The number of growing microtubules and nucleus-nucleus interactions uniquely regulate nuclear movement in Drosophila muscle

Nuclear movement is a fundamental process of eukaryotic cell biology. Skeletal muscle presents an intriguing model to study nuclear movement because its development requires the precise positioning of multiple nuclei within a single cytoplasm. Furthermore, there is a high correlation between aberrant nuclear positioning and poor muscle function. Although many genes that regulate nuclear movement have been identified, the mechanisms by which these genes act is not known. Using Drosophila melanogaster muscle development as a model system, and a combination of live-embryo microscopy and laser ablation of nuclei, we have found that phenotypically similar mutants are based in different molecular disruptions. Specifically, ensconsin (Drosophila MAP7) regulates the number of growing microtubules that are used to move nuclei whereas bocksbeutel (Drosophila emerin) and klarsicht (Drosophila KASH-protein regulate interactions between nuclei.


INTRODUCTION
Since the identification of the Linker of Nucleoskeleton and Cytoskeleton (LINC) complex (Crisp et al., 2006;Starr and Fridolfsson, 2010;Tapley and Starr, 2013), the question of how nuclei move has been a pressing question in biology. The process of moving this heavy organelle is conserved throughout evolution in all cell types (Mosley-Bishop et al., 1999;Tran et 5 al., 2001;Starr et al., 2001;Lee et al., 2002;Starr and Han, 2002;Del Bene et al., 2008;Zhang et al., 2009;Yu et al., 2011), thus magnifying the importance of understanding the underlying mechanism. Although many mechanisms have been described for mononucleated cells (Gundersen and Worman, 2013), how nuclei are moved in a syncytium has remained a mystery.
Many genes that regulate nuclear position in syncytial skeletal muscle cells have been identified 10 (Roman and Gomes, 2018), but how these genes contribute to nuclear movement and whether these genes regulate nuclear positioning through a single mechanism is not known.
In most contexts, nuclear movement is dependent on the microtubule cytoskeleton and its associated proteins which generate the force to move nuclei and the Linker of nucleoskeleton and cytoskeleton (LINC) complex which transmits force between the cytoskeleton and the 15 nucleus. This is indeed true during the development of the syncytial abdominal musculature of Drosophila melanogaster embryos and larvae. Several microtubule associated genes including ensconsin/MAP7 (Metzger et al., 2012), Bsg25D/Ninein (Rosen et al., 2019), and the motors kinesin and cytoplasmic dynein (Folker et al., 2012(Folker et al., , 2014 have been suggested to contribute to nuclear movement by regulating Kinesin activity (Metzger et al., 2012), microtubule stability 20 (Rosen et al., 2019), and the application of force both directly on (Folker et al., 2014) and at a distance from (Folker et al., 2012). Similar experiments have shown that the LINC complex components klarsicht, (Elhanany-Tamir et al., 2012;Collins & Mandigo et al., 2017), Msp300 (Elhanany-Tamir et al., 2012), and klaroid (Tan et al., 2018) along with the emerin homologs bocksbeutel and Otefin (Collins & Mandigo et al., 2017;Mandigo et al., 2019) are also critical for 25 4 nuclear positioning during muscle development. Despite identifying many of the factors that are critical for nuclear position, we know little about the mechanisms by which they support nuclear movement during muscle development.
The limited mechanistic understanding is in part driven by the complexity that many nuclei in a single cytoplasm creates. Although many studies investigating myonuclear 5 movement have been done in cell culture (Cadot et al., 2012;Wilson and Holzbaur, 2012), such in vitro systems lack the complex signaling cascades that provide directionality cues to nuclei as they translocate, highlighting the importance of studying nuclear movement in an organismal context (Folker et al., 2014). Consequently, most in vivo work has relied on describing nuclei as mispositioned with little, if any, distinction between phenotypes (Metzger et al., 2012;Collins & 10 Mandigo et al., 2017;Folker et al., 2012;Elhanany-Tamir et al., 2012). To better understand the mechanisms by which each gene regulates nuclear movement, it is critical to establish methods that can characterize nuclear phenotypes in vivo and distinguish between those that appear similar by a basic phenotypic scoring system. Here we describe a new analytical approach centered on live-embryo time-lapse microscopy and careful characterization of nuclear position 15 combined with new tools to provide the first direct evidence that some factors necessary for nuclear movement are required to apply force to nuclei whereas other factors are necessary for the utilization of that force to reach a specific position rather than to move.

Disruption of bocksbeutel and klarsicht have distinct effects on myonuclear positioning compared to ensconsin in the Drosophila embryo
As a first approach, we have investigated the contributions of bocksbeutel (Drosophila emerin), klarsicht (Drosophila KASH-protein), and ensconsin (Drosophila MAP7). Each gene 5 was zygotically removed in Drosophila embryos with the respective bocks DP01391 null (Collins & Mandigo et al., 2017), klar 1 null (Welte et al., 1998), or ens swo nonsense mutation (Metzger et al., 2012) alleles. Fixed images of Drosophila embryos showed that in controls, nuclei were in two clusters positioned at either end of the lateral transverse (LT) muscle whereas in bocks DP01391 and klar 1 embryos, most of the nuclei were clustered together in a single group near the ventral 5 end of the muscle (Fig. S1a), as we showed previously (Collins & Mandigo et al., 2017).
Qualitatively, this clustering phenotype was similar to nuclear positioning defects observed in ens swo embryos in which nuclei also failed to separate into distinct groups (Metzger et al., 2012).
To quantitatively evaluate myonuclear position, the distance of each nuclear cluster from the dorsal and ventral muscle poles was measured. Since the LT muscles in all three mutants were 10 significantly shorter (Fig. S2a, statistics summarized in Table S1), we measured the raw distance ( Fig. S2) and the distance as percent of muscle length (Fig. S1). Compared to controls, nuclei in bocks DP01391 and klar 1 embryos were positioned further from the dorsal muscle pole (Fig. S1b) yet closer to the ventral muscle pole (Fig. S1c), as previously described (Collins & Mandigo et al., 2017). However, nuclei in ens swo embryos were positioned significantly further 15 from both muscle poles when compared to controls or bocks DP01391 and klar 1 embryos.
Additionally, the distance between dorsal and ventral clusters was measured (Fig. S1d and Fig. S2d). The distance between clusters was significantly decreased in bocks DP01391 and klar 1 embryos because distinct clusters of nuclei formed in only a small fraction of muscles ( Fig. S2e and f). In contrast, since nuclei failed to separate in nearly all ens swo muscles, this distance was 20 approximately 0 µm. Finally, we measured the area of dorsal and ventral clusters to compare the distribution of nuclei as previously described (Collins & Mandigo et al., 2017). In controls, nuclei were evenly distributed between the two clusters, whereas more nuclei remained associated within the ventral cluster in bocks DP01391 and klar 1 embryos, thus significantly decreasing the nuclear separation ratio (Fig. S1e) Mandigo et al., 2017). Similarly, in the rare case in which nuclei separated in ens swo embryos, 6 there were more nuclei in the ventral cluster compared to the dorsal cluster. Although the total area occupied by nuclei was similar between controls, bocks DP01391 , and klar 1 , it was significantly reduced in ens swo embryos (Fig. S2i). However, the number of nuclei was the same between controls and ens swo embryos, indicating that fusion is not affected ( Fig. S3a and b). Additionally, the total volume occupied by nuclei is the same in both genotypes ( Fig. S3c and Movies S1 and 5 S2). Thus, the reduced area is due to nuclei occupying a greater depth in the ens swo embryos.
Based on these measurements, the most dominant phenotype observed in control embryos was nuclei that separated into two distinct groups of equal size. In bocks DP01391 and klar 1 embryos, nuclei either remained as a single cluster positioned near the ventral end of the muscle (Fig. S1f and g, "clustered" and "spread") or in two clusters in which the dorsal group 10 was significantly smaller than the ventral group ( Fig. S1f and g, "separated: unequal distribution"). Finally, the most dominant phenotype observed in ens swo embryos was a single cluster positioned near the center of the muscle ( Fig. S1f and h, "swoosh"). In total, these data indicate that while bocksbeutel, klarsicht, and ensconsin are all required for proper nuclear movement, the disruption of ens causes a distinct type of nuclear positioning defect compared 15 to the disruption of bocks and klar and suggest that these genes may regulate distinct aspects of nuclear movement.
Ensconsin is necessary for nuclear movement whereas bocksbeutel and klarsicht are necessary to separate nuclei 20 To investigate these phenotypes further, the position of nuclear clusters within the LT muscles was tracked over the course of 2 hours. In control muscles, once all nuclei separated into two distinct clusters, these clusters migrated toward opposite muscle ends, steadily increasing the distance between themselves ( Fig. 1a and Movie S3, left panel). However, 100% of all nuclei observed in ens swo muscles failed to separate over the time course (Fig. 1a, yellow 25 7 brackets and Movie S6, left panel), significantly reducing the separation speed to 0 μm/hr ( Fig.   1b and c). Similarly, nuclei that remained associated together in bocks DP01391 and klar 1 muscles also failed to separate (Fig. 1b blue data points and Movies S4 and S5, left panels). However, this non-separation phenotype was only observed in approximately 50% of muscles (Fig. 1c). In the other 50% of muscles, a single nucleus separated and migrated towards the dorsal end of 5 the muscle (Fig. 1a, yellow arrows), at a rate slightly faster than control nuclei (Fig. 1b, gray data points). Furthermore, the morphology of the single clusters was different in bocks DP01391 and klar 1 compared to ens swo . In ens swo clustered nuclei were spherical, whereas nuclear clusters in bocks DP01391 and klar 1 embryos were significantly elongated (Fig. 1d).
The trajectory of individual myonuclei within each cluster was then tracked over the 2- 10 hour time course (Fig. 1e). The total displacement of nuclei in bocks DP01391 and klar 1 embryos was similar to controls, even in ventral cluster where more nuclei were present ( Fig. 1f and Movies S3-5, right panels). Although the displacement was similar to controls, all of the nuclei within the cluster moved ventrally. However, nuclei that did stochastically separate from the ventral cluster did migrate dorsally suggesting that the interactions between nuclei within a 15 cluster is restricting the movement toward the ventral end of the muscle. Conversely, the displacement of nuclei in ens swo embryos was significantly decreased, as nuclei rotated within in the cluster but did not translocate ( Fig. 1f and Movies S6, right panel). Together these data suggest that in ens swo mutants, the ability of the cell to exert force on nuclei is reduced.
However, the movement and subsequent displacement of the nuclei in the klar 1 and bocks DP01391 20 suggests that force production is normal and that instead nuclei are being actively maintained in a single cluster.
Laser ablation of myonuclei demonstrates that the application of force onto nuclei is ensconsin-dependent 25 8 The fact that the nuclei were elongated in bocks DP01391 and klar 1 mutants compared to controls suggested that they may be under tension. To test this hypothesis, we used 2-photon laser ablation to remove individual nuclei and measure the response of the neighboring nuclei within the syncytium (Fig. S4a). When a nucleus was ablated in controls (1 s, yellow circle and Movie S7), the remaining nuclei within the cluster moved away from the ablation site, toward the 5 center of the muscle fiber (Fig. S4d, 2 -5 s). Nuclei in the opposite cluster also moved towards the muscle center. However, the nuclei in the neighboring LT muscles did not respond to the ablation. Furthermore, ablation did not affect the health of the muscle or the animal. Imaging of the transmitted light demonstrated that there was no gross damage to the embryo. Furthermore, three hours after ablation, nuclei returned to their proper position adjacent to the muscle end. 10 Similar movements of nuclei have been seen to occur due to muscle contractions, but in these cases, muscles detached from the tendon and formed a spheroid from which neither the muscle morphology or the nuclear position recovered (Auld et al., 2018b). Thus, the return of nuclei to the end of the muscle is consistent with the nuclei moving and not movement of the muscle ends due to contractions. Finally, ablation did not affect viability as embryos were able 15 to developmentally progress to stage 17, initiate muscle contraction and hatching (Fig. S4e), and crawl out of the field of view.
We then ablated nuclei in muscles of animals where nuclei had failed to separate into distinct clusters (Fig. 2a). When compared to controls, the area of the ventral clusters in bocks DP01391 (Movie S8) and klar 1 (Movie S9) embryos before ablation was significantly larger 20 ( Fig. 2b, before). After ablation, the remaining nuclei moved away from the ablation site and showed a 43% reduction in size in both genotypes (Fig. 2b, b' after). The dramatic decrease in size suggests that the stretching of nuclei, in addition to the greater number of nuclei present, contributed to the difference in the size of the clusters. In contrast, nuclei in ens swo embryos (Movie S10) moved only slightly after ablation (Fig. 2a) and their size was reduced by only 10%, 25 9 a value consistent with the removal of 1 out of 6-7 nuclei (Fig. 2b, b'). In addition, after ablation, clusters in bocks DP01391 and klar 1 embryos traveled a greater distance compared to controls while clusters in ens swo embryos traveled a shorter distance ( Fig. 2c and c'). Similarly, the clusters in bocks DP01391 and klar 1 had a greater initial velocity compared to controls whereas nuclei in ens swo embryos had a reduced initial velocity ( Fig. 2d and d'). Together, these data 5 demonstrate that nuclei in bocks DP01391 and klar 1 embryos are under more tension than nuclei in controls, while nuclei in ens swo embryos are under less tension. This is consistent with the hypothesis that ensconsin is necessary for the application of force to nuclei but that klarsicht and bocksbeutel are necessary for the directed movement of nuclei in response to that force. 10 Loss of bocksbeutel and klarsicht, and ensconsin are required for the organization of microtubules in Drosophila larval skeletal muscle Since myonuclei are physically linked to the microtubule cytoskeleton (Tassin et al., 1985;Espigat-Georger et al., 2016), ensconsin is a microtubule binding protein (Bulinski and Bossler, 1994;Gallaud et al., 2014), and nuclear envelope proteins have been demonstrated to 15 impact microtubule organization (Hale et al., 2008;Bugnard et al., 2005;Starr and Fridolfsson, 2010;Gimpel et al., 2017), we hypothesized that the differences in nuclear behaviors may be linked to variations in microtubule organization. For this analysis we used larvae in which the muscles are 100X larger and therefore provide greater resolution of microtubule organization.
Additionally, we used the ventral longitudinal muscle 3 (VL3) of stage L3 larvae (Fig. 3a), which 20 are a large, flat, rectangular muscle group that is at the top of a dissected larva. We focused on two distinct regions of microtubules that are uniquely organized. The first region pertained to areas of the muscle, distant from nuclei, where microtubules intersect to form a lattice (Fig. 3a, yellow box, and 3b) while the second region was adjacent to nuclei and consisted of microtubules that emanate directly from the nuclei (Fig. 3a, cyan box, and 3c). As previously 25 10 reported (Collins & Mandigo et al., 2017;Elhanany-Tamir et al., 2012), nuclei in bocks DP01391 and klar 1 larvae were mispositioned in a single row along the anterior-posterior axis of the muscle compared to nuclei in controls, which were evenly distributed in two parallel lines. Analysis of the lattice network of microtubules ( Fig. 3b) was performed using the Texture Detection Technique (TeDT), which detects the angles at which neighboring microtubules intersect (Liu 5 and Ralston, 2014). In controls, the dominant intersection angles were parallel (0°, 180°, 360°) to the anterior-posterior axis of the muscle (Fig. 3d, average in Fig. 3d'). Microtubules in bocks DP01391 , klar 1 , and ens swo larval muscles were highly disorganized, with an overall reduction in the frequency of microtubules intersecting at every 180° ( Fig. 3d').
To evaluate the organization of microtubules that extend off of nuclei, we counted the 10 percentage of nuclei that have a dense ring of microtubules on the nuclear periphery ( Fig. 3f) and measured the proportion of microtubules on the dorsal-ventral axis of the muscle versus the anterior-posterior axis (Fig. 3e). In controls, all nuclei had a ring of microtubules and the distribution ratio was close to 1.0, indicating that microtubules are uniformly emanating from nuclei. Although 85% of bocks DP01391 and 80% of klar 1 nuclei had a ring of microtubules (Fig. 3f), 15 the distribution ratio was reduced to 0.535 and 0.572 in bocks DP01391 and klar 1 larvae respectively ( Fig. 3e), indicating that more microtubules are extending along the dorsal-ventral axis compared to the anterior-posterior axis. However, only 20% of nuclei in ens swo mutants had rings (Fig. 3f) and there was a wide distribution in the proportion of microtubules on the dorsalventral and anterior-posterior axes compared to both controls, bocks DP01391 , and klar 1 mutants 20 (Fig. 3e). Together, these data indicate that although bocksbeutel, klarsicht, and ensconsin are necessary to maintain the link between myonuclei and microtubules, the disruption of bocks or klar results in the reorganization of microtubules around mispositioned nuclei whereas the 11 disruption of ens completely disrupts the general organization of microtubules throughout the muscle.
Our finding that microtubule organization is dependent on ensconsin differs from previous studies that suggested that the function of ensconsin was only to activate Kinesin (Barlan et al., 2013). To determine whether the disruption in microtubule organization was a 5 consequence of mispositioned nuclei or a contributor to nuclear movement, we examined the behavior of EB1 during embryonic muscle development when nuclei are actively moving. EB1 comets were tracked for 1 minute in the LT muscles (Movies S11 and S12) and the dorsal oblique (DO) muscles (Movies S13 and S14), a set of broad, flat muscles that are more amenable to fast, live-embryo imaging (Fig. 4a). The location from which EB1 emerged, their 10 direction of travel, and their speed was the same in controls and ens swo embryos in both muscle types ( Fig. 4b and c). However, the number of EB1 comets was significantly decreased in both LT and DO muscles of ens swo embryos (

20
All together, these data demonstrate that nuclear movement in a muscle syncytium requires both the transmission of force from the cytoskeleton to the nucleus and the separation of nuclei from their neighbors (Fig. 5). Disruption of these two separate processes produces superficially similar nuclear positioning phenotypes, but careful analysis of the precise position, shape, and movement of nuclei clearly indicates that there are distinct molecular underpinnings. 25 Consistent with this, we found that loss of ensconsin contributes to the application of force to nuclei by regulating the number of growing microtubules. Surprisingly, force was applied to nuclei in the absence of the KASH-domain protein klarsicht or the emerin homolog bocksbeutel.
Consequently, nuclei moved a similar total distance to those nuclei in control embryos.
However, nuclei remained attached rather than separating and therefore were all moved toward 5 the ventral end of the muscle. Interestingly in bocks DP01391 and klar 1 mutants, nuclei did rarely separate from the single cluster and move as individuals to the dorsal end of the muscle. This observation is consistent with the phenotype being based in aberrant associations between nuclei and not a disruption of directional cues. Finally, we use laser ablation of individual nuclei to demonstrate that nuclei in bocks DP01391 and klar 1 mutants are under increased tension 10 compared to controls whereas those in ens swo mutants are under decreased tension compared to controls to confirm that force is applied to nuclei in bocks DP01391 and klar 1 mutants but not in ens swo mutants. More broadly, these data present the first direct evidence that regulation of interactions between nuclei is a critical determinant of nuclear movement and that nucleusnucleus interactions are LINC complex-dependent. Thus, these data raise the possibility that 15 aligned nuclei in the center of a developing or regenerating muscle are physically linked and that this linkage is critical for nuclear functions.
The molecular mechanisms by which klarsicht and bocksbeutel regulate separation of nuclei from their neighbors and the molecular mechanisms by which ensconsin regulates the number of growing microtubules necessitate further investigation. However, we hypothesize that 20 ensconsin may contribute, either directly or indirectly, to microtubule nucleation and anchoring at the nuclear envelope. Recent work found that Bsg25D, the Drosophila homolog of Ninein, interacts with ensconsin and that Bsg25D contributed to ensconsin-dependent nuclear positioning (Rosen et al., 2019). Together with our data showing a reduction in the number of growing microtubules, we hypothesize that perhaps Bsg25D is recruiting ensconsin to 25 13 participate in microtubule nucleation. Alternatively, both Bsg25D and ensconsin may anchor microtubules to the nuclear envelope. Release of microtubule minus ends from the nuclear envelope may potentiate microtubule instability and the reduction in growing microtubules.
Indeed, Ninein does contribute to both nucleation and anchoring of microtubules to the centrosome (Delgehyr et al., 2005), and the loss of either function is consistent with the data 5 here and previously published (Rosen et al., 2019).
The molecular mechanism by which bocksbeutel and klarsicht regulate nuclear position is harder to predict. The simplest explanation might be that they are required to recruit microtubule motors as has been seen in other systems (Starr et al., 2001;Wilson and Holzbaur, 2012;Cadot et al., 2012). However, the phenotype seen here is distinct from the phenotypes 10 observed in animals null for either cytoplasmic dynein or kinesin (Folker et al., 2014).
Alternatively, work in C. elegans found that loss of nucleus anchoring resulted in a similar clustering of nuclei (Starr et al., 2001). But all of the data we present is from developmental stages that require active movement of nuclei rather than anchoring. When combined with our finding that the clusters of nuclei still move in these genotypes, the simplest explanation is that 15 these factors are required for nuclei to separate from one another. Because it is the loss of bocks or klar that results in the phenotype suggests that either the recruitment of a separation factor or a disruption in cytoskeletal organization is preventing the separation of nuclei. We speculate that this is based on variations in microtubule organization, consistent with our finding that microtubules are asymmetrically organized around nuclei in animals with mutations in either 20 gene. Furthermore, it is likely that the nuclei that emanate from adjacent nuclei can interact with each other and with other nuclei. Thus, the ablation of individual nuclei will ablate the associated 14 microtubule network. Thus, if the molecular glue is either the microtubules directly or indirectly, the data would be similar.
Altogether these data demonstrate that seemingly similar phenotypes are mechanically distinct and provide an approach along with some of the tools necessary to push beyond this basic understanding toward a molecular comprehension of how the movement of many nuclei is 5 coordinated within a single cytoplasm.

Drosophila genetics
All stocks were grown under standard conditions at 25°C. Stocks used were apRed 10 (Richardson et al., 2007), bocks DP01391 (Bloomington Drosophila Stock Center, 21846), klar 1 (Bloomington Drosophila Stock Center, 3256), ens swo (Metzger et al., 2012), and UAS-EB1.eYFP (Rogers et al., 2008). Mutants were balanced and identified using TM6b, DGY. The UAS-EB1.eYFP construct was specifically expressed in the mesoderm using the twist-GAL4, apRed driver. Flies carrying apRed express a nuclear localization signal (NLS) fused to the 15 fluorescent protein DsRed downstream of the apterous mesodermal enhancer. This results in the specific labeling of the myonuclei within the lateral transverse (LT) muscles of the Drosophila embryo (Richardson et al., 2007). Thus, only nuclei within the LT muscles are labeled using this reporter. The twist-GAL4, apRed Drosophila line was made by recombining the apRed promoter and the specific GAL4 driver, with both elements on the second 20 chromosome.

Immunohistochemistry
Embryos were collected at 25°C and washed in 50% bleach to remove the outer chorion membrane, washed with water, and then fixed in 50% formalin (Sigma, Product # HT501128)

Analysis of myonuclear position in Drosophila embryos
Embryos at stage 16 were selected to be imaged based on overall embryo shape, the 20 intensity of the apRed and tropomyosin signals, gut morphology, and the morphology of the trachea as previously described (Collins & Mandigo et al., 2017;Auld et al., 2018;Folker et al., 2012). Confocal z-stacks of fixed embryos were acquired on a Zeiss 700 LSM using a Plan-APOCHROMAT 40×, 1.4 NA oil objective with a 1.0× optical zoom. Images were processed as maximum intensity projections and oriented such that top is dorsal, bottom is ventral, left is 25 anterior, and right is posterior. Measurements were made using the Segmented Line tool in Fiji 16 software (Schindelin et al., 2012). Muscle length measurements were taken starting from the dorsal tip and following through the center of each LT muscle, down to the ventral tip. Dorsal and ventral end distances were taken from each LT muscle by measuring the distance between the closest group of nuclei to the dorsal or ventral muscle pole, respectively. Internuclear distances were taken by measuring the shortest distance in between the dorsal and ventral 5 clusters of nuclei within each LT muscle. Internuclear distances were also plotted according to relative frequency. All three measurements are reported as distances normalized to the muscle length ( Fig. S1) and as raw values (Fig. S2). All four LT muscles were measured in four hemisegments from each embryo. Statistical analysis was performed with Prism 4.0 (GraphPad). 10

Analysis of myonuclear cluster area in Drosophila embryos
Area of nuclear clusters were measured in fixed stage 16 embryos as previously described (Collins & Mandigo et al., 2017). In brief, the area of each cluster of nuclei near either the dorsal or ventral muscle pole was measured in Fiji (Schindelin et al., 2012). Total area of nuclear 15 clusters in each LT muscle was calculated by adding the dorsal and ventral areas. The nuclear separation ratio was calculated by dividing the area of the dorsal cluster by the area of the ventral cluster. Nuclear clusters from all four LT muscles were measured in four hemisegments from each embryo. Statistical analysis was performed with Prism 4.0 (GraphPad).
For qualitative nuclear phenotype analysis, embryos were scored on how nuclei were 20 positioned within the first three LT muscles of each hemisegment. LT 4 was excluded for this analysis due to its variable muscle morphology. Nuclear phenotypes were categorized as either "separated; equal distribution" (nuclei properly segregated into two distinct, even clusters with a nuclear separation ratio ≥ 0.85 and ≤ 1.15), "separated; unequal distribution" (nuclei that segregated into two disproportionate clusters with a nuclear separation ratio < 0.85 or > 1.15), 25 "central" (a nucleus that is not associated with either the dorsal or ventral group located in the 17 middle of the myofiber), "clustered" (nuclei remained in a single cluster toward the ventral end of the myofiber), "spread" (nuclei are distributed through the myofiber with no distinct dorsal or ventral clusters) or "swoosh" (nuclei remained in a single cluster within the middle of the myofiber). Linescans of DsRed intensity were performed on 10 LT muscles for each nuclear phenotype and averaged to determine the typical distribution of nuclei in bocks DP01391 and ens swo 5 genotypes compared to controls.

Volumetric imaging and analysis of nuclear clusters
Fixed stage 16 embryos were imaged on a Zeiss LSM 880 with Airyscan (super resolution acquisition, 2× Nyquist sampling) using a Plan-APOCHROMAT 40×, 1.3 NA oil objective at a 10 1.0× optical zoom and 0.15 µm step size interval through the entire depth of the muscle. Post processing of Airyscan images was completed in ZEN Blue 2016 software. Quantitative volumetric analysis was performed in Imaris version 9.2.1 (Bitplane AG). Images were first processed as maximum intensity projections of confocal z-stacks and oriented such that top is dorsal, bottom is ventral, left is anterior, and right is posterior. A volumetric rendering of each 15 nuclear cluster was created using the Surface Visualization tool of the DsRed channel. Volume measurements were automatically computed from the Surface renderings by Imaris. Statistical analysis was performed with Prism 4.0 (GraphPad). 20 Embryos for live-imaging were prepared as previously described (Collins & Mandigo et al., 2017;Auld et al., 2018). In brief, embryos were collected at 25°C, washed in 50% bleach to remove the outer membrane, washed with water, and mounted with halocarbon oil (Sigma, Product # H8898). For time-lapse imaging of nuclear movement, stage 15 embryos were selected for imaging based on gut morphology, the position of nuclei, and the intensity of the 25 apRed signal as previously described (Collins & Mandigo et al., 2017;Auld et al., 2018;Folker 18 et al., 2012) with the following modifications. Time-lapse images were acquired on a Zeiss 700 LSM using a Plan-APOCHROMAT 40×, 1.4 NA oil objective with a 1.0× optical zoom at an acquisition rate of 1 min/stack for 2 hours. Movies were processed in Fiji (Schindelin et al., 2012) as maximum intensity projections of confocal z-stacks and corrected for drift using the Correct 3D drift plugin. To calculate the separation speed of nuclei, the Line tool was used to 5 measure the distance between dorsal and ventral nuclear clusters at time 0 h and again at time 2 h. Separation speeds were also plotted according to relative frequency. The aspect ratio of ventral clusters was measured at time 0 h using the Shape Descriptors plugin, which calculates aspect ratio of an ellipse by dividing the major axis of the ellipse by its minor axis. An aspect ratio value closer to 1 indicates a more spherical cluster. Tracks following the movement of 10 individual nuclei within clusters were generated using the Manual Tracking plugin. The displacement of each nucleus was calculated as the difference between the final and initial position. Statistical analysis was performed with Prism 4.0 (GraphPad).

Live-embryo imaging and analysis
To assess for potential fusion defects, the number of nuclei in the LT muscles was counted from live stage 17 embryos when nuclei have separated and maximized their distance from their 15 neighbors. Nuclei within the LT muscles were identified by expression of DsRed. The number of nuclei were counted from all 4 LT muscles within a single hemisegment, with a total of 4 hemisegments analyzed for each embryo. 20 Embryos were collected at 25°C and were washed in 50% bleach to remove the outer membrane, washed with water, and mounted with halocarbon oil (Sigma, Product # H8898).

2-photon ablation of myonuclei
Stage 16 embryos were selected for ablation based on gut morphology, the position of nuclei, and the intensity of the apRed signal as previously described (Folker et al., 2012;Collins & Mandigo et al., 2017;Auld et al., 2018). Time-lapse images of embryos before, during, and after 25 ablation were acquired on a Zeiss 710 LSM using a Plan-APOCHROMAT 40×, 1.1 NA water 19 objective with a 1.0× optical zoom at an acquisition rate of 1 s/frame for 30 s. Ablation was performed using the Coherent Chameleon Ultra II femtosecond pulsed-IR laser at 860 nm with 15-17% laser power. As shown in Supplemental Figure 4, a nucleus was selected for ablation by drawing a region of interest (ROI) in ZEN Black 2012 software. For each ablation time-lapse, the first frame (time = 0 s) was taken before the ablation event. The next frame (time = 1 s), 5 shows the ablation of the targeted nucleus, followed by the subsequent response of the remaining nuclei present. Since no muscle marker is present, imaging with transmitted light was used to ensure that ablation did not destroy the surrounding tissue. An ablation was considered successful by the loss of the DsRed signal accompanied by the movement of nuclei. Nuclei that were simply photobleached were characterized by just the loss of DsRed fluorescence without 10 any subsequent response from the embryo (Fig. S4b). A failed ablation attempt that resulted in boiling of the embryo was identified by a hole burned through the membrane (Fig. S4c, arrowhead), as seen through the transmitted light channel.
Movies were processed in Fiji (Schindelin et al., 2012) as single confocal slices and oriented such that top is dorsal, bottom is ventral, left is anterior, and right is posterior. The area 15 of clusters in which a nucleus was ablated was measured before and after the ablation event.
The area of nuclear clusters before and after ablation were plotted as a percentage change. The displacement and velocity of nuclear clusters were measured using the centroid measurement, which calculates the center point of a cluster based on the average x and y coordinates of all pixels in the cluster. The total displacement of each cluster was calculated as the cumulative 20 distance traveled over the 30 s after ablation. The initial velocity was defined as the speed a cluster traveled the first second after ablation. Statistical analysis was performed with Prism 4.0 (GraphPad).

20
Confocal z-stacks of dissected stage L3 larvae were acquired on a Zeiss 700 LSM using a Plan-APOCHROMAT 40×, 1.4 NA oil objective lens at a 0.5× optical zoom for whole muscle images and at a 2.0× optical zoom for regions around myonuclei. Images were processed as maximum intensity projections and oriented such that top is dorsal, bottom is ventral, left is anterior, and right is posterior. Microtubule organization was assessed in two distinct regions of 5 interest within the ventral longitudinal muscle 3 (VL3). The first region consists of microtubules that intersect at regions between nuclei to form a lattice. For these regions, the Texture Detection Technique (TeDT) was used (Liu and Ralston, 2014). TeDT is a robust tool that can Images of nuclei were blindly scored for the presence or absence of a microtubule ring. A nucleus was considered to have a microtubule ring based on the contiguous presence of ɑtubulin intensity around the perimeter of the nucleus. Statistical analysis was performed with Prism 4.0 (GraphPad). 5

Analysis of microtubule dynamics in Drosophila embryos
Embryos for live imaging of EB1 comets were collected and prepared similarly. software. EB1 comets were imaged within the LT muscles as well as the dorsal oblique (DO) 15 muscles, which are a flatter muscle group, ideal for imaging quick dynamics. Movies were processed as single confocal slices in Fiji (Schindelin et al., 2012). Time-lapse images taken in the LT muscles were oriented such that top is dorsal, bottom is ventral, left is anterior, and right is posterior. Time-lapse images taken in the DO muscles were oriented such that top is posterior, bottom is anterior, left is dorsal, and right is ventral. Trajectories of EB1 comets were 20 made from time-lapse images using the Temporal-Color Code plugin, which sums up the first 15 consecutive frames (1 s each), and then overlays the resulting image to a blue-green-red color sequence, with each color representing a total of 5 seconds. All quantifications of EB1 dynamics was performed on temporal overlays by hand. Only comets that were visible for the full 15 seconds were used in this analysis. The starting position of each comet was categorized within 25 the LT muscles as either starting within the dorsal pole region, ventral pole region, or between 22 nuclei. Similarly, the starting position of each comet was categorized within the DO muscles as either starting within the anterior pole region, posterior pole region, or between nuclei. The direction of EB1 comets was also determined as either traveling dorsally/posteriorly or ventrally/anteriorly and whether the comets move toward or away from the nearest myotendinous junction. The length of EB1 trajectories over the 15 s timeframe was measured to 5 calculate EB1 comet velocity over the 1 min time-lapse. The number of EB1 comets was counted and normalized to the muscle area. Statistical analysis was performed with Prism 4.0 (GraphPad).     Error bars indicate the s.d. from 20 nuclei for each genotype from ≥10 VL3 muscles. One-way ANOVA with Tukey HSD post hoc test was used to assess the statistical significance of 15 differences in measurements between all experimental groups. (f) The frequency in which microtubule rings were observed around nuclei in each of the indicated genotypes. A total of 20 nuclei were analyzed for each genotype from ≥10 VL3 muscles.        showing the ablation of a single nucleus within the LT muscles of a stage 16 control embryo.
The first frame shows all the nuclei before the ablation event (0 s). The next frame (1 s) shows 5 the ablation of a single nucleus (yellow circle), followed by the subsequent response of the remaining nuclei present within the cluster after the ablation event (white arrows  Movie S1. Volumetric imaging of myonuclei in the lateral transverse muscle of a control Drosophila embryo. Movie of a three-dimensional volumetric rendering of the dorsal and ventral nuclear clusters within a single LT muscle from a stage 16 (16 hours AEL) control embryo. Muscles in magenta, myonuclei in green. Scale bar, 5 µm. The LT muscle is rotated 360° along the x-axis and 360° 5 along the y-axis. Time-lapse acquisition of the dorsal oblique muscles in a stage 16 (16 hours AEL) ens swo mutant embryo expressing EB1.eYFP. Time course, 60 s. Scale bar, 5 μm.