The diaphragm is a domed muscle between the thorax and abdomen essential for breathing in mammals. Diaphragm development requires the coordinated development of muscle, connective tissue, and nerve, which are derived from different embryonic sources. Defects in diaphragm development cause the common and often lethal birth defect, congenital diaphragmatic hernias (CDH). HGF/MET signaling is required for diaphragm muscularization, but the source of HGF and the specific functions of this pathway in muscle progenitors and effects on phrenic nerve have not been explicitly tested. Using conditional mutagenesis in mice and pharmacological inhibition of MET, we demonstrate that the pleuroperitoneal folds (PPFs), transient embryonic structures that give rise to the connective tissue in the diaphragm, are the source of HGF critical for diaphragm muscularization. PPF-derived HGF is directly required for recruitment of MET+ muscle progenitors to the diaphragm and indirectly (via its effect on muscle development) required for phrenic nerve primary branching. In addition, HGF is continuously required for maintenance and motility of the pool of progenitors to enable full muscularization. Localization of HGF at the diaphragm’s leading edges directs dorsal and ventral expansion of muscle and regulates its overall size and shape. Surprisingly, large muscleless regions in HGF and Met mutants do not lead to hernias. While these regions are likely more susceptible to CDH, muscle loss is not sufficient to cause CDH.
diaphragmHGFMETCDHmusclefibroblastscell cultureResearch organismMousehttp://dx.doi.org/10.13039/100000002National Institutes of HealthR01HD087360KardonGabriellehttp://dx.doi.org/10.13039/100000002National Institutes of HealthF32 HD093425SeftonElizabeth Mhttp://dx.doi.org/10.13039/100000002National Institutes of HealthK99HD101682SeftonElizabeth Mhttp://dx.doi.org/10.13039/100000002National Institutes of HealthR01 HD104317KardonGabrielleWheeler FoundationKardonGabrielleThe funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.Author impact statementHGF derived from fibroblasts is critical for complete muscularization of the mammalian diaphragm, and loss-of-function experiments reveal that partial muscularization of the diaphragm is not sufficient to cause congenital diaphragmatic hernias.Introduction
The diaphragm is an essential skeletal muscle and a defining feature of mammals (Perry et al., 2010). Contraction of the diaphragm, lying at the base of the thoracic cavity, powers the inspiration phase of respiration (Campbell et al., 1970). The diaphragm also serves an important passive function as a barrier separating the thoracic from the abdominal cavity (Perry et al., 2010). Respiration by the diaphragm is carried out by the domed costal muscle, composed of a radial array of myofibers surrounded by muscle connective tissue, extending laterally from the ribs and medially to a central tendon, and innervated by the phrenic nerve (Merrell and Kardon, 2013). Diaphragm development requires coordination of multiple embryonic tissues: (1) somites are well established as the source of the diaphragm’s muscle (Babiuk et al., 2003; Bladt et al., 1995; Dietrich et al., 1999), (2) the cervical neural tube gives rise to the phrenic nerve (Allan and Greer, 1997a; Allan and Greer, 1997b), and (3) the pleuroperitoneal folds (PPFs), paired mesodermal structures located between the thoracic (pleural) and abdominal (peritoneal) cavities, form the muscle connective tissue and central tendon (Merrell et al., 2015). Integration of these three tissues into a functional diaphragm is critical, but how their development is coordinated and integrated is largely unknown.
Defects in the development of the diaphragm cause congenital diaphragmatic hernias (CDH), a common (1 in 3000 births) and costly ($250 million per year in the US) birth defect (Pober, 2007; Raval et al., 2011; Torfs et al., 1992). CDH compromises the integrity of the diaphragm by effecting muscularization, leading to an incomplete barrier between the abdominal and thoracic cavities. As a result, the liver herniates into the thorax, impeding lung development and resulting in long-term morbidity and up to 50% neonatal mortality (Colvin et al., 2005). Correct innervation of the diaphragm by the phrenic nerve is also essential, as breathing must be functional by birth and fetal breathing movements are important for normal lung development (Jansen and Chernick, 1991).
Recruitment of muscle progenitors and targeting of phrenic nerve axons to the nascent diaphragm are essential first steps for correct diaphragm development. The receptor tyrosine kinase signaling cascade initiated by the binding of the ligand hepatocyte growth factor (HGF) to its receptor MET is a promising candidate pathway for regulating these steps of diaphragm development as HGF/MET signaling has been implicated in multiple aspects of muscle and motor neuron development (reviewed by Birchmeier et al., 2003; Maina and Klein, 1999). HGF binding to MET leads to MET phosphorylation and the activation of multiple downstream pathways, including JNK, MAPK, PI3K/Akt, and FAK (Organ and Tsao, 2011). HGF/MET signaling is a critical regulator of muscle progenitors migrating from somites (Bladt et al., 1995; Dietrich et al., 1999; Maina et al., 1996). HGF is also critical for innervation. HGF acts as a chemoattractant, required for correct guidance of MET+ motor neuron axons to target muscles in the developing limb (Ebens et al., 1996; Yamamoto et al., 1997), and MET signaling is required for distinct functions in different motor neuron pools, including axon growth in the latissimus dorsi and motor neuron survival in the pectoralis minor (Lamballe et al., 2011). However, these reports do not distinguish between the effect of HGF on muscle versus nerve or rely on in vitro experiments. Thus, HGF/MET signaling has complex, tissue-specific roles in regulating the neuromuscular system.
Here we dissect the role of HGF/MET signaling in muscularization and innervation of the diaphragm. In previous studies, Met mutations have been associated with CDH (Longoni et al., 2014) and Hgf is downregulated in mutants or pharmacological treatments that induce diaphragmatic hernias in rodents (Merrell et al., 2015; Takahashi et al., 2016). Furthermore, Met null mice lack all diaphragm musculature (Bladt et al., 1995; Dietrich et al., 1999; Maina et al., 1996). However, which cells are the source of HGF and what steps of diaphragm muscle development HGF/MET signaling regulates is unclear. Using conditional mutagenesis, pharmacological treatments, and an in vitro primary cell culture system (Bogenschutz et al., 2020), we demonstrate that the diaphragm’s connective tissue fibroblasts are a critical source of HGF that recruits and maintains MET+ muscle progenitors into and throughout the developing diaphragm. In addition, PPF-derived HGF, via its regulation of muscle, is required for defasciculation of the phrenic nerve. While either genetic or pharmacological inhibition of HGF/Met signaling results in large muscleless regions in the diaphragm, surprisingly these muscleless regions maintain their structural integrity and do not herniate. Thus, revising our previous conclusions (Merrell et al., 2015), we now show that muscle loss is not sufficient to induce CDH and additional connective tissue defects are required to weaken the diaphragm.
Results<italic>Hgf</italic> and <italic>Met</italic> are expressed in the developing diaphragm
To begin dissecting the precise role(s) of HGF/MET signaling in diaphragm development, we investigated the expression of Hgf and Met in the early diaphragm. We first examined mouse embryos at embryonic day (E) 10.5 when muscle progenitors are migrating from cervical somites to the nascent diaphragm and progenitors have already populated the forelimb (Sefton et al., 2018). At this stage, Hgf is expressed in the mesoderm lateral to the somites (but not in the somites themselves) and in the limb bud mesoderm (Figure 1A), while Met is expressed in the somites and the muscle progenitors in the limb bud (Figure 1D and Sonnenberg et al., 1993). At E11.5, muscle progenitors have migrated into the PPFs of the diaphragm (Sefton et al., 2018). During this stage, Hgf is expressed throughout the pyramidal PPFs (Figure 1B, arrows), while Met is expressed in a more restricted region in the PPFs (presumably in muscle progenitors; Figure 1E, arrows, Figure 1—figure supplement 1B), in limb muscle progenitors (Figure 1E and Sonnenberg et al., 1993), as well as the phrenic nerve (Figure 1F). By E12.5, the PPFs have expanded ventrally and dorsally across the surface of the liver (Merrell et al., 2015; Sefton et al., 2018). Strikingly, Hgf is restricted to the ventral and dorsal leading edges of the PPFs (Figure 1C, arrows and asterisks). Although Met is no longer detectable by whole-mount RNA in situ hybridization, qPCR indicates that it is still expressed at E12.5 at comparable cycle threshold values to Hgf and Pax7 (Figure 1—figure supplement 1A). These expression patterns suggest that HGF expressed in the mesoderm adjacent to the somites, PPF fibroblasts, and limb bud fibroblasts activates MET signaling in the diaphragm, limb muscle progenitors, and phrenic nerve.
Fibroblast-derived <italic>Hgf</italic> and somite-derived <italic>Met</italic> are required for muscularization of the diaphragm and limb.
(A) Lateral view at embryonic day (E) 10.5 of Hgf expression in lateral mesoderm adjacent to somites (arrow) and limb. (B) Cranial view of Hgf expression in E11.5 developing diaphragm (arrows) and limbs. (C) Cranial view of Hgf expression in E12.5 diaphragm at the leading edges of the pleuroperitoneal folds (PPFs) as they spread ventrally (arrows) and dorsally (asterisks). (D) Lateral view of E10.5 Met expression in muscle progenitors of limb and somites. (E) Cranial view of Met expression in E11.5 developing diaphragm (arrows) and limbs. (A–E) Expression via in situ hybridization. (F) MET and neurofilament immunofluorescence in transverse section through the phrenic nerve at E11.5. (G, I, K, M) E14.5 diaphragms stained for Myosin. (H, J, L, N) E14.5 forelimbs stained for Myosin and neurofilament. Deletion of Met in the Pax3 lineage (I, J; n = 3/3) or Hgf in Pdgfra lineage (tamoxifen at E8.5) (M, N; n = 3/3) leads to muscleless diaphragms and muscleless or partially muscularized limbs. Conversely, deletion of Hgf in Pax3 lineage (G, H; n = 3/3) or Met in Pdgfra lineage (tamoxifen at E9.5) (K, L; n = 3/3) results in normal diaphragm and limb muscle. Scale bars: (A–E) 500 μm; (F) 50 μm; (G, I, K, M) 500 μm; (H, J, L, N) 500 μm.
Limb and diaphragm phenotypes at embryonic day (E) 14.5 following deletion of <italic>Hgf</italic> and <italic>Met</italic> in <italic>Pax3</italic> and <italic>Pdgfra</italic> lineages.
<italic>Hgf</italic> and <italic>Met</italic> expression in the embryonic diaphragm and MET+ muscle progenitors are absent in <italic>Met<sup>Δ/Δ</sup></italic> mutant.
(A) Gene expression assayed by qPCR on pleuroperitoneal folds (PPFs) isolated from embryonic day (E) 12.5 WT+/+ embryos. Delta cycle threshold values (ΔCT), 18S RNA CT subtracted from CT of gene of interest, of Pax7, Met, and HGF are comparable at this stage (n = 3, no significant differences, as tested by one-way ANOVA; error bars represent standard error of the mean). (B) RNAScope of HGF and Met expression in E11.5 PPF. (C) MET immunofluorescence in cross section of the limb bud in E10.5 Met+/+ embryo (arrowheads). MET expression is absent in (D) MetΔ/Δ mutant. D, dorsal; L, lateral; M, medial; V, ventral. Scale bars: (B) 25 μm; (C, D) 100 μm.
qPCR for <italic>Pax7</italic>, <italic>Met</italic>, and <italic>Hgf</italic> from embryonic day (E) 12.5 pleuroperitoneal folds (PPFs).
<italic>Pdgfra</italic> is expressed within the developing diaphragm and <italic>Pdgfra<sup>CreERT2</sup></italic> allele targets the pleuroperitoneal folds (PPFs).
(A–D) mRNA in situ hybridization for Pdgfra at embryonic day (E) 10.5 in lateral view (A); E11.5 (B), E12.5 (C), and E13.5 (D) are in cranial view with dorsal to the top, dissected to diaphragm with the heart and lungs removed. Arrows indicate the leading edges of the PPFs. PdgfraCreERT2/+ labels PPF fibroblasts, including muscle connective tissue and central tendon, but not muscle. As expected, fibroblasts throughout the body wall are also labeled. Approximately E14.5 PdgfraCreERT2/+; Rosa26LacZ/+ embryos stained for LacZ, given different doses of tamoxifen at either E8.5 (E, F), E9.5 (G–, H), E10.5 (I, J), or E12.5 (K, L). Fibroblast labeling is stronger at higher doses of tamoxifen (6 mg) and fibroblasts are less strongly labeled with tamoxifen at E12.5. Scale bars: (A–D) 500 μm; (E, F, I, L) 500 μm; (G, H) 500 μm.
Control genotypes for embryos in <xref ref-type="fig" rid="fig1">Figure 1I–N</xref>.
Embryos are between embryonic day (E) 14.5 and E15.5. Top row: diaphragms stained for Myosin. Bottom row: forelimbs stained for Myosin and neurofilament. Muscle is formed in limb and diaphragm in all control genotypes. Scale bars: (A, C, E, G): 500 μm; (B, D, F, H) 500 μm.
Fibroblast-derived <italic>Hgf</italic> and somite-derived <italic>Met</italic> are required for diaphragm and limb skeletal muscle
The complete absence of diaphragm and limb muscles in mice with null-mutations for Met (Bladt et al., 1995; Dietrich et al., 1999; Maina et al., 1996) demonstrates that Met is critical for the development of these muscles. The spatially restricted expression of Met and Hgf suggests that MET signaling in muscle progenitors is activated by HGF in the PPF and limb fibroblasts. Surprisingly, the tissue-specific requirement of Met and Hgf has not been genetically tested in vivo. We tested whether the receptor is required in somite-derived diaphragm and limb myogenic cells by conditionally deleting Met (Huh et al., 2004) via Pax3Cre mice (Engleka et al., 2005), which recombines in the somites, including all trunk myogenic cells. Consistent with a hypothesized critical role of MET in myogenic cells, conditional deletion of Met in the somitic lineage results in a muscleless diaphragm and limbs (Figure 1I and J, Figure 1—figure supplement 1C and D). Additionally, we tested whether HGF derived from PPF and limb fibroblasts is critical via PdgfraCreERT2 mice (Chung et al., 2018). Pdgfra is expressed in the PPFs of the diaphragm (Figure 1—figure supplement 2B–D), and PdgfraCreERT2 drives Cre expression in the connective tissue fibroblasts of the diaphragm and limb, but not in muscle fibers (Figure 1—figure supplement 2E–L). When combined with Hgf fl (Phaneuf et al., 2004), PdgfraCreERT2/+; HgfΔ/fl mice given tamoxifen at E8.5 have a muscleless diaphragm (Figure 1M) and partial or complete loss of muscle in limbs (Figure 1N), demonstrating a crucial role for HGF derived from PPF and limb fibroblasts. We also tested the alternative hypotheses that HGF is produced by myogenic cells and MET signaling is active in fibroblasts, but Pax3Cre/+; HgfΔ/fl and PdgfraCreERT2/+; MetΔ/fl mice have normal diaphragm and limb musculature (Figure 1G, H, K and L; controls in Figure 1—figure supplement 3). In summary, these data establish that HGF derived from PPF and limb fibroblasts induces MET signaling in somite-derived myogenic cells, which is required for muscularization of the diaphragm and the limbs.
Diaphragm and shoulder muscle progenitors require fibroblast-derived <italic>Hgf</italic> during a similar temporal window
HGF/MET signaling is required for both diaphragm and limb muscle, but it is unclear whether Hgf is required during the same temporal window for development of diaphragm and forelimb muscles. This question is of particular interest because it has been proposed that a subset of shoulder muscle progenitors were recruited into the nascent PPFs during evolution, leading to the muscularization of the diaphragm in mammals (Hirasawa and Kuratani, 2013). If this were the case, shoulder muscle progenitors would be expected to migrate at a similar time and under the control of HGF/MET signaling as diaphragm progenitors in extant mammals. To dissect the temporal requirement for fibroblast-derived Hgf for diaphragm and forelimb muscles, we examined these muscles in E16.5–18.5 PdgfraCreERT2/+; HgfΔ/flox mice given tamoxifen at E9.5 or E10.5, when muscle progenitors are actively delaminating from the somites and migrating into the nascent diaphragm and forelimb. We initially gave 6 mg of tamoxifen at E9.5, but only two embryos survived. Based on these two embryos, there was no obvious difference between 6 mg versus 3 mg tamoxifen on muscle: one embryo given 6 mg had muscleless limbs and diaphragm (Figure 2Q–T), while the other had partial muscle in the diaphragm with normal limb muscle. Therefore, we used 3 mg of tamoxifen for all subsequent experiments and compared diaphragm and limb defects in individual embryos (each row in Figure 2 shows diaphragm and forelimb muscle from a single embryo). In the most mildly affected embryos, the diaphragm is missing a small ventral patch of muscle with normal forelimb muscles (Figure 2E–H; n = 3/10, compare with control in Figure 2A–D). In the most severely affected case, both the forelimb and diaphragm are muscleless (n = 1/10). A small number of mutants had muscleless diaphragms, but normal forelimb muscles (n = 2/10). Strikingly, a subset of mutants had muscleless (or nearly muscleless) diaphragms and displayed specific defects in shoulder musculature (Figure 2I–P; n = 4/10). The acromiodeltoid was absent or reduced and mispatterned (Figure 2L and P) and the spinodeltoid was strongly reduced in size, while other forelimb muscles appeared normal (Figure 2K and O). In all cases, the body wall muscles developed normally (e.g., Figure 2Q). When tamoxifen was given at E10.5, diaphragms had partial muscle, with normal forelimb muscles (n = 3/3; data not shown). In summary, muscleless limbs are always accompanied by a muscleless diaphragm, suggesting that these embryos had an early defect whereby muscle progenitors were unable to delaminate from the somites and migrate into nascent forelimbs and diaphragm. Partially muscularized diaphragms are associated with normal limb muscle. While a muscleless or nearly muscleless diaphragm may or may not have accompanying limb defects, loss of shoulder muscle was always associated with a muscleless or nearly muscleless diaphragm. These intermediate phenotypes indicate that most muscle progenitors migrate into the forelimb in advance of progenitors migrating into the diaphragm. However, based on their similar temporal sensitivity to HGF/MET signaling, the shoulder acromiodeltoid and spinodeltoid progenitors migrate at a similar time as the diaphragm progenitors. Thus continued expression of Hgf at this later developmental time point is required for recruitment of muscle cells necessary for development of diaphragm and shoulder muscles.
Reduced and mispatterned acromiodeltoid and spinodeltoid accompanies loss of diaphragm muscle following deletion of <italic>Hgf</italic> in the <italic>Pdgfra</italic> lineage.
(A–D) Diaphragm and limb musculature in control PdgfraCreERT2/+; Hgf fl/+ given 3 mg tamoxifen at embryonic day (E) 9.5. (E–H, I–L, M–P) Diaphragm and limb muscle in three mutant PdgfraCreERT2/+; HgfΔ/fl embryos given 3 mg tamoxifen at E9.5. In the mildest phenotype, loss of ventral diaphragm muscle (asterisk, E), but normal limb and shoulder muscles (F–H; n = 3/10). In the moderate phenotype, shoulder muscles were affected by the diaphragm (n = 4/10). Absence of diaphragm muscle (I) accompanied by reduced spinodeltoid and mispatterned acromiodeltoid (J–L). Similarly, loss of diaphragm muscle, except in ventral-most region (red dotted line, remaining purple is AP stain trapped in connective tissue layer, M) and normal limb muscle except spinodeltoid reduced and acromiodeltoid absent (asterisk, N–P). (Q–T) In PdgfraCreERT2/+; HgfΔ/fl embryo given 6 mg tamoxifen at E9.5 near complete loss of both diaphragm and limb muscle (n = 1/2). Embryos harvested between E16.5 and E18.5. All samples stained with Myosin antibody. ad, acromiodeltoid; sd, spinodeltoid; tbla, triceps brachii lateral; tblo, triceps brachii long. Scale bars: (A, E, I, M) 1 mm; (B, F, J, N, Q, R) 500 μm; (C, D, G, H, K, L, O, P, S, T) 500 μm.
Limb and diaphragm phenotypes at embryonic days (E) 16.5–E18.5 following deletion of <italic>Hgf</italic> in <italic>Pdgfra</italic> lineage.
HGF signaling can act as a neurotrophic factor and chemoattractant in spinal motor neurons and cranial axons (Caton et al., 2000; Ebens et al., 1996; Isabella et al., 2020). However, the function of HGF and MET in the development of the phrenic nerve, the sole source of motor innervation in the diaphragm, has not been examined. To test the role of HGF/MET signaling in the phrenic nerve, HgfΔ/Δ,, MetΔ/Δ, and Prrx1CreTg/+;Hgf fl/fl (Tg, transgene) mice were stained for neurofilament (Figure 3). By E12.0, in control mice the phrenic nerve has reached the PPFs and defasciculates into numerous small branchpoints prior to the full extension of the three primary branches (Figure 3A, arrows). However, in HgfΔ/Δ mutants, while the phrenic nerves reach the surface of the diaphragm, they do not correctly branch and defasciculate (Figure 3B). Instead of arborizing into numerous small branches, the right phrenic nerve bifurcates around the vena cava (n = 3/3; Figure 3B) and the left phrenic nerve fails to defasciculate to the same extent as in control embryos (Figure 3B). To test whether the PPF fibroblasts are a critical source of Hgf for branching of the phrenic nerve, Hgf was conditionally deleted using Prrx1CreTg (Logan et al., 2002), which robustly recombines in PPF-derived fibroblasts (Merrell et al., 2015). Consistent with PdgfraCreERT2/+;HgfΔ/fl mice, the diaphragms of Prrx1CreTg/+;HgfΔ/fl are muscleless (n = 5/12) or partially muscularized (n = 7/12). While the phrenic nerves reach the muscleless diaphragm, they lack primary and secondary branches (Figure 3C and D, arrows). Thus, loss of PPF-derived HGF leads to both muscle defects and phrenic nerve defasciculation defects. Similar defasciculation phenotypes are also present in MetΔ/Δ mutants. Confocal analysis of E11.5 diaphragms, when the phrenic nerve is just reaching the PPFs, reveals that defasciculation defects are present in MetΔ/Δ mutants by this early time point (Figure 3I and J). Comparison of E11.5 Met+/+, MetΔ/+, and MetΔ/Δ diaphragms reveals a dose-dependent requirement for Met as the number of fascicles is lower in heterozygotes and is further reduced in homozygous mutants (Figure 3E–K). Importantly, the reduced number of branches in MetΔ/+ nerves indicates that the reduced branching is not merely the result of the total loss of muscle, as MetΔ/+ embryos have normally muscularized diaphragms (e.g., see Figure 1K). One potential cause of the reduced defasciculation defect in MetΔ/+ embryos is developmental delay. Based on crown rump length and limb length, however, MetΔ/+ embryos are not developmentally delayed relative to Met+/+ embryos at E11.5 (Figure 3—figure supplement 1A–E). We also tested whether reduced defasciculation in MetΔ/+ embryos persists at later time points, but we found it resolves by E12.5 (Figure 3—figure supplement 1F–I).
Loss of <italic>Hgf</italic> and <italic>Met</italic> leads to defasciculation defects in the phrenic nerve.
Whole-mount neurofilament staining of the phrenic nerve in control (A, C, E, F) or Hgf (B, D) or Met mutants (G–J). Cranial view of dissected diaphragm region viewed with light microscopy (A–D) showing loss of phrenic nerve branches in HgfΔ/Δ (n = 3) diaphragm (arrows, A, B) and Prrx1CreTg/+; Hgf fl/fl diaphragm (arrows, C, D). Dorsal whole-mount view via confocal microscopy shows reduced phrenic nerve defasciculation in MetΔ/+ and MetΔ/Δ diaphragms (E–J). Phrenic nerve and C3-5 spinal nerves pseudocolored in light blue. (K) Quantification of phrenic nerve branchpoints at embryonic day (E) 11.5 in Met+/+ (n = 3), MetΔ/+ (n = 4), and MetΔ/Δ (n = 4) embryos. Significance tested with one-way ANOVA; error bars represent standard error of the mean (SEM). Scale bars: (A, B) 250 μm; (C, D) 1 mm; (E, G, I) 100 μm; (F, H, J) 100 μm.
Diaphragm and phrenic nerve phenotypes following deletion of <italic>Hgf</italic> or <italic>Met</italic>.
At embryonic day (E) 11.5, limb length and crown-rump length were not significantly different between <italic>Met<sup>+/+</sup></italic>, <italic>Met<sup>∆/+</sup></italic>, and <italic>Met<sup>∆/∆</sup></italic> embryos; branchpoints in <italic>Met<sup>∆/+</sup></italic> embryos are equivalent to <italic>Met<sup>+/+</sup></italic> at E12.5.
(A–C) Lateral view of unstained embryos (from the same cohort measured in Figure 3E–J). (D) Quantification of crown rump length. (E) Quantification of Limb length. (F) Quantification of branchpoints in sternocostal and dorsocostal phrenic nerve at E12.5. Significance tested with one-way ANOVA; error bars represent standard error of the mean (SEM). **p<0.01. (G–I) Cranial view of phrenic nerve stained with neurofilament at E12.5 in Met+/+, Met∆/+, and Met∆/∆. Embryos (n = 3). Scale bars: (A–C) 500 μm; (G, I) 100 μm.
qPCR for <italic>Pax7</italic>, <italic>Met</italic>, and <italic>Hgf</italic> from embryonic day (E) 12.5 pleuroperitoneal folds (PPFs).
Loss of <italic>Met</italic> in Olig2 lineage does not lead to defects in phrenic nerve branching and <italic>Hgf</italic> is not sufficient to rescue defasciculation defects in the absence of diaphragm muscle.
(A, B) Lateral view of the right phrenic nerve at embryonic day (E) 11.5 stained for neurofilament and imaged in whole mount by confocal microscopy. (C) Number of branches in phrenic nerve at E11.5 was not significantly different in Olig2Cre/+; Met∆/flox embryos (n = 4) compared to controls. (D, E) Anterior view of the right phrenic nerve at E11.5 stained for neurofilament and imaged in whole mount of Pax3SpD/SpD mutants compared to littermate controls. (F) Number of branches in phrenic nerve at E11.5 was significantly lower in Pax3SpD/SpD embryos (n = 3) compared to controls. Significance tested with unpaired t test; error bars represent standard error of the mean (SEM). ***p<0.001. (G, H) Hgf in situ hybridization on transverse section of E11.5 embryos. Hgf is expressed in Pax3+/+ (G) and muscleless Pax3SpD/SpD (H) pleuroperitoneal fold (PPF) (outline). Scale bars: (A, B, D, E) 100 μm; (G, H) 50 μm.
Number of axon branchpoints after deletion of <italic>Met</italic> in <italic>Olig2</italic> lineage and in <italic>Pax3<sup>SpD/SpD</sup></italic> mutants.
To test for the requirement of MET within the phrenic nerve, motor neuron-specific deletion of Met was performed using Olig2Cre (Zawadzka et al., 2010). However, defasciculation defects were not present at E11.5 in the phrenic nerve (Figure 3—figure supplement 2A–C). These data suggest that MET does not intrinsically regulate phrenic nerve branching, but instead PPF-derived HGF may regulate phrenic nerve branching indirectly via muscle. To test this, we analyzed at E11.5 the diaphragms of Pax3SpD/SpD embryos (Vogan et al., 1993), which are muscleless, but maintain Hgf expression (Figure 3—figure supplement 2G and H; Merrell et al., 2015). In Pax3SpD/SpD diaphragms, axon branchpoints are strongly reduced (Figure 3—figure supplement 2D–F), similar to MetΔ/Δ mutants. Thus, HGF in the absence of muscle is not sufficient to promote normal phrenic nerve defasciculation. Altogether these data demonstrate that PPF-derived HGF, via muscle, is required for normal phrenic nerve defasciculation and primary branching.
<italic>Hgf</italic> is required in fibroblasts to fully muscularize the diaphragm after delamination of muscle progenitors from somites
While HGF/MET signaling is critical for delamination of muscle progenitors from the somites (Dietrich et al., 1999), it is unclear whether HGF plays a later role in development of the diaphragm’s muscle. To test the later temporal requirement of HGF in PPF fibroblasts, we deleted Hgf via PdgfraCreERT2/+; HgfΔ/fl mice given tamoxifen at different time points. When PdgfraCreERT2/+; HgfΔ/fl mice were given tamoxifen at E9.0, prior to the onset of muscle precursor migration to the diaphragm (Sefton et al., 2018), the diaphragm lacks all muscle (Figure 4B; n = 3/3). This is likely due to a failure of muscle progenitors to delaminate and emigrate from the somites, as in Met-null diaphragms (Dietrich et al., 1999). When Hgf is deleted via tamoxifen at E9.5, when muscle progenitors are delaminating and migrating to the nascent diaphragm (Sefton et al., 2018), the diaphragm displays large ventral muscleless regions as well as dorsal muscleless patches at E14.5 (Figure 4C; n = 6/6). Notably, the phrenic nerves in these diaphragms only extend to the regions with muscle (Figure 4H). When Hgf is deleted via tamoxifen at E10.5, the muscle reaches its normal ventral extent in most E14.5 embryos (Figure 4D, n = 11/13). However, when these embryos are allowed to develop to E17.5 (when the muscle has normally expanded to the ventral midline), a large ventral muscleless region persists in mutant embryos (Figure 4F; n = 3/3). When mutants are given tamoxifen at E11.5, after migration of progenitors to the PPFs has completed (Sefton et al., 2018), a smaller muscleless region is present in the ventral diaphragm at E17.5 (Figure 4G; n = 4/4). Thus, these data demonstrate that after its initial requirement for muscle precursor delamination from the somites, PPF-derived Hgf is critical for muscularization of the ventral- and dorsal-most regions of the diaphragm. This role for HGF is consistent with its strong expression in these ventral- and dorsal-most regions (Figure 1C).
Loss of diaphragm muscle following timed deletion of <italic>Hgf</italic> in the pleuroperitoneal fold (PPF) fibroblast lineage.
(A–D) Upper panels: cranial view of embryonic day (E) 14.5 diaphragms stained for Myosin. Middle row panels: illustrations of muscle distribution. (A) Control PdgfraCreERT2/+; Hgf fl/+ diaphragm muscle forms two lateral wings that have not yet converged ventrally. (B) Muscleless diaphragm in PdgfraCreERT2/+; HgfΔ/flox when tamoxifen is administered at E9.0 (n = 3/3). (C) Large muscleless regions in ventral and dorsal diaphragm following tamoxifen administration at E9.5 (n = 6/6). (D) PdgfraCreERT2/+; HgfΔ/flox muscle reaches normal ventral extent when tamoxifen is administered at E10.5 (n = 11/13). (E–G) Unstained E17.5 diaphragms in cranial view. (E) Control diaphragm muscle has closed ventrally. (F) Large ventral muscleless region following HGF deletion at E10.5 (red bracket, n = 3/3). (G) Smaller ventral muscleless region following tamoxifen at E11.5 (red bracket, n = 4/4). (H) Cranial view of E14.5 PdgfraCreERT2/+; HgfΔ/flox mutant with tamoxifen administered at E9.5. Diaphragm stained for Myosin and neurofilament, indicating phrenic nerve tracks with regions of muscle (n = 3/3). Scale bars: (A–D, H) 500 μm; (E–G) 1 mm.
Timed deletion of <italic>Hgf</italic> in the pleuroperitoneal fold (PPF) fibroblast lineage.
Pleuroperitoneal folds (PPFs) spread normally and PPF fibroblasts are present in muscleless regions following deletion of <italic>Hgf</italic>.
(A–C) Cranial view of embryonic day (E) 13.5 Prrx1CreTg/+; Rosa26LacZ/+ diaphragms stained for LacZ. (A) By E13.5, PPFs have spread throughout the dorsal and all but the ventral-most region of the diaphragm. (B, C) PPF spread is unaffected in E13.5 embryos heterozygous or null for Hgf (n = 3). (D–F) Cranial view of E15.5 PdgfraCreERT2/+; Hgf∆/fl; Rosa26mTmG/+ diaphragm stained for GFP, Myosin, Pax7, and MyoD (MM, muscle markers). GFP+ fibroblasts are present in muscleless region (n = 4). Scale bars: (A–C) 500 μm; (D–F) 100 μm.
Sample information for spreading of pleuroperitoneal folds (PPFs) after deletion of <italic>Hgf</italic>.
We next sought to determine how PPF-derived HGF regulates development of the dorsal and ventral-most regions of the diaphragm muscle. Our previous studies (Merrell et al., 2015; Sefton et al., 2018) have shown that the PPFs expand dorsally and ventrally, carrying muscle as they expand, and therefore control overall morphogenesis of the diaphragm. Absence of dorsal and ventral muscle regions in Hgf mutants could result from a failure of the PPFs to expand dorsally and ventrally and thus lead to the consequent lack of dorsal and ventral diaphragm muscle. To test whether PPF expansion is aberrant following deletion of Hgf, we examined Prrx1CreTg/+; HgfΔ/fl; Rosa26LacZ/+ mice, in which we genetically labeled PPFs as they spread across the surface of the liver at E13.5. However, the PPFs reach their normal ventral extent at E13.5 following loss of Hgf (Figure 4—figure supplement 1A–C). To examine whether fibroblasts populate muscleless regions following deletion of Hgf, we stained for Pax7, MyoD, Myosin, and GFP in PdgfraCreERT2/+; HgfΔ/fl; Rosa26mTmG/+ mice at E15.5. GFP+ fibroblasts were present throughout large muscleless regions (Figure 4—figure supplement 1D–F). These data argue that the loss of ventral and dorsal muscle is not due to defects in PPF morphogenesis or survival of fibroblasts. Moreover, PPF expansion is not dependent on HGF or muscularization of the diaphragm.
Development of dorsal and ventral regions of diaphragm muscle requires continuous MET signaling
Our experiments conditionally deleting Hgf after emigration of myogenic progenitors from the somites indicate that HGF/MET signaling plays additional later roles in the development of the diaphragm’s muscle. To specifically test when MET signaling is required in myogenic cells, we first deleted Met using Pax7iCre/+ or tamoxifen-inducible Pax7CreERT2 mice (Keller et al., 2004; Murphy et al., 2011), which cause Cre-mediated recombination later than Pax3Cre in a subset of embryonic myogenic progenitors as well as all fetal and adult progenitors (Hutcheson et al., 2009). Neither Pax7iCre/+; MetΔ/fl nor Pax7CreERT2/+; MetΔ/fl embryos displayed any defects in diaphragm muscularization at E14.5 or P0 (Figure 5—figure supplement 1). This may indicate that Met is not required during fetal myogenesis as Pax7 is primarily expressed in fetal myogenic progenitors. However, Met derived from embryonic Pax3+Pax7- myogenic progenitors is likely present in the muscle of these mutant diaphragms and so does not permit analysis of the consequence of Met deletion in muscle.
As an alternate strategy to test when MET signaling is required, we turned to an ATP-competitive inhibitor of MET autophosphorylation, BMS777607 (as well as MET-related kinases RON and AXL; Schroeder et al., 2009), which inhibits phospho-Met in muscle progenitors (Figure 6—figure supplement 1A). We administered daily doses of BMS777607 to pregnant females to temporally inhibit MET signaling. In vehicle-treated controls harvested at E17.5, the diaphragm is completely muscularized (n = 12/12; Figure 5A). When BMS777607 was administered daily between E7.5 and E12.5, all embryos (n = 7/7) displayed bilateral dorsal muscleless patches and a ventral muscleless region (Figure 5B, arrows). When treated at E8.5–E9.5 or E9.5–10.5 (Figure 5C and D), all diaphragms had ventral muscleless regions (n = 14/14) and 35% had dorsal muscleless regions (n = 5/14). When treated at E11.5 and E12.5, after diaphragm progenitors have fully delaminated from the somites (Sefton et al., 2018), embryos had ventral muscleless regions (Figure 5E and F; n = 8/8) and dorsal muscleless regions (n = 3/8). Quantification of the size of the ventral muscleless region indicates that all MET inhibition strategies lead to muscleless regions, with the largest muscleless regions when MET is inhibited E7.5–E12.5 or E8.5–E9.5 (Figure 5F). These data demonstrate that MET is continuously required from E7.5 to E12.5 for complete muscularization of the diaphragm. The regions requiring continuous MET signaling are on the leading edges of the diaphragm: the bilateral dorsal muscle and ventral midline muscle. These are the last regions to receive muscle progenitors that differentiate into myofibers. For both Hgf deletion in fibroblasts and global MET inhibition, loss during muscle migration from somites at approximately E9.5 leads to dorsal and ventral muscleless regions, while later loss at E11.5 leads to primarily ventral muscleless regions. The ventral midline of the diaphragm does not fully close until E16.5, likely making it more susceptible to later perturbations.
Reduction of Met signaling through inhibitor BMS777607 results in muscleless dorsal and ventral regions of the diaphragm.
(A-E) Unstained E17.5 diaphragms in cranial view. (A) The left and right portions of the costal diaphragm meet in the ventral midline by embryonic day (E) 17.5 in vehicle-treated controls (n = 12/12). (B–E) Dorsal left, dorsal right (arrows), and ventral midline regions (brackets) are muscleless when treated with BMS777607 daily between E7.5 and E12.5 (B: n = 7; C: n = 6; D: n = 8; E: n = 9). (F) Width of ventral midline muscleless region is significantly larger than vehicle-treated controls when BMS777607 is administered at either early (E7.5–E8.5) or at later stages of diaphragm development (E11.5–E12.5). Significance tested with one-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars represent standard error of the mean (SEM). Scale bars (A–E) 1 mm.
Measurements of ventral muscle gap after timed treatments with BMS777607.
Normal muscularization of the diaphragm following deletion of <italic>Met</italic> in the <italic>Pax7</italic> lineage.
(A, B) Normal muscle, viewed via GFP, in P0 Pax7iCre/+; MetΔ/fl; Rosa26mTmG/+ and control diaphragms (n = 4). (C–, D) Normal muscle, stained with Myosin at embryonic day (E) 14.5 in Pax7CreERT2/+; MetΔ/fl and control diaphragms given 6 mg of tamoxifen at E9.5 (n = 3). Scale bars: (A, B) 1 mm; (C, D) 500 μm.
Sample information for <italic>Met</italic> deletion in <italic>Pax7</italic> lineage.
MET signaling is required for survival and motility of diaphragm muscle progenitors in vitro
While the regions most sensitive to MET inhibition are those that differentiate latest, it is unclear whether MET is required for proliferation, survival, differentiation, and/or motility. To investigate the function of MET in diaphragm muscle progenitors, we turned to an in vitro system to co-culture E12.5 diaphragm fibroblasts and myoblasts (Bogenschutz et al., 2020) in combination with BMS777607. PPF explants were dissected from E12.5 Pax3Cre/+; Rosa26nTnG/+ embryos (Engleka et al., 2005; Prigge et al., 2013), in which Pax3-derived myogenic nuclei are GFP+ and PPF fibroblast nuclei are Tomato+, and cultured them with either 10 μM BMS777607 or DMSO vehicle control (Figure 6—figure supplement 2). Overall, growth of GFP+ muscle progenitors was impaired with inhibitor treatment (Figure 6A–C). To assess effects of the inhibitor on the number of myoblasts, we examined MyoD. After 48 hr in culture, MyoD expression was reduced with inhibitor treatment (Figure 6D), and the percentage of cells co-expressing GFP and MyoD protein was similarly abrogated (Figure 6E and F). By contrast, expression of the PPF fibroblast marker Gata4 was not significantly changed following inhibitor treatment (Figure 6G). We tested whether the decreased growth of myogenic cells was due to decreased proliferation or increased apoptosis. Analysis of GFP+ myogenic cells labeled via EdU indicates that BMS777607 treatment does not significantly change the percentage of proliferating cells (Figure 6H and I). However, examination of apoptotic cells via staining for cleaved Caspase-3 showed that BMS777607 treatment significantly increased the number of apoptotic GFP+ myogenic cells (Figure 6J and K). To examine the relevant pathway(s) for increased apoptosis, we assayed the expression of Fas (a cell surface death receptor), tumor protein Trp53 (which encodes p53), and autophagy marker Map1l3ca in BMS777607 and vehicle-treated PPFs. Both Fas and Map1l3ca are significantly upregulated following treatment with BMS777607, while Trp53 is unaffected (Figure 6—figure supplement 1B–D). Inhibition of MET is also known to impair cell motility (reviewed by Birchmeier et al., 2003), and we found that the motility of GFP+ cells treated with BMS777607 was impaired, with reduced velocity and lower overall displacement (Figure 6L–M). We also assessed effects on cell morphology by examining diaphragm muscle progenitors, labeled with membrane-bound GFP (via Pax3Cre/+; Rosa26mTmG/+). Cells were significantly more circular with BMS777607 treatment, which is consistent with compromised survival and motility (Figure 6N and O). Overall, these data show that MET signaling is important for survival and motility of diaphragm muscle progenitors in vitro.
In vitro survival and motility of myogenic cells is impaired by pharmacological inhibition of Met signaling.
(A–O) Embryonic day (E) 12.5 pleuroperitoneal folds (PPFs) isolated from Pax3Cre/+; Rosa26nTnG/+ (A–M) or Pax3Cre/+; Rosa26mTmG/+ (N, O) embryos cultured with DMSO vehicle control or BMS777607 and imaged on the ImageXpress Pico. (A–C) The average ratio of GFP+ fold change (cell number at time T/cell number at time 0) ± SEM is plotted (A). The fold change (A) and total final count (B) of GFP+ cells are reduced following treatment with BMS777607 after 72 hr in culture (n = 4 biological replicates). Representative GFP images shown in (C) at 72 hr in culture. (D–F) MyoD expression (via qPCR, n = 3, D) and percentage of MyoD+ GFP+ cells (n = 4) (E) is significantly lower after 48 hr in culture with MET signaling inhibition. Representative images of GFP and MyoD expression (F). (G) Expression of PPF fibroblast marker Gata4 is not significantly affected by BMS777607 treatment (via qPCR, n = 3). (H, I) EdU labeling of GFP+ cells is not significantly changed by BMS777607 treatment (n = 4). (J, K) Cleaved Caspase-3 expression was significantly increased with BMS777607 treatment (n = 3). (L, M) Mean velocity and total displacement of peripheral GFP+ nuclei were decreased with BMS777607 treatment. GFP+ nuclei were imaged every 8 min over 14 hr to track cell motility (>500 nuclei measured from n = 3). (N, O) BMS777607-treated GFP+ cells were significantly more circular (>1000 cells measured from n = 3). Representative images (O). *p<0.05, **p<0.01, ****p<0.0001. Statistical changes in cell number over time (A) were determined using repeated-measures ANOVA on the log2 transformed fold change over time. Statistical changes determined with unpaired t test in (B, D, E, G, H, L, M, N, J). Error bars represent standard error of the mean (SEM). Scale bars: (C, F, I, K, O) 100 μm.
In vitro effects on diaphragm muscle after pharmacological inhibition of MET signaling.
BMS777607-mediated downregulation of phospho-Met and relevant pathways to cell survival in embryonic day (E) 12.5 diaphragms.
(A) Injured skeletal muscle satellite cells in the tibialis anterior (3 days post injury) were genetically labeled by Pax7CreERT2; Rosa26Pham/+, treated with BMS777607 or vehicle control, stained for PE conjugated phospho-Met, and analyzed via FACS. Median PE levels of GFP+ cells were downregulated from 1425–793 log-scale fluorescence intensity following a 4 hr BMS777607 treatment. (B–D) E12.5 pleuroperitoneal folds (PPFs) isolated from Pax3Cre/+; Rosa26nTnG/+ embryos cultured with DMSO vehicle control or BMS777607 and RNA isolated for qPCR (n = 3). Expression of Fas cell surface death receptor (B) and autophagy marker Map1lc3a (D) is significantly upregulated by BMS777607 treatment, while Trp53, which encodes tumor protein p53, is unchanged (C). Significance tested with unpaired t test; error bars represent standard error of the mean (SEM). **p<0.01, ***p<0.001.
qPCR data for <italic>Fas</italic>, <italic>Map1lc3a</italic>, and <italic>Trp53</italic>.
Initial muscle lineage and final nuclear density following vehicle and BMS777607 treatment.
(A–D) Embryonic day (E) 12.5 pleuroperitoneal folds (PPFs) isolated from Pax3Cre/+; Rosa26nTnG/+, and treated with DMSO (left column) or BMS777607 (right column). (A, B) GFP fluorescence at t = 0 hr of imaging. (C, D) Hoechst stain of fixed E12.5 PPFs after t = 72 hr of imaging. Scale bars: (A–D) 500 μm.
MET signaling is required for the population of muscle progenitors at the diaphragm’s leading edges and the consequent development of the dorsal and ventral-most muscle regions
Our in vivo studies show that HGF/MET signaling is required for the development of the dorsal and ventral-most regions of the diaphragm muscle, and our in vitro studies find that MET is required for muscle progenitor survival and motility. Based on these data, we hypothesized that in vivo loss of dorsal and ventral muscle regions is due to fewer muscle progenitors and/or myoblasts at the dorsal and ventral leading edges of the diaphragm when it is expanding. To test this, wild-type embryos were treated with BMS777607 daily E7.5–E11.5, harvested at E12.5, and stained for myogenic cells with a cocktail of PAX7, MyoD, and Myosin antibodies as well as for EdU, cleaved Caspase-3, and neurofilament. We found that the PPFs (identified and outlined in 3D by their unique morphology, viewed by autofluorescence) were more variable in size, but not significantly decreased in size from control diaphragms. We also found, consistent with our analysis of HgfΔ/Δ, and MetΔ/Δ mice, that nerve branching was strongly reduced by the inhibitor (Figure 7C, F, J and M; Video 1). Supporting our hypothesis, the inhibitor led to a reduction in the number of mononuclear progenitors and myoblasts at the ventral and dorsal leading edges of the muscle (Figure 7A, D, H and K, arrows; Video 1). Inhibitor-treated embryos also showed reduced numbers of EdU+ cells overall (Figure 7B, E and G) and an increased number of cleaved Caspase-3-positive cells within the PPFs (Figure 7I, L and N). To exclude early impacts of BMS777607 on muscle progenitor emigration from somites, BMS777607 or vehicle was also administered only E11.5–E12.5 to Pax3Cre/+; Rosa26mTmG/+ embryos, which were then harvested at E15.5. Although the number of mononuclear GFP+ cells were not significantly reduced, the number of Pax7/MyoD/Myosin-labeled cells was substantially reduced at the ventral leading edges with inhibitor treatment (Figure 7O–V). Thus, these data demonstrate that in vivo Met signaling is required to promote proliferation and survival of myogenic cells, and its inhibition leads to a loss of muscle progenitors and myoblasts at the leading edges of the PPFs (which express high levels of Hgf at E12.5) and results in a loss of the dorsal-most and ventral-most diaphragm muscle.
Inhibition of Met signaling in vivo alters cell proliferation, apoptosis, phrenic nerve morphology, and reduces muscle progenitors at the leading edge of the diaphragm.
(A–N) WT embryos were treated with BMS777607 or vehicle control daily between embryonic day (E) 7.5 and E11.5, harvested at E12.5, stained for Pax7/MyoD/Myosin (MM) and neurofilament, and imaged in whole-mount cranial view on the confocal. Treatment with BMS777607 leads to fewer mononuclear myogenic cells on the dorsal and ventral leading edges of the diaphragm (arrows; A, D, H, K), reduced total EdU-positive nuclei (n = 3 vehicle treated; n = 4 BMS777607 treated; B, E, G), increased cleaved-caspase-3 positive cells (n = 3; I, L, N), and aberrant phrenic nerve branching (F, M), where the right phrenic nerve wraps around vena cava (asterisk in F, M). Schematic of region imaged (black box) in (G). (O–V) Pax3Cre/+; Rosa26mTmG/+ embryos were treated with BMS777607 or vehicle at E11.5 and E12.5, harvested at E15.5 and stained for GFP and MM. Tomato is unlabeled. (O–T) Cranial view of leading ventral edges of the diaphragm at E15.5, with mononuclear muscle progenitors in region that will fill with muscle by E16.5. (U) Fewer mononuclear Pax7/MyoD/Myosin+ cells populate the leading edge following treatment with BMS777607 (n = 3). (V) Quantification of mononuclear GFP+ cells in vehicle or BMS777607-treated embryos. Significance analyzd with unpaired t test; error bars represent standard error of the mean (SEM). *p<0.05 **p<0.01.Schematic of region imaged (black box) in (V). Scale bars: (A–F, H–M) 100 μm; (O–T) 50 μm.
In vivo reduction diaphragm muscle at the leading ventral edge after pharmacological inhibition of MET signaling.
Inhibition of Met signaling in vivo reduces muscle progenitors at the leading edges of the diaphragm.
Embryos were treated with BMS777607 or vehicle control daily between embryonic day (E) 7.5 and E11.5, harvested at E12.5, stained for Pax7, MyoD, Myosin (muscle markers in red), and neurofilament (blue) and imaged in whole-mount cranial view on the confocal. Fewer mononuclear myogenic cells are present on the dorsal and ventral leading edges of the diaphragm at E12.5 and the phrenic nerve displays abnormal branching after treatment with BMS777607.
Discussion
The diaphragm is an essential mammalian skeletal muscle, playing a critical role in respiration and serving as a barrier that separates the thorax from the abdomen. Not only is the diaphragm a functionally important muscle, but it serves as an excellent system to study muscle patterning and morphogenesis, since it is a flat muscle that largely develops in two dimensions. Development of the diaphragm, like other skeletal muscles, requires the integration of muscle, connective tissue, and nerve that arise from different embryonic sources. Our study establishes that the connective tissue fibroblasts are the source of a molecular signal, HGF, that directly controls the recruitment, survival, and expansion of MET+ muscle progenitors and indirectly, via muscle, regulates phrenic nerve branching (Figure 8).
Model of HGF-MET signaling in skeletal muscle and phrenic nerve in the diaphragm.
(A) Our data support a model where pleuroperitoneal fold (PPF) fibroblast-derived HGF directly regulates recruitment, survival, and expansion of MET+ muscle progenitors (in red) and indirectly, via muscle, regulates phrenic nerve (in yellow) branching and outgrowth. (B) Met null mutations or Met deletion in muscle progenitors (prior to embryonic day [E] 9.0) lead to muscleless limbs and diaphragm and reduced defasciculation of the phrenic nerve. Pharmacological inhibition of MET between E7.5 and E11.5 or later time points (i.e., E11.5–12.5) results in reduced muscle progenitors (via increased apoptosis and reduced motility) at the diaphragm’s leading edges at E12.5 and muscleless regions in the dorsal and ventral diaphragm at E17.5. (C) Early deletion of fibroblast-derived HGF (tamoxifen at E9.5) at E14.5 results in large muscleless regions, including dorsal and ventral muscle regions, while later mutations (tamoxifen at E11.5) lead to ventral muscleless regions. CM, crural muscle; CT, central tendon; So, somite.
Development of muscle and its innervating motor neurons must be tightly integrated to produce a functional muscle. The connective tissue is an ideal candidate tissue to orchestrate this process as it enwraps myofibers and neuromuscular junctions (Nassari et al., 2017; Sefton and Kardon, 2019). In the diaphragm, the PPFs are the source of the diaphragm’s connective tissue fibroblasts and critical for overall diaphragm morphogenesis (Merrell et al., 2015). Thus, the PPFs are likely to coordinately regulate muscle and the phrenic nerve. A number of previous studies also suggested that HGF/MET is a key signaling pathway for this coordination since it has been found to regulate both muscle development and innervation (Bladt et al., 1995; Dietrich et al., 1999; Ebens et al., 1996; Lamballe et al., 2011; Maina et al., 1996; Yamamoto et al., 1997). In addition, previous studies established that Met is required for diaphragm development (Bladt et al., 1995; Dietrich et al., 1999; Maina et al., 1996). Here, we used conditional mutagenesis to specifically target Hgf and Met in PPF-derived connective tissue fibroblasts or muscle progenitors to genetically dissect the role of these cells and HGF/MET signaling in diaphragm development (Figure 8). First, as expected we found that HGF/MET signaling is required for the initial delamination and migration of MET+ muscle progenitors from the somites to the nascent diaphragm. More surprisingly, we found that HGF expressed at the dorsal and ventral margins of the expanding PPFs is continuously required for proliferation, survival, and motility of muscle progenitors and the consequent expansion and full muscularization of the diaphragm. Our experiments also showed that, unlike the muscle progenitors, HGF/MET signaling is not required for the recruitment and targeting of phrenic nerve axons to the nascent diaphragm. However, PPF-derived HGF, through its regulation of muscle, is required for phrenic nerve defasciculation, primary branching, and subsequent branch outgrowth throughout the diaphragm muscle. Previous studies have identified how phrenic motor neurons are specified in the motor column (Dasen et al., 2008; Dasen et al., 2003; Dasen et al., 2005; Jung et al., 2010; Rousso et al., 2008) and later arborize (Philippidou et al., 2012) to form neuromuscular junctions (Burden, 2011; Li et al., 2008; Wang et al., 2003; Yumoto et al., 2012). Only one study (Uetani et al., 2006) has identified molecular regulators, Receptor Protein Tyrosine Phosphatases σ and δ, of phrenic nerve defasciculation and primary axon outgrowth. Altogether our study demonstrates that PPF-derived connective tissue fibroblasts and HGF directly control diaphragm muscle recruitment and expansion and indirectly, via muscle and the neurotrophic factors it expresses, control phrenic nerve branching and outgrowth. Connective tissue and HGF are also likely to regulate the development of other muscles and their motor neurons as connective tissue and HGF/MET signaling have been implicated in the development and innervation of limb and back muscles (Caruso et al., 2014; Helmbacher, 2018).
Our study also provides insights into how the diaphragm might have evolved. The diaphragm muscle is unique to mammals and how it evolved in mammals is a major unanswered question (Perry et al., 2010). Evolution of the diaphragm involved the acquisition of developmental innovations in mammals that are absent from birds and reptiles. Comparison of these groups suggests the following important developmental innovations: formation and expansion of PPFs across the liver to separate the thoracic and abdominal cavities, recruitment of muscle progenitors to the PPFs and their expansion and differentiation into the radial array of myofibers, and recruitment and targeting of motor neurons to the diaphragm muscle (Hirasawa et al., 2016; Hirasawa and Kuratani, 2013; Sefton et al., 2018). HGF/MET signaling has important, conserved functions in the development of most vertebrate hypaxial muscles (Adachi et al., 2018; Haines et al., 2004; Okamoto et al., 2019). Here we identify that Hgf expressed by the PPFs is crucial for recruitment and expansion of diaphragm muscle progenitors. Our experiments conditionally deleting Hgf also indicate that migration of diaphragm and shoulder (spinodeltoid and acromiodeltoid) muscle progenitors is Hgf-dependent and occurs contemporaneously. Recruitment of shoulder muscle progenitors to the nascent diaphragm has been proposed as important for the evolutionary origin of the mammalian diaphragm (Hirasawa and Kuratani, 2013). Thus, our experiments suggest that the evolutionary acquisition of Hgf expression in the PPFs may have been a key event that allowed a subset of shoulder muscle progenitors to be recruited to the nascent diaphragm. Our experiments also indicate that continued Hgf expression is critical for the full muscularization of the diaphragm. Therefore, once Hgf was expressed in the PPFs there may have been selection for its continued expression at the PPF’s leading edges to enable expansion of the muscle and complete separation of the thoracic and abdominal cavities. Localization of HGF at the leading edges provides a mechanism to regulate the directional expansion and shape of muscle by promoting survival and motility of muscle progenitors at the dorsal and ventral edges of the muscle. Interestingly, our data indicate that the evolutionary acquisition of Hgf expression in the PPFs is not sufficient to recruit motor neurons to the PPFs and so some other signal(s) must be involved in the evolutionary recruitment of the phrenic nerves. Also, still unknown are the developmental and evolutionary mechanisms driving formation and morphogenetic expansion of the PPFs.
Finally, our study elucidates some of the cellular mechanisms underlying the etiology of CDH. CDH is characterized by defects in the muscularization of the diaphragm, and two sites where muscle is commonly absent are the dorsal-most region of the diaphragm (designated posterior in humans and hernias in this area are called Bochdalek hernias) and the ventral-most diaphragm (anterior in humans and hernias here are called Morgagni hernias) (Ackerman et al., 2012; Irish et al., 1996; Kardon et al., 2017). Our analysis of diaphragms in which HGF/MET signaling is perturbed has found that these two regions are the most likely to have muscularization defects and suggests mechanistically why the dorsal-most and ventral-most diaphragm are most susceptible to muscularization defects. Mutations or variants in any gene or signaling pathway that, similar to HGF/MET, regulates the proliferation, survival, or motility of muscle progenitors will lead to a depletion of the pool of muscle progenitors and the consequent loss of the dorsal and ventral-most diaphragm muscle since these regions develop last (Figure 8). Most surprisingly, our analysis revealed that the muscleless connective tissue regions in mice with deletion of Hgf in the PPFs or pharmacological inhibition of MET signaling do not herniate. We previously conducted a detailed analysis of mice in which the transcription factor Gata4 was deleted in the PPFs (Merrell et al., 2015). In these mice, muscleless connective tissue regions develop, but these regions always herniate and give rise to herniated tissue covered by a connective tissue sac. Based on our study of Gata4, we had proposed that such ‘sac’ hernias (Pober, 2007) develop when localized regions of amuscular tissue develop in juxtaposition with muscularized tissue; the biomechanical difference in strength between these regions allows the abdominal tissues to herniate through the weaker amuscular regions. However, the lack of herniation in the Hgf mutants demonstrates that the formation of amuscular connective tissue regions is not sufficient to cause ‘sac’ hernias. While the formation of amuscular regions is likely a critical step in the formation of ‘sac’ hernias, other defects in connective tissue integrity are likely necessary to cause these susceptible amuscular regions to actually herniate. Thus, herniation may be a multistep process involving loss of muscle followed by defects in connective tissue strength or elasticity that allow the liver to herniate into the thoracic cavity. Comparison of amuscular connective tissue that maintains its structural integrity to herniated connective tissue may provide further insight into these processes and reveal therapeutic targets in the future.
Materials and methods
Reagent type (species) or resource
Designation
Source or reference
Identifiers
Additional information
Genetic reagent (Mus musculus)
PdgfraCreERT2
Chung et al., 2018
RRID:IMSR_JAX:032770
Dr. Brigid Hogan (Duke University Medical School)
Genetic reagent (M. musculus)
Prrx1CreTg
Logan et al., 2002
RRID:IMSR_JAX:005584
Dr. Clifford Tabin (Harvard Medical School)
Genetic reagent (M. musculus)
Pax3Cre
Engleka et al., 2005
RRID:IMSR_JAX:005549
Dr. Kurt Engleka (University of Pennsylvania)
Genetic reagent (M. musculus)
Pax7iCre
Keller et al., 2004
RRID:IMSR_JAX:010530
Dr. Mario Capecchi (University of Utah)
Genetic reagent (M. musculus)
Pax7CreER
Murphy et al., 2011
RRID:IMSR_JAX:017763
Dr. Gabrielle Kardon (University of Utah)
Genetic reagent (M. musculus)
Olig2Cre
Zawadzka et al., 2010
RRID:IMSR_JAX:025567
Dr. William Richardson (University College London)
Genetic reagent (M. musculus)
Hprt-cre
Tang et al., 2002
RRID:IMSR_JAX:004302
Dr. Jeffrey Mann (Monash University)
Genetic reagent (M. musculus)
Rosa26LacZ
Soriano, 1999
RRID:IMSR_JAX:003309
Dr. Philippe Soriano (Mount Sinai School of Medicine)
Genetic reagent (M. musculus)
Rosa26nTnG
Prigge et al., 2013
RRID:IMSR_JAX:023537
Dr. Edward Schmidt (Montana State University)
Genetic reagent (M. musculus)
Rosa26mTmG
Muzumdar et al., 2007
RRID:IMSR_JAX:007576
Dr. Liqun Luo (Stanford University)
Genetic reagent (M. musculus)
Rosa26Pham/+
Pham et al., 2012
RRID:IMSR_JAX:018385
Dr. David Chan (California Institute of Technology)
Genetic reagent (M. musculus)
Hgffl
Phaneuf et al., 2004
MGI:3574633
Dr. James Wilson (University of Pennsylvania)
Genetic reagent (M. musculus)
Metfl
Huh et al., 2004
RRID:IMSR_JAX:016974
Dr. Snorri Thorgeirsson (National Institutes of Health)
Anti-PHOSPHO-MET, PE Conjugate (rabbit monoclonal)
Cell Signaling
Cat# 12468
Working concentration:0.5 µg/ml
Other
TaqMan Gata4
ThermoFisher
Mm00484689_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 84
Other
TaqMan Myod1
ThermoFisher
Mm00440387_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 86
Other
TaqMan Met
ThermoFisher
Mm01156972_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 74
Other
TaqMan Hgf
ThermoFisher
Mm01135193_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 68
Other
TaqMan Pax7
ThermoFisher
Mm01354484_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 68
Other
TaqMan 18S rRNA
ThermoFisher
4333760T
Oligonucleo-tides for qPCR; cDNA product size (bp): 187
Other
TaqMan Fas
ThermoFisher
Mm01204974_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 76
Other
TaqMan Trp53
ThermoFisher
Mm01731287_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 133
Other
TaqMan Map1lc3a
ThermoFisher
Mm00458724_m1
Oligonucleo-tides for qPCR; cDNA product size (bp): 63
Chemical compound, drug
BMS777607
Selleckchem
Cat# S1561
In vivo: 0.05 mg/g of body weightIn vitro: 10 μM
Mice and staging
All mouse lines have been previously published. We used Prrx1CreTg (Logan et al., 2002), PdgfraCreERT2 (Chung et al., 2018), Pax3Cre (Engleka et al., 2005), Pax7iCre (Keller et al., 2004), Pax7CreER (Murphy et al., 2011), Olig2Cre (Zawadzka et al., 2010), and Hprt-cre (Tang et al., 2002) Cre alleles. Cre-responsive reporter alleles included Rosa26LacZ (Soriano, 1999), Rosa26nTnG (Prigge et al., 2013), Rosa26mTmG (Muzumdar et al., 2007), and Rosa26Pham/+ (Pham et al., 2012). The Hgf fl (Phaneuf et al., 2004) conditional allele (B6;129-Hgftm1Jmw/Mmnc, identification number 423-UNC) was obtained from the Mutant Mouse Regional Resource Center, an NIH-funded strain repository, and was donated to the MMRRC by S. E. Raper, Ph.D., University of Pennsylvania Medical Center. We also used the Metfl (Huh et al., 2004) conditional allele. Hgf Δ/+ and Met Δ/+ mice were generated by breeding Hgf fl and Metfl mice to Hprt-cre mice. Embryos were staged as E0.5 at noon on the day dams presented with a vaginal plug. Mice were backcrossed onto a C57Bl/6J background. Experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee at the University of Utah.
Immunohistochemistry, immunofluorescence, and in situ hybridization
For section immunofluorescence, optimal cutting temperature (OCT)-embedded tissues were sectioned to 10 μm thickness and fixed for 5 min in 4% paraformaldehyde (PFA). Tissue sections were blocked for 60 min in 5% goat serum in phosphate-buffered saline (PBS), incubated overnight at 4°C in primary antibodies. Sections were washed in PBS, incubated with secondary fluorescent antibodies (used at 1–5 μg/ml; Jackson Laboratories or Thermo Fisher) for 2 hr at room temperature (RT), washed with PBS, stained for 5 min with Hoechst to label nuclei, post-fixed in 4% PFA, rinsed in water and mounted with Fluoromount-G (Southern Biotech). Primary antibodies are listed in Key resources table. Sections were imaged on an Olympus BX63.
For immunofluorescence on PPF cell cultures, cells were fixed in 4% PFA for 20 min at RT, washed in PBS, blocked for 60 min in 5% goat serum with 0.1% Triton X-100 in PBS, and stained overnight for primary antibodies (see Key resources table). Cells were then washed in PBS, incubated for 2 hr in secondary antibodies, washed in PBS, incubated in Hoechst to label nuclei, washed in PBS, and rinsed in water and mounted in Fluoromount. EdU (Life Technologies) was applied to cells 1 hr prior to fixation and detected after secondary labeling based on the manufacturer’s instructions with Alexa647 picolyl azide. Stained cells were imaged with ImageXPress Pico automated cell imager (Molecular Devices).
Whole-mount embryos were fixed for 24 hr in 4% PFA at 4°C, dissected, incubated for either 2 hr at RT or overnight at 4°C in Dent’s bleach (1:2 30% H2O2:Dent’s fix) and stored in Dent’s fix (1:4 DMSO:methanol) for at least 5 days at 4°C. Embryos were washed in PBS, blocked for 1 hr in 5% goat serum and 20% DMSO, incubated in primary antibodies (see Key resources table) for 48 hr, washed in PBS, incubated in secondaries for 24–48 hr, washed in PBS, and cleared BA:BB (33% benzyl alcohol, 66% benzyl benzoate) at RT. Embryos labeled with AP-conjugated anti-Myosin heavy chain were heat-inactivated at 65°C for 1 hr, incubated in primary antibody for 48 hr, and detected with 250 μg/ml NBT and 125 μg/ml BCIP (Sigma) in alkaline phosphatase buffer. For detection of HRP-conjugated secondary antibodies, embryos were incubated in 10 mg diaminobenzidine tetrahydrochloride in 50 ml PBS with 7 μl hydrogen peroxide for approximately 20 min.
For whole-mount EdU analysis in embryos, 10 μg/g of body weight of EdU was administered to pregnant females 1 hr prior to harvest via IP injection.
Whole-mount in situ hybridization was performed as previously described (Riddle et al., 1993). For whole-mount β-galactosidase staining, embryos were fixed overnight in 1% PFA at 4°C and 2 mM MgCl2. Diaphragms were dissected, washed in PBS and then in LacZ rinse buffer (100 mM sodium phosphate, 2 mM MgCl2, 0.01% sodium deoxycholate, and 0.02% Ipegal), and stained for 16 hr at 37°C in X-gal staining solution (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, and 1 mg/ml X-gal).
Microscopy and three-dimensional rendering
Whole-mount fluorescent images were taken on a Leica SP8 confocal microscope. Optical stacks of whole-mount images were rendered and structures highlighted using FluoRender (Wan et al., 2009). To highlight features (such as the phrenic nerve in Figure 4), objects were selected in FluoRender based on morphology on individual Z optical sections using the paintbrush tool, and then these objects were extracted, rendered, and pseudo-colored. For EdU and cleaved Caspase-3 labeling, PPFs were first selected based on morphology and the total PPF area measured. Individual nuclei from PPFs were then counted using the Component Analyzer Tool.
Cell culture, media, and reagents
E12.5 embryos were dissected from pregnant Rosa26mTmG/mTmG or Rosa26nTnG/nTnG females mated with Pax3Cre/+ males. Embryos and PPFs were dissected as previously described (Bogenschutz et al., 2020). Briefly, embryos were dissected from yolk sacs in DMEM/F-12 GlutaMAX (Invitrogen) pre-warmed to 37°C. To isolate the trunk region, embryos were cut posterior to the forelimbs and just anterior to the hindlimbs, leaving liver largely attached to the trunk. Heart and lungs were removed from the thoracic cavity with forceps and the trunk trimmed to expose the PPFs sitting cranial to the liver. Trunks were pinned to a 6 mm dish coated in Sylgard and PPF pairs manually isolated with forceps from the body wall, septum transversum, and underlying liver. Each PPF explant pair was then placed in a single well from a 96-well plate with 100 μl of media. Growth of both PPF fibroblasts and myogenic cells was promoted using DMEM/F-12 GlutaMAX (Invitrogen), 10% FBS, 50 μg/ml gentamicin, and 0.5 nM FGF. PPFs were grown in a 37°C incubator overnight and then imaged on an ImageXPress Pico automated cell imager (Molecular Devices) or a Leica SP8 confocal microscope, for 1–4 days, changing media in the wells every 2 days.
Chemical treatments
For BMS777607 administration to pregnant females, 0.05 mg/g of body weight (e.g., 1 mg BMS777607 for a 20 g mouse) in 70% PEG-300 in PBS was administered via oral gavage. Vehicle alone (1% DMSO in 70% PEG300 in PBS) was administered to control pregnant dams. For cell culture experiments, BMS777607 was used at 10 μM concentration in 0.001% DMSO, and vehicle controls were treated with 0.001% DMSO.
Cell growth, motility, and shape analysis
Proliferating myogenic cells were imaged using the ImageXPress Pico that took GFP, Tomato, and phase images every 3 hr of the entire PPF sample for 72 hr total. CellReporterXpress (Molecular Devices) software was then used to count GFP+ cells per time point to calculate growth of myogenic cells over time. To control for differences in initial number of myogenic cells per well, fold changes of cell growth were calculated by dividing each treatment by the initial cell number at time 0. For cell motility analysis of Pax3Cre/+; Rosa26nTnG/+ nuclei, GFP and phase images were taken every 8 min for 14 hr. Tracking, cell velocity, and displacement of peripheral cells (as an analog for the leading edge of the PPFs) were determined using TrackMate (Tinevez et al., 2017). For cell shape analysis on Pax3Cre/+; Rosa26mTmG/+ membranes, GFP and phase images were imaged every 4 min apart for 68 min total. Circularity of peripheral cells was analyzed in Fiji.
Tamoxifen injections and muscle injury
Pax7CreERt2/+Rosa26Pham/+ mice (Pham et al., 2012) were given five 2 mg doses (10 mg total) of tamoxifen (Cayman Chemical, 13258) (TAM) by intraperitoneal injection prior to injury. Barium chloride (25 μl 1.2% in sterile demineralized water) was injected into the tibialis anterior muscle of each mouse with a Hamilton syringe similar to Murphy et al., 2014.
FACS cell isolation and sorting
Isolation of mononuclear myogenic cells from adult tibialis anterior muscle was performed as described previously (Murphy et al., 2014). Tibialis anterior muscles were dissected, minced, and digested for 1 hr at 37°C in 100 μl of 5 mg/ml liberase (Sigma-Aldrich, 5401127001) and 25 μl of 10 U/µl DNAseI (Sigma-Aldrich, 4716728001) in 3 ml Ham’s F12 media (Thermo Fisher Scientific, 11765054). Samples were passed through 70 μm and 40 μm filters, centrifuged at 1800 rpm for 10 min, aspirated supernatant, and pellet resuspended in satellite cell growth media (15% horse serum [Gibco, 16050-122], 1:1000 50 mg/ml gentamicin [Thermo Fisher Scientific, 15750060] in F12 media). Myogenic mononuclear cells were isolated and sorted via GFP on Propel Labs Avalon (Bio-Rad).
Phospho flow cytometry
Sorted GFP+ cells were washed with Ham’s F12 media, centrifuged at 800 × g for 5 min at 4°C and aspirated supernatant. Cells were fixed with 1.5% PFA for 15 min at RT, washed with PBS, centrifuged at 800 × g for 5 min at 4°C, and supernatant was aspirated. The cell pellet was resuspended in 100% methanol cooled to –20°C, vortexed for 30 s, and incubated on ice for 30 min. Cells were washed with 0.5% BSA in PBS with sodium azide for 5 min, centrifuged at 800 × g for 5 min at 4°C, aspirated supernatant. Cells were washed with 0.5% BSA with sodium azide for 5 min, centrifuged at 800 × g for 5 min at 4°C, aspirated supernatant, and resuspended into 0.5% BSA with sodium azide. pMET conjugated to phycoerythrin (PE) (Cell Signaling, 12468) primary antibody 1:50 dilution was added to the samples for 20 min at RT. Samples were washed with PBS, centrifuged at 800 × g for 5 min at 4°C, aspirated supernatant, and resuspended in PBS for analysis on BD FACSCanto II HTS (BD Biosciences). Median PE frequency from FloJo was used for statistical analysis.
RNA extraction, cDNA synthesis, and quantitative polymerase chain reaction (qPCR)
The Quick-RNA Microprep Kit (Zymo, Irvine, CA) was used to extract total RNA according to the manufacturer’s protocol. Applied Biosystems High-Capacity RNA-to-cDNA kit (Thermo Fisher) was used to synthesize cDNA from purified RNA according to the manufacturer’s protocol. qRT-PCR was used to analyze expression of Met, Pax7, Hgf, MyoD1, and Gata4 using pre-validated primer sets (TaqMan, Thermo Fisher; Key resources table). 10 μl reaction volumes were prepared using TaqMan Fast Advanced Master Mix (Thermo Fisher). The following conditions were used for amplification: 20 s at 95°C followed by 40 cycles at 95°C for 1 s, and 60°C for 20 s. Gene expression levels were normalized against 18S ribosomal RNA for each sample and fold changes calculated using 2-Δ/ΔCt method (Schmittgen and Livak, 2008) by setting expression levels of each gene in DMSO-treated cell culture as 1. Data from three biological replicates were calculated and plotted as average fold changes with standard error of the mean (SEM).
Statistical analysis
Data are presented as ± SEM. For growth comparison between chemical treatments, repeated ANOVA analysis was run on the log2 fold change of GFP+ cells to normalize the distribution of cell growth over time. Unpaired two-tailed t-tests or one-way ANOVA were used for other statistical analyses.
Additional informationCompeting interests
No competing interests declared
No competing interests declared
Author contributions
Conceptualization, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
Resources (Figure 3 – figure supplement 2 Panels G and H)
Conceptualization, Resources, Data curation, Formal analysis, Supervision, Writing - original draft, Project administration, Writing - review and editing
Ethics
Experiments were performed in accordance with protocols (#1435) approved by the Institutional Animal Care and Use Committee at the University of Utah.
Additional filesData availability
Numerical and source data used to generate figures have been included in the Source Data file for Figures 1-7; Figure 1-figure supplement 1, Figure 3-figure supplement 1, Figure 3-figure supplement 2, Figure 3-figure supplement 1, Figure 4-figure supplement 1, Figure 5-figure supplement 1, and Figure 6-figure supplement 1.
Acknowledgements
We thank N Burns for critical reading of the manuscript and Y Wan for help with FluoRender analysis. We are grateful to P Pandey for technical assistance. qPCR experiments were performed by the University of Utah Genomics Core Facility and confocal imaging was performed at the University of Utah Cell Imaging Core with assistance of Xiang Wang. This work was supported by the National Institutes of Health R01HD087360 to GK, F32 HD093425 and 1K99HD101682 to EMS and Wheeler Foundation to GK.
ReferencesAckermanKGVargasSOWilsonJAJenningsRWKozakewichHPWPoberBR2012Congenital diaphragmatic defects: proposal for a new classification based on observations in 234 patientsPediatric and Developmental Pathology1526527410.2350/11-05-1041-OA.122257294AdachiNPascual-AnayaJHiraiTHiguchiSKurodaSKurataniS2018Stepwise participation of HGF/Met signaling in the development of migratory muscle precursors during vertebrate evolutionZoological Letters41810.1186/s40851-018-0094-y29946484AllanDWGreerJJ1997aDevelopment of phrenic motoneuron morphology in the fetal ratThe Journal of Comparative Neurology38246947910.1002/(SICI)1096-9861(19970616)382:4<469::AID-CNE4>3.0.CO;2-0AllanDWGreerJJ1997bEmbryogenesis of the phrenic nerve and diaphragm in the fetal ratThe Journal of Comparative Neurology38245946810.1002/(SICI)1096-9861(19970616)382:4<459::AID-CNE3>3.0.CO;2-1BabiukRPZhangWClugstonRAllanDWGreerJJ2003Embryological origins and development of the rat diaphragmThe Journal of Comparative Neurology45547748710.1002/cne.1050312508321BirchmeierCBirchmeierWGherardiEVande WoudeGF2003Met, metastasis, motility and moreNature Reviews. Molecular Cell Biology491592510.1038/nrm126114685170BladtFRiethmacherDIsenmannSAguzziABirchmeierC1995Essential role for the c-Met receptor in the migration of myogenic precursor cells into the limb budNature37676877110.1038/376768a07651534BogenschutzELSeftonEMKardonG2020Cell culture system to assay candidate genes and molecular pathways implicated in congenital diaphragmatic herniasDevelopmental Biology467303810.1016/j.ydbio.2020.07.01332827499BurdenSJ2011Snapshot: neuromuscular junctionCell14482610.1016/j.cell.2011.02.03721376240CampbellEJMAgostoniENewsom DavisJ1970The Respiratory Muscles: Mechanics and Neural ControlLondonLloye-LukeCarusoNHerberthBLamballeFArce-GorvelVMainaFHelmbacherF2014Plasticity versus specificity in RTK signalling modalities for distinct biological outcomes in motor neuronsBMC Biology125610.1186/s12915-014-0056-625124859CatonAHackerANaeemALivetJMainaFBladtFKleinRBirchmeierCGuthrieS2000The branchial arches and HGF are growth-promoting and chemoattractant for cranial motor axonsDevelopment1271751176610.1242/dev.127.8.175110725250ChungMIBujnisMBarkauskasCEKobayashiYHoganBLM2018Niche-mediated BMP/Smad signaling regulates lung alveolar stem cell proliferation and differentiationDevelopment145dev16301410.1242/dev.16301429752282ColvinJBowerCDickinsonJESokolJ2005Outcomes of congenital diaphragmatic hernia: a population-based study in Western AustraliaPediatrics116e356e36310.1542/peds.2004-284516140678DasenJSLiuJPJessellTM2003Motor neuron columnar fate imposed by sequential phases of hox-c activityNature42592693310.1038/nature0205114586461DasenJSTiceBCBrenner-MortonSJessellTM2005A Hox regulatory network establishes motor neuron pool identity and target-muscle connectivityCell12347749110.1016/j.cell.2005.09.00916269338DasenJSDe CamilliAWangBTuckerPWJessellTM2008Hox repertoires for motor neuron diversity and connectivity gated by a single accessory factor, FOXP1Cell13430431610.1016/j.cell.2008.06.01918662545DietrichSAbou-RebyehFBrohmannHBladtFSonnenberg-RiethmacherEYamaaiTLumsdenABrand-SaberiBBirchmeierC1999The role of SF/HGF and c-Met in the development of skeletal muscleDevelopment1261621162910.1242/dev.126.8.162110079225EbensABroseKLeonardoEDHansonMGBladtFBirchmeierCBarresBATessier-LavigneM1996Hepatocyte growth factor/scatter factor is an axonal chemoattractant and a neurotrophic factor for spinal motor neuronsNeuron171157117210.1016/s0896-6273(00)80247-08982163EnglekaKAGitlerADZhangMZhouDDHighFAEpsteinJA2005Insertion of Cre into the Pax3 locus creates a new allele of Splotch and identifies unexpected Pax3 derivativesDevelopmental Biology28039640610.1016/j.ydbio.2005.02.00215882581HainesLNeytCGautierPKeenanDGBryson-RichardsonRJHollwayGEColeNJCurriePD2004Met and HGF signaling controls hypaxial muscle and lateral line development in the zebrafishDevelopment1314857486910.1242/dev.0137415342468HelmbacherF2018Tissue-Specific activities of the Fat1 cadherin cooperate to control neuromuscular morphogenesisPLOS Biology16e200473410.1371/journal.pbio.200473429768404HirasawaTKurataniS2013A new scenario of the evolutionary derivation of the mammalian diaphragm from shoulder musclesJournal of Anatomy22250451710.1111/joa.1203723448284HirasawaTFujimotoSKurataniS2016Expansion of the neck reconstituted the shoulder-diaphragm in amniote evolutionDevelopment, Growth & Differentiation5814315310.1111/dgd.1224326510533HuhCGFactorVMSánchezAUchidaKConnerEAThorgeirssonSS2004Hepatocyte growth factor/c-met signaling pathway is required for efficient liver regeneration and repairPNAS1014477448210.1073/pnas.030606810115070743HutchesonDAZhaoJMerrellAHaldarMKardonG2009Embryonic and fetal limb myogenic cells are derived from developmentally distinct progenitors and have different requirements for beta-cateninGenes & Development23997101310.1101/gad.176900919346403IrishMSHolmBAGlickPL1996Congenital diaphragmatic hernia. A historical reviewClinics in Perinatology2362565310.1016/s0095-5108(18)30199-48982561IsabellaAJBarshGRStonickJADubrulleJMoensCB2020Retinoic acid organizes the zebrafish vagus motor topographic MAP via spatiotemporal coordination of HGF/Met signalingDevelopmental Cell5334435710.1016/j.devcel.2020.03.01732302545JansenAHChernickV1991Fetal breathing and development of control of breathingJournal of Applied Physiology701431144610.1152/jappl.1991.70.4.14312055820JungHLacombeJMazzoniEOLiemKFGrinsteinJMahonySMukhopadhyayDGiffordDKYoungRAAndersonKVWichterleHDasenJS2010Global control of motor neuron topography mediated by the repressive actions of a single Hox geneNeuron6778179610.1016/j.neuron.2010.08.00820826310KardonGAckermanKGMcCulleyDJShenYWynnJShangLBogenschutzESunXChungWK2017Congenital diaphragmatic hernias: from genes to mechanisms to therapiesDisease Models & Mechanisms1095597010.1242/dmm.02836528768736KellerCHansenMSCoffinCMCapecchiMR2004Pax3: FKHR interferes with embryonic Pax3 and Pax7 function: implications for alveolar rhabdomyosarcoma cell of originGenes & Development182608261310.1101/gad.124390415520281LamballeFGenestineMCarusoNArceVRichelmeSHelmbacherFMainaF2011Pool-specific regulation of motor neuron survival by neurotrophic supportThe Journal of Neuroscience31111441115810.1523/JNEUROSCI.2198-11.201121813676LiXMDongXPLuoSWZhangBLeeDHTingAKLNeiswenderHKimCHCarpenter-HylandEGaoTMXiongWCMeiL2008Retrograde regulation of motoneuron differentiation by muscle beta-cateninNature Neuroscience1126226810.1038/nn205318278041LoganMMartinJFNagyALobeCOlsonENTabinCJ2002Expression of Cre recombinase in the developing mouse limb bud driven by a prxl enhancerGenesis33778010.1002/gene.1009212112875LongoniMHighFARussellMKKashaniATracyAAColettiCMHilaRShamiaAWellsJAckermanKGWilsonJMBultCJLeeCLageKPoberBRDonahoePK2014Molecular pathogenesis of congenital diaphragmatic hernia revealed by exome sequencing, developmental data, and bioinformaticsPNAS111124501245510.1073/pnas.1412509111MainaFCasagrandaFAuderoESimeoneAComoglioPMKleinRPonzettoC1996Uncoupling of Grb2 from the Met receptor in vivo reveals complex roles in muscle developmentCell8753154210.1016/s0092-8674(00)81372-08898205MainaFKleinR1999Hepatocyte growth factor, a versatile signal for developing neuronsNature Neuroscience221321710.1038/631010195212MerrellAJKardonG2013Development of the diaphragm -- a skeletal muscle essential for mammalian respirationThe FEBS Journal2804026403510.1111/febs.1227423586979MerrellAJEllisBJFoxZDLawsonJAWeissJAKardonG2015Muscle connective tissue controls development of the diaphragm and is a source of congenital diaphragmatic herniasNature Genetics4749650410.1038/ng.325025807280MurphyMMLawsonJAMathewSJHutchesonDAKardonG2011Satellite cells, connective tissue fibroblasts and their interactions are crucial for muscle regenerationDevelopment1383625363710.1242/dev.06416221828091MurphyMMKeefeACLawsonJAFlygareSDYandellMKardonG2014Transiently active Wnt/β-catenin signaling is not required but must be silenced for stem cell function during muscle regenerationStem Cell Reports347548810.1016/j.stemcr.2014.06.01925241745MuzumdarMDTasicBMiyamichiKLiLLuoL2007A global double-fluorescent CRE reporter mouseGenesis4559360510.1002/dvg.2033517868096NassariSDuprezDFournier-ThibaultC2017Non-Myogenic contribution to muscle development and homeostasis: the role of connective tissuesFrontiers in Cell and Developmental Biology52210.3389/fcell.2017.0002228386539OkamotoEMoriyamaYKurakuSKaiK-ITanakaM2019Involvement of HGF/Met signaling in appendicular muscle development in cartilaginous fishDevelopment, Growth & Differentiation619710310.1111/dgd.1259130644548OrganSLTsaoMS2011An overview of the c-Met signaling pathwayTherapeutic Advances in Medical Oncology3S7S1910.1177/175883401142255622128289PerrySFSimilowskiTKleinWCoddJR2010The evolutionary origin of the mammalian diaphragmRespiratory Physiology & Neurobiology17111610.1016/j.resp.2010.01.00420080210PhamAHMcCafferyJMChanDC2012Mouse lines with photo-activatable mitochondria to study mitochondrial dynamicsGenesis5083384310.1002/dvg.2205022821887PhaneufDMoscioniADLeClairCRaperSEWilsonJM2004Generation of a mouse expressing a conditional knockout of the hepatocyte growth factor gene: demonstration of impaired liver regenerationDNA and Cell Biology2359260310.1089/dna.2004.23.59215383179PhilippidouPWalshCMAubinJJeannotteLDasenJS2012Sustained hox5 gene activity is required for respiratory motor neuron developmentNature Neuroscience151636164410.1038/nn.324223103965PoberBR2007Overview of epidemiology, genetics, birth defects, and chromosome abnormalities associated with CDHAmerican Journal of Medical Genetics. Part C, Seminars in Medical Genetics145C15817110.1002/ajmg.c.3012617436298PriggeJRWileyJATalagoEAYoungEMJohnsLLKundertJASonstengKMHalfordWPCapecchiMRSchmidtEE2013Nuclear double-fluorescent reporter for in vivo and ex vivo analyses of biological transitions in mouse nucleiMammalian Genome24e946910.1007/s00335-013-9469-824022199RavalMVWangXReynoldsMFischerAC2011Costs of congenital diaphragmatic hernia repair in the United states-extracorporeal membrane oxygenation foots the billJournal of Pediatric Surgery4661762410.1016/j.jpedsurg.2010.09.04721496527RiddleRDJohnsonRLLauferETabinC1993Sonic hedgehog mediates the polarizing activity of the ZPACell751401141610.1016/0092-8674(93)90626-28269518RoussoDLGaberZBWellikDMorriseyEENovitchBG2008Coordinated actions of the forkhead protein FOXP1 and Hox proteins in the columnar organization of spinal motor neuronsNeuron5922624010.1016/j.neuron.2008.06.02518667151SchmittgenTDLivakKJ2008Analyzing real-time PCR data by the comparative CT methodNature Protocols31101110810.1038/nprot.2008.73SchroederGMAnYCaiZ-WChenX-TClarkCCorneliusLAMDaiJGullo-BrownJGuptaAHenleyBHuntJTJeyaseelanRKamathAKimKLippyJLombardoLJManneVOppenheimerSSackJSSchmidtRJShenGStefanskiKTokarskiJSTrainorGLWautletBSWeiDWilliamsDKZhangYZhangYFargnoliJBorzilleriRM2009Discovery of N- (4- (2-amino-3-chloropyridin-4-yloxy) -3-fluorophenyl) -4-ethoxy-1- (4-fluorophenyl) -2-oxo-1,2-dihydropyridine-3-carboxamide (BMS-777607), a selective and orally efficacious inhibitor of the Met kinase superfamilyJournal of Medicinal Chemistry521251125410.1021/jm801586s19260711SeftonEMGallardoMKardonG2018Developmental origin and morphogenesis of the diaphragm, an essential mammalian muscleDevelopmental Biology440647310.1016/j.ydbio.2018.04.01029679560SeftonEMKardonG2019Connecting muscle development, birth defects, and evolution: an essential role for muscle connective tissueCurrent Topics in Developmental Biology13213717610.1016/bs.ctdb.2018.12.00430797508SonnenbergEMeyerDWeidnerKMBirchmeierC1993Scatter factor/hepatocyte growth factor and its receptor, the c-Met tyrosine kinase, can mediate a signal exchange between mesenchyme and epithelia during mouse developmentThe Journal of Cell Biology12322323510.1083/jcb.123.1.2238408200SorianoP1999Generalized lacZ expression with the ROSA26 CRE reporter strainNature Genetics21707110.1038/50079916792TakahashiTFriedmacherFZimmerJPuriP2016Decreased expression of hepatocyte growth factor in the nitrofen model of congenital diaphragmatic herniaPediatric Surgery International3296797310.1007/s00383-016-3944-827480986TangSHESilvaFJTsarkWMKMannJR2002A cre/loxp-deleter transgenic line in mouse strain 129S1/svimjGenesis3219920210.1002/gene.1003011892008TinevezJYPerryNSchindelinJHoopesGMReynoldsGDLaplantineEBednarekSYShorteSLEliceiriKW2017TrackMate: an open and extensible platform for single-particle trackingMethods115809010.1016/j.ymeth.2016.09.01627713081TorfsCPCurryCJBatesonTFHonoréLH1992A population-based study of congenital diaphragmatic herniaTeratology4655556510.1002/tera.14204606051290156UetaniNChagnonMJKennedyTEIwakuraYTremblayML2006Mammalian motoneuron axon targeting requires receptor protein tyrosine phosphatases sigma and deltaThe Journal of Neuroscience265872588010.1523/JNEUROSCI.0386-06.200616738228VoganKJEpsteinDJTraslerDGGrosP1993The splotch-delayed (SPD) mouse mutant carries a point mutation within the paired box of the Pax-3 geneGenomics1736436910.1006/geno.1993.13338406487WanYOtsunaHChienCBHansenC2009An interactive visualization tool for multi-channel confocal microscopy data in neurobiology researchIEEE Transactions on Visualization and Computer Graphics151489149610.1109/TVCG.2009.11819834225WangZZWashabaughCHYaoYWangJMZhangLOntellMPWatkinsSCRudnickiMAOntellM2003Aberrant development of motor axons and neuromuscular synapses in myod-null miceThe Journal of Neuroscience235161516910.1523/JNEUROSCI.23-12-05161.200312832540YamamotoYLivetJPollockRAGarcesAArceVdeLapeyrièreOHendersonCE1997Hepatocyte growth factor (HGF/SF) is a muscle-derived survival factor for a subpopulation of embryonic motoneuronsDevelopment1242903291310.1242/dev.124.15.29039247333YumotoNKimNBurdenSJ2012Lrp4 is a retrograde signal for presynaptic differentiation at neuromuscular synapsesNature48943844210.1038/nature1134822854782ZawadzkaMRiversLEFancySPJZhaoCTripathiRJamenFYoungKGoncharevichAPohlHRizziMRowitchDHKessarisNSuterURichardsonWDFranklinRJM2010CNS-resident glial progenitor/stem cells produce Schwann cells as well as oligodendrocytes during repair of CNS demyelinationCell Stem Cell657859010.1016/j.stem.2010.04.0022056969510.7554/eLife.74592.sa0Editor's evaluationBirchmeier-KohlerCarmenReviewing Editorhttps://ror.org/04p5ggc03Max Delbrueck Center for Molecular Medicine (MDC) in the Helmholtz SocietyGermany
It was previously known that HGF and Met control delamination and migration of muscle progenitor cells that colonize the diaphragm. The article by Sefton and coworkers confirms and extends these observations using conditional mouse lines in which the HGF gene was targeted by Cre/loxP recombination, and Met inhibitors that are applied at different stages of development. Together these new data show that HGF derived from the pleuroperitoneal folds is directly required for the recruitment of Met+ muscle progenitors to the diaphragm and continues to be essential for the survival of a pool of progenitors that eventually form the diaphragm muscle. Moreover, the authors show the effects of the diaphragm muscle on the development of the phrenic nerve that innervates this muscle, in particular, in the absence of muscle deficits in the branching of the phrenic nerve are observed. Overall, the technical quality of the data on diaphragm muscle development and its effect on the branching of the phrenic nerve are excellent.
10.7554/eLife.74592.sa1Decision letterBirchmeier-KohlerCarmenReviewing Editorhttps://ror.org/04p5ggc03Max Delbrueck Center for Molecular Medicine (MDC) in the Helmholtz SocietyGermanyBirchmeier-KohlerCarmenReviewerhttps://ror.org/04p5ggc03Max Delbrueck Center for Molecular Medicine (MDC) in the Helmholtz SocietyGermanyBraunThomasReviewerhttps://ror.org/0165r2y73Max-Planck-Institute for Heart and Lung ResearchGermany
Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.
Decision letter after peer review:
Thank you for submitting your article "Fibroblast-derived HGF integrates muscle and nerve development during morphogenesis of the mammalian diaphragm" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Carmen Birchmeier-Kohler as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Didier Stainier as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Thomas Braun (Reviewer #2).
The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.
Summary
Sefton et al., analyze 'Fibroblast-derived HGF integrates muscle and nerve development during morphogenesis of the mammalian diaphragm'. The role of Met in the development of the muscle has been previously investigated by many papers, and roles in delamination of progenitor cells form the somites, in migration, survival, proliferation and differentiation have been described. The novel finding of the paper is a in depth analysis of the development of the diaphragm muscle, and the role of Met and HGF in the process. The authors provide evidence for the expression of HGF in pleuroperitoneal folds and for its requirement for muscle progenitor expansion and maintenance during diaphragm muscle formation. The reviewers found the manuscript clearly written and the experimental approach overall of high standard. However, there were several deficiencies noted that need to be addressed. The data supporting that HGF/Met function during development of the phrenic nerve was found to be less well supported and requiring additional experimental data to be included in the manuscript.
Essential revision
1) A major concern is the limited data on the role of Met in the development of the phrenic nerve. While it is well documented that HGF acts as a trophic factor for motor neurons in culture, its role in development of motor neurons has been highly debated. Moreover, careful genetic analyses previously demonstrated indirect mechanisms of Met during motor neuron development. Despite the weakness of the data on the role of Met in phrenic development, the role of HGF/Met in is strongly emphasized in abstract /intro/discussion. Data relying on motor neurons specific ablation of met need to be included in order to strengthen this point. Additional suggestions how to improve the analysis of phrenic development in Met-/+ animals are provided in the individual reviews appended below.
2) The authors should exclude a connective tissue phenotype in conditional HGF/Met mutants. Why is Prx1Cre used instead of PDGFRaCre to trace PPFs? In addition, the reason for the differences in phenotypes regarding dorsal/ventral diaphragm muscles after either HGF inactivation or Met-inhibitor treatment should be explained, or at least discussed. Is this due to timing of the mutation/inhibition, or to efficacy of ablation/inhibition?
3) The authors propose a role of Met in myogenic commitment, based on the co-culture experiments. Myogenic commitment implies a major role in myogenic lineage progression or differentiation. The evidence for that claim is weak and the reduced differentiation might be the consequence of increased cell death, reduced proliferation or other causes that change myoblast density. It is mandatory to address this issue appropriately, by analyzing both, fibroblast and myoblast cell numbers, proliferation, apoptosis, myogenic differentiation (i.e. ratio of MyoD/MyoG-positive cells).
4) A potentially interesting conclusion form the manuscript is hat hernia does only develop when both, connective tissue and muscle of the diaphragm are affected. Although this explanation is intriguing, the authors have to make sure that body wall muscles are fully developed, to exclude that the abdominal pressure is not reduced in mice conditional Met/HGF mutations. The images in Figures 1 and 2 seem to indicate normal development of body wall muscles but a dedicated statement in this respect would be helpful.
5) Treatment of WT embryos from E7.5 will impact on myogenic progenitor delamination, and this is also expected for Pax3CreKIMetnull/fl mutation. The authors have to dissociate the Met-related somite delamination, a migration of the progenitors to the diaphragm and the potential additional role of Met signaling in the colonization and expansion of myogenic progenitor cells in the anlage of the diaphragm.
6) Validation for the drug BMS777607 effectiveness in inhibiting MET downstream pathway should be performed and shown.
7) Comments referring to missing controls, scale bars, statistics need to be addressed.
Additional points: The authors should address wherever possible the other comments raised by the three reviewers (see below).
Reviewer #1 (Recommendations for the authors):
It was previously shown that HGF and Met controls development of the diaphragm muscle. In particular, the signal induces delamination and migration of muscle progenitor cells that colonize the diaphragm. The present manuscript by Sefton and coworkers confirms and extends these observations using (i) conditional mouse lines in which the HGF gene was targeted by Cre/loxP recombination in the pleuroperitoneal folds (Prx1-cre) and at other sites PdgfraCreERT2, and of (ii) Met inhibitors. Overall, the technical quality of the data on diaphragm muscle development is excellent; the conceptual advance over previous work is not exceptional; the evidence for Met/HGF-dependent development of the phrenic nerve is marginal and needs to be strengthened.
The data show that fibroblasts provide HGF signals received by Met in muscle progenitor cells that is essential for diaphragm development. The PdgfraCreERT2 line was used to demonstrate that HGF produced by fibroblasts but not by muscle progenitors is essential for diaphragm development. Moreover, development of dorsal and ventral regions of diaphragm muscle requires continuous MET signaling. Thus, HGF is not only required for the delamination of progenitors, but also for proliferation and survival of those muscle progenitors that reached the anlage of the diaphragm.
My major concern is the limited data on the HGF-dependent development of the phrenic nerve (defasciculation). While it is well documented that HGF acts as a trophic factor for motor neurons in culture, its role in development of motor neurons was highly debated due to the fact that some changes observed in Met or HGF mutant mice in vivo are also present in other mutants that lack the muscle groups derived from migrating muscle progenitors. Moreover, careful genetic analyses previously demonstrated indirect mechanisms of Met during motor neuron development, i.e. a non-cell-autonomous function of Met during the recruitment of motor neurons to PEA3-positive motor pools (Helmbacher et al., Neuron 2003).
Sefton et al., provide an analysis of a single time point, one histological picture (3G, magnified in 3H) that indicate that in Met+/- animals defasciculation of the phrenic nerve does not occur correctly. This is accompanied by a quantification that barely reaches significance (Figure 3K). Data shown in Figure 7 using Met inhibitors show a major change in phrenic nerve branching, which is presumably due to the major change in diaphragm development, as conceded by the authors.
Despite this weakness on the experimental side, the role of HGF/Met in phrenic nerve development is strongly emphasized in abstract /intro/discussion (e.g. line 414: However, PPF-derived HGF is crucial for the defasciculation and primary branching of the nerve, independent of muscle). The data need to be strengthened in order to conclude that HGF coordinates both, diaphragm muscle and phrenic development. I expect that the defasciculation of the phrenic nerve is highly dependent on the developmental stage. The authors should provide data that show different stages of phrenic nerve development, i.e. the time course of the of defasciculation in wildtype animals, explain how the staging was done, and compare different stages in the Met mutants and analyze whether the defasciculation is resolved at later stages. Met mutations also affect placental development, resulting in developmental delays that in turn might lead to an apparent small change in the time course of defasciculation. The authors should exclude that indirect effects cause the small change in phrenic nerve morphology, for instance by examining conditional Met mutations that are restricted to motor neurons. Good Cre lines that target motor neurons are available.
Reviewer #2 (Recommendations for the authors):
Since the authors observed a correlation between loss of should muscle and loss of diaphragm muscularization, which is related to the timing of migration of shoulder muscle and diaphragm muscle progenitors, they claim a "closer relationship" between should muscles and diaphragm, which was further extended in the discussion. I was not convinced by this conclusion. Is there a "closer relationship" between the muscles just because the progenitor cells migrate roughly at the same time? They authors may modify or delete this statement, although I agree that a broader expression of HGF may facilitate enhanced recruitment of muscle progenitor cells, required for formation of the diaphragm.
Based on co-cultures of PPFs and myoblasts the authors describe a function of MET in myogenic commitment. In my opinion the evidence for such a function is weak. The authors observed no change in proliferation but reduced motility and a higher rate of apoptosis after pharmacological inhibition of MET. Increased apoptosis and reduced aggregation of myoblasts at distinct locations may easily interfere with myogenic differentiation. A general function of MET in myogenic commitment does not seem very likely, since myoblasts that do not undergo prior long-range migration differentiate normally in the absence of Met.
Pharmacological inhibition of MET increases the rate of apoptosis in numerous cell types, which has been studied extensively in cancer. It seems appropriate to explore the mechanism of increased apoptosis in myoblasts following MET inhibition more closely. Previous reports suggest increased expression of p53, increased sensitivity to Fas-mediated apoptosis or increased autophagy, among others, as potential causes for increased apoptosis after MET inhibition. Which pathway is relevant in myoblasts?
Obviously, it would be great to learn more about the mechanisms that control Hgf expression in fibroblasts within and derived from PPTs. Such knowledge may also help to better understand the reasons leading to Congenital Diaphragmatic Hernias (CDH). Unfortunately, the authors did not to go any further in this direction.
The authors observed that branching of the phrenic nerve was reduced in heterozygous Met mutants with normal diaphragm musculature, suggesting a direct role of Met in phrenic nerve branching. Timed inactivation of Met specifically in motoneurons would greatly increase the impact of this finding and allow a more specific analysis of the role of MET in phrenic nerve development and branching.
Surprisingly, the authors did not observe CDH in mutant mice with muscle-less diaphragms, from which they conclude that additional defects in the connective tissue are necessary to allow hernia formation. Although this explanation is intriguing, the authors have to make sure that abdominal pressure is not reduced in mice without muscularized diaphragms, e.g. demonstrate that body wall muscles are fully functional. The images in Figures 1 and 2 seem to indicate normal development of body wall muscles but a dedicated statement in this respect would be helpful.
Scale bars are missing in some panels.
According to the methods part, Student's t-test was used for the statistical analysis shown in Figure 3. A pairwise comparison of WT, heterozygous and homozygous Met mutants is not appropriate when all three genotypes are compared with each other. An ANOVA test should be used as in Figure 5F.
Reviewer #3 (Recommendations for the authors):
The present manuscript addresses questions on the role of HGF/MET signaling in diaphragm formation once myogenic progenitors have already migrated to the PPFs. In addition, it identifies the PPFs as the source of HGF. The study is interesting and developed with rigor. However, the role of HGF is not clearly dissociated from the presence/absence of the fibroblasts/connective tissue itself. Also, the authors do not link in the results, for example, how the muscleless diaphragms and HGF itself relate to the hernia phenotype mentioned in the abstract.
Figure 1
1C) Co-staining for Hgf (PPFs) and Met (migrating progenitors) should be provided for a clear visualization of ligand-expressing cells and receptor-expressing cells.
1F) Co-staining for Met and migrating progenitors (PAX3 or LBX1 for example) should be provided for a clear visualization of migrating progenitors versus general Met expression in other cell types.
1M) The claim that fibroblast-derived HGF is required for diaphragm muscle development cannot be addressed with only this experimental analysis. What is the connective tissue phenotype in PDGFaCreER;HGFnull/fl embryos? Lack of connective tissue could affect muscle migration and development rather than HGF expression on its own. Are fibroblasts still present?
1G, I, K, M) Control genotypes should be analyzed and added in the Figure or as Supplementary Data.
1I) Pax3CreKIMetnull/fl originates a muscleless diaphragm but this could be associated with lack of delamination and migration from the somites rather than a specific MET requirement for diaphragm muscle formation once progenitors have colonized this area as suggested by the authors.
Figure 3
It is not clear whether HGF controls phrenic nerve formation independently of muscle. In the end of this section the authors mention that Met heterozygous embryos have normal muscles referring to Figure 1K (which is not a picture referring to a Met het embryo). The authors should confirm if in Met null embryos there is a direct effect on the nerve bifurcation and brunching or if the lack of muscle is leading to this observation. Is the phenotype dose dependent for muscle formation? This is not properly shown by the authors. Could the authors perform a conditional KO of Met in the nerves?
This concern is further supported by the Figure 4 data where phrenic nerve only extends to regions with muscle.
Figure 4
The role of HGF versus the presence of fibroblasts/connective tissue remains vague. Is the lack of ventral muscles associated with lack of migration of progenitors within the forming diaphragm towards an HGF source? Or due to the lack of connective tissue scaffold itself? Why is Prx1Cre used instead of PDGFRaCre (the one used in the actual experiments) to trace PPFs?
4E) Please provide a control with a Cre allele to compare putative secondary effects for the presence of the Cre in the cKO embryos.
Figure 5
Validation for the drug BMS777607 effectiveness in inhibiting MET downstream pathway should be performed and shown.
The phenotype with the inhibition of Met signaling (lack of dorsal muscles in addition to ventral) is distinct to the one observed in Figure 4 (less ventral muscle formation in PDGFaCreER;HGFnull/fl). How do the authors explain the dorsal phenotype when inhibiting Met (since migration should have not been affected at this time-point) and Hgf inhibition in Figure 4 is not leading to this phenotype?
Figure 6
What is the time point after treatment in B and C? What is the total nuclei number in the cultures (since these are co-cultures)?
The authors should provide pictures together with nuclear staining for a global view of the cell density in the cultures. Also T0 time point pictures should be added for comparison, in particular in C.
In these experiments the authors performed co-culture (fibroblasts + myogenic progenitors) derived from PPfs. However, fibroblast phenotype is not addressed. If there is less GFP+ cells is there more fibroblasts in the culture? Could this impact on the phenotype observed and linked by the authors to the Met inhibitor treatment? The only piece of evidence is Gata4 qPCR which is not sufficient to address fibroblast phenotype in the culture.
Figure 7
Treatment of WT embryos from E7.5 will impact on myogenic progenitor migration. The authors have to dissociate the Met-related somite migration phenotype and the potential additional role of Met signaling in the diaphragm muscle formation.
[Editors’ note: further revisions were suggested prior to acceptance, as described below.]
Thank you for resubmitting your work entitled "Fibroblast-derived HGF controls recruitment and expansion of muscle during morphogenesis of the mammalian diaphragm" for further consideration by eLife. Your revised article has been evaluated by Didier Stainier (Senior Editor) and a Reviewing Editor.
The manuscript has been much improved but there are a few remaining issues that need to be addressed, as outlined below:
Reviewer #1 (Recommendations for the authors):
The majority of my concerns were appropriately addressed by the authors. There are a few additional points that need to be addressed, and most of these concern the text/wording.
Unfortunately, the changes in the text have not been highlighted in red in the uploaded version of the manuscript, as it was mentioned in the rebuttal letter.
1. While the quality of the figures is overall good, many of the panels in Figure 2 are very dark. It should be easy to modify this.
2. While the authors show very convincingly that the effects on the phrenic nerve observed are caused indirectly by the loss of muscle, and not directly by the loss of HGF/Met signaling, this is not always made clear in the text. Furthermore, the mechanism that causes branching deficits should be clearly stated in the Abstract.
Line 41-42...and indirectly required for phrenic nerve primary branching.
Please mention the specific indirect mechanisms, i.e. via the effect on muscle development.
Line 81…HGF is also critical for innervation.
The reports cited do not distinguish between the effect of the nerve/muscle or report in vitro experiments. The text should take this into account.
238: Thus, PPF-derived HGF is necessary for phrenic nerve defasciculation.
Additional experiments shown in the next paragraph indicate that it is the absence of the muscle that causes defasciculation. Please rephrase.
3) Met inhibition using BMS777607
BMS777607 acts as an AXL, RON and Met tyrosine kinase inhibitor. This should be mentioned. Are effects of RON/AXL on skeletal muscle development described?
4) Antibody specificity: Antibody specificity should be tested by analysis of the phrenic nerve on Met mutants.
Reviewer #2 (Recommendations for the authors):
Sefton et al., have submitted a revised version of a study, in which the role of fibroblast-derived HGF for recruitment and expansion of muscle during morphogenesis of the mammalian diaphragm was investigated. The authors have changed the title to cope with new findings, indicating that reduced primary branching of the phrenic nerve in Met-mutants is not due direct effects of HGF on the nerve but most likely caused by indirect effects resulting from reduced muscle formation in Met-mutants.
The authors did an excellent job to deal with the reviewers' criticisms. In particular, by specifically deleting Met in motoneurons and by analysis of splotch mutants they demonstrate that Met does not intrinsically regulate phrenic nerve branching. Analysis of splotch mice, which show normal expression of HGF PPFs but display muscle-less diaphragms, revealed reduced branching of the phrenic nerve, similar to Met-mutants, clearly suggesting a critical role of muscle fibers in the diaphragm for phrenic nerve branching. Furthermore, Sefton et al., now demonstrate that fibroblasts expand and populate the diaphragm in the absence of muscle, which essentially excluded a connective tissue phenotype in conditional Hgf/Met mutants. They also clarified the statement about what I understood was meant to claim a role of Hgf/Met in myogenic commitment. Additional controls (effectiveness of BMS777607 for inhibition of MET, treatment with BMS777607 at additional timepoints) were done as requested, providing additional interesting insights into a role of Met for ventral expansion of the diaphragm at relatively late developmental timepoints.
The authors argue that "PPF-derived HGF, via muscle, controls phrenic nerve defasciculation". Well, this is formally correct but exaggerates the role PPF-derived for phrenic nerve defasciculation in my view. Probably, deletion of any gene that prevents diaphragm muscle formation will have similar effects on phrenic nerve defasciculation. I would prefer a more neutral statement: "Muscle formation, requiring PPF-derived HGF, controls phrenic nerve defasciculation", or something similar. The authors may consider modifying the statement.
I do not have any further objections.
10.7554/eLife.74592.sa2Author response
Essential Revision
1) A major concern is the limited data on the role of Met in the development of the phrenic nerve. While it is well documented that HGF acts as a trophic factor for motor neurons in culture, its role in development of motor neurons has been highly debated. Moreover, careful genetic analyses previously demonstrated indirect mechanisms of Met during motor neuron development. Despite the weakness of the data on the role of Met in phrenic development, the role of HGF/Met in is strongly emphasized in abstract /intro/discussion. Data relying on motor neurons specific ablation of met need to be included in order to strengthen this point. Additional suggestions how to improve the analysis of phrenic development in Met-/+ animals are provided in the individual reviews appended below.
In response to comments from the reviewers, we have more thoroughly investigated the role of Met in the development of the phrenic nerve and include two new sets of genetic experiments. In our first submission, we found a decreased number of phrenic nerve branches at E11.5 in Met Δ/ Δ and Met Δ/+ compared with Met+/+ embryos. In the Met Δ/ Δ embryos, no muscle is present in the diaphragm. Therefore, the greatly reduced branching in these embryos is likely a secondary effect of the requirement of Met in muscle progenitors for diaphragm muscularization. Of particular interest is the reduced branching in the Met Δ/+ embryos. Because the diaphragm is muscularized in these embryos, this suggested that Met may be required intrinsically in the phrenic nerve. One reviewer suggested that the reduced branching in the Met Δ/+ embryos could be due to a developmental delay in the whole embryo. However, we found that Met Δ/ Δ and Met Δ/+ embryos are not overall delayed relative to Met+/+ embryos (as measured by crown rump length or limb length; Figure 3—figure supplement 1). Also, to increase the robustness of these data, we added additional embryos to the analysis. We then extended our analysis of Met Δ/ Δ, Met Δ/+ and Met+/+ embryos to E12.5 (Figure 3—figure supplement 1) to see whether the branching phenotype persisted; we found that while the of Met Δ/ Δ embryos continue to have very few branches, the number of branches in Met Δ/+ embryos recovers and matches that of Met+/+ embryos.
To explicitly test whether Met is required within the phrenic nerve, we used Olig2Cre/+to conditionally delete Met. This line was chosen for its early expression in motor neurons (Zawadzka et al., 2010). We examined Olig2Cre/+;MetΔ/flox embryos compared to Olig2Cre/+; Metflox/+ embryos. We chose to include Olig2Cre in our controls because the Olig2Cre is a knock-in/knock-out and Olig2 has important roles in nerve development. However, deletion of Met did not affect the number of branches at E11.5 (Figure 3—figure supplement 2) or E12.5 (data not shown). These data suggest that Met does not intrinsically regulate phrenic nerve branching. This suggests that PPF-derived HGF regulates phrenic nerve branching indirectly via muscle. To test if HGF is sufficient to promote early stages of nerve branching in the absence of muscle, we turned to Pax3SpD/SpD mutants in which a point mutation in Pax3 prevents migration of muscle progenitors into the diaphragm (Figure 3—figure supplement 2). In these embryos, the diaphragm is muscleless, but the PPFs still express HGF. In these diaphragms the number of branches at E11.5 is severely reduced. These data demonstrate that in the absence of muscle the presence of HGF in the PPF fibroblasts is not sufficient to support diaphragm branching.
Altogether our data demonstrate that PPF-derived HGF, via its regulation of muscle, controls the primary branching of phrenic nerve. The Met Δ/+ data demonstrate that Met controls phrenic nerve branching at E11.5 in a dose-dependent manner, but this effect is lost by E12.5. Although we see no obvious defects in muscle of Met Δ/+ diaphragms at later stages, the most parsimonious explanation of the reduced phrenic nerve branching at E11.5 is that this is due to fewer muscle progenitors at this time point.
We thank the reviewers for prompting us to look at the role of HGF/Met in the phrenic nerve more closely. Our revised conclusions are presented in the Results and Discussion. We show that PPF-derived HGF is critical for integrating both muscle and phrenic nerve development, but now demonstrate that HGF’s regulation of phrenic nerve branching is via muscle, which is well-known to express multiple trophic factors required by motor neurons.
2) The authors should exclude a connective tissue phenotype in conditional HGF/Met mutants. Why is Prx1Cre used instead of PDGFRaCre to trace PPFs? In addition, the reason for the differences in phenotypes regarding dorsal/ventral diaphragm muscles after either HGF inactivation or Met-inhibitor treatment should be explained, or at least discussed. Is this due to timing of the mutation/inhibition, or to efficacy of ablation/inhibition?
Prx1Cre and PDGFRaCreER both label connective tissue fibroblasts in the developing diaphragm. We have included additional panels to Figure 4—figure supplement 1 of PDGFRaCreER/+; HGFdelta/flox; RosamTmG/+embryos demonstrating the presence of GFP+ fibroblasts in the muscleless regions. Thus, fibroblasts do expand and appropriately populate the diaphragm even in the absence of muscle.
The phenotype after genetic deletion of HGF via Prx1Cre or PDGFRaCreER and after MET-inhibitor treatment is quite similar (see slightly revised model in Figure 8). Genetic HGF deletion or MET inhibition early leads to ventral muscleless regions and two bilateral dorsal muscleless regions (Figures 4 C, H and 5 B-D). Genetic HGF deletion or MET inhibition later consistently leads to ventral muscleless regions and less commonly to dorsal muscleless regions (Figures 4 G and 5 E). We have provided additional clarification in the text and in model in Figure 8.
3) The authors propose a role of Met in myogenic commitment, based on the co-culture experiments. Myogenic commitment implies a major role in myogenic lineage progression or differentiation. The evidence for that claim is weak and the reduced differentiation might be the consequence of increased cell death, reduced proliferation or other causes that change myoblast density. It is mandatory to address this issue appropriately, by analyzing both, fibroblast and myoblast cell numbers, proliferation, apoptosis, myogenic differentiation (i.e. ratio of MyoD/MyoG-positive cells).
We apologize for the misunderstanding here and have altered the text to indicate that we do not propose a role for Met in myogenic commitment, but rather that Met regulates the number of MyoD+ cells by promoting their survival. Using our co-culture methods, unfortunately we are unable to quantify fibroblast cell numbers, as they are too dense to distinguish using either our software or manual counting.
4) A potentially interesting conclusion form the manuscript is hat hernia does only develop when both, connective tissue and muscle of the diaphragm are affected. Although this explanation is intriguing, the authors have to make sure that body wall muscles are fully developed, to exclude that the abdominal pressure is not reduced in mice conditional Met/HGF mutations. The images in Figures 1 and 2 seem to indicate normal development of body wall muscles but a dedicated statement in this respect would be helpful.
We have added a statement in the Results section indicating the body wall muscles are normal.
5) Treatment of WT embryos from E7.5 will impact on myogenic progenitor delamination, and this is also expected for Pax3CreKIMetnull/fl mutation. The authors have to dissociate the Met-related somite delamination, a migration of the progenitors to the diaphragm and the potential additional role of Met signaling in the colonization and expansion of myogenic progenitor cells in the anlage of the diaphragm.
In Figure 5, we have delivered the Met inhibitor at various stages to determine when Met is required. Delivery at E7.5-E12.5 gives the most pronounced phenotype; muscleless regions in the dorsal and ventral regions of the diaphragm. To exclude that inhibition of Met is affecting the initial delamination of progenitors from the somites and migration to the PPFs, we also delivered the inhibitor at E11.5-E12.5 (Figure 5E). Previously we showed that migration to the PPFs is completed by E11.5 (Sefton et al., 2018). Thus the appearance of dorsal and ventral muscleless regions in these diaphragms demonstrates that Met has an additional later role.
Our original analysis of the mechanistic effects of Met inhibition in vivo (in Figure 7) were based on experiments administering BMS777607 E7.5-E12.5 because this dosage gave the most highly penetrant and consistent phenotype. To exclude possible effects due to early delamination and migration defects, we have repeated these experiments by administering inhibitor at E11.5-E12.5 and analyzing the ventral-most expansion of diaphragm muscle at E15.5 (Figure 7O-V). We find that Pax7/MyoD/Myosin+ cells are substantially reduced in the ventral leading edge of the diaphragm following BMS777607 treatment. This demonstrates that Met is required at later timepoints for the ventral expansion of the diaphragm.
6) Validation for the drug BMS777607 effectiveness in inhibiting MET downstream pathway should be performed and shown.
Due to limited number of cells in our PPF cultures, quantitative validation of pathways is difficult in the embryonic diaphragm. As such, we have validated the effectiveness of BMS777607 in inhibiting phosphorylation of MET, which is essential for downstream pathway signaling, using adult skeletal muscle stem cells, where there are sufficient cells to perform phospho-specific flow cytometry (Figure 6—figure supplement 1). Moreover, the drug BMS777607 has been previously validated in downregulating pAkt, pERK, and pPLC by Western blot (Faham, Zhao and Welm, 2018 doi:10.1038/s41523-018-0091-5). BMS777607 from Selleck Chem has been cited by 47 publications.
7) Comments referring to missing controls, scale bars, statistics need to be addressed.
We have addressed comments about controls, scale bars, and statistics below and thank reviewers for their careful and thoughtful reviews.
Reviewer #1 (Recommendations for the authors):
It was previously shown that HGF and Met controls development of the diaphragm muscle. In particular, the signal induces delamination and migration of muscle progenitor cells that colonize the diaphragm. The present manuscript by Sefton and coworkers confirms and extends these observations using (i) conditional mouse lines in which the HGF gene was targeted by Cre/loxP recombination in the pleuroperitoneal folds (Prx1-cre) and at other sites PdgfraCreERT2, and of (ii) Met inhibitors. Overall, the technical quality of the data on diaphragm muscle development is excellent; the conceptual advance over previous work is not exceptional; the evidence for Met/HGF-dependent development of the phrenic nerve is marginal and needs to be strengthened.
The data show that fibroblasts provide HGF signals received by Met in muscle progenitor cells that is essential for diaphragm development. The PdgfraCreERT2 line was used to demonstrate that HGF produced by fibroblasts but not by muscle progenitors is essential for diaphragm development. Moreover, development of dorsal and ventral regions of diaphragm muscle requires continuous MET signaling. Thus, HGF is not only required for the delamination of progenitors, but also for proliferation and survival of those muscle progenitors that reached the anlage of the diaphragm.
My major concern is the limited data on the HGF-dependent development of the phrenic nerve (defasciculation). While it is well documented that HGF acts as a trophic factor for motor neurons in culture, its role in development of motor neurons was highly debated due to the fact that some changes observed in Met or HGF mutant mice in vivo are also present in other mutants that lack the muscle groups derived from migrating muscle progenitors. Moreover, careful genetic analyses previously demonstrated indirect mechanisms of Met during motor neuron development, i.e. a non-cell-autonomous function of Met during the recruitment of motor neurons to PEA3-positive motor pools (Helmbacher et al., Neuron 2003).
Sefton et al., provide an analysis of a single time point, one histological picture (3G, magnified in 3H) that indicate that in Met+/- animals defasciculation of the phrenic nerve does not occur correctly. This is accompanied by a quantification that barely reaches significance (Figure 3K). Data shown in Figure 7 using Met inhibitors show a major change in phrenic nerve branching, which is presumably due to the major change in diaphragm development, as conceded by the authors.
Despite this weakness on the experimental side, the role of HGF/Met in phrenic nerve development is strongly emphasized in abstract /intro/discussion (e.g. line 414: However, PPF-derived HGF is crucial for the defasciculation and primary branching of the nerve, independent of muscle). The data need to be strengthened in order to conclude that HGF coordinates both, diaphragm muscle and phrenic development. I expect that the defasciculation of the phrenic nerve is highly dependent on the developmental stage. The authors should provide data that show different stages of phrenic nerve development, i.e. the time course of the of defasciculation in wildtype animals, explain how the staging was done, and compare different stages in the Met mutants and analyze whether the defasciculation is resolved at later stages. Met mutations also affect placental development, resulting in developmental delays that in turn might lead to an apparent small change in the time course of defasciculation. The authors should exclude that indirect effects cause the small change in phrenic nerve morphology, for instance by examining conditional Met mutations that are restricted to motor neurons. Good Cre lines that target motor neurons are available.
In response to comments from the reviewers, we have more thoroughly investigated the role of Met in the development of the phrenic nerve and include two new sets of genetic experiments. In our first submission, we found a decreased number of phrenic nerve branches at E11.5 in Met Δ/ Δ and Met Δ/+ compared with Met+/+ embryos. In the Met Δ/ Δ embryos, no muscle is present in the diaphragm. Therefore, the greatly reduced branching in these embryos is likely a secondary effect of the requirement of Met in muscle progenitors for diaphragm muscularization. Of particular interest is the reduced branching in the Met Δ/+ embryos. Because the diaphragm is muscularized in these embryos, this suggested that Met may be required intrinsically in the phrenic nerve. One reviewer suggested that the reduced branching in the Met Δ/+ embryos could be due to a developmental delay in the whole embryo. However, we found that Met Δ/ Δ and Met Δ/+ embryos are not overall delayed relative to Met+/+ embryos (as measured by crown rump length or limb length; Figure 3—figure supplement 1). Also, to increase the robustness of these data, we added additional embryos to the analysis. We then extended our analysis of Met Δ/ Δ, Met Δ/+ and Met+/+ embryos to E12.5 (Figure 3—figure supplement 1) to see whether the branching phenotype persisted; surprisingly we found that while the of Met Δ/ Δ embryos continue to have very few branches, the number of branches in Met Δ/+ embryos recovers and matches that of Met+/+ embryos.
To explicitly test whether Met is required within the phrenic nerve, we used Olig2Cre/+to conditionally delete Met. This line was chosen for its early expression in motor neurons (Zawadzka et al., 2010). We examined Olig2Cre/+;MetΔ/flox embryos compared to Olig2Cre/+; Metflox/+ embryos. We chose to include Olig2Cre in our controls because the Olig2Cre is a knock-in/knock-out and Olig2 has important roles in nerve development. However, deletion of Met did not affect the number of branches at E11.5 (Figure 3—figure supplement 2) or E12.5 (data not shown). These data suggest that Met does not intrinsically regulate phrenic nerve branching. This suggests that PPF-derived HGF regulates phrenic nerve branching indirectly via muscle. To test if HGF is sufficient to promote early stages of nerve branching in the absence of muscle, we turned to Pax3SpD/SpD mutants in which a point mutation in Pax3 prevents migration of muscle progenitors into the diaphragm (Figure 3—figure supplement 2). In these embryos, the diaphragm is muscleless, but the PPFs still express HGF. In these diaphragms the number of branches at E11.5 is severely reduced. These data demonstrate that in the absence of muscle the presence of HGF in the PPF fibroblasts is not sufficient to support diaphragm branching.
Altogether our data demonstrate that PPF-derived HGF, via its regulation of muscle, controls the primary branching of phrenic nerve. The Met Δ/+ data demonstrate that Met controls phrenic nerve branching at E11.5 in a dose-dependent manner, but this effect is lost by E12.5. Although we see no obvious defects in muscle of Met Δ/+ diaphragms at later stages, the most parsimonious explanation of the reduced phrenic nerve branching at E11.5 is that this is due to fewer muscle progenitors at this time point.
We thank the reviewers for prompting us to look at the role of HGF/Met in the phrenic nerve more closely. Our revised conclusions are presented in the Results and Discussion. We show that PPF-derived HGF is critical for integrating both muscle and phrenic nerve development, but now demonstrate that HGF’s regulation of phrenic nerve branching is via muscle, which is well-known to express multiple trophic factors required by motor neurons.
Reviewer #2 (Recommendations for the authors):
Since the authors observed a correlation between loss of should muscle and loss of diaphragm muscularization, which is related to the timing of migration of shoulder muscle and diaphragm muscle progenitors, they claim a “closer relationship” between should muscles and diaphragm, which was further extended in the discussion. I was not convinced by this conclusion. Is there a “closer relationship” between the muscles just because the progenitor cells migrate roughly at the same time? They authors may modify or delete this statement, although I agree that a broader expression of HGF may facilitate enhanced recruitment of muscle progenitor cells, required for formation of the diaphragm.
This statement has been modified (lines 206-211).
“However, based on their similar temporal sensitivity to HGF/MET signaling, the shoulder acromiodeltoid and spinodeltoid progenitors migrate at a similar time as the diaphragm progenitors. Thus continued expression of HGF at this later developmental time point may facilitate recruitment of muscle cells necessary for development of diaphragm and shoulder muscles.”
Based on co-cultures of PPFs and myoblasts the authors describe a function of MET in myogenic commitment. In my opinion the evidence for such a function is weak. The authors observed no change in proliferation but reduced motility and a higher rate of apoptosis after pharmacological inhibition of MET. Increased apoptosis and reduced aggregation of myoblasts at distinct locations may easily interfere with myogenic differentiation. A general function of MET in myogenic commitment does not seem very likely, since myoblasts that do not undergo prior long-range migration differentiate normally in the absence of Met.
We apologize for the misunderstanding here and have altered the text to indicate that we do not propose a role for Met in myogenic commitment, but rather that Met regulates the number of MyoD+ cells by promoting their survival.
Pharmacological inhibition of MET increases the rate of apoptosis in numerous cell types, which has been studied extensively in cancer. It seems appropriate to explore the mechanism of increased apoptosis in myoblasts following MET inhibition more closely. Previous reports suggest increased expression of p53, increased sensitivity to Fas-mediated apoptosis or increased autophagy, among others, as potential causes for increased apoptosis after MET inhibition. Which pathway is relevant in myoblasts?
We have performed qPCR on vehicle and BMS777607-treated PPFs for Fas, Trp53, and autophagy marker Map1l3ca. We found increased expression of Fas and Map1l3ca in PPFs following treatment with the Met inhibitor, while Trp53 did not change (Figure 6—figure supplement 1).
Obviously, it would be great to learn more about the mechanisms that control Hgf expression in fibroblasts within and derived from PPTs. Such knowledge may also help to better understand the reasons leading to Congenital Diaphragmatic Hernias (CDH). Unfortunately, the authors did not to go any further in this direction.
Understanding the mechanisms regulating HGF expression in the PPFs is critical for understanding why in development and in evolution muscle is recruited to the nascent diaphragm. This will be explored in future studies.
The authors observed that branching of the phrenic nerve was reduced in heterozygous Met mutants with normal diaphragm musculature, suggesting a direct role of Met in phrenic nerve branching. Timed inactivation of Met specifically in motoneurons would greatly increase the impact of this finding and allow a more specific analysis of the role of MET in phrenic nerve development and branching.
Please see response to Essential Revision Point 1. We have now included a more detailed analysis of the role of Met in phrenic nerve branching, including conditional deletion in motor neurons and analysis of Pax3SpD/SpD mice.
Surprisingly, the authors did not observe CDH in mutant mice with muscle-less diaphragms, from which they conclude that additional defects in the connective tissue are necessary to allow hernia formation. Although this explanation is intriguing, the authors have to make sure that abdominal pressure is not reduced in mice without muscularized diaphragms, e.g. demonstrate that body wall muscles are fully functional. The images in Figures 1 and 2 seem to indicate normal development of body wall muscles but a dedicated statement in this respect would be helpful.
We have added a dedicated statement to the manuscript indicating that no defects in abdominal wall muscle development were observed.
Scale bars are missing in some panels.
Thank you for this observation and we have checked for scale bars in figure panels. Please note that some figures contain one scale bar for multiple panels.
According to the methods part, Student’s t-test was used for the statistical analysis shown in Figure 3. A pairwise comparison of WT, heterozygous and homozygous Met mutants is not appropriate when all three genotypes are compared with each other. An ANOVA test should be used as in Figure 5F.
We had used a one-way ANOVA for the statistical analysis used in Figure 3, but did not specifically state this in the methods. This has now been added to the methods and Figure legend.
Reviewer #3 (Recommendations for the authors):
The present manuscript addresses questions on the role of HGF/MET signaling in diaphragm formation once myogenic progenitors have already migrated to the PPFs. In addition, it identifies the PPFs as the source of HGF. The study is interesting and developed with rigor. However, the role of HGF is not clearly dissociated from the presence/absence of the fibroblasts/connective tissue itself.
We have added to a supplemental figure demonstrating that fibroblasts are indeed present in muscleless regions following loss of HGF (Figure 4—figure supplement 1D-F).
Also, the authors do not link in the results, for example, how the muscleless diaphragms and HGF itself relate to the hernia phenotype mentioned in the abstract.
We now discuss how the partially muscleless diaphragms relate to herniation in the abstract. While loss of muscle is necessary for development of hernias, it is not sufficient; additional defects in the connective tissue are necessary to weaken the diaphragm and lead to CDH.
Figure 1
1C) Co-staining for Hgf (PPFs) and Met (migrating progenitors) should be provided for a clear visualization of ligand-expressing cells and receptor-expressing cells.
We have included an additional panel in Figure 1—figure supplement 1 showing coexpression of HGF and Met in E11.5 PPF.
1F) Co-staining for Met and migrating progenitors (PAX3 or LBX1 for example) should be provided for a clear visualization of migrating progenitors versus general Met expression in other cell types.
We have previously shown this in a supplemental figure analyzing the migration pathway of diaphragm muscle progenitors, co-staining MET and GFP abelling in Pax3CreKI/+; RosamTmG/+ embryos in cross section (Figure S2 in Sefton et al., 2018 doi: 10.1016/j.ydbio.2018.04.010).
1M) The claim that fibroblast-derived HGF is required for diaphragm muscle development cannot be addressed with only this experimental analysis. What is the connective tissue phenotype in PDGFaCreER;HGFnull/fl embryos? Lack of connective tissue could affect muscle migration and development rather than HGF expression on its own. Are fibroblasts still present?
We have demonstrated that fibroblasts are still present in muscleless regions (see point 2 above and Figure 4—figure supplement 1).
1G, I, K, M) Control genotypes should be analyzed and added in the Figure or as Supplementary Data.
We have added a supplemental figure with control embryos stained for myosin (Figure 1—figure supplement 3).
1I) Pax3CreKIMetnull/fl originates a muscleless diaphragm but this could be associated with lack of delamination and migration from the somites rather than a specific MET requirement for diaphragm muscle formation once progenitors have colonized this area as suggested by the authors.
Yes, we agree. It is the primary reason we chose to use the pharmacological inhibitor BMS777607 to control the timing of MET signaling inhibitions.
Figure 3
It is not clear whether HGF controls phrenic nerve formation independently of muscle. In the end of this section the authors mention that Met heterozygous embryos have normal muscles referring to Figure 1K (which is not a picture referring to a Met het embryo). The authors should confirm if in Met null embryos there is a direct effect on the nerve bifurcation and brunching or if the lack of muscle is leading to this observation. Is the phenotype dose dependent for muscle formation? This is not properly shown by the authors. Could the authors perform a conditional KO of Met in the nerves?
This concern is further supported by the Figure 4 data where phrenic nerve only extends to regions with muscle.
Please see response to Essential Revision Point 1. We have now included a more detailed analysis of the role of Met in phrenic nerve branching, including conditional deletion in motor neurons and analysis of Pax3SpD/SpD mice.
Figure 4
The role of HGF versus the presence of fibroblasts/connective tissue remains vague. Is the lack of ventral muscles associated with lack of migration of progenitors within the forming diaphragm towards an HGF source? Or due to the lack of connective tissue scaffold itself? Why is Prx1Cre used instead of PDGFRaCre (the one used in the actual experiments) to trace PPFs?
We have lineage data demonstrating that PDGFRa-derived fibroblasts are present in the muscleless regions. The Prx1Cre lineage is also used in supplemental experiments for HGF deletion and in previous work from the lab. We believe it is useful to demonstrate the finding that fibroblasts are present with multiple lines of evidence and have left this in the supplemental figure (Figure 4—figure supplement 1).
4E) Please provide a control with a Cre allele to compare putative secondary effects for the presence of the Cre in the cKO embryos.
The panel in 4E has been replaced with a control that has the Cre allele.
Figure 5
Validation for the drug BMS777607 effectiveness in inhibiting MET downstream pathway should be performed and shown.
Due to limited cell number, this is difficult to directly measure in embryonic diaphragms either via Western or phospho-flow cytometry. As such, we took a strategy to measure phospho-Met levels in adult skeletal muscle satellite cells. We have demonstrated that phospho-Met levels are downregulated in skeletal muscle satellite cells following treatment with BMS777607 (please see point 6 above; Figure 6—figure supplement 1A).
The phenotype with the inhibition of Met signaling (lack of dorsal muscles in addition to ventral) is distinct to the one observed in Figure 4 (less ventral muscle formation in PDGFaCreER;HGFnull/fl). How do the authors explain the dorsal phenotype when inhibiting Met (since migration should have not been affected at this time-point) and Hgf inhibition in Figure 4 is not leading to this phenotype?
There is quite a bit of variability in the PDGFRaCreER/+; HGFnull/fl embryos and we do indeed often find dorsal muscleless regions in PDGFRaCreER/+; HGFnull/fl embryos given tamoxifen at E9.5 (see panels 4C and 4H, where myofibers are absent in the dorsal regions of the diaphragm). We suspect that slight differences in the timing of Cre-mediated deletion of HGF deletion likely have a substantial impact on the extent of muscle loss and result in the variable muscle phenotypes. We have also clarified the summary of the phenotype of PDGFRaCreER/+; HGFnull/fl diaphragms in the model in Figure 8.
Figure 6
What is the time point after treatment in B and C? What is the total nuclei number in the cultures (since these are co-cultures)?
The timepoint after treatment is 72 hr. This has been added to the figure legend. With our co-culture method, we have not been able to maintain myogenic progenitors after passaging. As such, these experiments are done on freshly explanted PPFs that are in a pyramid shape on the plate. Because the fibroblasts are tightly packed together in 3-dimensions prior to passaging, we cannot count total nuclei accurately with the automated software or evenly manually. The myogenic progenitors are less dense and can be accurately counted. As such, unfortunately we do not have total numbers of nuclei.
The authors should provide pictures together with nuclear staining for a global view of the cell density in the cultures. Also T0 time point pictures should be added for comparison, in particular in C.
We have added photos of nuclear staining of a 72 hr culture to a new supplemental figure (Figure 6 —figure supplement 2). While we cannot get accurate quantitative data from this, it does give a qualitative picture of overall nuclear density. T0 pictures, using GFP fluorescence of live muscle progenitors have also been added to this supplemental figure.
In these experiments the authors performed co-culture (fibroblasts + myogenic progenitors) derived from PPfs. However, fibroblast phenotype is not addressed. If there is less GFP+ cells is there more fibroblasts in the culture? Could this impact on the phenotype observed and linked by the authors to the Met inhibitor treatment? The only piece of evidence is Gata4 qPCR which is not sufficient to address fibroblast phenotype in the culture.
Although it is possible that fibroblasts are affected by Met inhibition, our in vivo deletion of Met in fibroblasts (Figure 1K) does not suggest that Met inhibition would negatively impact PPF fibroblasts in culture. Because of technical limitations (see above), it is difficult to explicitly quantify the number of fibroblasts unless we were to passage the fibroblasts.
Figure 7
Treatment of WT embryos from E7.5 will impact on myogenic progenitor migration. The authors have to dissociate the Met-related somite migration phenotype and the potential additional role of Met signaling in the diaphragm muscle formation.
Our original analysis of the mechanistic effects of Met inhibition in vivo (in Figure 7) were based on experiments administering BMS777607 E7.5-E12.5 because this dosage gave the most highly penetrant and consistent phenotype. To exclude possible effects due to early delamination and migration defects, we have repeated these experiments by administering inhibitor at E11.5-E12.5 and analyzing the ventral-most expansion of diaphragm muscle at E15.5 (Figure 7O-V). We find that Pax7/MyoD/Myosin+ cells are substantially reduced in the ventral leading edge of the diaphragm following BMS777607 treatment. This demonstrates that Met is required at later timepoints for the ventral expansion of the diaphragm.
[Editors’ note: further revisions were suggested prior to acceptance, as described below.]
Reviewer #1 (Recommendations for the authors):
The majority of my concerns were appropriately addressed by the authors. There are a few additional points that need to be addressed, and most of these concern the text/wording.
Unfortunately, the changes in the text have not been highlighted in red in the uploaded version of the manuscript, as it was mentioned in the rebuttal letter.
1. While the quality of the figures is overall good, many of the panels in Figure 2 are very dark. It should be easy to modify this.
Thank you for this observation. We have brightened many of the panels in Figure 2.
2. While the authors show very convincingly that the effects on the phrenic nerve observed are caused indirectly by the loss of muscle, and not directly by the loss of HGF/Met signaling, this is not always made clear in the text. Furthermore, the mechanism that causes branching deficits should be clearly stated in the Abstract.
Line 41-42...and indirectly required for phrenic nerve primary branching.
Please mention the specific indirect mechanisms, i.e. via the effect on muscle development.
We have explicitly noted the indirect effect is mediated by muscle in the abstract.
Line 81…HGF is also critical for innervation.
The reports cited do not distinguish between the effect of the nerve/muscle or report in vitro experiments. The text should take this into account.
A sentence has been added indicating the reports do not distinguish between effect on muscle and nerve.
238: Thus, PPF-derived HGF is necessary for phrenic nerve defasciculation.
Additional experiments shown in the next paragraph indicate that it is the absence of the muscle that causes defasciculation. Please rephrase.
This has been rephrased as “Thus, loss of PPF-derived HGF leads to both muscle defects and phrenic nerve defasciculation defects.”
3) Met inhibition using BMS777607
BMS777607 acts as an AXL, RON and Met tyrosine kinase inhibitor. This should be mentioned. Are effects of RON/AXL on skeletal muscle development described?
We have added a note that BMS777607 also inhibits AXL and RON in addition to MET. There is limited information about the role of RON/AXL on skeletal muscle development. A recent study found that Axl is a survival and growth receptor in mouse myoblasts during skeletal muscle regeneration (Al-Zaeed et al., 2021; doi: 10.1038/s41419-021-03892-5). A published meeting abstract states that double knock out Gas6-Axl mice have reduced hindlimb skeletal muscle mass (doi: https://doi.org/10.1096/fasebj.2020.34.s1.09757).
4) Antibody specificity: Antibody specificity should be tested by analysis of the phrenic nerve on Met mutants.
The antibody we used is the goat polycolonal (#AF527 R&D Systems) which was generated against a recombinant protein that encompasses Glu25-Asn929. This region includes the extracellular semaphorin domain, furin cleavage site, plexin semaphorin domain, and the four IPT domains. The mutant Met allele we are using is from Huh et al., (PNAS 2004; doi: 10.1073/pnas.0306068101) and deletes exon 16 which contains a critical ATP-binding site in the intracellular tyrosine kinase domain. Western analysis from Webster et al., (Figure 1C; PLoS 2013; doi: https://doi.org/10.1371/journal.pone.0081757) and the Huh et al., 2004 paper (Figure 1 L) show that the MET pTyr 1234/1235 domain and downstream pAKT are inactivated with this mutant allele, respectively. However, the pre-processed MET protein is still present in these mutants (Figure 1K of Huh et al., and Figure 1B of Webster et al.,). The R&D antibody that we and others use recognizes the extracellular domain that is present in the preprocessed protein (see Figure 1B of Webster et al). Thus when we examine MetD/D mutants using the R&D antibody, we still see labeling (see Author response image 1). Therefore the Huh et al., Met allele that we use is not able to test the specificity of the R&D antibody. However, over 24 references have used this antibody (https://www.rndsystems.com/products/mouse-hgfr-c-met-antibody_af527#product-citations). We have not exhaustively searched these references to determine whether any tested the specificity of the R&D antibody (e.g. that it does not cross-react with the closely related RON protein).
The extracellular region of MET protein (using AF527 R&D antibody) is still detected in <italic>Met<sup>D/D</sup></italic> mutants.
Transverse cross section of E10.5 embryos with dorsal to the top; lateral is on the left side in left panel; lateral is on the right side in right panel. NT, neural tube.
Reviewer #2 (Recommendations for the authors):
Sefton et al., have submitted a revised version of a study, in which the role of fibroblast-derived HGF for recruitment and expansion of muscle during morphogenesis of the mammalian diaphragm was investigated. The authors have changed the title to cope with new findings, indicating that reduced primary branching of the phrenic nerve in Met-mutants is not due direct effects of HGF on the nerve but most likely caused by indirect effects resulting from reduced muscle formation in Met-mutants.
The authors did an excellent job to deal with the reviewers' criticisms. In particular, by specifically deleting Met in motoneurons and by analysis of splotch mutants they demonstrate that Met does not intrinsically regulate phrenic nerve branching. Analysis of splotch mice, which show normal expression of HGF PPFs but display muscle-less diaphragms, revealed reduced branching of the phrenic nerve, similar to Met-mutants, clearly suggesting a critical role of muscle fibers in the diaphragm for phrenic nerve branching. Furthermore, Sefton et al., now demonstrate that fibroblasts expand and populate the diaphragm in the absence of muscle, which essentially excluded a connective tissue phenotype in conditional Hgf/Met mutants. They also clarified the statement about what I understood was meant to claim a role of Hgf/Met in myogenic commitment. Additional controls (effectiveness of BMS777607 for inhibition of MET, treatment with BMS777607 at additional timepoints) were done as requested, providing additional interesting insights into a role of Met for ventral expansion of the diaphragm at relatively late developmental timepoints.
The authors argue that "PPF-derived HGF, via muscle, controls phrenic nerve defasciculation". Well, this is formally correct but exaggerates the role PPF-derived for phrenic nerve defasciculation in my view. Probably, deletion of any gene that prevents diaphragm muscle formation will have similar effects on phrenic nerve defasciculation. I would prefer a more neutral statement: "Muscle formation, requiring PPF-derived HGF, controls phrenic nerve defasciculation", or something similar. The authors may consider modifying the statement.
I do not have any further objections.
We appreciate the positive response to our revisions. We have altered the sentence as suggested.