The Hippo pathway controls myofibril assembly and muscle fiber growth by regulating sarcomeric gene expression

Skeletal muscles are composed of gigantic cells called muscle fibers, packed with force-producing myofibrils. During development the size of individual muscle fibers must dramatically enlarge to match with skeletal growth. How muscle growth is coordinated with growth of the contractile apparatus is not understood. Here, we use the large Drosophila flight muscles to mechanistically decipher how muscle fiber growth is controlled. We find that regulated activity of core members of the Hippo pathway is required to support flight muscle growth. Interestingly, we identify Dlg5 and Slmap as regulators of the STRIPAK phosphatase, which negatively regulates Hippo to enable post-mitotic muscle growth. Mechanistically, we show that the Hippo pathway controls timing and levels of sarcomeric gene expression during development and thus regulates the key components that physically mediate muscle growth. Since Dlg5, STRIPAK and the Hippo pathway are conserved a similar mechanism may contribute to muscle or cardiomyocyte growth in humans.


Introduction 38
Mammalian skeletal muscles are built from gigantic cells called muscle fibers, up to 39 several centimetres long, that mechanically link distant skeletal elements. Muscle forces 40 are produced by highly regular molecular arrays of actin, myosin and titin filaments 41 Large muscle fibers contain many parallel myofibrils, which are laterally aligned to a 46 cross-striated pattern to effectively power animal locomotion (Gautel, 2008;Schiaffino et 47 al., 2013). How muscle fibers grow to these enormous sizes and how their growth is 48 coordinated with the assembly and growth of the individual myofibrils within the muscle 49 is a challenging biological problem that is not well understood. 50 Muscle fibers are built during animal development. Initially, many small 51 myoblasts fuse to myotubes, whose long-ends then mechanically connect to tendon cells 52  Table 1). Inspection of the thoraces of these animals revealed complete 153 flight muscle atrophy in pupae at 90 h APF ( Figure 1B). Expression of a UAS-Dlg5-GFP 154 but not a UAS-GFP-Gma control construct, was able to rescue the number of muscle 155 fibers of Dlg5 knock-down (Dlg5-IR-1) flies to wild type providing further strong 156 evidence for the specificity of the knock-down phenotype ( Figure 1D). We conclude that 157 Dlg5 and Slmap are two conserved genes essential for flight muscle morphogenesis 158 during pupal stages. 159 To identify the developmental time point when Dlg5 and Slmap are required, we 160 analysed pupal stages and found that at 24 h APF all flight muscles are present after Dlg5 161 or Slmap knockdown. However, the fibers are more than 20% longer than wild type and 162 fail to compact at 32 h APF when myofibrils normally assemble ( Figure 1B,C, 163 Supplementary Table 1). Interestingly, after 32 h APF, when wild-type myofibers 164 strongly grow in length, Dlg5 and Slmap knock-down fibers undergo complete flight 165 muscle atrophy until 48 h APF ( Figure 1B). Taken together, these data demonstrate that 166 Dlg5 and Slmap play an essential role during stages of myofibril assembly and muscle 167 fiber growth. 168 169 Dlg5 interacts with Slmap, a STRIPAK complex member, in Drosophila muscle 170 To define a molecular mechanism for how Dlg5 regulates muscle morphogenesis we 171 performed GFP immunoprecipitation followed by mass-spectrometry, using the 172 functional UAS-Dlg5-GFP, which was expressed in pupal muscles with Mef2-GAL4. 173 Interestingly, we not only identified Slmap as a binding partner of Dlg5 in muscle, but 174 also Fgop2, GckIII, Striatin (Cka), and the catalytic subunit of PP2A phosphatase (Mts) 175 ( Figure 2A  In contrast to the myofiber compaction defect of yorkie knock-down muscles, 201 expression of an activated form of Yorkie (yorkie-CA), which cannot be phosphorylated 202 by Warts, results in premature muscle fiber compaction already at 24 h APF and strongly 203 hyper-compacted muscle fibers at 32 h APF ( Figure 3A,B). The increased cross-sectional 204 area is particularly obvious in cross-sections of yorkie-CA fibers (Figure 3 supplement 205 1B). Importantly, we observed the same phenotypes after knock-down of each of the two 206 kinases hippo and warts, both negative regulators of Yorkie nuclear entry ( Figure 3A,B). 207 This strongly suggests that the Hippo pathway, by regulating phosphorylation of the 208 transcriptional co-activator Yorkie, is essential for the correct developmental timing of 209 flight muscle morphogenesis: too much active Yorkie accelerates myofiber compaction, 210 while too little active Yorkie blocks it. Importantly, these GAL80ts Mef2-GAL4 Dlg5 and yorkie knock-down muscles do 240 display the same fiber compaction defect as observed with Mef2-GAL4 resulting in 241 longer but thinner fibers with grossly comparable volumes at 24 h APF (yorkie-IR is 242 slightly smaller) ( Figure 4A The wild-type flight muscle fibers grow in volume from 24 h to 48 APF (see Figure 1), 259 while yorkie or Dlg5 knock-down fibers undergo atrophy after 32 h APF. As the Yorkie 260 activity is known to suppress apoptosis in epithelial tissues (Harvey and Tapon, 2007) we 261 asked if we could rescue fiber atrophy by over-expressing the apoptosis inhibitor Diap1. 262 Indeed, over-expression of Diap1 during pupal stages in GAL80ts Mef2-GAL4 (hereafter 263 abbreviated as GAL80ts) yorkie and Dlg5 knock-down fibers substantially rescues fiber 264 atrophy, often resulting in the normal number of six muscle fibers at 48 h APF ( Figure  265 5A, compare to Figure 4C). This demonstrates that apoptosis contributes to flight muscle 266 fiber atrophy in yorkie and Dlg5 knock-down muscles. 267 Presence of muscle fibers at 48 h APF enabled us to quantitatively investigate the 268 role of the Hippo pathway during the post-mitotic muscle fiber growth. As in Figure 4, Concomitant with the myofibril assembly defect, we also found that the spacing of the 295 nuclei is defective. In control muscle fibers the nuclei are present mainly as single rows 296 located between myofibril bundles, whereas in yorkie and Dlg5 knock-down muscles 297 they form large centrally located clusters (Figure 6 supplement 1B). This indicates that at 298 32 h APF, Hippo signalling is required within the muscle fibers to trigger proper 299 myofibril assembly and nuclear positioning. 300 The myofibril defect becomes even more pronounced at 48 h APF when control 301 myofibrils have matured and sarcomeres are easily discernable ( Figure 6A), while no 302 organised sarcomeres are present in GAL80ts yorkie and Dlg5 knock-down muscles and 303 myofibril traces remain short ( Figure 6A,B, Figure 6 supplement 1A). Furthermore, cryo 304 cross-sections revealed that not only the cross-sectional area but also the total number of 305 myofibrils is strongly reduced in GAL80ts yorkie and Dlg5 knock-down muscles 306 compared to control ( Figure 6C,D, Figure 6 supplement 1C). These data demonstrate that 307 the Hippo pathway controls both the morphological quality of the myofibrils at the 308 assembly and maturation stages as well as their quantity. As myofibrils occupy most of 309 the muscle fiber space, their reduced amount likely causes the reduced muscle size in 310 Dlg5 or yorkie knock-down fibers. 311 312

Yorkie is a transcriptional co-regulator in muscle fibers 313
It was recently shown that the transcription of most sarcomere key components is tightly 314 regulated starting shortly before myofibril assembly and being strongly boosted during 315 myofibril maturation (Spletter et al., 2018). Thus, we reasoned that Yorkie activity may 316 be involved in this transcriptional regulation step to control the timing of 317 myofibrillogenesis. However, it had also been recently shown that Yorkie can regulate 318 myosin contractility directly at the cell membrane without entering into the nucleus (Xu 319 et al., 2018). As we have thus far failed to unambiguously locate Yorkie protein in 320 muscle fibers during development, we used genetic tools to address this important point. 321 To test whether Yorkie may play a role outside of the nucleus, we manipulated Yorkie 322 levels by over-expressing different Yorkie variants post-mitotically using Act88F-GAL4 323 and investigated the consequences at 24 h and 32 h APF. Over-expression of either 324 Yorkie-CA, whose import into the nucleus is uncoupled from the Hippo pathway, or 325 wild-type Yorkie, whose nuclear import is regulated by Hippo, both result in premature 326 muscle fiber compaction at 24 h APF, with seemingly normal actin filaments ( Figure 6E). 327 Strikingly, the muscle fiber hyper-compaction at 32 h APF coincides with a chaotic 328 organisation of the myofibrils, with many myofibrils not running in parallel but in various 329 directions ( Figure 6E, Figure 6 supplement 1D). This suggests that the hyper-compaction 330 phenotype upon Yorkie over-expression is likely caused by uncontrolled and premature 331 force production of the chaotically assembling myofibrils. 332 In contrast, over-expression of a membrane-anchored myristoylated form of 333 Yorkie, which has been shown to activate myosin contractility at the epithelial cell cortex   We applied the selection criteria log2FC > 1 and adjusted p-value < 0.05 to 358 identify differentially expressed genes compared to wild type (Supplementary Table 2). 359 Applying FlyEnrichr (Kuleshov et al., 2016) on the differential data sets, we found a 360 strong enrichment for muscle and, in particular, for sarcomere and myofibril Gene 361   Table 2). This strongly suggests that 375 nuclear entry of Yorkie contributes to the transcriptional induction of sarcomeric protein 376 coding genes as well as genes important to link the nuclei to the sarcomeres. This subunit UQCR-C2, which are all required to boost ATP production during muscle fiber 405 growth ( Figure 7A,B, Supplementary Table 2). Taken together, these data strongly 406 suggest that the Hippo pathway regulates the correct expression dynamics of many key 407 muscle components, most prominently mRNAs coding for core sarcomeric and 408 mitochondrial proteins to enable myofibril assembly and mitochondrial maturation. Consistent with the transcriptomics data, we found a mild reduction of Mhc and Act88F 420 protein levels in yorkie knock-down muscles at 24 h APF, which became more 421 pronounced at 32 h APF ( Figure 8A-C). This is consistent with the myofibril assembly 422 defects found in yorkie knock-down muscles at 32 h APF. The closest mammalian homolog to insect flight muscles is the mammalian heart. 470 Similar to flight muscles, the heart is a very stiff muscle using a stretch-modulated UTRs of mRNA to regulate their translation efficiencies (Webster et al., 1997). Another 507 RNA binding protein that requires the Hippo pathway for normal expression is Imp (IGF-508 II mRNA-binding protein, see Figure 7B). Imp has been shown to regulate the stability 509

Fly strains and genetics 572
Fly strains were maintained using standard conditions. Unless otherwise stated, all 573 experiments were performed at 27°C to improve RNAi efficiency. When applying 574 temperature sensitive Tub-GAL80ts, fly crosses were kept at 18°C to suppress GAL4 575 activity and the white pre-pupae (0-30 min APF) were shifted to 31°C to allow GAL4 576 activity only at pupal stages. The pupae were then raised at 31°C until the desired age.   Table 1). 592 593

Pupal dissection and flight muscle stainings 594
For 24 h, 32 h and 48 h APF pupal dissections, the pupa was stabilized on a slide by 595 sticking to double-sticky tape. The pupal case was removed with fine forceps. Using 596 insect pins 2 or 3 holes were made in the abdomen to allow penetration of the fixative 597 and pupae were fixed in 4% PFA (Paraformaldehyde) in PBST (PBS with 0.3% Triton-598 X) in black embryo glass dishes for 15 min at room temperature (RT). After one wash 599 with PBST the pupae were immobilised using insect pins in a silicone dish filled with 600 PBST and dissected similarly as described previously . 601 Using fine scissors the ventral part of each pupa was removed, then the thorax was cut To count nuclei numbers of flight muscle fibers, half thoraces were stained with 637 phalloidin (actin) and DAPI (nuclei) and imaged first using 10x objective to quantify the 638 entire length of the fiber and then with 63x oil objective to visualize details. The acquired 639 63x z-stacks contain the entire muscle depth and about half of the muscle fiber length. 640 Using Fiji's multi-point tool all the nuclei in each z-stack were counted manually, using 641 actin labelling as a landmark to visualize the borders of the fiber. Nuclei number for the 642 entire fiber was calculated using the length of the entire fiber from the 10x image. 643 644

Cryo cross-sections 645
Cryo cross-sections were performed as described previously (Spletter et al., 2018). Pupae of the different genotypes were dissected, stained and imaged on the same day in 663 parallel under identical settings (master mixes for staining reagents, identical laser and 664 scanner settings for imaging). Per hemi-thorax one or two different areas were imaged 665 using 63x oil objective, zoom factor 2. In Fiji, five z planes at comparable positions in the 666 muscles were selected for an average intensity projection of the volume into a 2D plane. 667 In the 2D plane, two or three regions of 50 µm 2 occupied by myofibrils (based on actin 668 labelling) were selected. Mean intensities of each of these regions were averaged to 669 calculate one value per hemi-thorax used for the quantification graphs.  The funders had no role in study design, data collection and analysis, decision to publish, 783 or preparation of the manuscript. 784 785

Competing interests 786
The authors declare to have no competing interests relevant to this study.