The following stocks were used: kon (kon ^F1-3 ) (Estrada et al., 2007); LanB1 ^DEF and LanB1 ^1P3 (Urbano et al., 2009); 5053-GAL4 (Bloomington Drosophila Stock Center: 2702) (Swan et al., 2004); UAS-src:GFP (Lesch et al., 2010). The CyO, twist-GAL4, UAS-2EGFP (Halfon et al., 2002), CyO, ftz-LacZ (Knoblich and Lehner, 1993), and CyO, act:GFP (Bloomington Drosophila Stock Center) balancer chromosomes, were used to identify homozygous mutants. In both Figure 1 and Figure 3, we used the kon ^F1-3 , LanB1 ^DEF (Rec8) recombinant chromosome to analyze double kon and LanB1 mutation.

Embryo antibody stainings were carried out as described previously (Carmena et al., 1998). The following primary antibodies were used: 1:5,000 rabbit anti-β-galactosidase (Cappel, 55,976); 1:5,000 rabbit anti-GFP (Invitrogen, A-6455), 1:5,000 mouse anti-GFP (Invitrogen, A-11120), and 1:400 rat anti-Tropomyosin (Babraham Bioscience Technologies, MAC141). The following secondary antibodies were used: 1:200 goat anti-rabbit-Cy2 (Jackson, 111-225-144), 1:200 goat anti-rat-Cy5 (Jackson, 112-175-143), 1:200 goat anti-mouse-488 (Life Technologies, A-1101), and 1:200 goat anti-rabbit-Cy5 (Jackson, 111-175-144). Confocal images were obtained using SP2 and Stellaris five confocal microscopes from Leica, and processed with Adobe Illustrator and ImageJ.

The identification of specific embryonic stages was performed by following previous characterization (Campos-Ortega and Hartenstein, 1997). Stage 13–14 embryos showed head involution with retracting clypeolabrum, germ band retraction, yolk sac protruding dorsally and dorsal closure of midgut and epidermis; Stage 15 embryos showed retracted clypeolabrum, completion of dorsal closure, and growth of the hindgut; Stage 16 embryos showed midgut constrictions and final pattern of somatic musculature was developed. In LanB1 mutants there is a failure in midgut constriction so that staging of these embryos was based on the other morphologies. Quantification of phenotypes in VL1 muscle morphology was performed by counting the number of each type of phenotype per embryo. Image analysis was performed from both confocal z-stacks and maximal projections using ImageJ software.

Our VL1 muscles data consisted of relative proportions of WT and several types of muscle phenotypes across genotypes. Since the relative proportion of muscles in each category was not statistically independent from each other, we first conducted an overall compositional data analysis (Aitchison, 1982) using the package CoDaPack (Comas and Thió-Henestrosa, 2011). Visual inspection of the data distribution across the tetrahedron (Figure 3A) suggested that genotypes differed markedly in the proportion of abnormal muscles, which was confirmed statistically by a Multiple Analysis of Variance (MANOVA). Confirmed these overall differences across genotypes, we then tested for statistical differences between genotypes or within genotypes across stages in the proportion of specific muscle phenotypes. We tested such comparisons fitting Generalized Linear Models with a binomial error distribution and a logit link function, where the dependent variable was number of muscles with a given phenotype vs total number of muscles scored in a given embryo.

To study the role of laminins and their interaction with kon, during the development of the MTJ, we first studied the pattern of the body wall musculature at the end of embryogenesis, stage 16 (st. 16) (Figure 1A), in both control and mutant embryos. To study kon function, we analyzed embryos for the null mutant allele kon ^F1-3 (hereafter referred as kon). In agreement with previous findings (Estrada et al., 2007; Schnorrer et al., 2007), we found that, while control embryos (kon/+) did not show any defect in the musculature (Figure 1B), some muscles from kon embryos, particularly the ventral-longitudinal ones (VL), formed myospheres, a term used to describe rounded shape muscles due to detachment from their tendon cells (Wright, 1960) (arrowhead in Figure 1C). To study laminin function, we analyzed embryos carrying a deficiency that eliminates the only Drosophila gene encoding the laminin ß-subunit, LanB1, hereafter referred as LanB1 embryos (Figure 1D). Similar to kon embryos, LanB1 mutant muscles also formed myospheres, particularly the VL ones (Urbano et al., 2009) (arrowhead in Figure 1D). Since loss of function of either kon and LanB1 showed myospheres, we studied their cooperation during MTJ development by performing genetic interaction experiments. To do this, we generated flies with a recombinant chromosome for both mutations. Double kon, LanB1 heterozygous mutants showed a wild-type muscle pattern (Figure 1E), suggesting that a single functional allele for each gene is enough to properly establish the muscle pattern. However, we found that the muscle detachment phenotype of LanB1 embryos that were also kon heterozygotes, was dramatically enhanced compared with either kon or LanB1 single mutants (Figure 1F). Moreover, complete loss of both kon and LanB1 (hereafter referred as kon, LanB1 embryos) led to a generalized presence of myospheres (Figure 1G) (quantification below). The high number of detached muscles in the mutant combination kon, LanB1 (either partial or complete), resulted in the formation of gaps in the muscle pattern (Figures 1F,G and Supplemetary Figure S1). Since LanB1 deficiency also removes the 5’ UTR of the adjacent CG72143 gene, we validated our results by recombining kon ^F1-3 with a null allele for LanB1, LanB1 ^1P3 (Urbano et al., 2009). Consistently, these recombinant mutant embryos showed the same phenotype as the kon, LanB1 embryos (Supplemetary Figure S1). Altogether, this data supports a cooperative role for LanB1 and kon in the formation of the MTJ.

The muscle detachment phenotype observed at the end of embryogenesis in LanB1 embryos (Figure 1D) (Urbano et al., 2009) might be caused by defects in either muscle guidance and/or muscle attachment to tendon cells. To distinguish between these two possibilities, and to further characterize the role of LanB1 in the formation of the MTJ, we characterized muscle migration and attachment in both control and LanB1 embryos. Previous studies have shown that as muscles migrate towards their tendon cells they extend projections towards future attachment positions (Volk, 1999; Schnorrer and Dickson, 2004). The visualization of the complete body wall musculature (Figure 1) complicates the analysis of the morphology of individual muscles as they migrate, as well as their cellular protrusions. To solve this, we decided to label just one specific muscle, VL1, using 5053-GAL4 driver (Swan et al., 2004) to express a GFP-tagged plasma membrane marker, src:GFP (Lesch et al., 2010) (Figure 2). wild type (wt) muscles (Figures 2A,B) (Schnorrer et al., 2007) first project protrusions anteriorly towards their future attachment positions at the anterior segment border (st. 13–14). The posterior end does not migrate, as it is already placed near the posterior segment border, the other attachment point. Then muscles initiate (st. 15) and stabilize (st. 16) the attachment to tendon cells (Figure 2A). In LanB1 embryos (Figure 2C), at early stages (13–14), 6% of the VL1 muscles showed misoriented projections, that is, muscles with projections oriented perpendicularly to the normal posterior to anterior direction of migration (Figure 2A). At st. 15, LanB1 loss caused a few cases of myospheroid-shape muscles (rounded, and detached muscles) (2%) and of muscles with misshaped projections (2%), muscles whose projections were properly oriented but longer than normal, and presenting a small contact surface with the attachment site (Figure 2A), which reflects a defect in the initial contact between muscle and tendon cells. Moreover, at st. 16, 5% of the muscles were myospheres, and some muscles showed misoriented or misshaped projections (2% each). Our data suggests that LanB1 regulates muscle targeting to tendon cells at early stages and stabilization of muscle attachments later in embryogenesis. The fact that LanB1 muscles send projections suggests that LanB1 is not required for the overall formation of filopodia during muscle migration, although we cannot distinguish if they are intact at this resolution.

As in LanB1 embryos, we had observed that kon embryos also showed myospheres at the end of muscle development (Figure 1) (Estrada et al., 2007; Schnorrer et al., 2007). However, the specific analysis of VL1 muscle morphology at early and late stages revealed striking differences between both genotypes (Figures 2C,D) (Estrada et al., 2007; Schnorrer et al., 2007). At st. 13-14, kon mutants showed a remarkable proportion of muscles with misshaped projections (63%), already indicating an altered ability to initiate contacts with tendon cells. However, unlike LanB1 mutants, loss of kon did not produce muscles with misoriented projections at these early stages (Figure 2D) (Estrada et al., 2007; Schnorrer et al., 2007). Later, at st. 15, kon embryos showed an increase in the number of misshaped projections (from 63 to 86%), supporting the known role of kon in mediating muscle-tendon attachment. Remarkably, kon mutants also showed some misoriented muscles at this stage (5%), likely consequence of failed attachment, as previously reported (Schnorrer et al., 2007). Finally, at st. 16, 69% of kon mutants showed misshaped projections and 21% myospheres, indicating respectively incomplete and failed muscle attachments. Together, this data supports that the myospheres observed in late kon embryos could be due to defects in muscle-tendon migration and/or attachment formation, while LanB1 additionally showed a specific role in projection orientation at the early stages of muscle migration.

To better understand the specific role of LanB1 and kon and their interaction, we first analyzed, independently of the stage, VL1 muscle morphology in different dosages of mutant alleles for LanB1 and kon. The spatial distribution of the observed muscles with different phenotypes within a polyhedron showed that there is a statistically significant difference in their distribution among genotypes. Specifically, loss of one or two copies of LanB1 in a kon background caused the displacement of the kon phenotypes towards the misoriented muscle morphology, suggesting an interaction between LanB1 and kon in muscle migration and attachment formation. This statistical analysis (Figure 3A and Supplementary Movie) (see Material and Methods for more details) allowed us to carry out the specific pairwise comparisons of the phenotypes among genotypes in different stages.

At st. 13-14, we observed that while embryos with only one functional copy of kon showed no alterations in muscle development (Figure 2B), the additional loss of one LanB1 allele (double heterozygous) (Figure 3B) showed around one third (28%) of all muscles with misoriented projections. Interestingly, this was significantly higher than what we found in either LanB1 (6%) or kon (0%) single mutant embryos (Figures 2C,D), supporting a cooperation between both kon and LanB1 in muscle targeting. Moreover, kon, LanB1 mutants (Figure 3C), presented a similar amount of misoriented projections than double heterozygote mutants (Figure 3B; 34 and 28% respectively), but the former additionally showed misshaped muscles and myospheres (likely unable to send projections). Together, this resulted in that more than the half (57%) of the muscles in kon, LanB1 embryos presented some type of muscle phenotype at early stages. Therefore, our data supports a role for LanB1 in muscle guidance and its cooperation with kon in this process.

At st. 15, comparing kon, LanB1 (Figure 3C) with kon (Figure 2D) embryos, they both showed a similar proportion of total number of muscles with any of the described phenotypes (86 and 91% respectively), while this proportion was only 3% in LanB1 mutants (Figure 2C). However, kon, LanB1 mutants showed 37% myospheroids and 23% misoriented muscles, phenotypes which were respectively non-observed or milder (5%) in kon mutants. Therefore, the early formation of myospheres and the relatively high number of misoriented muscles in kon, LanB1 embryos, suggest a cooperation between both genes to regulate the later stages of muscle migration and the attachment formation between muscle and tendon cells.

At st. 16, all the muscles were affected in kon, LanB1 mutants (Figure 3C), where 95% were myospheres and 5% misshaped projections. The remarkable difference of the myospheres/misshaped projections ratio between kon (21/70) (Figure 2D) and kon, LanB1 mutants (95/5) (Figure 3C), supports a strong cooperation between kon and LanB1 in the stabilization of the myotendinous junction.

The transmembrane proteoglycan Kon is a cell adhesion receptor that mediates the interaction between the muscle cell and the ECM (Estrada et al., 2007; Schnorrer et al., 2007; Pérez-Moreno et al., 2017). Therefore, the observed cooperation between kon and LanB1 during the formation of the MTJ might suggests a potential role for Kon in localizing LanB1 at the MTJ. As previously reported in wt condition (Urbano et al., 2009; Wolfstetter and Holz, 2012), we observed that control embryos showed LanB1 enrichment at the MTJ (arrowheads in Figure 4A and Supplementary Figure S2A). In kon embryos, despite the decreased MTJ surface of attachment of VL muscles, we still detected accumulation of LanB1 where muscles were attached (arrowheads in Figure 4B and Supplementary Figure S2B). This result indicates that Kon is not essential to recruit LanB1 at the MTJ.