Aplip1, the Drosophila homolog of JIP1, regulates myonuclear positioning and muscle stability

Actively positioning organelles has been demonstrated to be an important process in the development of naturally polarized cells such as epithelial cells, cell migration and neuronal outgrowth among other cellular functions (van Bergeijk et al., 2016). One of the more dramatic examples of cellular organization is the precise positioning of the many nuclei within the skeletal muscle syncytium. Beyond the striking appearance of evenly distributed nuclei throughout the periphery of a large cell, the mechanism of myonuclear positioning has gained interest in recent years because mispositioned nuclei in muscle correlate with a variety of muscle diseases (Iyer et al., 2017; Jeannet et al., 2004; Meinke et al., 2014). Although this correlation has been noted for decades (Spiro et al., 1966), the mechanisms and functions of myonuclear movement in muscle have only recently begun to emerge.

During muscle development, myonuclei undergo a complex set of movements to maximize the internuclear distance at the periphery of the muscle cell (Bruusgaard et al., 2006; Cadot et al., 2012; Wilson and Holzbaur, 2012, 2015). Although experiments with mammalian cell culture systems have shed light on some of the mechanisms that govern myonuclear positioning, they do not recapitulate the constraints of development in an organism. In vivo analysis of myonuclear movement during muscle development in mammals is difficult because functional muscles are a complex set of bundled bundles of muscle cells. To overcome this limitation, we use embryonic muscle development in Drosophila melanogaster as a model system. In addition to being a powerful genetic system, Drosophila is a powerful cell biological system in which each individual muscle cell is a fully functional muscle. This simplicity makes Drosophila muscle development optically tractable. Despite the differences in the organization of the muscle cells relative to one another, the structure of the individual muscle cells is highly conserved between Drosophila and mammals (Bruusgaard et al., 2006; Metzger et al., 2012; Rai et al., 2014; Weitkunat et al., 2014).

Nuclear movement in Drosophila is a multi-step process that can be divided into at least three distinct phases. After fusion, nuclei are clustered off-center toward the ventral end of the muscle [10:20–11:20 h after egg laying (AEL)]. The first movement is the splitting of this large cluster of nuclei into two separate groups of nuclei. Second, the two separate groups move directionally, with one moving toward the ventral end of the muscle and the other toward the dorsal end of the muscle (11:20–16 h AEL). Third, after the distance between the two groups of nuclei has reached its maximum, the nuclei move back into the center and evenly distribute throughout the muscle (16–20 h AEL) (Folker et al., 2012; Metzger et al., 2012).

Previous work in Drosophila has demonstrated that myonuclear movement is dependent on two spatially segregated Kinesin and Dynein pathways (Folker et al., 2012, 2014; Schulman et al., 2014). One pathway, termed the proximal pathway, involves Kinesin and Dynein exerting force directly on the nucleus. Kinesin pulls the front of the nucleus to create a leading edge while Dynein acts on the lagging edge of the nucleus. This polarization of Kinesin and Dynein activity drives dynamic changes in the shape of the nucleus as it moves toward the muscle end (Folker et al., 2014). The second pathway, termed the cortical pathway, involves the cytoskeleton-dependent generation of force from the muscle end. Briefly, plus-ends of microtubules that are nucleated at the myonuclear envelope interact with the cell cortex in a CLIP-190-dependent manner. Kinesin transports Dynein to the cell cortex along these microtubules where Dynein is anchored by Raps (also known as Pins). This allows Dynein to pull the minus-ends of microtubules toward the muscle pole. These two mechanisms appear to occur simultaneously. Therefore, how each pool of motors is specified and how each motor is regulated within a pathway are crucial questions to understanding the mechanisms of myonuclear movement.

Kinesin and Dynein are regulated by several mechanisms in distinct signaling pathways, including JNK signaling. Specifically, Kinesin and Dynein are coordinated by JNK-interacting proteins (JIPs). JIP-dependent motor regulation is critical for transport of cargoes in axons (Cavalli et al., 2005; Fu and Holzbaur, 2013; Horiuchi et al., 2005). Additionally, JIP3 (also known as Syd in flies and MAPK8IP3 in mammals) contributes to myonuclear postion in Drosophila via regulation of Kinesin-dependent transport of Dynein (Schulman et al., 2014). In muscle, Syd facilitates Kinesin-dependent transport of Dynein to the muscle end (Schulman et al., 2014). However, these data left many questions, including how Kinesin and Dynein are coordinated to apply force directly to the nucleus, and how Dynein activity is regulated when it is at the muscle end.

Another cytoskeleton-dependent behavior that is regulated at the end of the muscle is the development of the myotendinous junction (MTJ). This specialized cell–cell attachment is essential for the stability of the muscle and the transmission of force from the muscle cell to the skeleton. As the cell grows, it extends filopodia-like projections from the muscle pole to connect with the correct tendon cell (Weitkunat et al., 2014). Once bound, transmembrane proteins, such as β-integrins, cluster at the muscle end and make a stable attachment (Bao et al., 1993; Pines et al., 2012; Volk et al., 1990). Cytoskeleton-dependent MTJ development occurs at the same time and place that the cytoskeleton-dependent force is being generated to move nuclei. Thus, whether some molecules might be used in, and coordinate both processes, is an intriguing question.

We therefore investigated the function of Aplip1 (the Drosophila JIP1; JIP1 is also known as MAPK8IP1 in mammals) during muscle development. Aplip1 is one of two JIP proteins in Drosophila (Syd is the other) (Taru et al., 2002). Aplip1/JIP1 only has one motor-binding domain, which is regulated to shuffle JIP1 between Kinesin and Dynein binding states. This makes Aplip1 a candidate for organizing the microtubule motors to move a large organelle like the nucleus, as the coordination of both motors is important in progressively moving the nucleus in one direction (Fu and Holzbaur, 2013; Horiuchi et al., 2005, 2007).

Here, we demonstrate that Aplip1 regulates both myonuclear positioning and muscle attachment. With respect to myonuclear positioning, Aplip1 genetically interacts with Khc and Raps suggesting that it operates within the already described pathways for the regulation of myonuclear movement. However, Aplip1 does not interact with any of the known regulators of myonuclear movement to regulate muscle attachment. Thus, Aplip1 may represent a critical node that coordinates the positioning of nuclei and the structure of the myotendinous junction.

To determine where Aplip1 is localized during early muscle development in the Drosophila embryo, we expressed Aplip1–eGFP via the UAS system specifically in the mesoderm by using Twsit-Gal4 (Greig and Akam, 1993). Aplip1–eGFP localized to the cell ends near the MTJs. Embryos were stained for both Aplip–eGFP and β-PS-integrin (also known as Myospheroid). The signals were closely associated and confirmed that there was an accumulation of Aplip1 near the MTJ (Fig. 1A–C). Live embryo time-lapse microscopy of Aplip1–eGFP indicated that Aplip1 localized in puncta at the MTJ that were dynamic (Fig. 1D; Movie 1). Specifically, Aplip1–eGFP puncta extended away from the main cell body and back. These extensions suggest that, at late embryonic stages, after the MTJ has matured, the muscle continues to send dynamic projections and that Aplip1 is one protein component of these projections.

During muscle development, filopodia-like structures extend from the muscle cell and are essential to the establishment of the cell–cell attachment that is the MTJ. A rigid structure is then built and the MTJ matures to transmit force from the muscle to the skeleton and maintain muscle integrity during contraction (Weitkunat et al., 2014). Because Aplip1–eGFP localized to MTJs, we tested whether Aplip1 contributed to muscle stability. To do this we used the Aplip1^DG20707 allele that has a p-element insertion in the 3′ untranslated region (UTR) and reduced the expression of Aplip1 transcript (Fig. 2A) and counted the number of embryos that were missing lateral transverse (LT) muscles. 19.4% of control embryos were missing at least one LT muscle compared to 30% of Aplip1^DG20707/+ heterozygous embryos and 58.7% of Aplip1^DG20707 homozygous embryos (Fig. 2B,D).

Missing muscles could result from either a block in the formation of the muscle during early muscle development or compromised muscle stability that causes the muscle degradation later in development. To distinguish between these possibilities, we stained late stage 16 embryos and counted the number of visibly collapsed muscles. Collapsed muscles were evident in 30% of control embryos compared to 40% of Aplip1^DG20707/+ heterozygous embryos and 60% of Aplip1^DG20707 homozygous embryos (Fig. 2C,E). To support the fixed embryo data, we expressed apRed, a reporter that specifically labels the nuclei of the LT muscles (Richardson et al., 2007), and performed live-embryo time-lapse microscopy. In these movies, nuclei separated and moved toward the muscle end as has been previously described (Metzger et al., 2012; Movie 2). Often, in Aplip1 mutant embryos, after nuclei had moved toward the muscle end, one cluster of nuclei would move rapidly toward the other cluster of nuclei (Fig. 2F; Movie 3). This movement is faster than previously reported nuclear movements in muscle (Folker et al., 2014) suggesting that the movement of these nuclei is not merely irregular nuclear movement. Furthermore, clustered nuclei were never seen in a muscle with two appropriate attachments. This suggests that the muscles have detached and collapsed, and that nuclei did not move back into a cluster. Taken together, these data suggest that the missing muscles and the small muscles that are attached at only one end in the Aplip1^DG20707 homozygous animals result from an inability of the muscle to form a stable MTJ.

In addition to the missing muscles and the muscles that were attached to only one end, the muscles were thinner in Aplip1^DG20707 embryos. We measured the width of the LTs from the anterior edge of first LT to the posterior edge of third LT. We standardized this measure to the width of the hemisegment by dividing the width of the LTs by the width of the segment boarder. By this measure, the LT muscles in Aplip1^DG20707 homozygous mutant animals were significantly thinner (Fig. 2G,H) compared to both control and Aplip1^DG20707 heterozygous animals. Taken together, these data show that Aplip1 regulates several features of muscle morphology.

Another feature of muscle development that is regulated from the muscle end is myonuclear positioning. Cytoplasmic Dynein is localized to the muscle end in a Kinesin-dependent manner. From the end of the muscle, Dynein pulls nuclei toward the muscle end (Folker et al., 2012). Because Aplip1–eGFP is localized to the muscle end and can regulate Dynein and Kinesin during axonal trafficking (Fu and Holzbaur, 2013; Horiuchi et al., 2005; Klinedinst et al., 2013; Siebert et al., 2015), we examined whether Aplip1 regulates myonuclear position. We first determined the position of nuclei in the LT muscles of stage 16 embryos as previously described (Folker et al., 2012). For these analyses, we ignored collapsed muscles and focused only on muscles that were attached at both ends. In control embryos, myonuclei were positioned in two distinct clusters adjacent to either the dorsal end of the muscle or the ventral end of the muscle as previously described (Folker et al., 2012). In Aplip1^DG20707 homozygous embryos the myonuclei were significantly further from the dorsal muscle end compared to the distance in either the Aplip1^DG20707 heterozygous or control embryos (Fig. 3A,B). However, the distance between the ventral end of the muscle and the nearest nucleus was the same in each genotype (Fig. 3A,C). This is likely because the myonuclei are at first clustered closer to the ventral end, and therefore the nuclei that move to the ventral end have a much shorter distance to travel. We did not see a significant difference in the LT muscle length comparing Aplip1^DG20707 to controls (Fig. 3D). Thus, Aplip1 regulates myonuclear positioning in the Drosophila embryo.

Because Aplip1 function had been explored primarily in the nervous system, we next examined whether the effects on nuclear position were muscle autonomous. Fluorescence in situ hybridization demonstrated that Aplip1 is indeed expressed at low levels in the developing mesoderm (Fig. S1). Additionally, muscle specific expression of a UAS-RNAi construct that targeted Aplip1 caused an increase in the distance between either the dorsal muscle end or the ventral muscle end and the nearest nucleus without impacting muscle length (Fig. 3E–H). Taken together, these data indicate that Aplip1 functions within the muscle cell to regulate myonuclear position.

To determine the mechanism that underlies mispositioned nuclei in Aplip1^DG20707 embryos, we used time-lapse microscopy to evaluate the movement of nuclei during embryonic myogenesis. As myonuclei move toward the muscle ends they develop a leading edge and a lagging edge by a Kinesin- and Dynein-dependent mechanism that is characterized by dynamic changes in the shape of nuclei (Folker et al., 2014). The average shape of nuclei [as determined by measuring the ratio of the length to the width (the aspect ratio) as described in the Materials and Methods], was the same in Aplip1^DG027070 and control embryos (Fig. 4A,B). However, the myonuclei in Aplip1^DG027070 embryos were less dynamic than those in controls. In control embryos, as myonuclei move toward the muscle end, they continually change between an elongated and spherical shape with aspect ratios that range from ∼3 to ∼1. However, the maximal aspect ratio of myonuclei in Aplip1^DG20707 embryos is ∼2 and the minimum aspect ratio is ∼1.5 (Fig. 4A,C,D). These measurements result in a significant change in the difference between the greatest and lowest aspect ratio for each nucleus (Fig. 4E). Furthermore, nuclei exhibit fewer shape changes per hour in Aplip1^DG20707 embryos compared to controls (Fig. 4F). These data suggest that Aplip1 is needed for normal myonuclear movement during early muscle development.

To further determine the mechanism by which Aplip1 regulates myonuclear position, we evaluated genetic interactions between Aplip1 and known regulators of myonuclear movement. Because Aplip1 regulates both Kinesin and Dynein during axonal transport in neurons (Fu and Holzbaur, 2013; Horiuchi et al., 2005; Klinedinst et al., 2013; Siebert et al., 2015), we first tested whether Aplip1 interacted with either motor. To do this we carried out double heterozygous experiments using the Aplip1^DG20707 allele described above, a Kinesin heavy chain null mutant (Khc⁸) and a Dynein heavy chain null mutant (Dhc64C^4-19). The myonuclei in doubly heterozygous Khc⁸/+; +/Aplip1^DG20707 embryos were significantly further from both the dorsal (Fig. 5A,C) and ventral (Fig. 5A,D) ends of the muscle when compared to the singly heterozygous embryos, indicating a functional interaction between Khc and Aplip1 for the purpose of myonuclear positioning. However, the length of the LT muscles in the Khc⁸/+; +/Aplip1^DG20707 doubly heterozygous embryos was similar to that in the Khc⁸/+ heterozygous embryos indicating that Aplip1 does not interact with Khc to regulate muscle length (Fig. 5E). To ensure that the nuclear positioning phenotype was not a result of nuclear collapse, we counted the number of embryos with missing LTs. In +/Aplip1^DG20707 and Khc⁸/+ the percentage of embryos missing LTs was 24% and 12%, respectively. However, the Khc⁸/+; +/Aplip1^DG20707 double heterozygous embryos only had 15% of embryos missing LT muscles showing that Aplip1 does not interact with Khc to regulate muscle stability (Fig. S2A).

The same approach was used to determine whether Aplip1 also genetically interacts with Dhc64C to regulate myonuclear positioning. The position of myonuclei in Dhc64C^4-19/Aplip1^DG20707 double heterozygotes was not significantly different from that in +/Aplip1^DG20707 heterozygous embryos at either the dorsal (Fig. 5B,F) or ventral end of the muscle (Fig. 5B,G), indicating that Aplip1 and Dhc64C do not genetically interact to regulate myonuclear position. Additionally, the length of the muscles in Dhc64C^4-19/Aplip1^DG20707embryos was not significantly different from that in the +/Aplip1^DG20707 heterozygous embryos, indicating that Aplip1 and Dhc64C do not interact to regulate muscle length (Fig. 5H). Finally, we looked at muscle stability and saw that 4% of Dhc64C^4-19/+ embryos, 24% of +/Aplip1^DG20707 embryos, and 29% of Dhc64C^4-19/Aplip1^DG20707 embryos had fewer than 4 LTs. This suggests that Aplip1 and Dhc64C do not genetically interact to regulate muscle stability (Fig. S2B). These data suggest that Aplip1 interacts with Khc but not Dhc64C to position myonuclei in the Drosophila embryo, and that Aplip1 regulates muscle length and attachment by a mechanism that is independent of both Kinesin and Dynein.

The localization of Dynein to the muscle end for the purpose of myonuclear movement is Raps-dependent (Folker et al., 2012). From the muscle end, Dynein pulls on microtubules in a CLIP-190-dependent manner, to move the myonuclei toward the muscle pole (Folker et al., 2012, 2014). We used doubly heterozygous embryos to determine whether Aplip1 interacts with Raps (Fig. 6A) or CLIP-190 (Fig. 6B) for the purposes of nuclear positioning. The myonuclei were positioned further from both the dorsal end (Fig. 6C) and the ventral end (Fig. 6D) of the muscle in Raps¹⁹³/Aplip1^DG20707 compared to the distance in either +/Aplip1^DG20707 or Raps¹⁹³/+ heterozygous embryos. Furthermore, the muscles were longer in Raps¹⁹³/Aplip1^DG20707 compared to those in either +/Aplip1^DG20707 or Raps¹⁹³/+ singly heterozygous embryos (Fig. 6E). Additionally, 24% of +/Aplip1^DG20707 embryos, 14% of Raps¹⁹³/+ embryos, and 26% Raps¹⁹³/Aplip1^DG20707 had fewer than four LT muscles suggesting that Aplip1 does not interact with Raps to regulate muscle attachment (Fig. S2C). These data indicate that Aplip1 interacts with Raps to regulate both myonuclear position and muscle length. Similar evaluation of CLIP-190^KG06490/+; +/Aplip1^DG20707 doubly heterozygous embryos indicated that Aplip1 and CLIP-190 did not interact to regulate muscle length myonuclear positioning or muscle stability (Fig. 6F–H; Fig. S2D). Taken together, these data indicate that Aplip1 interacts with Raps but not CLIP-190 to regulate myonuclear position.

To further evaluate the mechanism of Aplip1-dependent myonuclear position, we examined the distribution of microtubules (Fig. 7), Dynein and Kinesin (Fig. 8) in Aplip1 mutant embryos. We used the intensity of microtubule immunofluorescence as an indicator of microtubule density in a given region and focused on the region near the muscle end because changes in microtubule organization there were previously correlated with mispositioned nuclei (Folker et al., 2012). This analysis showed an increase in microtubule abundance near the muscle end in Aplip1 mutant embryos compared to the distance in controls (Fig. 7A–C).

Similarly, mispositioned nuclei were correlated with decreased Dynein immunofluorescence at the muscle end (Folker et al., 2012) and decreased Kinesin immunofluorescence intensity near the nuclei (Folker et al., 2014). We measured the intensity of Dynein and Kinesin immunofluorescence intensity near both the nuclei and near the muscle end. The levels of Dynein were reduced both near the muscle end (Fig. 8A,C) and near the nuclei (Fig. 8B,D). Kinesin levels, however, were reduced only near the muscle end (Fig. 8E,G) and not near the nuclei (Fig. 8F,H). These data suggest that mispositioned nuclei in Aplip1 mutants are based in changes in the organization of the microtubule network and in the distribution of microtubule motors.

Pervious work suggested that two separate pools of Kinesin and Dynein work together to move myonuclei during muscle development in Drosophila. One pool of the two motors exerts force directly on the nuclei (proximal pathway) (Folker et al., 2014) and the other pool pulls on nuclei from the muscle end via microtubules (cortical pathway). Kinesin-dependent transport of Dynein to the muscle end is regulated by the JNK-interacting protein (JIP) Sunday Driver (Syd, the Drosophila JIP3 homolog) (Schulman et al., 2014). However, beyond the role of Syd, the mechanisms by which these spatially distinct pools of Kinesin and Dynein are specified and regulated are not known. Here, we show a role for the other JIP, Aplip1, in regulating myonuclear position, myonuclear dynamics and muscle integrity.

Exogenously expressed Aplip1–eGFP localized to the muscle end (Fig. 1). This localization suggested that Aplip1 may regulate Dynein-mediated pulling forces generated from the muscle end (Folker et al., 2012). In the cortical pathway, it is thought that Dynein is transported to the muscle end by Kinesin and anchored to the cell cortex by Raps. The anchored Dynein then pulls microtubule minus-ends that are attached to the myonuclear envelope toward the muscle end (Fig. S3). Aplip1 genetically interacts with Kinesin and Raps to position myonuclei but does not genetically interact with Dynein (Figs 5 and 6). One interpretation of these data is that Aplip1 specifically regulates the ability of Dynein to pull on microtubules. Under this mechanism, Aplip1 would interact with any gene that reduced the amount of Dynein at the muscle end, known functions of Khc and Raps (Folker et al., 2012). This also suggests that sufficient Dynein is localized to the muscle end in Dhc64C^4-19/+ heterozygotes, a suggestion that is consistent with the lack of phenotype in those embryos.

The genetic interaction between Aplip1 and Raps, and the strong localization of Aplip1–eGFP to the muscle end, suggest that Aplip1 functions within the cortical pathway (Fig. S3). But, Aplip1 also genetically interacted with Khc to position myonuclei (Fig. 5). Although Aplip1/JIP1 biochemically interacts with Kinesin (Horiuchi et al., 2007), there is no evidence that Kinesin exerts force on myonuclei from the muscle end. However, we see a marked reduction in the abundance of Kinesin near the muscle end in Aplip1 mutant embryos. This correlation may indicate that Kinesin contributes directly to the generation of force from the muscle end or may be an indication that a cargo that is transported in a Kinesin-dependent manner is essential.

Alternatively, Aplip1 may regulate Kinesin-dependent movement of nuclei in the proximal pathway. Although the Dynein–Aplip1 interaction may be promoted by JNK, the Aplip1–Kinesin interaction is inhibited by JNK activation (Horiuchi et al., 2005, 2007). This variation would allow Aplip1 to exist in two states in two distinct cellular locations for two distinct purposes. Specifically, Aplip1 near the muscle end may experience high JNK signaling that promotes interactions with Dynein whereas the Aplip1 near the cell center may experience reduced JNK signaling that promotes interactions with Kinesin.

Consistent with a role in regulating Kinesin activity within the proximal pathway, dynamic changes in nuclear shape are disrupted in Aplip1^DG20707 homozygous embryos. Changes in nuclear shape are driven by the proximal pathway and are independent of the cortical pathway. In the absence of Kinesin, nuclei remain spherical and do not change shape (Folker et al., 2014). Although some changes in nuclear shape are evident in Aplip1 mutants, they are reduced in speed and magnitude. This reduction in efficiency is consistent with Aplip1 functioning to regulate motor activity within the proximal pathway.

Beyond its contribution to myonuclear movement, Aplip1 regulates muscle width and muscle attachment. Furthermore, Aplip1 genetically interacts with Raps to regulate muscle length (Fig. 6E). Increased levels of Dynein at muscle ends correlates with shorter muscles (Folker et al., 2012). However, the genetic interaction between Aplip1 and Raps results in embryos with longer muscles. This suggests that the loss of Dynein anchors (Raps) combined with reduced Aplip1 activity at the muscle end may increase Dynein retrograde movement that is correlated with increased muscle growth. More generally, these data suggest that there may be competition for Dynein between the pathways that regulate muscle length and the pathways that regulate myonuclear position.

Finally, Aplip1 regulates muscle attachment through a Kinesin- and Dynein-independent mechanism. The localization of Aplip1 to cellular extensions is similar to JIP1 localization in non-myogenic cells in vitro (Yasuda et al., 1999). Furthermore, the dynamic localization of Aplip1–eGFP puncta to the MTJ that seems to extend toward the tendon cell and retract back, is consistent with known mechanisms of MTJ formation (Weitkunat et al., 2014). This suggests that despite a smooth and static appearance, Drosophila MTJs are dynamically maintained throughout embryogenesis.

It remains unclear whether the effect on muscle structure is a result of Aplip1 playing a structural role, or if the loss of Aplip1 is contributing to a change in JNK signaling. In a dystrophic mouse muscle, overexpression of JIP1 reduced JNK signaling and improved muscle structure (Kolodziejczyk et al., 2001). Perhaps in this case, the loss of Aplip1 leads to increased activation of JNK and therefore has a detrimental effect on muscle structure.

Nevertheless, we have demonstrated that Aplip1 is necessary for several aspects of myogenesis including MTJ development and myonuclear movement. Its ability to regulate myonuclear movement, but not MTJ development, occurs through known microtubule-dependent pathways (Klinedinst et al., 2013). Future experiments aimed at identification of the mechanism of Aplip1-dependent MTJ formation will be critical to understanding the function of Aplip1 in muscle development. More importantly, these experiments will inform potential crosstalk in the regulation of MTJ formation and myonuclear movement.

The following stocks were grown under standard conditions: apRed (apME-NLS::DsRed) (Richardson et al., 2007), Aplip1^DG20707 [21745, Bloomington Drosophila Stock Center (BDSC), Bloomington, IN], Dhc64C^4-19 (Gepner et al., 1996), Khc⁸ (Brendza et al., 1999), raps¹⁹³ (Parmentier et al., 2000), CLIP-190^KG06490 (14493,BDSC, Bloomington, IN); UAS-Aplip1 RNAi (26024, BDSC, Bloomington, IN) and UASp-Aplip1.EGFP (24634, BDSC, Bloomington, IN). All UAS lines were driven specifically in the mesoderm by twist-GAL4, apRed which was made as previously described (Richardson et al., 2007). For the RNAi experiment mCherry RNAi (35785, BDSC, Bloomington, IN) was driven with twist-GAL4 as a control. Mutants were balanced and identified using CyO, DGY or TM6b, DGY.

In double heterozygous experiments, controls were Aplip1^DG20707 males crossed with virgins that expressed apRed in a w* genetic background. This cross was necessary to ensure that the proper genetic backgrounds were compared for all analyses. In all other cases, the singly heterozygous embryos were generated by crossing virgins of the indicated genotype with apRed male flies in a yw background to achieve the proper genetic background. These crosses were necessary to control for differences between the w* background that most of the alleles were carried in and the yw background that Aplip1^DG20707 was carried in. For Fig. 2, 3 and 4, Aplip1^DG20707 virgin females were crossed with apRed male flies in the yw background to achieve Aplip1^DG20707 heterozygous embryos that were in the same genetic background as the Aplip1^DG20707 homozygotes.

Total RNA was extracted from 20 stage 16 embryos using TRIzol (15596026, Invitrogen). The RNA concentration was then determined by using NanoDrop2000 system (Thermo Fisher Scientific Inc.). Equivalent concentrations of RNA from control and Aplip1^DG20707 were then reverse transcribed by using the SuperScript VILO cDNA synthesis kit according to the manufacturer's instructions (11754050, Invitrogen, Carlsbad, CA). The Aplip1 primer pairs used were: Aplip1-F, 5′-CCAGCAACGTCAAGGAAATTA-3′ and Aplip1-R, 5′-TGTAGATGTCCTCGATGGGATA-3′. For a loading control, RP49 was amplified with primers RP49-F, 5′-TACAGGCCCAAGATCGTGAA-3′ and RP49-R, 5′-GGTATCGACAACAGAGTGCGTC-3′.

Embryos were washed in 50% bleach for 3 min to remove the chorion membrane and then washed thoroughly with water. Embryos were fixed in a 1:1 dilution of heptane and 10% formalin (Sigma, Product # HT501128). Fixation was completed for 20 min on an orbital shaker at 300 rpm. For microtubule analysis, embryos were fixed in 37% formaldehyde for 10 min. After fixation, embryos were devitellinized by vortexing for 1 min in a 1:1 solution of heptane and methanol. Embryos were stained with rat anti-tropomyosin (1:200, Abcam ab50567), dsRed (1:400, Clontech 632496), rabbit anti-GFP (1:400, Torrey Pines Biolabs, catalog # TP-401), anti-integrin βPS [1:50, Developmental Studies Hybridoma Bank (DSHB), CF.6G11], mouse anti-Kinesin (1:50, DSHB SUK4), mouse anti-Dynein (1:50, DSHB 2C11-2) and mouse anti-α-tubulin (1:100, Sigma T6199) antibodies. Secondary antibodies used were Alexa Fluor 488-conjugated donkey anti-rat-IgG (1:200, A21208), Alexa Fluor 555-conjugated donkey-anti-rabbit-IgG (1:200, A31572) and Alexa Fluor 647-conjugated donkey-anti-mouse-IgG (1:200, A31571) (all Life Technologies). Embryos were mounted in ProLong Gold (Life Technologies, P36930).

For Dynein, Kinesin and microtubule imaging a Zeiss LSM 880 Airyscan with a C-Plan-APOCHROMAT 40×1.3 NA objective was used. Embryos stained for Dynein and Kinesin were imaged using the ‘resolution versus sensitivity’ setting due to the low signal. For microtubule staining, the superresolution setting was used. All embryos were put through manual Airyscan processing in order to compare images. For all other images, a Zeiss LSM 700 microscope using an APOCHROMAT 40×, 1.4 NA objective was used. The zoom for each experiment is stated in the relevant Materials and Methods section.

Embryos were imaged on a Zeiss LSM 700 microscope using an APOCHROMAT 40×, 1.4 NA objective with a 1.0× optical zoom. Stage 16 embryos were identified as previously described (Folker et al., 2012). The distance from the dorsal or ventral tip of the four lateral transverse (LTs) muscles to the nearest nucleus was measured by using the line tool in ImageJ (NIH). The distances between the muscle end and the nearest nucleus were averaged from all four LT muscles from between two to four hemisegments for each data point that is presented. Nuclear position was measured in at least two independent experiments that included ≥20 embryos from each genotype. The minimum n-value required was determined by using software at https://www.stat.ubc.ca/~rollin/stats/ssize/n2.html. Statistical analysis was completed with Prism 4.0 (GraphPad) using a Students t-test with <0.05 considered significant.

Embryos were imaged with an APOCHROMAT 40×, 1.4 NA objective with a 1.0× optical zoom. LT muscle width was measured from the anterior edge of LT1 to the posterior edge of LT3, at the center of the muscles on the dorsal–ventral axis. This measure was normalized to the width of the hemisegment, which was measured as the distance between two myotendinous junctions (MTJs) using the line tool in ImageJ (NIH). LT width measurements were taken in at least two independent experiments that included ≥20 embryos from each genotype. Statistical analysis was completed with Prism 4.0 (GraphPad) using a Student's t-test with <0.05 considered significant.

For the missing LT count, each microscope slide was scanned using the Zeiss LSM 700 microscope in each respective genotype and an embryo was scored as either normal or having missing LTs. A total of ≥20 embryos were counted for each genotype. For collapsed muscle analysis images were taken with an APOCHROMAT 40×, 1.4 NA objective with a 1.0× optical zoom. An embryo was then scored as either having visibly collapsed muscles or not. For this analysis, all muscles were considered, rather than just the LT muscles. A total of ≥20 were counted for each genotype.

To measure the colocalization of Aplip1 and β-PS-integrin, Aplip1–eGFP and β-integrin were imaged in Drosophila embryos that expressed Aplip1–eGFP. Images were acquired on a Zeiss LSM700 microscope using an APOCHROMAT 40×, 1.4 NA objective and a step size of 0.5 μm. Images at the MTJ of the Dorsal (DO) muscles were acquired with a 2.2× optical zoom, whereas images of the LT muscle ends were acquired with a 3.6× optical zoom. For the DO2 muscle (Fig. 1B) colocalization analysis, a 5 μm line-scan was performed through the MTJ in both the β-integrin and Aplip1–eGFP channel. The raw intensity data was then overlaid from both channels.

Embryos were dechorionated with a 3 min incubation in 50% bleach. Embryos were then washed in water to remove all the bleach and mounted on a gas-permeable membrane with halocarbon oil (Sigma, product # H8898). Late stage 15 embryos were selected as previously described (Folker et al., 2012) and time lapse images were taken at a rate of 2 min/stack (except for Fig. 1D and Movie 1 which were taken at 10 s/stack at a zoom of 2.2×) and a step size of 1 μm for 1 h using an APOCHROMAT 40×, 1.4 NA objective with a 1.6× optical zoom on a Zeiss 700 LSM.

The aspect ratio of each nucleus was determined by measuring the length of each nucleus on the dorsal-ventral axis and dividing it by the length of each nucleus on the anterior-posterior axis by using the line tool in ImageJ (NIH). A nucleus was judged to change shape if the aspect ratio changed by ≥0.5. A total of ≥36 nuclei were examined from ≥7 movies per genotype indicated and from at least three independent experiments. Statistical analysis was carried out with Prism 4.0 (GraphPad) using a Student's t-test with <0.05 considered significant.

Stage 16 embryos, stained with either the anti-Dynein, -Kinesin or -α-tubulin antibody, were aged as previously described (Folker et al., 2012). For Dynein, Kinesin or microtubule analysis at the muscle pole, the tropomyosin signal was used to select Z-stacks to ensure all of the Dynein, Kinesin or microtubule signal was fully within the muscle. Maximum projections were then made and a box of set dimensions (2×3 µm) was drawn over the end of the muscle. The mean fluorescence of Kinesin, Dynein or microtubules was measured. The value was then normalized to the background signal (immunostaining against DsRed) within the same box. A total of ≥50 muscles were examined from at least two independent experiments.

To measure Dynein and Kinesin localization near the nucleus the apRed signal was used to crop each cluster of nuclei from the top to the bottom. Z-stacks were then converted into maximum intensity projections. The mean fluorescence of Dynein or Kinesin was then measured around each nuclear cluster and normalized to the mean apRed fluorescence. These measurements were taken using a custom MATLAB program based on the image segmentation tool (MathWorks), which can be obtained from the author upon request. Briefly, using the apRed channel the perimeter of each cluster of nuclei was established and the mean apRed fluorescence was measured from inside this area. Based on where the coordinate of points in the perimeter were located with respect to the center of the nucleus, the perimeter was expanded outward by 5 pixels. It was from this expanded area that the mean fluorescence of Kinesin or Dynein was taken. A total of ≥26 clusters of nuclei were examined from at least two independent experiments.

Dynein, Kinesin and microtubules were imaged using the Zeiss LSM 880 Airyscan with a C-Plan-APOCHROMAT 40×1.3 NA objective. Dynein images were taken at a zoom of 5×, Kinesin images were taken at a zoom of 6× and microtubule images were taken at a zoom of 7.2×. Statistical analysis was carried out with Prism 4.0 (GraphPad) using a Student's t-test, with <0.05 considered significant.

RNA fluorescence in situ hybridization (FISH) was carried out as previously described (Legendre et al., 2013). Briefly, cDNA was made as described above from stage 16 embryos. Antisense and sense (used as a negative control) probes of Aplip1 were created by using T7 polymerase (FEREP0111, Fisher) and labeled with DIG-UTP nucleotide mix (11209256910, Sigma). Embryos were fixed as described above and each probe was hybridized for 14 h. Post hybridization the probe was detected using a biotin-conjugated mouse anti-DIG (200-062-156, Jackson ImmunoResearch) and streptavidin-conjugated horseradish peroxidase (HRP; S911A, Fisher). To enable fluorescence, a cyanine 3 tyramide reagent kit was used (SAT704A001EA, Perkin Elmer).

We thank the Bret Judson and the Boston College Imaging Core for infrastructure and support. Additionally, Casey Kraft from Carl Zeiss Microscpy, LLC provided critical support in performing localization studies using the Zeiss AiryScan microscope. Drosophila stocks obtained from the Bloomington Drosophila Stock Center (NIHP400D018537) were used in this study. The integrin betaPS antibody developed by D. Brower at Harvard Medical School was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at the University of Iowa, Department of Biology, Iowa City, IA 52242.