Collective nuclear behavior shapes bilateral nuclear symmetry for subsequent left-right asymmetric morphogenesis in Drosophila

Directional left-right (LR) asymmetry, which is evident in the external and internal morphology of many animals, is genetically determined (Davison, 2020; Hobert et al., 2002; Inaki et al., 2018a,b; Kuroda, 2015). Recent studies show that the mechanisms determining LR asymmetry are evolutionarily divergent (Davison, 2020; Hobert et al., 2002; Inaki et al., 2018a,b; Kuroda, 2015). In vertebrates, several different mechanisms contribute to LR-asymmetric development, including nodal flow, LR-asymmetric proton influx and LR-asymmetric cell migration; some of these mechanisms have parallel functions (Hamada and Tam, 2020; Vandenberg and Levin, 2013). In Lophotrochozoa and Ecdysozoa, intrinsic cell chirality plays a key role in LR-asymmetric development. For example, cell chirality in snail and nematode blastomeres determines their subsequent LR-asymmetric organ and body development (Davison, 2020; Kuroda, 2015; Pohl and Bao, 2010). In Drosophila, the LR-asymmetrical development of several organs also relies on cell chirality, which is controlled by the myosin 1D gene (Hatori et al., 2014; Inaki et al., 2018a; Ishibashi et al., 2019; Lebreton et al., 2018; Sato et al., 2015; Taniguchi et al., 2011). Importantly, chiral cells are also found in vertebrates and are thought to contribute to their LR-asymmetric development (Tee et al., 2015; Wan et al., 2011). However, the molecular mechanisms of LR-asymmetric development in invertebrates remain largely unclear. Drosophila is an excellent model system for studying these mechanisms (Inaki et al., 2016, 2018b; Uechi and Kuranaga, 2018).

At least one other mechanism besides cell chirality is responsible for creating LR-asymmetry in Drosophila (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). The first detectable LR-asymmetry in the developing Drosophila anterior midgut (AMG), observed in the visceral muscles overlying the epithelial tube of the midgut, occurs independently of cell chirality (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). Initially, the long axis of nuclei in these visceral muscle cells is aligned perpendicular to the midline; however, this angle changes and becomes LR-asymmetrical in ventral-region nuclei at stage 13-14, just before overall LR-asymmetric morphological changes begin (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). These visceral muscles play a crucial role in AMG LR-asymmetric development (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). We have previously shown that when the long axis of the nuclei failed to undergo this asymmetric rearrangement (due to augmented JNK signaling or reduced Wnt signaling in the visceral muscles), LR-asymmetry of the AMG also failed (Kuroda et al., 2012; Taniguchi et al., 2007). We also showed that myosin II (MyoII) is essential for both LR-asymmetric AMG development and the LR-asymmetric rearrangement of the long axis of the nuclei in the visceral muscles, which suggests that the change in the angle of the axis is controlled mechanically (Okumura et al., 2010). However, the dynamics and underlying mechanisms of this rearrangement remain elusive.

The location of the nucleus, which is the largest organelle in the cell, changes as needed for various cellular contexts and functions (Gundersen and Worman, 2013). For example, to permit efficient cell migration, the nucleus remains behind the center of the cell, away from the leading edge (Calero-Cuenca et al., 2018). The position of the nucleus can differ with tissue morphology and integrity (Razafsky and Hodzic, 2015; Roman and Gomes, 2018), and defects in nuclear positioning are connected with muscular dystrophy and centronuclear myopathy in humans (Azevedo and Baylies, 2020; Folker and Baylies, 2013). Nuclear migration events depend on LINC (linker of nucleoskeleton and cytoskeleton) complex, which physically links nuclei and F-actin/microtubules (Infante et al., 2018).

Here, we have studied the movement of nuclei in the visceral muscle overlying the midgut in stage 13-14 wild-type Drosophila using three-dimensional (3D) time-lapse movies and quantitative imaging analysis. We found that the nuclei of the visceral muscles were positioned LR symmetrically in distinct regions along the anterior-posterior axis in wild-type embryos; hereafter, we refer to this distribution as proper nuclear positioning. The densely crowded nuclei in these regions actively rearranged their positions relative to neighboring nuclei; hereafter, we refer to this as collective nuclear behavior. Dally-like protein (Dlp), a component of Wnt signaling, was essential for both proper nuclear positioning and collective nuclear behavior. MyoII and the LINC complex were required for proper nuclear positioning but not for collective nuclear behavior. Unexpectedly, however, the nuclei aligned LR asymmetrically in mutants with disrupted MyoII or LINC complex, although the AMG developed LR symmetrically. Our results show that the positioning of the nuclei in the visceral muscles is accomplished via multiple regulatory machineries, including Wnt signaling, MyoII and the LINC complex, and that the LR-symmetric positioning of the nuclei is important for the LR-asymmetric development of the AMG.

The midgut is composed of the epithelial tube and the overlying visceral muscles (Fig. 1A). The visceral muscle cells, which are binucleated and bipolar, align LR symmetrically at the lateral sides of the embryo with the long axis of each nucleus perpendicular to the midline (Fig. 1A,B) (Klapper et al., 2002; Kuroda et al., 2012; Schroter et al., 2006). In stages 13-14, the leading edges of the visceral muscles extend dorsally and ventrally toward the dorsal and ventral midlines, respectively, and eventually merge at the midlines at late stage 14 (Fig. 1A) (Kuroda et al., 2012). Studies show that the first detectable LR asymmetry in the AMG is a difference between the right and left sides in the angle between the long axis of the nuclei and the midline in the ventral side of this organ at stage 14 (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). As these studies were conducted in fixed embryos, the events leading to the LR asymmetry of the visceral muscle nuclei and the AMG are still unclear.

Therefore, to examine the process by which nuclei are arranged, we obtained 3D time-lapse movies of the midgut in developing embryos from stage 13 to 14 using a confocal laser scanning microscope. We used the GAL4/UAS system to drive the visceral muscle-specific expression of UAS-RedStinger, which encodes a nuclear DsRed, and of UAS-lifeact-EGFP, which encodes a GFP with an actin-binding peptide (Haralalka et al., 2014). The time-lapse movies were obtained from the ventral side of the embryo (Fig. 1B,B′, Movie 1). We designated the time point when the leading edges of the visceral muscles merged at the midline (approximately corresponding to the end of stage 14) as T4; we set T1, T2 and T3 at 30, 20 and 10 min before T4, respectively (Fig. 1B,B′).

In wild-type embryos, the nuclei were densely aligned in distinct regions along both sides of the anterior-posterior axis, creating a region visually similar to the mammalian rib cage, from T1 to T4, in all cases examined (n=10) (Fig. 1B,B′). Collectively, the proper nuclear positioning, referring to the overall positioning of the nuclei with respect to the midline, was maintained from T1 to T4; however, the positions of the individual nuclei changed relative to one another (Fig. 1C-C″, Movie 1). By tracking the position of individual nuclei over time, we found that the nuclei actively moved and adjusted their position relative to each other in all wild-type embryos examined (n=10) (Fig. 1C-C″, Movie 1). We plotted the position of individual nuclei every 2.41 min, starting at T1; at higher magnification, the time-lapse images revealed small movements of the individual nuclei along disparate paths (blue lines in Fig. 1D, Movie 1) and that nuclei changed position relative to each other (arrowheads, Fig. 1E,F), even though cell position changed very little over the 20 min imaging period. Furthermore, despite the dense grouping of the nuclei, they were clearly separated from each other by F-actin along the anterior-posterior axis (Fig. 1G,G′). Thus, the changes in the relative positions of the nuclei were due to the movement of the nuclei within the cells, rather than the rearrangement of entire muscle cells, and we defined this novel collective positioning behavior as collective nuclear behavior. We speculated that, as with other specific nuclear behaviors, collective nuclear behavior is under the control of genetic pathways and may contribute to the LR-asymmetric development of the embryonic midgut (Azevedo and Baylies, 2020; Calero-Cuenca et al., 2018; Folker and Baylies, 2013; Gundersen and Worman, 2013; Razafsky and Hodzic, 2015; Roman and Gomes, 2018).

We conducted a genetic screen that identified a new allele, dlp³, as a mutation that affects the LR-asymmetric development of the AMG (Fig. 2A,B) (the genetic screen will be reported elsewhere). Our sequence analysis revealed that dlp³ carries a nonsense mutation that introduces a stop codon at the 133rd amino acid residue. Embryos homozygous for dlp³ or dlp^MH20 (an amorphic dlp allele), or trans-heterozygous for dlp³ and dlp^MH20 showed similar defects in AMG LR asymmetry, including inverted LR asymmetry and bilateral symmetry (Fig. 2C). The dlp gene encodes a core protein of Drosophila glypicans, a family of heparan sulfate proteoglycans (Kim et al., 2011; Williams et al., 2010) (including Dlp) that is involved in regulating several cell-signaling pathways, including the Wnt, transforming growth factor β and fibroblast growth factor signaling pathways (Kim et al., 2011; Lin, 2004).

We have previously shown that Wnt4 signaling must be active in the visceral muscle of the AMG for this organ to develop proper LR asymmetry (Kuroda et al., 2012). Thus, we speculated that LR-asymmetric AMG development also requires dlp function in the visceral muscle. To test this possibility, we overexpressed UAS-dlp specifically in the visceral muscles of the midgut, using the GAL4/UAS system driven by hand, 65E04 or 24B, to see whether it could rescue LR defects in dlp³ homozygotes (Kuroda et al., 2012). Control embryos carrying only UAS-dlp (no driver) showed LR defects of the AMG (52% frequency), as did dlp³ homozygotes (54%) (Fig. 2D) (Kuroda et al., 2012). As expected, UAS-dlp overexpression markedly suppressed these LR defects when driven by hand (frequency of LR defects 12%), 65E04 (19%) or 24B (12%) (Fig. 2D). In contrast, the frequency of LR defects was not suppressed by overexpressing UAS-dlp in the midgut epithelium (NP5021, 60%) or nervous system (Elav-Gal4, 42%), when compared with control (Fig. 2D). Although arm-GAL4 is used to drive ubiquitous expression, including in visceral muscles, dlp expression driven by arm-GAL4 in dlp³ homozygotes did not suppress LR defects (Fig. 2D). We speculated that this might be due to potential LR defects associated with dlp misexpression in some tissues. Indeed, the ubiquitous misexpression of UAS-dlp driven by arm-Gal4 in wild-type Drosophila causes LR defects, whereas control embryos carrying UAS-dlp but no Gal4 driver had no LR defects (Fig. 2D). Taken together, our results show that wild-type dlp is required in the visceral muscles for normal LR-asymmetric development of the AMG, which is consistent with our previous finding that normal LR-asymmetric AMG development requires activated Wnt4 signaling in the visceral muscles of the midgut (Kuroda et al., 2012).

Furthermore, recent studies show that Dlp associates with Wnt4 and regulates Wnt signaling in germline cells (Tu et al., 2020; Waghmare et al., 2020). Therefore, we hypothesized that dlp contributes to Wnt4 signaling in the visceral muscles, and thus contributes to LR-asymmetric AMG morphogenesis. To test this possibility, we specifically overexpressed UAS-disheveled (dsh), which can cell-autonomously activate Wnt signaling, in the visceral muscles (driven by hand) or midgut epithelium (driven by NP5021) of dlp³ homozygotes, and examined the effect on LR defects (Fig. 2E) (Morel and Arias, 2004; Wehrli et al., 2000). Compared with control embryos (carrying UAS-dsh but no Gal4 driver), the frequency of LR defects associated with the dlp³ mutant decreased when UAS-dsh was overexpressed in the visceral muscle (12%) but not when overexpressed in the midgut epithelium (40%) (Fig. 2E). The different frequencies of LR defects in the no-driver controls with dlp³ homozygous background (carrying only UAS-dlp or UAS-dsh) could be explained by the distinct effects of leaky expression from these UAS lines (Fig. 2D,E). Taken together, these findings suggest that dlp is required for the activation of Wnt4 signaling in the visceral muscles, and this activation is essential for normal AMG LR-asymmetric development.

Wnt signaling plays multiple roles in embryonic development (DasGupta et al., 2005; Swarup and Verheyen, 2012). Thus, mutants of genes that encode the core components of Wnt signaling show a broad range of phenotypes, including gut deformation, in addition to defects in LR asymmetry (Bejsovec, 2018; Swarup and Verheyen, 2012). Nonetheless, the structure of the midgut in dlp³ mutants was largely normal except for LR randomization, suggesting a specific function for dlp in LR-asymmetric morphogenesis (Fig. 2A-C). For example, the extension of the leading edge of the midgut visceral muscles toward the midline is normal in dlp mutant embryos, demonstrating that dlp is dispensable for this extension (Fig. 3A,B). Therefore, in the following studies of nuclear behavior in AMG visceral muscles, we used the dlp³ mutant to study the visceral muscle-specific depletion of Wnt4 signaling.

To reveal potential defects in the positioning of visceral muscle nuclei in dlp mutant embryos, we examined 3D time-lapse movies of the AMG of wild-type and dlp³ homozygous embryos from T1 to T4. When examining 2D snapshots projected from the 3D time-lapse movies, we noticed that the nuclei were more dispersed in dlp³ mutants than in wild-type embryos (Fig. 3A,B). To track the behavior of nuclei in the visceral muscles in the midgut, which is a thick, rounded organ, we used a surface-modeling approach (Fig. 3C-F, Movie 2). In the surface-modeling analyses, visceral muscles are outlined in green, representing the outer surface of lifeact-EGFP distribution driven by 65E04-Gal4, a visceral muscle-specific Gal4 driver (Fig. 3C, left). Nuclear position was defined as the center of the surface-modeled nucleus (Fig. 3C, right). The outline of the visceral muscles (green) was merged with the position of the nuclei (red spheres) using image analysis software (Fig. 3C). In our previous studies relying on fixed embryos, the first indication of LR-asymmetric changes was found in nuclei in the posterior part of the AMG (Kuroda et al., 2012; Okumura et al., 2010; Taniguchi et al., 2007). Therefore, in this study, we selected nuclei located 40-80 μm from the anterior tip of the midgut for further analysis (Fig. 3D, shown in magenta).

To detect potential defects in nuclear positioning, we measured the position of nuclei relative to the midline of the AMG. In the surface model, the midline (red) was placed along the merged points of the left and right visceral muscles at T4 (Fig. 3D). We then measured the distance from the center of each nucleus to the midline (Fig. 3E). Considering potential differences in the size of the AMG, we normalized nucleus-midline distances as a ratio (percentage) relative to the maximum width of the AMG and calculated the mean of the normalized distances in each embryo (width of the blue box in Fig. 3E). Thus, our procedures assumed an involvement of potential biological scaling in nuclear postponing. Values and standard deviations were calculated for T1-T4.

We used this procedure to analyze the distance between the nuclei and the midline in wild-type and dlp³ homozygous embryos. We then averaged the mean values from 10 embryos (the average number of nuclei in each embryo was 20.1±4.8, Fig. S1) and defined this as the distance between the nuclei and the midline, and found that the distance from the nucleus to the midline was significantly less in the visceral muscles of dlp³ mutants, on both the right and left sides, than in wild-type embryos, at T1-T4 (Fig. 4A,B). Importantly, the specific overexpression of UAS-dlp in the visceral muscles, driven by 65E04, rescued this defect in both the left and right sides in dlp³ homozygotes (Fig. 4A,B). Therefore, the loss of Wnt4 signaling in the midgut visceral muscles caused mispositioning of the nuclei, such that they approached the midline more closely (on both the left and right sides) than in wild-type visceral muscles. Thus, Wnt4 signaling is required for proper nuclear positioning.

In 3D time-lapse movies, nuclei appeared more dispersed in the visceral muscles of dlp³ mutants compared with wild-type embryos (Fig. 3A,B). To measure defects in collective nuclear behavior, we calculated a collectivity index to represent the mean distances between each nucleus and its nearest posterior neighbor, normalized as a percentage of the maximal width of the midgut (Fig. 3F). We then averaged the collectivity index values from 10 embryos and calculated the standard deviations at T1-T4 (Fig. 4C,D). The collectivity index of the left and right visceral muscles was higher in dlp³ homozygotes than in wild-type embryos at stage T1-T4; this difference was statistically significant at T2 to T4 for the left side and at T2 for the right side (Fig. 4C,D). These results suggest that collective nuclear behavior depends on Wnt4 signaling. Indeed, collective nuclear behavior defects were rescued in the visceral muscle of dlp³ mutant embryos overexpressing UAS-dlp, as their collectivity index was similar to that of wild-type embryos at T2-T4 on the right side (Fig. 4D). Although the rescue effect was weaker on the left side, the collectivity index did not differ significantly between wild-type and rescued embryos (P values for T1-T4 ranged from 0.11 to 0.95) (Fig. 4C). Therefore, Wnt4 signaling in the visceral muscle regulates both proper nuclear positioning and collective nuclear behavior.

Considering our observations that the nuclei actively moved and changed their positions relative to each other in wild-type embryos (Fig. 1C,D, Movie 1), we speculated that the reduced collectivity of nuclei in dlp mutants could be due to augmented movement. We used 3D time-lapse movies to track the migration of nuclei in wild-type and dlp³ homozygous embryos by determining their position at 5 min intervals over a 30 min period beginning at T1 (Fig. 4E, Movie 1). Mean values calculated for migration distance (μm) and averaged for three embryos demonstrated that nuclei migrated farther in both left and right sides of dlp³ mutants than in wild-type embryos; the difference in the right side was statistically significant (Fig. 4E). Thus, accelerated migration may be responsible for the dispersion of nuclei in dlp mutants.

We next examined the mechanisms underlying proper nuclear positioning and collective nuclear behavior. The LINC complex, which consists of KASH- and SUN-domain proteins, physically links the nuclear envelope and the cytoskeleton and plays crucial roles in nuclear migration in several species, including Drosophila (Calero-Cuenca et al., 2018). Muscle-specific protein 300 kDa (Msp300), a Drosophila KASH-domain protein (Nesprin-like protein), is required for the proper positioning of the nuclei in skeletal muscles and the eye imaginal disc (Patterson et al., 2004; Volk, 2013; Xie and Fischer, 2008). Therefore, we investigated potential roles for Msp300 in the LR asymmetry of the AMG and in proper nuclear positioning and collective nuclear behavior.

We analyzed AMG LR asymmetry using Msp300^ΔKASH, a Msp300 loss-of-function allele that encodes a mutant protein lacking the KASH domain required for its activity (Xie and Fischer, 2008). The predominant LR defect in Msp300^ΔKASH homozygous embryos was a no-laterality phenotype in the AMG (18%), whereas LR defects were rare in wild-type or Msp300^ΔKASH heterozygous embryos (Fig. 5A). Interestingly, surface-modeling analyses revealed that the distance between the nuclei and the midline in in the right-side visceral muscles of Msp300^ΔKASH embryos at T1-T4 was significantly less in than in wild type (Fig. 5D,E). Thus, the requirement for Msp300 in proper nuclear positioning was LR asymmetric, unlike for dlp, which was required for both the right and left sides (Fig. 5D,E). We also analyzed collective nuclear behavior in Msp300^ΔKASH homozygotes, and found that, despite the defect in proper nuclear positioning, the collectivity index did not differ significantly from that of wild-type embryos at T1-T4, revealing that collective nuclear behavior was not markedly disrupted (Fig. 5B,C,F,G). However, the standard deviation for the right side was significantly larger in Msp300^ΔKASH mutants compared with wild type (P values: T1, 0.001; T2, 0.0005; T3, 0.01; T4. 0.03), which suggests that the collectivity index varied among the individual embryos (Fig. 5G).

MyoII contributes to LINC complex-dependent nuclear migration in various systems by physically linking F-actin (Gundersen and Worman, 2013). We previously reported that zipper² (zip²), a mutant of the gene encoding MyoII heavy chain, produced a symmetrical AMG phenotype reminiscent of the Msp300^ΔKASH mutant phenotype at a frequency of 60%, whereas LR defects rarely occurred in wild-type or zip² heterozygous embryos (Fig. 5A) (Okumura et al., 2010). Moreover, MyoII is required in AMG visceral muscles for normal LR-asymmetric development of the organ (Okumura et al., 2010). Given the relevance of aberrant nuclear positioning to the LR defects we observed, we analyzed collective nuclear behavior in zip² homozygotes. As in Msp300^ΔKASH mutants, the average distance between the nuclei and the midline was decreased in zip² mutants compared with wild-type embryos at T2-T4, but only in the right-side visceral muscles (Fig. 5D,E). Thus, nuclear positioning was LR-asymmetric in Msp300^ΔKASH and zip² mutants, although it was LR symmetric in wild-type embryos (Fig. 5F,G). In other words, Msp300 and zip may be required only in the right-side visceral muscles in wild-type embryos. However, the average collectivity index did not differ significantly between zip² and wild-type embryos, except for a slight reduction in the zip² mutants at T1 (Fig. 5F,G). Based on these results, we speculated that proper nuclear positioning is controlled by a MyoII-dependent mechanical force applied to the nuclear envelope via physical links between F-actin and the LINC complex. However, these mechanical processes may be irrelevant to collective nuclear behavior. Nevertheless, in mutants with defects in proper nuclear positioning, the LR-asymmetry of the AMG was also disturbed; but this was not always the case with defects in collective nuclear behavior. Therefore, proper nuclear positioning in the visceral muscles may be a prerequisite for establishing normal LR asymmetry, but it may also be integral to the mechanism of LR-asymmetrical development.

In the developing embryo, the breaking of bilateral symmetry is the primary cue that initiates the cell signaling, gene expression and morphological changes that support LR-asymmetric development (Grimes and Burdine, 2017; Hamada et al., 2002; Hirokawa et al., 2006). In the mouse embryo, the clockwise rotation of the nodal cilia breaks bilateral symmetry by inducing the leftward flow of the extra-embryonic fluid (Shinohara et al., 2012). In snails and nematodes, blastomere chirality breaks the bilateral symmetry of the embryo at early cleavage stages and drives the subsequent LR-asymmetric events (Davison, 2020; Kuroda, 2015; Pohl and Bao, 2010). In these scenarios, the initial cue that initiates LR asymmetry is gradually amplified to achieve the LR-asymmetric development of the whole body. However, our present study revealed a different strategy, in which achieving LR symmetry is a crucial step toward establishing LR-asymmetry (Fig. 6).

Here, we have demonstrated that the bilaterally symmetric arrangement of the nuclei in the visceral muscles of the AMG is required for the LR-asymmetric development of this organ (Fig. 6). In the absence of MyoII or a LINC-complex component, the nuclei align LR asymmetrically but the AMG develops LR symmetrically (Fig. 6A,C). We previously determined the phenocritical period of zip in the LR-asymmetric development of the AMG and found that zip is required at stage 11-13 (this includes the time period when proper nuclear positioning occurs) but is dispensable from stage 14 onward (this includes the T4 period, when LR-asymmetric morphogenesis begins) (Okumura et al., 2010). Taken together, these results suggest an indispensable role for MyoII in preventing asymmetric nuclear positioning through stage 13, rather than in driving the subsequent LR-asymmetric morphogenesis that begins in stage 14. Thus, MyoII and the LINC complex play important roles in the LR-symmetric rearrangement of the nuclei, which is required for or coupled with the subsequent LR-asymmetric morphogenesis (Fig. 6D). We speculate that such translation between lateral symmetry and asymmetry may act as an additional layer of regulatory steps, and that this multi-layered regulation allows multiple mechanisms to contribute to LR-asymmetric development in a species. In such cases, complex LR-asymmetric structures can be built with a limited number of machineries.

Although a requirement for proper nuclear positioning in LR asymmetry has not been reported previously, defective nuclear positioning has been connected to human diseases (Azevedo and Baylies, 2020; Folker and Baylies, 2013). Mutations of genes that encode key molecules for nuclear positioning, such as the LINC complex, are associated with Emery-Dreifuss muscular dystrophy and cerebellar ataxia (Gundersen and Worman, 2013), and genetic analysis in model animals has revealed that the LINC complex plays key roles in the development of these diseases (Azevedo and Baylies, 2020). In Drosophila optic epithelium and in vertebrate neuroectoderm, defects in nuclear migration and positioning affect the pattern of cell division (Del Bene et al., 2008; Patterson et al., 2004). However, it is unlikely that cell division initiates the LR-asymmetric development of the AMG, because cell propagation is complete before the collective nuclear rearrangement and LR-asymmetric development of this organ begins (Gunage et al., 2017). On the other hand, defective nuclear positioning may mechanically influence AMG morphogenesis. The nucleus can act as a piston that physically compartmentalizes the cytoplasm and provides hydrostatic pressure toward the direction of nuclear migration (Calero-Cuenca et al., 2018; Petrie et al., 2014). Given that the nucleus can provide this type of dynamic force, the positioning of the nuclei may help to create mechanical forces that promote LR-asymmetric morphogenesis. The midgut is also subject to mechanical forces from rhythmic contractions in the visceral muscles of the developing embryo; however, these muscle pulses begin at stage 16, after the onset of LR-asymmetric morphogenesis of the AMG at stage 14 (Movie 1), and thus are not part of any mechanical interactions involved in initiating LR-asymmetrical development.

Here, we revealed two distinct events that control nuclear location, proper nuclear positioning and collective nuclear behavior (Fig. 6A,B,C), both of which require Wnt4 signaling (Fig. 6B). However, the LINC complex and MyoII are required for proper nuclear positioning but not for collective nuclear behavior (Fig. 6C), demonstrating that the two events depend on distinct underlying mechanisms. MyoII provides contractile force to F-actin, which is involved in LINC complex-dependent nuclear movement in other systems (Gundersen and Worman, 2013). Considering that nuclei in the right-side visceral muscles shifted toward the midline in the absence of MyoII or a LINC complex component, these two factors may introduce an ability to resist a pulling force from the midline (Fig. 6D). Resistance to such a pulling force might also derive from the counteracting forces of LR-asymmetric tissue deformation (Fig. 6D). This idea is consistent with our observation that AMG morphogenesis was bilaterally symmetrical in the absence of MyoII or a LINC complex component. We have previously demonstrated that the angle of the long axes of the nuclei relative to the midline changed in the right side of wild-type embryos but not in various mutants in which AMG LR asymmetry is disrupted, including Wnt4 and MyoII mutants (Kuroda et al., 2012; Okumura et al., 2010). Although this LR-asymmetrical alteration in axial angle can be reproducibly demonstrated in fixed embryos, it was not detected in our live-imaging analysis in the present study. This inconsistency is likely due to the dynamic movement of the nuclei and consequent fluctuations in the axial angle in living embryos. Alternatively, the LR-asymmetric developmental force as in our proposed model might be visualized by the fixing process as LR asymmetry of the nuclear axis (Fig. 6D). The generation of a biosensor that can measure forces in the visceral muscle by live imaging would allow us to find answers to this question.

We also found that Wnt4 signaling is required for collective nuclear behavior in the visceral muscles (Fig. 6B). In the wild-type embryo, the nuclei are densely packed into a limited area in each lateral half of the ventral region of the AMG (Fig. 6A). However, when Wnt4 signaling was interrupted, as in dlp mutants, the nuclei were sparsely distributed over a larger area and migrated more actively (Figs 2F and 6B). This observation suggests that Wnt4 signaling might organize the collective movement of the nuclei in wild-type embryos by downregulating nuclear migration. A specific association between the LR-randomization phenotype and defects in Wnt4 signaling suggests that defective nuclear placement, shown by a more dispersed distribution and a failure to preserve a distance from the midline, might contribute to LR randomization (Fig. 6B) (Kuroda et al., 2012). If this is the case, proper placement of the nuclei may be important for directing the LR polarization of the mechanical force driving AMG morphogenesis. Our results also suggest that the degree of the collective nuclear behavior varied between individual Msp300^ΔKASH mutants, even though their collectivity index did not differ significantly from that of wild-type embryos. However, Drosophila has multiple KASH genes, and their redundant functions may make it difficult to fully ascertain the role of Msp300 in collective nuclear behavior (Technau and Roth, 2008). At this point, it is not clear whether there is some functional interplay between Wnt4 signaling and MyoII/LINC complex. However, LINC complex regulates Wnt signaling in mammals, so a similar biochemical interaction may contribute to proper nuclear positioning in the Drosophila embryonic visceral muscle (Uzer et al., 2018).

In this analysis, we demonstrated that nuclear position is crucial in forming LR asymmetry. Considering that non-skeletal muscles – which are, like Drosophila visceral muscles, formed of multi-nucleated cells – contribute to LR-asymmetric organs and tissues such as the heart, blood vessels and digestive organs in vertebrates and other organisms, the contribution of nuclear positioning to LR-asymmetric development may be evolutionarily conserved.

We used Canton-S as the wild-type control strain. We also used Drosophila lines with the following genotypes: dlp³, a loss-of-function allele (induced by ethyl methane sulfonate in this study); dlp^MH20, a null allele (Franch-Marro et al., 2005); zip², an amorphic allele (Zhao et al., 1988); and Msp300^ΔKASH, a loss-of-function allele (Xie and Fischer, 2008). We used the UAS lines UAS-lifeact-EGFP (Haralalka et al., 2014), UAS-RedStinger (Barolo et al., 2004), UAS-dlp (Wang and Page-McCaw, 2014) and UAS-dsh (Kuroda et al., 2012), and the following Gal4-driver lines: arm-Gal4, which is an ubiquitous driver (Sanson et al., 1996); hand-Gal4, which is specific to circular visceral muscles and cardiac cells (Popichenko et al., 2007); 65E04, which is specific to circular visceral muscles (Jenett et al., 2012); 24B, which is specific to the somatic, circular visceral and longitudinal visceral muscles (Michelson, 1994); NP5021, which is specific to the endodermal epithelium (Hozumi et al., 2006); and Elav-Gal4, a pan-neuronal driver (Lin and Goodman, 1994).

Mutations on the second chromosome were balanced with Cyoβ. Mutations on the third chromosome were balanced with TM2β. All genetic crosses were performed at 25°C on standard Drosophila culture media.

Immunostaining of embryos was as previously described (Hozumi et al., 2008). We used the following primary antibodies: mouse anti-β-galactosidase (Promega, z3781; 1:1000), rabbit anti-GFP (MBL, 598; 1:1000), rat anti-HA (3F10; Roche Diagnostics, 1:1000), mouse anti-connectin [C1.427, Developmental Studies Hybridoma Bank (DSHB), 1:5], mouse anti-Crumbs (Cq4, DSHB, 1:30) and mouse anti-FasIII (7G10, DSHB, 1:100). We used Cy3-conjugated anti-mouse IgG (Jackson ImmunoResearch, 715-165-151; 1:500) and biotinylated anti-mouse IgG (Vector Labs, BA2000; 1:200) as secondary antibodies. We used the Vectastain ABC kit (Vector Labs) for biotin-staining reactions.

Embryos were fixed and stained with anti-Fas3 (DSHB, 1:50) as described previously (Kuroda et al., 2012). Images were obtained with an LSM 700 scanning laser confocal microscope (Carl Zeiss). The LR asymmetry of the AMG in the fixed embryo was scored based on the position of the joint between the proventriculus and the AMG relative to the midline, as previously described (Kuroda et al., 2012). Briefly, if the joint was to the left of the midline, the phenotype was scored as normal; if it was to the right, it was scored as inverse; and if overlapping the midline, it was scored as no laterality.

Embryos with the following genotypes were collected before stage 13: 65E04, UAS-RedStinger/UAS-lifeact-EGFP (used as wild-type); 65E04-GAL4,dlp3,UAS-RedStinger/65E04-GAL4 dlp3,UAS-lifeact-EGFP-p10 (referred to as dlp³ mutant); UAS-RedStinger/+;dlp^MH20,65E04-gal4,UAS-Lifeact-GFP/UAS-dlp, dlp^MH20 (referred to as dlp³ rescued); Msp300^ΔKASH/Msp300^ΔKASH; 65E04, UAS-RedStinger/65E04, UAS-lifeact-EGFP (referred to as Msp300^ΔKASH mutant); and zip²/zip²; 65E04, UAS-RedStinger/65E04, UAS-lifeact-EGFP (referred to as zip² mutant). Embryonic eggshells were removed by immersion in 50% bleach for 1 min, followed by a wash in water. Stage 13 embryos of the appropriate genotype were selected under florescence microscopy, mounted ventral-side up on double sticky tape on a glass slide, placed between 0.25 mm thick spacers made from coverslips, mounted in oxygen-permeable Halocarbon oil 27 (Sigma) and covered with a coverslip (Inaki et al., 2018a). 3D time-lapse movies of embryos at 18°C were taken every 10 min for 2 h using the LSM 880 (Carl Zeiss). At stage 13, the anterior-posterior axis of the embryo (identified by head and tail structures), was manually reoriented to the y-axis of the image. We obtained 3D time-lapse movies using z stacks (13-15 images at 5 µm intervals). To reduce phototoxicity, we used a relatively low laser power (488 nm laser, 0.03-1; 561 nm laser, 0.3-5). The time-lapse images were saved as LSM files in ZEN software (2012 SP1 black edition, Release Version 8.1, Carl Zeiss).

The time-lapse movies were 3D-reconstructed using Imaris image analysis software (Bitplane). The positions of the surface-modeled nuclei in 3D coordinates were determined for each time point using the Spot function. The 3D surface models of the visceral muscles were constructed using the Surface function. The 3D surface models and the positions of the center of each nucleus were saved as VRML (Virtual Reality Modeling Language) files, to be analyzed simultaneously.

To easily obtain measurements in 3D space, Maya version 2018 (Autodesk, San Rafael, CA), a computer animation and modeling software, was used to animate the 3D-surface model (Park et al., 2019). To use Maya, the VRML files, in which the 3D-surface models of visceral muscles and the centers of the nuclei were integrated, were converted to 3DS (native file format of the old Autodesk 3D Studio DOS) using Meshconv (https://www.patrickmin.com/meshconv).

The 3D space files were imported into Maya version 2018 (Autodesk). The surface models of the visceral muscles were transparently colored and then added to the layers (Fig. 3C). Although 65E04-driven UAS-RedStinger expression is highly specific to the visceral muscles, some nuclei outside the visceral muscles were also labeled. Therefore, any nuclei that lay outside the surface-modeled visceral muscles were manually deleted. In previous studies, we detected the LR-asymmetric rearrangement of the nuclei in the posterior half of the AMG at stage 15. Thus, in this study, the nuclei were divided by whether they were in the anterior or posterior region of the AMG, corresponding to 0-40 µm and 40-80 µm from the anterior tip of the midgut, respectively (Fig. 3D), and nuclei in the posterior region were analyzed further.

Distances between the nuclei and the midline of the AMG were measured using Maya version 2018 (Autodesk) (Park et al., 2019). To define the midline, we used Maya's Convert function to fit the surface-modeled AMG in a minimal cuboid placed along the anterior-posterior axis of each embryo (Fig. 3D). Thus, the width of the cuboid corresponded to the maximal width of the AMG. The midline was manually placed in the 3D surface model as a line parallel to the anterior-posterior grids of the cuboid that connected the points where the left- and right-side visceral muscles merge at T4 (Fig. 3D). This midline was used retrospectively for data captured at T1 to T3.

We measured distances between nuclei in the posterior part of the surface model (40-80 μm from the anterior tip) and the midline by measuring the length of the line drawn perpendicularly from the midline to the nucleus (Fig. 3E). To adjust for differences in the size of the AMG, the values were normalized as a ratio (percentage) of the maximum width of the cuboid, and the mean of the normalized distances was calculated for each embryo. This procedure was carried out automatically using Python Script in Maya and NumPy library (see supplementary information). We then averaged the mean values and calculated standard deviations for T1-T4.

To analyze the collectivity of nuclear arrangement, we measured the distance between each nucleus and its nearest most-posterior neighbor, and calculated the mean distance for each embryo. We used Python Script and NumPy library in Maya version 2018 (Autodesk) to automatically connect each nucleus in the lower region of the surface model (40-80 µm below the anterior tip) to its nearest most-posterior neighbor and to calculate the length of the connecting line (Fig. 3F). This protocol allowed us to obtain all relevant distances while measuring the distance between any two neighboring nuclei only once. Distances were measured separately for the right and left sides of the visceral muscles, and mean values were obtained for each embryo (Fig. 3F). We averaged the mean values and calculated standard deviations for T1-T4 using the AVERAGE and STDEV.P functions in Excel. To normalize differences in embryo size, values are presented as a percentage of the maximum width of the AMG (the width of the cuboid) (Fig. 3F).

We measured migration distance by tracing the paths traveled by the nuclei. We used 3D time-lapse movies obtained at 5 min intervals over a 30 min period. The 3D movies, composed of 13-15 z stacks, were converted to 2D-sequence image files using the maximum intensity projection feature in ZEN software (Carl Zeiss). 2D-sequence image files were imported into Maya version 2018 (Autodesk) and displayed to track the migrating nuclei. We manually traced the path of each nucleus through the sequence of images using the EP curve tool in Maya. The length of each traced path (μm) was then automatically measured using a Python script in Maya (see supplementary information). From these measurements, we calculated the average migration distance and standard deviation.

Statistical processing was carried out in Maya version 2018 (Autodesk) and Excel 2013. The calculated values were copied to Excel, and the average of the mean values and their standard deviations were calculated using the AVERAGE and STDEV.P functions. To evaluate the statistical significance of the differences between phenotypes, we conducted F-tests and paired t-tests using Excel's F.TEST and T.TEST functions. Before these tests, we confirmed that all data had normal distributions. In Fig. 2D, a χ2 test was conducted using the the CHITEST function in Excel.

We thank the Bloomington Drosophila Stock Center (Indiana University), the Drosophila Genetic Resource Center (Indiana University), and the Kyoto Stock Center (Kyoto Institute of Technology) for fly stocks, and the Developmental Studies Hybridoma Bank (University of Iowa) for antibodies.

This work was supported by Japan Society for the Promotion of Science KAKENHI (18H02450 and 15H05857). Deposited in PMC for immediate release.