apoptosis, downregulation of fwe or upregulation of Sparc, increased DNA damage was also seen in TnI-depleted cells (Fig. S8). By contrast, the downregulation of Dronc did suppress the DNA damage, as monitored by γH2AV signal as well as apoptosis as monitored through the Caspase 3 signal (Fig. 7). This dual suppression was also confirmed by analyzing cells with concomitant downregulation of hid, rpr and grim (Fig. 7).

It is worth noting that cells exhibiting DNA damage and rescued from apoptosis are still able to proliferate. This is consistent with equivalent experiments in which chromosome instability had been elicited by alternative methods such as mutations in genes involved in the spindle assembly checkpoint (bub3 and rod), spindle assembly (asp), chromatin condensation (orc2) or cytokinesis (dia) (Dekanty et al., 2012). Recently, the protein kinase p38a has been identified as a factor that limits chromosome instability by interfering with the activation of the ATR kinase and homologous recombination repair mechanisms in breast cancer cells (Canovas et al., 2018). Ultimately, depletion of p38a increases replication stress (Haahr et al., 2016). Thus, we attempted to promote DNA damage and apoptosis by downregulating p38a. Neither γH2AV nor Caspase 3 signals decreased in TnI-depleted cells with downregulated p38a (Fig. 7). This result suggests that the DNA damage elicited by TnI depletion corresponds to the double-strand DNA breaks rather than the replication stress. Pending a future study of this issue in more detail, however, we use here the generic term of DNA damage.

In summary, the events triggered by TnI depletion follow the order: apico-basal cell polarity loss, chromosomal instability/DNA damage, structural abnormalities in the cytoskeleton and adherens junctions defects. Cells can be eliminated from the epithelium, either by extrusion or apoptosis. The last step can be prevented allowing cells to proliferate to some extent, but they acquire a suboptimal state which is prone to displaying morphological abnormalities after differentiation. In addition, chromosomal instability seems to be mechanistically different from the rest of AB polarity-related events. The scheme of Fig. 8 summarizes the main features of the role of TnI in AB cell polarity and the known genes that contribute to the cellular processes studied here.

This study shows that TnI contributes to the AB cell polarity and binds directly certain proteins which can accumulate in the apical or in the basolateral region of the cell. The binding partners for TnI, however, are selective according to the cell type. Thus, the basal Dlg of epithelial cells requires TnI for its preferential localization, but it is not perturbed if TnI is depleted in neuroblasts where Dlg is not polar. These features unveil a diversity of mechanisms to allocate specific proteins to polar domains that was previously unsuspected. In addition to the previously known AB polarity factors, TnI is revealed as an apical factor in its own right. Our preliminary observations in salivary gland cells also seem to indicate a still unexplored apical localization of TnI within the cell nucleus.

The mechanism by which TnI mediates the localization of apical (Baz/Par3) and basolateral (Dlg) factors in epithelial cells includes their direct binding, albeit in different complexes. That is, TnI associates with different partners according to the cargo destination and cell type. Since TnI is an Actin-binding protein, this feature is likely to contribute to the molecular machinery that generates the force needed to move polarity factors within a cell. The involvement of Tropomyosin proteins (Tm proteins), another key component of the muscle sarcomere, was documented in our previous TnI study (Casas-Tintó et al., 2016). Although not analyzed yet in sufficient detail, it seems that the contractile properties of the sarcomere and the AB cell polarity of epithelial cells result from very similar protein complexes. If Tm proteins turn out to act in concert with TnI to achieve AB cell polarity, it could transmit, to the actin filaments, the force required to move Baz, Dlg and other polarity proteins. Also, the fact that TnI is involved in this process helps to explain the regulation of force from the polarity complexes with that of the neighboring cells, through the adherence junctions that these complexes help to establish (Wang et al., 2012). Similar to its role during muscle sarcomere contraction, TnI would regulate Actin filaments in their sliding along, most likely, a membrane-anchored isoform of Myosin. Following the analogy to the sarcomere mechanisms, it would be appropriate to incorporate Troponin C and Troponin T in future studies of AB epithelial polarity. The role of PI3K in the TnI-mediated mechanisms regulating AB polarity seems to be different between vertebrates and Drosophila, although this kinase clearly participates in AB cell polarity in the fly. The phosphorylated substrate of PI3K in this context remains to be identified. However, a particular form of Myosin, Myosin-1s, has been reported to target membrane phosphoinositides in addition to F-Actin (Rajendraprasad et al., 2018). This Myosin-1s scenario would be suitable to integrate PI3K, TnI, F-Actin and Myosin into a force-generating complex to move AB polarity factors. Finally, phosphorylation by PI3K could provide the necessary mechanism to render the TnI complex reversible, and hence subject to regulation.

TnI-depleted cells lose AB polarity and reduce proliferation. Thus, TnI represents a link between these two cellular properties. TnI would regulate Actin fibers that control cell shape. Loss of polarity and reduced proliferation cause some form of unfitness with deleterious effects. In wing discs, cell unfitness may result in apoptosis or extrusion. One of the proposed mechanisms for wing cell extrusion is dependent on Actin/Myosin (Rosenblatt et al., 2001; Slattum et al., 2009). In RasV12-transformed vertebrate epithelial cells, extrusion is mediated by a mechanism that depends on EPLIN (also known as LIMA1), and that activates myosin II and PKA (Ohoka et al., 2015). Cell competition, as defined in Drosophila, is considered to maintain organ shape through extrusion of the so-called loser cells. The probability of loser cell extrusion correlates with the extent of their surface interaction with the surrounding winner cells (Levayer et al., 2015). The process, however, is likely to be more heterogeneous than initially thought, since cell elimination can also be triggered away from the loser–winner interface by means of cell crowding, a phenomenon named ‘super-competition’ (Levayer et al., 2016). Elimination of non-fit loser cells is also observed in vertebrates, particularly in epithelia, where it has been renamed as ‘epithelial defense against cancer’ (EDAC) (Kajita and Fujita, 2015). Mechanisms for apoptosis are also diverse across species, as some pro-apoptosis signals including hid, rpr and grim, are not conserved while others, such IAP1 and Dronc, are (Behura et al., 2011).

The cell phenotypes reported here are compatible with a causal link between loss of AB polarity and cell competition, extrusion or apoptosis. An additional feature, DNA damage, has been proposed to be part of this sequence of events; in particular to explain genetic heterogeneity in tumor metastases (reviewed in McGranahan and Swanton, 2017). In this study on TnI, we have attempted to determine whether DNA damage is indeed a consequence of AB polarity loss, or at least the polarity loss mediated by TnI. Experiments in which TnI-depleted cells are prevented from undergoing extrusion or apoptosis by means of protecting them with p35, Dronc, Sparc or fwe expression, clearly indicate that DNA damage is a process that is mechanistically different from cell competition due to cell polarity loss, and it does not necessarily lead to cell death. Furthermore, cells exhibiting DNA damage are able to proliferate to some extent, and, actually, this feature is in agreement with other experiments in which DNA integrity, aneuploidy in particular, was altered by other methods (Dekanty et al., 2012). Although TnI mutant embryos show chromosomal fragments (see fig. 5F in Sahota et al., 2009), the criterion for DNA damage used here is the γH2AV immunosignal. That signal may be mechanistically independent from apoptosis and regulated by the Dronc caspase (Khan et al., 2017). We observed that Dronc is required for the TnI-dependent increase of the γH2AV signal, and also of the Caspase 3 signal. Thus, the DNA damage elicited by TnI depletion may trigger a cellular response towards apoptosis, extrusion or to some sort of repair which renders the cell alive and proliferative, although not fit. Heterogeneities in the cellular response to quantitative changes in the levels of caspase activities have been proposed previously (Florentin and Arama, 2012).

The apical complex of the fly is conserved in vertebrates (Harris and Peifer, 2005; Zhu et al., 2011), and cell polarity genes are also tumor suppressors (Hanada et al., 2000; Humbert et al., 2008; Iden et al., 2012; Zhu et al., 2014). As shown previously, TnI enhances the oncogenic properties of lgl, N and Ras in Drosophila and it is at the origin of certain types of cancers in humans (Casas-Tintó et al., 2016). Thus, it would be expected the functional link between AB polarity and proliferation through TnI is also conserved. Given the selectivity of the TnI contribution to AB polarity and its requirement for cell proliferation, it may become a suitable target for cell type-specific therapies.

The fly stocks used are from the Bloomington Stock Center (see FlyBase) except as indicated. As a null TnI mutation we used a rearrangement located in the regulatory region of the wupA gene, Df(1)TnI²³⁴³⁷ (Marín et al., 2004). To induce TnI loss-of-function cell mosaics, we used the Bloomington line P{TRiP.JF02172}attP2 (BL#31893) following a comparative study with other available RNAi lines (Casas-Tintó et al., 2016). Constructs UAS-p35, UAS-fweLoseA/B^RNAi and UAS-Sparc were obtained from Eduardo Moreno (Champalimaud Center, Lisbon, Portugal) (Portela et al., 2010; Rhiner et al., 2010), UAS-DE-Cadherin-GFP (Gorfinkiel and Arias, 2007) (BL#58445) and w; worniu-Gal4; Ase-Gal80 were from Cayetano González (IRB, Barcelona, Spain). Line pEYFP.asl^FL marks the centriole-specific Spindle abnormal assembly protein 6 and was obtained from C. González (Varmark et al., 2007). Line UAS-bsk^DN (BL#6409) was from our fly collection, and UAS-Bazooka-GFP was a gift from Antoine Guichet (Institut Jacques Monod, Paris, France). To down- or up-regulate PI3K, we used the dominant negative form PI3K-92E^D954A or the constitutively active form UAS-PI3K^92E.CAAX, respectively. Line p38a^RNAi (#KK102484) was from the VDRC repository. FLP/FRT mosaics were obtained by heat-shocking larvae carrying the constructs actin-FRT-yellow-Stop-FRT-Gal4, UAS-GFP or actin-FRT-yellow-Stop-FRT-Gal4, UAS-RFP combined with the desired UAS-x construct and hsFLP. The expression of the flipase after a heat shock (AHS) excises the FRT-flanked ‘yellow’ cassette and generates Gal4-expressing clones. These clones overexpress the respective UAS-constructs driven by the ubiquitous actin promoter (act Gal4) and are marked by the GFP or RFP reporters. See Fig. S1 for details. We used a GFP–Scrib fusion protein to visualize Scrib together with a RFP–histone reporter to visualize nuclei: Scribbled-GFP his1Av-mRFP. To downregulate the expression of rpr, hid and grim pro-apoptosis genes, we used the construct originally referred to as UAS-RHG miRNA (Siegrist et al., 2010), here denoted as miR.

Two cell systems were used in this study, imaginal wing discs and larval neuroblasts. Wing discs are two flat sac-shaped structures located in each side of the larvae where cells proliferate by regular mitoses, show epithelial arrangement, and differentiate into adult notum and wings upon metamorphosis. By contrast, neuroblasts proliferate by asymmetrical divisions during the larval period, yielding neural and glial cells, which undergo substantial rewiring through metamorphosis. Crosses were set with 10 females and 10 males per vial at 25°C and vials were changed every 72 h to avoid overcrowding. FLP-out clones were induced by a heat shock (8 min at 37°C) during the second-larval instar stage followed dissection at the third-larval instar stage 48–72 h afterwards. Control cultures were run in parallel.

Tissue samples were fixed in formaldehyde 4% for 25 min and stained according to standard protocols. Antibodies used were: anti-active caspase 3 (1:100, cat. no. 9661s, Cell Signaling), mouse anti-γH2AV [1:100, cat. no. UNC93-5.2.1, Developmental Studies Hybridoma Bank (DSHB)], rabbit anti-γH2AV (1:500, cat. no. GTX48733, Genetex), anti-Lamin (1:100, cat. no. LC28.26, DSHB), anti-Actin (1:100, cat. no. JLA20, DSHB), anti-Tub (1:100, cat. no. T7451, Sigma), anti-Dlg (1:20, cat. no. 4F3, DSHB), anti-Par3/Baz (1:50, cat. no. P4A1, DSHB), anti-Crumbs (1:100, cat. no. Cg4, DSHB), anti-Scrib (1:20, cat. no. Dc-20, Santa Cruz Biotechnology), anti-Arm (1:100, cat. no. N27A1, DSHB), anti-Miranda (1:100, gift from Rita Sousa Nunes, Kings College, London, UK) and anti-β-galactosidase (1:50, cat. no. 40-1a, DSHB) antibodies. Fluorescent secondary antibodies conjugated to Alexa Fluor 488, 569 and 647 were used (1:200, Invitrogen). All images were obtained with a LEICA TCS-SP5 confocal microscope and processed with Imaris software (Bitplane) for measurements. The SC1 mouse monoclonal Troponin-I antibody was raised against the peptide PDGDPSKFAS (Abmart Inc.; Casas-Tintó et al., 2016). TnI-Dlg co-immunoprecipitation was detected in whole fly extracts using specific mouse SC1 anti-TnI (1:100), mouse anti-Dlg (1:10), mouse anti-Baz (Par3) (1:100, 1:50, DSHB) (antibody details as above), rabbit anti-aPKC (1:100, a gift from Sonsoles Campuzano, CBMSO, Madrid, Spain) and mouse anti-GFP (1:100, cat. no. 11814460001, Roche) as a negative control, coupled to G-Sepharose beads (GE Healthcare). Pulldown assays were performed with Canton-S strain protein extracts. Antibody immobilization to Sepharose beads was performed for 8 h at 4°C and subsequent incubation with protein extracts was conducted in the presence of 0.25% BSA for 2 h at room temperature. Immunoprecipitated proteins were subjected to western blotting developed with True Blot secondary antibodies (eBioscience). Negative controls (Neg Ctrl) are G-Sepharose beads without primary antibodies. Input is 1:10 of the original protein extract.

Western blots were obtained by standard procedures (Invitrogen) and protein bands were quantified by densitometry using ImageJ software with the ‘GelAnalyzer’ option (see further details in http://www.di.uq.edu.au/sparqimagejblots).

The intensity of the fluorescent signal in Figs 4 and 5 was measured using ImageJ 1.44a; an intensity plot was drawn and the area below the graphic was quantified. The area of the signal for C3 and γH2Av in Fig. 7 was measured using Imaris (Bitplane) volume quantification and filtered for GFP intensity to distinguish anterior and posterior compartments. Statistical significance was calculated with the two-tailed Student's t-test. Significance levels are indicated as *P<0.05, **P<0.005, ***P<0.001. Number of samples was n>8 animals in all cases.

We appreciate fly strains from Bloomington Stock Center and the VDRC repository. We would like to thank Dr M. Milán, Dr S. Campuzano, Dr L. A. Baena-López and Dr O. Fernández-Capetillo for helpful discussion and reagents.