A polarized nucleus-cytoskeleton-ECM connection in migrating cardioblasts controls heart tube formation in Drosophila

The genetic network that regulates heart development is highly conserved from flies to humans with orthologous transcription factor (TF) families and signaling pathways essential for heart development, such as Tinman/Nkx2.5, Pannier/GATA4-6, Tailup/Isl1, Dorsocross H15 Mid/Tbx, Hand, Dpp/BMP or Wg/Wnt (Bryantsev and Cripps, 2009; Busser et al., 2015; Jin et al., 2013; Junion et al., 2012; Lovato and Cripps, 2016). However, although mechanistic insights have been gained on the specification of cardiac progenitors, less is known about genes controlling subsequent events of cardiogenesis.

As in mammals, formation of the Drosophila heart starts by the migration of bilateral strips of cardiac progenitor cells to the midline and the assembly of a linear heart tube. Cardioblasts (CB) migration occurs concurrently with dorsal closure, a process leading to the spreading of the epidermis dorsally over an extra-embryonic epithelium, the amnioserosa. Migrating CBs extend projections from their dorsal surface and towards the leading edge of the epidermis suggesting coupling of these two tissues (Haack et al., 2014; Rugendorff et al., 1994). However, during the late phase of migration, CBs move cell autonomously. They send out filopodia that contact contralateral CBs and ensure correct cell matching and linear heart tube assembly (Haack et al., 2014; Medioni et al., 2008; Zhang et al., 2018).

Some factors in this process have been identified, including the transmembrane protein Toll (Wang et al., 2005) and Faint sausage, which participates in adhesion between ipsilateral CBs (Haag et al., 1999).

Proteins involved in setting or maintaining polarity of CBs have also been implicated in collective migration. Among them, the integrin dimer expressed by the CBs (αPS3 and βPS), its ligand Laminin-A and cytoplasmic integrin adapter Talin are all required for proper CBs alignment (Stark et al., 1997; Vanderploeg and Jacobs, 2015; Vanderploeg et al., 2012).

Moreover, the coordinated migration and alignment of CBs requires the Slit morphogen and its receptor Robo (MacMullin and Jacobs, 2006; Qian et al., 2005). Slit function in cardiac cell migration is conserved from Drosophila to vertebrates. In zebrafish, Slit2 knockdown embryos show disruption of endocardial collective cell migration, with individual cells migrating faster and with loss of directionality (Fish et al., 2011).

Recent findings have shown a genetic interaction between αPS3 gene scb and Slit/Robo, suggesting an integrated function of cell membrane-associated factors in CB migration polarization during cardiac tube formation (Vanderploeg et al., 2012). The link between cell adhesion and actin cytoskeleton remodeling is also fundamental for CB migration and requires intermediate proteins to mediate connections and to transduce signals. Among them, two members of the Dock Guanine nucleotide Exchange Factors (GEFs) containing SH3 domain, Myoblast city and Sponge, are required during CB migration through activation of Rac and Rap1, respectively (Biersmith et al., 2015). Mutants for these two genes have defects in the migrating CB rows, with gaps and clusters of multilayered CBs. Moreover, the Rho GTPase CDC42 is required in Tinman-positive CBs for the contact with the opposite row (Swope et al., 2014; Vogler et al., 2014) by modulating localization of non-muscle myosin II Zipper (Vogler et al., 2014). Thus, cytoskeleton modulations allow filopodia formation at the CBs leading edge and ensure coordinated, tension-mediated modification of cell shape to keep migrating CBs aligned.

Here, we identify the SORBS Cbl Associated Protein (CAP) and Muscle Specific Protein 300 (MSP300, also known as Msp300) as novel mediators of the transition from mitotic state to CBs migration. In vitro, SORBS proteins are generally targeted to focal adhesion sites, where they play diverse roles in mechano-transduction, contributing to stiffness sensing and contractility (Ichikawa et al., 2017). CAP is the only SORBS family member in Drosophila and has been described as an adaptor protein that makes links between different partners at the membrane-cytoskeletal interface of stretch-sensitive structures via the vinculin-integrin complex (Bharadwaj et al., 2013).

MSP300 is a large protein of the Nesprin family with several protein domains, including spectrin and nuclear anchoring KASH domains. Nesprin provides a link between actin cytoskeleton and perinuclear layer via Linker of Nucleoskeleton and Cytoskeleton (LINC) (Mellad et al., 2011). Recently, MSP300 has been shown to display stretching capacity participating in the elasticity of a myonuclear scaffold to protect myonuclei from mechanical stress (Wang et al., 2015).

Nucleus deformability is necessary when cells are squeezing through small constrictions during their migration (Friedl et al., 2011). The role of the nucleus in cell migration is suggested to be dependent on its connections with components of the cytoskeleton mediated by LINC and indirectly with extracellular matrix, including focal adhesion complex (Fruleux and Hawkins, 2016).

In the developing heart, our data show that CAP and MSP300 accumulate in a punctate manner on apical and basal sides of migrating CBs. We also demonstrate that polarized accumulation of nuclei-anchored MSP300 and cell membrane associated CAP promote post-mitotic state and are required for CB alignment, coordinated migration and proper CB nuclei shapes. These findings imply the requirement of polarized nucleus-cytoskeleton-ECM connections during collective CB migration.

Because coordinated CB migration towards the midline involves stretch-induced morphological changes of cells, we asked whether CAP and MSP300 could be involved in these processes as their functions have not yet been investigated in Drosophila heart. Orthologous genes, SORBS family and Syne1, respectively, exhibit conserved cardiac expression in mammals (Banerjee et al., 2014; Dixon et al., 2015; Hong et al., 2014; Puckelwartz et al., 2010).

We first tested expression patterns of CAP and MSP300 by immunostaining and confirmed that they are both expressed in migrating CBs, starting from late embryonic stage 13 (Fig. 1A-A″). CAP and MSP300 are co-expressed in a stereotyped punctate pattern on the apical and basal sides of each CB, clearly visible from stage 15 (Fig. 1B-B″ and arrowheads in Fig. 1C-C″). This expression is maintained when the two rows of CBs have reached the dorsal midline (Fig. 1D-E″). Although expression of CAP seems to be also diffuse under cell membranes (arrow in Fig. 1E′), MSP300 is largely colocalized with CAP protein (arrowheads in Fig. 1E″). Transverse sections of 3D reconstructed developing hearts confirmed colocalization of CAP and MSP300 on basal and apical sides of CBs in aorta, as well as in the heart proper (arrows in Fig. 1F-F″, Fig. S1A,B). To validate CAP and MSP300 localization in the apical side of the luminal domain, we performed co-immunostaining experiments with CAP and Slit antibodies (Fig. S1C-F″). We confirmed that CAP and MSP300 are enriched apically at the level of the lumen (arrowheads in Fig. S1C-E) but absent at the CB adhesion sites characterized by Armadillo expression (Fig. S1G,H). Unexpectedly, we also observed Slit localization in a punctate pattern at the basal side of CBs at stage 15-16, reminiscent of CAP and MSP300 expression patterns (arrows in Fig. S1C-C″). This dotty accumulation of Slit is transient as it becomes polarized at the luminal side at late stage 16 (Fig. S1F-F″).

Next, we wondered whether the punctate localizations of CAP and MSP300 are interdependent. MSP300 is a multidomain scaffold protein. Therefore, we hypothesized that it could participate in the recruitment and accumulation of CAP into subcellular compartments, creating a repeated punctate pattern. To test this, we performed immunostaining in both CAP and MSP300 mutants. We observed a clear loss of punctate accumulation of CAP on basal and apical CB sides in MSP300^SZ75 mutant embryos, suggesting that MSP300 is required for CAP localization in CBs (arrows, Fig. 1G,H). Similarly, in dCAP49e mutants (in which a 2.9 kb region downstream of the CAP^CA06924 P-element is deleted), the MSP300 expression pattern is severely affected. However, as some MSP300 expression in CBs persists, we conclude that CAP is only partially required for the MSP300 punctate pattern (arrows, Fig. 1I,J). We also performed co-immunoprecipitation experiments using GFP-tagged MSP300 transgenic Drosophila line but could not find any interaction between MSP300-GFP and CAP under our experimental conditions, indicating that several intermediate partners might be involved in the complex.

In summary, we observed that during the late phase of CBs migration, CAP, MSP300 and Slit colocalize in polarized dots on each side of the CBs in the basal and luminal domains. These findings suggest that CAP and MSP300 are part of a protein complex as their punctate localization appears to be largely interdependent.

We next examined CAP and MSP300 functions during CB migration. Immunostaining with antibodies against the muscle differentiation transcription factor Mef2 revealed misalignments of CBs in both MSP300 and CAP mutants (Fig. 2A-C″). Cardioblasts formed clusters (arrowheads in Fig. 2B,C) and multiple cell layers (brackets, Fig. 2B,C). We scored these alignment defects and found that 80% of MSP300 (44/55) and 64% of CAP mutant embryos (34/52) present a multilayered row of CBs, suggesting that MSP300 and CAP are both involved in maintaining cohesion of migrating CBs (Fig. 2K).

Slit and its Robo receptors are known factors in CB migration, alignment and polarization (MacMullin and Jacobs, 2006; Medioni et al., 2008; Qian et al., 2005; Santiago-Martinez et al., 2006). Slit- and Robo-mediated cardiac morphogenesis is modulated by the transmembrane integrin receptors, their ligands and intracellular interactors (Vanderploeg et al., 2012). Given the link between CAP and the integrin adhesion complex, as well as the observed alignment defects in dCAP49e embryos, we tested whether Slit localization is affected in CAP and MSP300 mutants.

We found clear Slit localization defects in both mutants. (Fig. 2A-C″). Slit was ectopically expressed in misaligned CBs, creating a meshwork between CBs (arrows in Fig. 2B′,C′). Similar ectopic enrichment of Slit around mislocalized CBs has also been observed in cardiac-specific knockdowns of CAP (miR strategy; Bharadwaj et al., 2013) and MSP300 (short hairpin mediated strategy) (Fig. 2K, Fig. S2A-B″). Collectively, these data show that CAP and MSP300 ensure correct Slit localization in CBs. They also suggest a potential interaction between CAP, MSP300 and the integrin complex in migrating CBs.

We then tested whether the CB cell shape is affected in CAP mutants. α-Spectrin staining, which marks cell membrane subdomain, showed that CAP^49e CBs are misarranged compared with the wild type (where CBs are perfectly aligned along AP axis and matching with their contralateral partner) (arrows and arrowheads in Fig. 2D). In CAP^49e hearts, we observed changes in both the shapes and the orientation of CBs along the A-P axis (arrowhead in Fig. 2D′), CB misalignment/matching defects (arrows and arrowheads, Fig. 2D,D′) and mispositioning within rows of cells (star in Fig. 2D′), suggesting mis-regulation of CB migration directionality.

We also tested the genetic interaction between CAP and MSP300 using trans-heterozygous dCAP49e/MSP300^SZ75 mutant embryos. In contrast to the single heterozygous control, CB alignment is affected in trans-heterozygotes (18% versus 59%; Fig. 2K, Fig. S2C-C″), showing that CAP and MSP300 genetically interact and play convergent roles during CB migration. Finally, we checked whether changes in total cardiac cells number could account for MSP300 and CAP mutant phenotypes. We performed immunostaining experiments using antibodies against H15, a transcription factor specific to all CBs, and Seven-up (Svp), a marker of a CB subset that gives rise to ostia. These two markers allow the four Tin-positive CBs (Svp negative) to be distinguished from the two Svp-positive CBs in each abdominal hemi-segment A2-A7. We found that knockout of either MSP300 or CAP both led to a significant increase in CB number (Fig. 2E-J,L). Interestingly, only Tin-positive CB levels were increased in homozygous mutant or cardiac-specific knockdowns of either CAP or MSP300, suggesting a specific function of these genes in maintaining the post-mitotic state of Tin⁺ CBs (Fig. 2E-J,M, Fig. S5) thought to undergo symmetric cell divisions. Alignment defects also occur in hemi-segments without supernumerary CBs, but this appears to be concomitant with an increase in CBs in other hemi-segments. (arrows in Fig. 2G,H).

Considering the impact of MSP300 loss of function on embryonic heart morphology and that MSP300^sz75 homozygous mutants are embryonic lethal, we wondered whether cardiac-specific knockdown of MSP300 could lead to some defects in the adult heart. We first validated the efficiency of an MSP300 RNAi line in embryos by immunostaining (Fig. S3A-B′) and analyzed semi-intact preparations of the fly hearts to measure heart physiological parameters after knockdown using SOHA software (Ocorr et al., 2007).

MSP300 RNAi flies show a significant increase in systolic diameter leading to a reduction in fractional shortening (Fig. 3A,A′). MSP300 RNAi hearts stained for F-actin also show disorganization and gaps (arrows in Fig. 3C′) in the circumferential myofibrils in comparison with highly organized and densely packed myofibrils observed in control hearts (arrowheads in Fig. 3B′-C′). This disturbed cardiac architecture suggests a potential discontinuity in the arrangement of cardiomyocytes, leading to failure in coordinated contraction.

CAP protein interacts directly with Vinculin (Vinc) at the muscle attachment sites, where it is recruited to the integrin complex to modulate the Actin cytoskeleton (Bharadwaj et al., 2013). Therefore, we asked whether Vinc is also present in CBs during embryonic development and could be involved in CAP-regulated processes. We took advantage of a recently generated Vinc-GFP knock-in allele allowing us to follow Vinc localization in vivo. We detected restricted and similar Vinc-GFP signal at the apical and basal membranes of CBs, consistent with CAP and MSP300 expression (arrows, Fig. 4A-B‴) as well as the localization of other integrin adhesion complex members, such as βPS-Integrin (Myospheroid) (arrows, Fig. S4A-A″). Interestingly, we confirmed that this accumulation of Vinc coincides with F-actin localization at basal (arrowheads, Fig. 4B-B‴) and apical sides, revealed with Phalloidin staining (arrowheads, Fig. 4C-C‴, Fig. S4B-B‴).

To test whether Vinc is required for CAP localization, we analyzed the impact of heart-specific Vinc knockdown on both CAP and MSP300 expression (Fig. 4D-E″). We first validated the efficiency of Vinc knockdown by immunostaining in embryos (Fig. S3C-D′) and observed that the heart proper morphology contains inclusion of CBs (Fig. 4E, arrowhead) indicating their misalignment during migration. Dots of co-accumulation of CAP and MSP300 are still present; however, their pattern is disorganized, suggesting that polarization of CBs is defective (arrows, Fig. 4E-E″). Cardiac-specific knockdown of Vinc also induces a meshwork of Slit (Fig. 4F-F″) and an increase in Tin-positive CBs in some hemi-segments, which is consistent with CAP and MSP300 loss of function (Fig. 5A-D,G-I). Similar results are also obtained by cardiac-specific knockdown of βPS-Integrin (mys) (Fig. 5E,G-I). Interestingly, transheterozygous mutant embryos with MSP300sz75 combined with the cardiac α-PS3-Integrin mutation scb⁰¹²⁸⁸ show disorganized CB alignment and an increased number of Tin-positive CBs (Fig. 5F), showing their involvement in the same pathway. These results are reminiscent of the genetic interaction recently identified between CAP and scb (Jammrath et al., 2020), suggesting related roles of CAP, MSP300 and the integrin complex in controlling the postmitotic state of Tin-positive CBs. Finally, we tested whether the integrin ligand Laminin A (LamA), which is also known to interact genetically with slit (MacMullin and Jacobs, 2006), is affected by the loss of MSP300, Vinc and βPS-integrin. Interestingly, we observed a disorganized pattern of LamA in MSP300 homozygous mutant embryos with breaks in the apical side and the basal lamina, suggesting that Nesprin-like proteins are required for ECM structure (Fig. 6A-B′). These results are consistent with defects in LamA pattern observed after Vinc and βPS-integrin knockdown (Fig. 6A″-B‴). A similar requirement for MSP300 and the integrin complex was also observed for the correct localization of another ECM component Perlecan (Trol), a secreted heparan sulfate proteoglycan (Fig. 6C-D‴). Overall, the polarized location of the CAP-MSP300-Vinc/integrin complexes plays an important role in determining the appropriate number and arrangement of CBs, and the localization of ECM proteins.

CAP and MSP300 might function by locally binding Vinc adhesion complexes to the actin cytoskeleton and the nucleus. This would generate a directional internal tension that would keep the CBs polarized, maintain their postmitotic state and their alignment, and thus allow correct lumen formation and compaction of the heart tube. MSP300/Syne 1-2 proteins contain an actin-binding domain, whereas CAP binds to actin-binding proteins, including Vinc. This suggests that F-actin might be a potential co-partner associated to the MSP300-CAP-Vinc dots during CB migration. To confirm the involvement of F-actin in bridging the nucleus to the integrin complex, we analyzed null mutation and cardiac-specific KD of WASp. WASp regulates the nucleation activity of the Arp2/3 complex, facilitating the assembly of branched actin filament networks with the spatiotemporal control required to orchestrate cellular processes (Padrick et al., 2011).

Consistent with our previous findings, cardiac (Hand-GAL4) or Tin-positive CBs with specific knockdown of WASp have defects in alignment of cardiac cells and an increased number of Tin-positive CBs in some hemi-segments (Fig. 7A-D,F-H and Fig. S5). These defective arrangements of CBs also occur in hemi-segments bearing the correct number of Tin CBs (arrows in Fig. 7D), suggesting that clustering could occur as a consequence of exiting the post-mitotic state of CBs in a different hemi-segment. Finally, we assessed the genetic interaction between MSP300 and WASP by comparing single heterozygous mutant for each gene with the transheterozygous mutant. Our results clearly demonstrate a significant increase in CBs alignment defects and Tin-positive CB numbers in transheterozygous embryos, suggesting that MSP300 and WASp function in a similar pathway during heart closure to regulate actin dynamics (Fig. 7E,F-H). Interestingly, removing WASp function had a negative impact on the localization of the ECM proteins LamA and Trol that was similar to that produced by MSP300 and integrin complex loss of function (Fig. 6A″″-D″″), strengthening their involvement in the same mechanism.

To further confirm the link between F-actin and MSP300 localization, we performed Phalloidin staining to reveal the F-actin pattern in migrating CBs. Interestingly, F-actin accumulates on basal and apical sides of CBs, very much like CAP and MSP300 (Fig. 8A-A‴ and Fig. S2B-B‴). Hence, we tested whether MSP300 was required to stabilize F-actin accumulation at the interface of the nucleus and the CAP-Focal adhesion complex. Indeed, MSP300 loss of function induced a strong decrease in localized F-actin expression in CBs (arrowheads, Fig. 8B-D). Strikingly, this F-actin phenotype was stronger for Tin-positive CBs, as Svp-positive CBs still accumulated F-actin in MSP300 mutants (arrows, Fig. 8B,C), suggesting a specific function for MSP300 in Tin-positive CBs.

Consistent with its known function, WASp³ mutant embryos show a significant reduction of F-actin levels in a similar manner to that in MSP300 mutant hearts (Fig. 8D). Taken together, these data suggest that CAP and MSP300 control the cell cycle exit and alignment of CBs by facilitating a F-actin linkage of the nucleus with the CB cell membrane. This function is limited during migration to the ‘leading CB’ population expressing Tin.

F-actin stress fibers have been shown to exert mechanical tension on the nucleus, providing mechano-sensing properties to orient the nucleus and modulate its shape during cell migration. Thus, we measured nucleus shape in CAP and MSP300 mutants. We observed that CAP and MSP300 null mutants or cardiac-specific CAP knockdown induce a significant decrease of nuclei volume as well as an increase in flatness relative to controls (Fig. 9A,B). Consistent with our findings, WASp³ mutant embryos show the same effect on nuclei shape, suggesting the requirement of actin stress fibers to maintain nuclei morphology during CBs migration.

In this study, we have analyzed expression and function of two novel regulators of cardiac morphogenesis: CAP, an integrin complex adaptor protein; and MSP300, a Nesprin protein belonging to the LINC complex. Our study highlights a crucial role for CAP and MSP300 in maintaining the nucleus-cell membrane connection via the integrin/vinculin adhesion complex, which is ultimately required in locking cell division processes during the migration progression of Tin-positive cardiomyocytes. (Fig. 8C). We also provide evidence that this nucleus-cytoskeleton-cell membrane link is important for maintaining CB polarity, contact between ipsilateral and contralateral CBs, and proper nucleus shape.

These mechanisms appear to be coupled and more work will be needed to determine whether the quiescent stage exit is a cause or a consequence of polarity and alignment defects. Slit localization is also affected in MSP300 and CAP mutants, suggesting that Slit-mediated directionality of CB migration might be affected. Indeed, a conserved aspect of Slit/Robo signaling is its involvement in cell guidance mechanisms (Fish et al., 2011; Raza and Jacobs, 2016; Wu et al., 1999). In Drosophila, Slit is known to interact genetically with several members of the integrin complex (αPS3, βPS), and the integrins are required for apical localization of Slit and leading-edge activity. The direct coupling between the cell membrane and the ECM at integrin-based adhesion sites (focal adhesion) allows cells to mechanically sense their physical environment and tune downstream signaling events that regulate the organization of the cytoskeleton. Vinculin is a key player in focal adhesion, where it interacts directly with Talin and ensures a link between adhesion proteins and actin cytoskeleton (Burridge and Mangeat, 1984; Humphries et al., 2007; Rothenberg et al., 2018). In vertebrates and in Drosophila, CAP has the potential to directly interact with Vinculin (Bharadwaj et al., 2013). In CBs, loss of Vinculin leads to disorganized CAP and irregular MSP300 localization, probably owing to the loss of Vinc-anchoring points at the cell membrane, resulting in changes in the number, polarity and motility of affected CBs. We propose that MSP300 and CAP ensure the connection of nuclei to adhesion complex via actin cytoskeleton and Vinculin, and could thus modulate the coordination between cell division arrest and the initiation of migration. However, even though both rows of CBs appear to reach the dorsal midline in CAP and MSP300 mutant embryos, their expression pattern, which is maintained throughout cardiac tube formation, could suggest a potential secondary function in the fine regulation of the dynamics of CB migration.

MSP300 possesses spectrin repeats, allowing interaction with actin. In this study, we demonstrate the genetic interaction of MSP300 with CAP and WASp, an activator of the Arp2/3 complex, suggesting that the branched actin network is involved in connecting the nucleus to the cell membrane (Padrick et al., 2011).

Transition from cell cycle exit to cell migration is a coordinated process that requires multilevel controls in terms of cell polarity, cell adhesion and differentiation involving changes in gene expression programs. LINC complexes are known to be important in transmitting mechanical stresses from the cytoskeleton to the inside of the nucleus that ultimately lead to changes in nuclear structure and organization, and to regulation of cell cycle progression (Uroz et al., 2018). Recently, the LINC complex has been shown to transmit the forces from the β1-integrin-engaged actin network to the nuclear lamina to directly regulate the expression of epidermal differentiation genes (Carley et al., 2021). Our results support these views, although the outcome might depend on the context. One of our interesting observations is that CAP-MSP300 dots accumulate at both sides of migrating CBs. Recent evidence suggests the existence of an actin structure called a perinuclear actin-cap that interacts with specific actin-cap-associated focal adhesion components at two positions in the direction of the aligned actin-cap fibers: at the leading lamellipodial edge and at the trailing edge of the cell. This newly identified actin structure is linked to the nucleus via the LINC complex (Kim et al., 2012; Maninova et al., 2017) and might be necessary for coordinating quiescence, cell polarity and nucleus shape during migration. Furthermore, nuclei deformations observed in CAP, MSP300 and WASp mutant nuclei are reminiscent of a previously described model based on results obtained after Nesprin 1 siRNA treatment of endothelial cells (Anno et al., 2012). Indeed, as in Nesprin 1 knockdown, CB nuclei seem more sensitive to deformation, and become smaller and flattened, suggesting that wild-type nuclei might be subjected to directional tensile forces along the apical-basal axis during CB migration.

Additional work will determine whether F-actin bound to the nucleus through MSP300 and to focal adhesions leads to a sustainable force transmission to CB nuclei and whether direct transmission of mechanical force to the nucleus can influence the exit of CBs from the cell cycle. Recent findings have shown that nesprin-1 mutations lead to increased activation of the ERK pathway in mouse heart tissue (Zhou et al., 2017). In vertebrates, CAP/Ponsin has been also shown to repress the MAPK pathway (Zhang et al., 2006). In Drosophila, late EGFR-mediated ERK activation has been shown to promote a Tin-positive CB subpopulation (Schwarz et al., 2018). According to our data, one hypothesis is that CAP and MSP300 would contribute to the maintenance of the correct number of Tin-positive migrating CBs by counteracting the EGF pathway. Further investigations will be needed to clarify the involvement of CAP, F-actin and MSP300 in controlling these processes.

Our physiological analysis of MSP300 mutant adult hearts indicate that the cardiac contractile function is affected. Fractional shortening in adults were strongly affected by MSP300 knockdown and was linked to locally defective morphology of adult hearts.

During the publication process, a new study has shown that a similar phenotype occurs after CAP mutation in larvae and adult heart (Jammrath et al., 2020), suggesting that the MSP300-CAP connection might be conserved for the correct architecture and function in the adult heart. These preliminary results open new perspectives concerning the possible implication of this connection in the formation of the primitive cardiac tube of vertebrates.

The Drosophila CAP deletion mutant dCAP49e has a deleted 2.9 kb region downstream of the CAP^CA06924 P-element. The dCAP49e mutant and UAS-CAPmiR construct were generated according to Bharadwaj et al. (2013). dCAP49e and P(UAS-CAP.miRNA)3 were from the Bloomington Stock Center. The MSP300^SZ75 mutant was a gift from Talila Volk (Weizmann Institute of Science, Rehovot, Israel). It carries truncation of the C-terminal region of MSP300 (Rosenberg-Hasson et al., 1996). MSP300 TRIP lines (32377 and 32848) myospheroid TRIP lines (27735 and 33642), WASp TRIP lines (25955 and 36647), a scb⁰¹²⁸⁸ mutant line (11035) and a EGFP-MSP300 protein trap (59757) were from the Bloomington Stock Center. Vinculin RNAi lines (v34585 and v34586; Kaushik et al., 2015) were obtained from the Vienna Drosophila Research Center. The Vinculin-GFP line is tagged at its N-terminus with superfolded GFP using a pFlyFos025866 Fosmid that encompasses the 8 kb of Vinculin gene along with 23.4 kb upstream and 6.8 kb downstream regions integrated in the attp2 site (Kale et al., 2018). Cardiac GAL4 drivers used in this study are Hand^4.2 and TinC Δ4.

Drosophila embryos were processed for immunostaining as described previously for most of the experiments (Patel, 1994). Heat fixation were also performed on wild-type dechorionated embryos for 10 s in boiling water followed by a 5 min devitellinization in a 1:1 mix of heptan and methanol on a shaker. Microscopic observations were performed by confocal microscopy to produce images using Leica SP8. All images were acquired using a 40× oil objective with a numerical aperture of 1.3. The following antibodies were used: rabbit anti-CAP 1:1000 (a gift from A. L. Kolodkin, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Bharadwaj et al., 2013), rabbit anti-Mef2 1:1000 (a gift from M. V. Taylor, School of Biosciences, Cardiff University, Cardiff, UK), guinea pig anti-MSP300 1:2000 (Rosenberg-Hasson et al., 1996) and guinea pig anti-Laminin A 1:1000 (a gift from T. Volk, Weizmann Institute of Science, Rehovot, Israel), guinea pig anti-H15 1:2000 (a gift from J. B. Skeath, Washington University School of Medicine, St Louis, MO, USA), rabbit anti-Trol 1:1000 (a gift from S. Baumgartner Lund University, Lund, Sweden), mouse anti-Slit at 1:100 (DSHB), anti-Seven up at 1:100 (DSHB), anti-Armadillo at 1:100 (DSHB), anti-α-Spectrin 1:100 (DSHB), anti-Vinculin 1:50 (Santa Cruz N-19) and Alexa fluor 488 Phalloidin 1:300 (Invitrogen A12379).

IMARIS software 7.6.5 has been used to open images in a 2D display mode (Slice view). In the menu bar, we selected ‘Edit – Crop 3D’. A rectangle, with yellow borders, will overlay on all three views. It represents the region of interest and it can be modified in size and position, resulting in the cross-sections of the selected part of the heart tube.

The Surpass mode is used to create a 3D reconstruction of the Drosophila embryonic heart tubes; setting the pointer on Navigate mode allows the resulting object to be rotated through a range of angles. This was carried out in two different regions: aorta and the heart proper.

Semi-intact preparations of the fly hearts were made as described previously (Vogler and Ocorr, 2009). Female flies are used for technical reasons (females are bigger). Briefly, flies (20 each genotype) were dissected in oxygenated artificial hemolymph (AHL) to expose the linear tube-like heart. After a 15-20 min recovery in fresh oxygenated AHL, 30 s high-speed movies (∼140 fps) of contracting hearts were captured with a 10× immersion objective. These movies were analyzed with the SOHA software (sohasoftware.com; Ocorr et al., 2007; Vogler and Ocorr, 2009; Fink et al., 2009). The following key heart function parameters were measured: diastolic interval (DI), systolic interval (SI), diastolic diameter (DD) and systolic diameter (SD). Functional parameters were calculated based on these measures including heart period (HP, DI+SI). Fractional shortening (FS) as a measure of contractility was calculated as FS=[(EDD−ESD)/EDD], where EDD is the DD and ESD is the SD.

F-actin was detected and quantified using the Weka segmentation tool in ImageJ in both heart proper and aorta in each embryo. The same classifier has been used on every image. The values are normalized to the total area of the heart.

Microscopic observations were performed by confocal microscopy to produce images using Leica SP8. All images were acquired using a 40× oil objective with a numerical aperture of 1.3, a pixel size of 0.054 µm (x,y) and voxel depth of 0.2 µm. Furthermore, all initial anisotropic voxels are converted to an isotropic voxel (i.e. cubic, xyz=0.1 µm) prior to calculation. The ImageJ plug-in NucleusJ was used to characterize nuclear morphology (Dubos et al., 2020; Poulet et al., 2015). A detailed description of the quantitative parameters generated by NucleusJ can be found in Poulet et al. (2015).

Statistical analyses were performed using Xlstat. The significance of the difference was determined using an unpaired Wilcoxon-Mann-Whitney test. All P-values are two tailed. P<0.05 was considered significant.

We thank Sophie Desset, Tristan Dubos and Caroline Vachias (the GReD institute) for their technical assistance, as well as CLIC facility.