ReviewCell Biology

The Hippo signal transduction network in skeletal and cardiac muscle

See allHide authors and affiliations

Science Signaling  05 Aug 2014:
Vol. 7, Issue 337, pp. re4
DOI: 10.1126/scisignal.2005096

Abstract

The discovery of the Hippo pathway can be traced back to two areas of research. Genetic screens in fruit flies led to the identification of the Hippo pathway kinases and scaffolding proteins that function together to suppress cell proliferation and tumor growth. Independent research, often in the context of muscle biology, described Tead (TEA domain) transcription factors, which bind CATTCC DNA motifs to regulate gene expression. These two research areas were joined by the finding that the Hippo pathway regulates the activity of Tead transcription factors mainly through phosphorylation of the transcriptional coactivators Yap and Taz, which bind to and activate Teads. Additionally, many other signal transduction proteins crosstalk to members of the Hippo pathway forming a Hippo signal transduction network. We discuss evidence that the Hippo signal transduction network plays important roles in myogenesis, regeneration, muscular dystrophy, and rhabdomyosarcoma in skeletal muscle, as well as in myogenesis, organ size control, and regeneration of the heart. Understanding the role of Hippo kinases in skeletal and heart muscle physiology could have important implications for translational research.

Introduction

Skeletal and cardiac muscle function to generate body movement and blood flow. Both skeletal and cardiac muscle cells have sarcomeric motor proteins that convert the chemical energy of nutrients into work and heat. Skeletal and cardiac muscles develop from differentiation of the mesoderm, but express different genes and vary in their morphological and physiological properties. In addition, skeletal muscle has the ability to regenerate after chemical or mechanical damage (1): a property conferred by the activation of resident stem cells known as satellite cells (25). In contrast, adult mammalian cardiac muscle does not effectively regenerate after cardiac injury, such as a heart attack in humans (6). However, neonatal mammalian (7) and adult zebrafish hearts (8) can regenerate, suggesting that the underlying molecular program responsible for regeneration exists in the heart and may be activated for therapeutic purposes. Both heart and skeletal muscle are subject to progressive degeneration, as seen in aging (9) and muscular dystrophy (10), and skeletal, but not cardiac, muscle can undergo oncogenic transformation, for example, in patients with rhabdomyosarcoma (1113). Several signal transduction pathways have been linked to muscle development, regeneration, and disease. Here, we review the evidence for the emerging role of the Hippo signal transduction network in these processes.

Discovery of the Hippo signal transduction network

The Hippo pathway is an important signal transduction pathway involved in development, stem cell function, regeneration, and organ size in multiple tissues in various species, and underlies several human pathologies, including cancer (1418). Genetic screens designed to identify tumor suppressors in the fruit fly Drosophila melanogaster led to the discovery of the Hippo pathway, including the kinases Hippo and Warts, the kinase-binding proteins Salvador and Mob1, and several upstream kinase activators. Consistent with a role as tumor suppressors, loss-of-function mutations in the genes encoding these proteins typically increase cell proliferation and decrease apoptosis (19, 20). Several genetic epistasis and protein interaction studies indicate that these proteins constitute a distinct, integrated signal transduction pathway (21).

As early as the late 1980s, transcriptional regulators, which are now known to be part of the Hippo pathway, were studied independently from the work in flies, often in the context of muscle research. The transcription factor Tead1 (TEA domain containing 1, also referred to as Tef-1 in early studies) binds to a CATTCC DNA binding motif (22), which is variously referred to as a muscle CAT (MCAT) motif (23), GTIIC motif (22), or Hippo response element (24). A conserved 66– to 68–amino acid region known as the TEA/ATTS (TEA) domain in Teads (Tead1 to Tead4) binds to MCAT elements (Fig. 1, A and B) (25). The activity of Teads requires binding to transcriptional coactivators (26), including Yap (yes-associated protein, encoded by the gene Yap1) (27, 28), Taz (transcription coactivator with PDZ binding motif, encoded by the gene Wwtr1) (2931), and Vglls (Vestigial-like proteins) (26, 32, 33). The discovery that the Hippo pathway inhibits the transcriptional cofactor Yorkie (34), which is the fly homolog of Yap and Taz, connected the Hippo pathway to the transcriptional regulators Yap, Taz, and Teads.

Fig. 1 Protein interactions in the Hippo pathway.

(A) The Hippo pathway comprises a cassette of serine and threonine kinases, including Mst1 or Mst2 (Mst1/2) and Lats1 or Lats2 (Lats1/2), that phosphorylates the functionally redundant transcriptional coactivators Yap and Taz (Yap/Taz). The key domains and motifs that enable protein-protein and protein-DNA interactions are WW, Tondu, and Tea domains and PPxY or PPxF motifs (PPxY/F). P, phosphorylation; Yap/Taz, Yap or Taz. Credit: H. McDonald/Science Signaling. (B) Structural model of the TEA domain of Tead (red, yellow) binding to DNA (white). Image adapted with permission from (210). (C) Structural model of the β strands in the WW domain (green) of YAP in complex with a LATS1 peptide (yellow) containing the PPxY motif. The side-chain moieties of amino acids within the WW domain (red) and the consensus residues within the PPxY motif (blue) (assigned P0, P + 1, and Y + 3) are engaged in intermolecular contacts. Credit: A. Farooq/University of Miami. (D) Structural model of the transactivation domain of Tead4 (red) binding to Yap (orange) at three interfaces or Vgll1 (green) at two interfaces. Image reproduced with permission from (32).

In recent years, the list of genes and proteins linked to the Hippo pathway has expanded greatly. Research demonstrates that the Hippo pathway is only one of several signal transduction modules that target the Hippo transcriptional regulators Yap, Taz, Teads, and Vglls. Additionally, the Hippo transcriptional regulators interact with many downstream signaling proteins (35, 36). Therefore, to most appropriately describe the structure of the entire signaling system, we refer to it as the “Hippo signal transduction network.”

Hippo pathway

The central Hippo pathway comprises the mammalian STE20-like protein kinases 1 and 2 (Mst1 and Mst2, encoded by the genes Stk4 and Stk3, orthologs of Hippo in flies) and the large tumor suppressor kinases 1 and 2 (Lats1 and Lats2, paralogous to Warts in flies) of the NDR (nuclear dbf2-related) family (Fig. 1A) (37). Multiple proteins in the Hippo signal transduction network interact using WW domains, which are defined by two highly conserved tryptophans located 20 to 22 amino acids apart that bind to proline-rich motifs [PPxY or PPxF (PPxY/F)] (38). Mst1 or Mst2 (Mst1/2) binds to the auxiliary protein Sav1 (Salvador homolog 1), presumably through interactions of WW domains in Sav1 and noncanonical PPxF motifs in Mst1/2 (39, 40), and this complex directly phosphorylates a threonine in the kinase domain of Lats1 or Lats2 (Lats1/2) (Thr1041 of Lats1 or Thr1079 of Lats2) (41). Moreover, Mst1/2 bound to Sav1 binds to and phosphorylates Mob1α or Mob1β (Mob kinase activator 1), which then binds to the autoinhibitory loop of Lats1/2 to promote autophosphorylation (Ser909 of Lats1 or Ser872 of Lats2) (41, 42). The PPxY motif of Lats1/2 binds to the WW domains of Yap or Taz (43, 44) (Fig. 1C). Yap is alternatively spliced to produce variants with either one or two WW domains (45), and Taz has one WW domain (30). Phosphorylation and activation of Lats1/2 by Mst1/2 promote the ability of Lats1/2 to phosphorylate Yap or Taz (30, 34, 46, 47). Many cell adhesion and cell junction proteins can affect the activity of Mst1/2 and Lats1/2; however, these proteins primarily have been studied in epithelial cells (14, 48), and it is unclear whether they have similar functions in skeletal and cardiac muscle cells.

Phosphorylation of Yap or Taz by Lats1/2 occurs on HXRXXS amino acid motifs (46). Human YAP has five (Ser61, Ser109, Ser127, Ser164, and Ser381) and human TAZ has four (Ser66, Ser89, Ser117, and Ser311) HXRXXS motifs, all of which are phosphorylated by Lats1/2 (49). Phosphorylation of the best-characterized phosphosite of YAP (Ser127) leads to sequestration in the cytosol by 14-3-3 proteins (50). Mutating YAP Ser127 to alanine prevents phosphorylation and promotes constitutive activation of its transcriptional function (47, 51). Except in intestinal stem cells (52), expression of YAP S127A increases cell proliferation and inhibits apoptosis (47, 51, 53, 54). Ser89 of TAZ is analogous to Ser127 of YAP, and the TAZ S89A mutant is also a constitutive transcriptional activator (30). Yap can be activated independently of Lats1/2 by phosphorylation by the tyrosine kinase Yes (27, 55). Moreover, the PPxY motif of the nonreceptor tyrosine phosphatase Ptpn14 binds to the WW domain of Yap and inhibits nuclear localization of Yap (5658). The serine and threonine phosphatase PP1 activates Taz by dephosphorylation of Ser89 and Ser311, which induces nuclear localization and stabilization of Taz (59). Monomethylation of Yap at Lys494 by Setd7 (SET domain containing lysine methyltransferase 7) leads to cytoplasmic localization and inhibition of Yap (60). Moreover, murine Yap phosphorylated at Ser112, which is homologous to Ser127 in human YAP, localizes to both the cytoplasm and the nucleus of cells grown at low density in culture (61). Thus, Yap and Taz are inhibited and activated by proteins other than Lats kinases, suggesting that the phosphorylation state of YAP Ser127 and TAZ Ser89 may not be sufficient to indicate inhibition of their transcriptional activity.

In the absence of coactivators, Teads repress expression of target genes (25). Yap and Taz bind to three interfaces on Teads (32) (Fig. 1D) and thereby relieve repression and promote transcription in a manner analogous to transcription coactivators in other developmental signaling pathways. Chromatin immunoprecipitation studies in breast epithelial cells show that Yap and Tead occupy about 80% of the same genomic loci (62). However, Tead-dependent transcription can also occur independently of coactivation by Yap or Taz. Like Yap and Taz, binding of some Vgll proteins can activate Tead-based transcription (32, 33). In contrast, Vgll4 is a repressor of Teads (26, 63). The Tondu domains of Vgll proteins interact with Teads (32, 33) and may compete for common binding interfaces with Yap and Taz (26) (Fig. 1D). Vgll4 has two Tondu domains, whereas other Vgll paralogs have only one.

Crosstalk with the Hippo pathway

The Hippo pathway extends beyond a simple kinase cascade, leading to inhibition of Yap and Taz. Several proteins have been discovered to either completely or partially bypass the Hippo pathway to target Yap and Taz. Likewise, Yap and Taz can influence the activity of DNA binding transcription factors other than Teads (Fig. 2).

Fig. 2 Schematic of key signaling modules within the Hippo signal transduction network.

As detailed in the main text, evidence suggests that crosstalk between the Hippo pathway and other signaling modules occurs in parallel to the core kinase cassette. P, phosphorylation; M, methylation; LPA, lysophosphatidic acid; S1P, sphingosine 1-phosphate. Credit: H. McDonald/Science Signaling.

Cells grown on stiff substrates or at low density form stress fibers (64, 65) composed of filamentous (F)–actin, myosin II, and α-actin (66). Actin polymerization is required for Yap and Taz nuclear localization and activation by mechanical signals or low-density cell culture (61, 64, 65). Several angiomotin proteins, which are present in skeletal muscle and the heart in addition to other tissues (67), bind to F-actin and Yap and are required for inhibition of Yap by mechanical stimuli (68). Disruption of actin polymerization can activate Lats (69), and expression of a kinase-dead form of Lats2 can prevent the ability of the actin depolymerizing agents to promote the cytosolic localization of Yap (69). However, knockdown of Lats1 and Lats2 does not rescue Yap and Taz inhibition in cells grown on a soft extracellular matrix (65), suggesting that inhibition of Yap and Taz by mechanical stimuli can function in parallel with or independent of Lats.

G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors (GPCRs) are a large family of seven-transmembrane receptor proteins (70), which can regulate the nuclear translocation of Yap and Taz. Exposing cultured cells to fetal bovine serum or GPCR ligands, including lysophosphatidic acid or S1P, which activates Gα12/13-, Gαq/11-, and Gαi/δ-coupled receptors, inhibits phosphorylation of Lats, Yap, and Taz, and promotes nuclear localization of Yap (69). Disruption of the actin cytoskeleton by exposing cells to latrunculin B prevents the ability of lysophosphatidic acid or fetal bovine serum to activate Yap and Taz (69), suggesting that the cytoskeleton may integrate mechanotransduction and GPCR signaling. In contrast to Gα12/13-, Gαq/11-, and Gαi/δ-coupled receptors, the activation of GαS-coupled receptors by epinephrine or other ligands increases inhibitory phosphorylation and cytosolic localization of Yap (69). GαS increases the production of cyclic AMP (adenosine 5′-monophosphate), which activates protein kinase A (PKA), and exposing cells to forskolin, which increases cyclic AMP, increases phosphorylation of Lats1/2 and Yap and promotes cytosolic localization of Yap (69).

Multiple studies provide evidence for crosstalk between the Wnt–β-catenin and Hippo signaling pathways (35). For example, Yap can form a complex with β-catenin, Yes, and Tbx5 (T-box transcription factor 5) and is required in colorectal cancer cell lines with high β-catenin activity for proliferation and colony formation in culture and tumor-forming potential in orthotopic xenografts in mice (55). Additionally, members of the β-catenin destruction complex can promote the cytosolic sequestration of Yap and Taz and the ubiquitin-dependent degradation of Taz and thereby inhibit Yap- and Taz-dependent gene expression (71, 72).

Yap and Taz bind and coactivate transcription factors other than Teads (36). Of these, Smads (sma gene mothers against decapentaplegic peptide) and Tbx5 are especially relevant to skeletal and cardiac muscle biology. The WW domains of Yap bind to the PPxY motifs of the phosphorylated linker region of the bone morphogenetic protein (BMP) signaling–associated receptor Smad, Smad1, to promote Smad1-dependent transcription (73). The WW domains of Yap can also bind to the PPxY motif of Smad7, an inhibitor of transforming growth factor–β (TGFβ) signaling (74). The C terminus of Taz binds to complexes of the TGFβ signaling–associated Smads, Smad2 and Smad3, and the DNA binding Smad, Smad4, in the nucleus and promotes the expression of Smad2- and Smad3-dependent genes (75). Moreover, Yap and Taz can bind to Tbx5 and promote Tbx5-dependent gene expression (55, 76).

Hippo signaling can influence mTOR (mammalian target of rapamycin) kinase signaling. Yap drives the expression of miR-29, which promotes the degradation of the phosphatase Pten. Pten inhibits Akt, and Akt indirectly activates mTOR. Knockdown of Lats1 and Lats2 in cells or overexpression of Yap in the skin of mice increases mTOR activity (77). Mst1/2 can bind in a complex with Akt1, and knockdown of Mst1/2 reduces the activating phosphorylation of Akt1 at Ser473 and reduces phosphorylation of Akt substrates (78). Moreover, Akt phosphorylates Mst1 at Thr120 and Thr387, leading to inhibition of Mst1 activity (7981). Because Yap and Taz promote cell proliferation and because mTOR signaling is required for protein synthesis during cell proliferation, the mechanistic connection between Hippo and mTOR signaling suggests that cell proliferation and protein synthesis could be coordinated through this crosstalk.

There is evidence that other signaling pathways regulate Hippo signaling by altering the transcription of genes encoding proteins in the Hippo pathway. For example, exposing cerebellar neuronal precursors to the secreted glycoprotein Shh (sonic hedgehog) increases Yap1 expression (82). Moreover, the Notch-associated transcription factor Rbpj (recombination signal binding protein for immunoglobulin kappa J) directly induces transcription of Yap1 and Tead2 in mouse cortical neural stem cells (83). Sox2 (sex determining region box 2) directly binds to the Yap1 promoter and increases its expression in mesenchymal stem cells (84). Moreover, the promoter of Yap1 contains several binding sites for the transcription factor GA-binding protein, which also promotes Yap1 expression (85).

The abundance of Yap and Taz protein is controlled by phosphorylation-dependent degradation. Phosphorylation of Ser381 by Lats1 or Lats2 primes YAP for phosphorylation by casein kinases CK1δ or CK1ε, which creates a phosphodegron motif that leads to Yap degradation by the proteasome (49). Similarly, glycogen synthase kinase 3 of the β-catenin destruction complex phosphorylates Taz, creating a phosphodegron motif and targeting it for degradation (86, 87).

Several other signaling proteins crosstalk with the Hippo pathway. For example, the kinase Lkb1 (also known as serine and threonine protein kinase 11) (88, 89) and integrin-linked protein kinase (90) interact with Hippo transcriptional regulators and may be relevant to skeletal and cardiac muscle biology. Moreover, recent proteomic studies in several cell types have identified previously uncharacterized protein interactions with members of the Hippo signal transduction network. For example, Yorkie can interact with proteins of lysosomal pathway, and knockdown of these proteins or neutralization of lysosomal pH increases the activity of a Yorkie transcriptional reporter (91). Likewise, interaction with coiled-coil domain-containing protein 85C regulates the localization of Yap (92). Striatin-interacting phosphatase and kinase (STRIPAK) complexes bind Mst1 and Mst2, but the functional implications of this interaction are poorly understood (93, 94). Collectively, this diversity of crosstalk supports the notion that the Hippo pathway does not function in isolation but participates in various other pathways as part of a wider Hippo signal transduction network.

Hippo signaling and skeletal muscle myogenesis

Several studies implicate the Hippo signal transduction network in muscle development, regeneration, and disease. Skeletal muscle myogenesis begins around embryonic day 8 (E8) in mice, when Pax3 (paired box protein-3)–positive cells in the dermomyotome begin to undergo epithelial-to-mesenchymal transitions, delaminate, and migrate (95). Migratory Pax3-positive cells begin to express the myogenic regulatory transcription factors Myf5 and MyoD (MyoD is expressed later than Myf5) (9698), which promote the specification of Pax3-positive progenitor cells into mononucleated myoblasts (98). Overexpression of Myf5 or MyoD in non-muscle cells is sufficient to initiate myogenesis in vivo (99), suggesting that these factors function redundantly (100, 101). MyoD induces the expression of the transcription factors myogenin and Mrf4, causing myoblasts to undergo terminal differentiation, in which they fuse into multinucleated myotubes that mature into striated muscle fibers (98). MyoD has been reported to bind more than 20,000 DNA loci in myoblasts and multinucleated myotubes (102). Although some of these loci might be detected due to nonspecific MyoD binding (103), the large number of sites is consistent with the idea that MyoD acts as a “pioneer” transcription factor. Pioneer transcription factors stably bind and “preselect” a set of genes to be expressed in a given cell lineage (104). Consistent with this model, MyoD recruits histone methyltransferases and acetyltransferases to the enhancers and promoters of myoblast-associated genes, which prepares the chromatin for active transcription (105, 106).

There is both direct and indirect evidence that the Hippo signal transduction network is involved in the regulation of myogenic differentiation in cultured myoblasts in vitro and during embryonic myogenesis in vivo. Studies suggest that Yap is active in myoblasts and inhibits terminal differentiation into myotubes (107, 108). In Xenopus laevis embryos, overexpression of constitutively active Yap increases proliferation of neural progenitor cells and reduces the expression of markers of somatic muscle differentiation, including MyoD (109). The evolutionarily conserved ECR111 enhancer, which contains an MCAT motif that binds Teads, is located ~111 kilo–base pairs upstream of the Myf5 gene and is required for the expression of Myf5 in ventral somatic compartments (110). Consistent with a potential role of Hippo signaling in activating ECR111, overexpression of YAP S127A in C2C12 myoblasts increases Myf5 expression (107). It is unknown whether Yap1 is required for Myf5 expression during developmental myogenesis in vivo because mice with global knockout of Yap1 die around E8.5 (111). Global knockout of Wwtr1, the gene encoding Taz (112), does not result in an obvious skeletal muscle phenotype, and in-depth analysis of this tissue was not the focus of that study.

Additional circumstantial evidence suggests that inhibition of Yap is essential for myoblast differentiation into multinucleated myotubes. Cell culture conditions optimal for differentiating myoblasts into myotubes are similar to those that inhibit Yap activity in other cell types. Myoblasts grown to a high degree of confluence maximize cell-cell contact, which is known to inhibit Yap in other cell types (46, 65) and enhances myoblast differentiation (113). Similarly, myoblast differentiation is enhanced when cells are cultured on soft hydrogel substrates with a stiffness of about 12 kPa similar to the stiffness of muscle, rather than directly on plastic, which is much stiffer (114). Reducing the concentration of serum in the medium, which should inhibit Yap by reducing activation of GPCR signaling (69), also promotes terminal differentiation of C2C12 myoblasts (113). Together, these results imply that inhibition of Yap may be a key event for the terminal differentiation of myoblasts into myotubes.

In contrast to the observation that active Yap prevents the differentiation in myoblasts (107) and activated satellite cells (108), some genes with MCAT motifs are expressed in differentiated muscle. In zebrafish, activation of a transgenic reporter for Tead-dependent transcription is highly abundant in differentiated trunk muscles at 2 and 3 days after fertilization (Fig. 3A) (115). Similarly, the expression of genes that encode proteins characteristic of differentiated muscle, such as cardiac troponin T (Tnnt2) (116) and α-actin (117) in mouse, relies at least partially on MCAT response elements (118). One possible explanation is that different Tead, Yap, Taz, and Vgll complexes may selectively target myoblast or differentiated muscle genes, leading to differential expression. Similar target specificity occurs in flies where the homologs of Vgll and Teads homologs form complexes that target different genes than complexes of Yorkie and Tead homologs (26, 119). In mammals, Yap1 is highly expressed in myoblasts and inhibits differentiation (107, 108), whereas Wwtr1 [Taz (120122)], Vgll2 (123126), and Tead4 (127) are more highly expressed in differentiated muscle and promote differentiation. A second possibility is that different Tead, Yap, Taz, and Vgll complexes target the same genes but exert different effects. During neurogenesis, the transcription factors Sox2, Sox3, and Sox11 are expressed at different stages of neuronal differentiation and bind to identical DNA loci; however, the function of the different isoforms can be either to preselect genes for expression or to actively promote their expression (128). A third possibility is that Yap or Taz could not only coactivate MCAT elements but also bind and coactivate additional transcription factors such as Smads (75, 129) or Tbx5 (55, 76) to drive myoblast or myotube and differentiated muscle-specific gene expression.

Fig. 3 The Hippo signal transduction network and skeletal muscle.

(A) Green fluorescent protein (GFP) fluorescence in a 4xGTIIC:dGFP (MCAT reporter) zebrafish at 2 days after fertilization. Note the intense GFP signal in skeletal muscle (arrowheads). Image reproduced with permission from (115). (B) Transmission electron micrograph of a satellite cell (sc) between the plasmalemma and basal lamina. Image reproduced with permission from (4). (C) Immunofluorescence of ex vivo muscle fibers showing that Yap abundance (red) is higher in activated (48 hours, MyoD+) than in quiescent (0 hours, Pax7+) satellite cells. Image reproduced with permission from (108).

The Hippo pathway in satellite cells and skeletal muscle regeneration

In skeletal muscle, nuclei within differentiated muscle fibers do not divide and, thus, do not contribute to regeneration. Resident Pax7-expressing stem cells, called satellite cells (4), proliferate and differentiate in response to injury to give rise to new muscle (130). Satellite cells reside between the basal lamina and the plasma membrane of differentiated muscle cells (4, 5, 131) (Fig. 3B). The nuclei of satellite cells account for between 1.4 and 7.3% of all nuclei within an adult human muscle (132). Satellite cells undergo self-renewing cell divisions and, when stimulated, can differentiate into muscle fibers (2). Satellite cells are essential for regeneration (3), but not for short-term hypertrophy after overload (133).

Hippo signaling is likely to play a role in the proliferation and differentiation of satellite cells. The expression of Yap1 increases about threefold during activation of mouse satellite cell grown in mitogen-rich medium (108) (Fig. 3C), and overexpression of human YAP S127A in activated satellite cells increases proliferation and inhibits differentiation (108), consistent with the activation and function of Yap in other stem and progenitor cells (51, 53, 54). Additional evidence suggests that Yap activity in satellite cells may be regulated by pathways that crosstalk to the Hippo pathway. S1P, which activates Yap through GPCR signaling (69, 134), promotes the proliferation of satellite cells (135). Notch, which can increase the expression of Yap1 in cortical neural stem cells (83), promotes satellite cell proliferation (136). Recent reports suggest that Notch can also promote satellite cell self-renewal (137) and niche colonization (138). Wnt signaling regulates myogenesis during development and differentiation of satellite cells (139). Yap activates the expression of Bmp4 in satellite cells (108), and Bmp4 protein promotes proliferation and inhibits differentiation of satellite cells (140). Moreover, overexpression of YAP S127A in muscle satellite cells increases the expression of genes encoding proteins angiomotin-like 2 and Frmd6 (also known as Willin) and decreases the expression of genes that encode GPCRs (107). Because these changes in gene expression should result in inhibition of Yap activity (68, 141, 142), this observation suggests that negative feedback mechanisms may serve to limit Yap-dependent cell proliferation. Thus, these data suggest that Yap may be an important regulator of the proliferation of activated satellite cells, and future studies should test the role of Yap on skeletal muscle regeneration in vivo.

The Hippo pathway in terminally differentiated skeletal muscle

In adult humans, muscle fibers can be up to ~20 cm long (143), and a single fiber may contain several tens of thousands of nuclei (144). The human vastus lateralis of young males comprises ~400,000 to 900,000 muscle fibers, and the number of fibers decreases during aging (145). Fibers can be distinguished into slow type I, intermediate type IIa, and fast type IIx and IIb fibers based on the presence of myosin heavy chain isoforms and on the abundance and isoforms of other motor, metabolic, and mitochondrial proteins (146, 147). The number of fibers and the percentage of different fiber types vary greatly both within and between differing muscles of the body and among individuals (145, 148, 149). Muscle fibers hypertrophy in response to overload (for example, as a result of resistance training), and increase mitochondrial biogenesis and change the concentrations and isoforms of motor and metabolic proteins in response to endurance training (147). Thus, differentiated skeletal muscle has a high degree of plasticity, enabling it to change its force production and metabolic capacity in response to various types of stimuli.

Hippo signaling may regulate gene expression in differentiated skeletal muscle. Genes and reporter genes with MCAT elements are actively expressed in differentiated skeletal muscle (115117, 150). Acute resistance exercise increases expression of the genes encoding cysteine-rich angiogenic inducer 61 and connective tissue growth factor (Ctgf) (151), the latter of which has three MCAT elements in its proximal promoter (62). These genes are frequently used as marker genes for Yap and/or Taz activity [for example (152)], suggesting that Yap or Taz may be activated by acute resistance exercise. Teads have been shown to regulate α-actin promoter activity in a model of stretch overload–induced hypertrophy in chicken (117). Overexpression of Tead1 in muscle fibers in mice causes a fast-to-slow fiber type transition, but not hypertrophy (153). Moreover, denervation of fast, but not slow, skeletal muscle induces atrophy and increases the expression of Mst1 (154). Mst1 can phosphorylate Ser207 of the forkhead transcription factor Foxo3a, promoting muscle atrophy in mice (154) and thereby increasing the expression of genes encoding skeletal muscle atrophy–regulating factors known as atrogins (155). Future studies are needed to clarify the role of the Hippo pathway in overload-induced hypertrophy, the regulation of muscle fiber type–specific gene expression, muscle atrophy, and other related phenomena.

Hippo pathway and myopathies

Perturbation of Hippo signaling may contribute to the pathology of different myopathies, including muscular dystrophies (10). We recently found that the expression of human YAP S127A in muscle fibers in adult mice causes a “fulminant” myopathy characterized by atrophy, signs of centronuclear myopathy, deterioration, and, after several weeks, death of the mice (156). The WW domain–containing protein Bag3 (Bag family molecular chaperone regulator 3) interacts directly with the Hippo pathway proteins angiomotin 1, angiomotin 2, and Lats1, and is a positive regulator of Yap and Taz and of Ctgf expression (157). Bag3 knockout causes a fulminant muscular dystrophy in mice (158), and in humans, loss-of-function mutations of Bag3 are associated with severe childhood muscular dystrophy (159). Loss-of-function mutations in the WW domain–containing protein dystrophin (DMD) causes either Duchenne’s (severe) or Becker’s (mild) muscular dystrophy in humans (10, 160). It is unknown whether DMD, or the related protein utrophin, uses WW domains to interact with the Hippo pathway and whether mutations in these proteins perturb Hippo signaling and thereby contribute to the pathology of muscular dystrophies. However, considering the relatively small number of human proteins with WW domains (39), it is intriguing that mutations in these proteins cause human muscular dystrophy.

Other studies further support a role for perturbation of Hippo signaling in myopathies. The Hippo pathway target gene CTGF is highly expressed in muscles of patients with muscular dystrophy (161). Overexpression of Ctgf in mouse skeletal muscle is sufficient to cause muscular dystrophy (162), whereas knockout of Ctgf in a mouse model of muscular dystrophy partially ameliorates the pathology (163). Laminopathies, which can result in muscular dystrophy, are diseases that result from mutation in LMNA, which encodes the nuclear lamina protein lamin A/C (164). Human myoblasts with LMNA mutations have impaired ability to produce appropriate cytoskeletal rearrangements in response to mechanical stimuli, such as the changes in substrate stiffness (165). Consistent with the role of Hippo in mechanotransduction (64, 65), YAP is more active, and CTGF expression is higher in LMNA mutant compared to wild-type myoblasts cultured on substrates with the same stiffness (165). Collectively, these results suggest that dysregulation of Hippo signaling in skeletal muscle due to the mutations in DMD, BAG3, or LMNA may result in changes in expression of Hippo target genes such as CTGF and thereby contribute to pathology.

Myopathies resulting from nongenetic causes may also involve Hippo signaling. Statins are widely prescribed drugs that act on the mevalonate pathway, the activation of which increases Yap and Taz transcriptional activity (166, 167). Statins induce myopathies with symptoms ranging from muscle weakness to rhabdomyolysis in as many as 1.5 million patients per year (168). Thus, investigating whether the inhibition of Yap and Taz contributes to statin-induced myopathies may have important clinical ramifications.

Hippo and rhabdomyosarcoma

Rhabdomyosarcomas, diagnosed on the basis of the presence of rhabdomyoblasts, are the most common soft tissue sarcomas in children and adolescents (11). Rhabdomyosarcomas are classified as embryonal (ERMS), alveolar (ARMS), and pleomorphic (anaplastic). ERMS occur in infants and young children, ARMS occur in adolescents and young adults and have a poorer prognosis than ERMS, and pleomorphic rhabdomyosarcomas occur in adults and are relatively rare (11). About 70 to 80% of ARMS express PAX3-FOXO1 or PAX7-FOXO1 fusion genes, which indicates a poorer prognosis than other ARMS (169). Expression of PAX3-FOXO1 in mice with homozygous loss of the gene encoding tumor protein 53 (p53) or homozygous loss of the gene encoding cyclin-dependent kinase inhibitor 2A (CDKN2A, also known as p16/Ink4A) gives rise to tumors that resemble human ARMS (170), supporting the tissue-specific, oncogenic potential of this fusion gene.

Given that YAP S127A overexpression drives proliferation and inhibits differentiation of C2C12 myoblasts (107) and activated satellite cells (108), deregulation of Hippo signaling may contribute to the pathogenesis of rhabdomyosarcoma. The Ras association (RalGDS/AF-6) domain family (Rassf) of proteins contain SARAH (Salvador, Rassf, and Hippo) domains that bind to and inhibit Mst1 and Mst2 (171). The expression of the tumor suppressor RASSF4 is increased in PAX3-FOXO1–positive ARMS (172). The PAX3-FOXO1 fusion protein directly binds to the 5′ enhancer of RASSF4, and knockdown of RASSF4 in cultured ARMS cell lines reduces cell proliferation and the survival of mice with ARMS xenotransplants (172). Similar to other Rassf proteins, RASSF4 binds and inhibits MST1 in human myoblast and ARMS cell lines (172); however, a functional connection between RASSF4 and YAP in ARMS has not been identified.

Hippo signaling may also be important in the development of ERMS. YAP is more abundant in the nucleus of ERMS compared to ARMS patient samples, which, in some cases, may be explained by increased copy number of the YAP1 locus (173). Overexpression of YAP S127A in activated, but not quiescent, satellite cells causes ERMS-like tumors in mice with high penetrance and a short latency to tumor onset. Cessation of YAP S127A transgene expression in YAP S127A–driven ERMS-like tumors in mice or a knockdown of YAP in human ERMS cells causes differentiation of tumor cells into myosin heavy chain–expressing muscle fibers. Knockdown of YAP reduces proliferation and anchorage-independent growth in human ERMS cells in culture and decreases the tumor burden in mice with human ERMS xenotransplants. Combined analyses of Yap and Tead1 genome-wide chromatin immunoprecipitation and quantitative reverse transcription polymerase chain reaction (RT-qPCR) studies and cDNA microarrays of YAP S127A–driven ERMS-like tumors suggest that in mouse ERMS myoblasts, Yap and Tead1 bind to and increase the expression of genes that regulate cell proliferation (Ccnd1 and Cdc6), as well as oncogenes and cancer-related genes (Met, Myc, and Birc5), and conversely repress the expression of genes typically expressed in terminally differentiated skeletal muscle (Myl4, Myh2, Mybph, and Tnnc2) (173). Other studies provide additional evidence for a connection between Hippo signaling and rhabdomyosarcoma. One case of a spindle cell variant of ERMS contained a TEAD1-NCOA2 fusion gene (174), and knockdown of the tyrosine kinase Yes, which binds and phosphorylates Yap (27, 55), reduces proliferation of ARMS and ERMS cell lines (175). Collectively, these studies suggest that Hippo pathway dysregulation causes or contributes to rhabdomyosarcoma, identifying the Hippo pathway as a treatment target for these cancers.

The Hippo pathway in heart development

The heart is a heterogeneous organ comprising cardiomyocytes, endocardial cells, valvular components, connective tissues, cells of the electrical conduction system, as well as the smooth muscle and endothelial cells of the coronary arteries and veins (176). It develops from mesodermal progenitor cells located in the anterior region of the primitive streak and the proepicardium (176). These cardiac progenitor populations migrate away from the primitive streak and give rise to the primary and secondary heart fields. Cells in the primary heart field differentiate and form a linear heart tube that begins to beat. The heart tube grows unevenly, using cells from the primary heart field to form the lower bulk of the heart. Cells from the secondary heart field migrate to the heart tube to form the outflow tract. Heart looping enables the developing heart tube to fold within the pericardial cavity (6, 176). The growth of the heart during embryonic development is driven by cardiomyocyte and precursor cell proliferation, but just after birth, cardiomyocyte proliferation stops and heart growth occurs primarily by cardiomyocyte hypertrophy (176). These processes are regulated through complex intrinsic and extrinsic signaling events among cells in both heart fields.

Several lines of evidence suggest that Hippo signaling is involved in heart development. Pathways that crosstalk with the Hippo pathway, including BMP, Wnt, Notch, and Shh (51, 54, 71, 177, 178), regulate transcription factors critical for heart development. The genes encoding the transcription factors Gata4 and Nkx2.5 are expressed in the primary and secondary heart fields, along with the gene encoding Tbx5, which is expressed only in the primary heart field, and these proteins work in concert to promote cardiomyocyte differentiation and heart maturation (179). Tbx5, which can be coactivated by both Yap and Taz (55, 76), is a critical mediator of embryonic heart development (180), as mutations in TBX5 cause Holt-Oram syndrome, which is characterized by heart and limb abnormalities (181). Conditional deletion of Sav1, Lats2, or Mst1 and Mst2 in Nkx2.5-positive cardiomyocytes in mice increases proliferation, leading to cardiomegaly and perinatal lethality (Fig. 4A) (182). Crossing Sav1 conditional knockout mice to those with heterozygous deletion of Ctnnb1, which encodes β-catenin, rescues cardiomyocyte hyperproliferation, indicating that Wnt–β-catenin signaling acts downstream of Hippo signaling (182). Cardiomyocytes from Sav1 conditional knockout mice have increased expression of Wnt–β-catenin target genes, including the transcription factors Sox2 and Snail2. Because Sox2 increases the expression of Yap (84), this could represent a positive feedback mechanism. Moreover, using either Nkx2.5- or Tnnt2-Cre–mediated excision to create conditional deletion of Yap1 in cardiomyocytes during embryonic development results in lethality between E10.5 and E16.5 (183, 184). Hearts of these mice have normal cardiac looping and chamber formation, but reduced cardiomyocyte proliferation and smaller, thinner ventricles, indicating that Yap is an important regulator of embryonic cardiac growth.

Fig. 4 The Hippo signal transduction network and the heart.

(A) Conditional cardiomyocyte-specific knockout of Sav1 (Salv) results in cardiac hypertrophy (cardiomegaly) associated with increased cardiomyocyte proliferation. ra, right atrium; la, left atrium; rv, right ventricle; lv, left ventricle. Image reproduced with permission from (182). (B) Serial sections of Masson’s trichrome–stained wild-type (WT) and αMHC promoter–driven Yap S112A transgenic (Tg1 and Tg2) hearts showing scar tissue (blue) 21 days after transient ligation of the left anterior descending artery at postnatal day 7. Image reproduced with permission from (189).

The Hippo pathway in cardiac regeneration and growth

The potential for cardiomyocyte proliferation in adult mammals is low and declines with age (185187). Therefore, after injury, the adult mammalian myocardium replaces lost cardiomyocytes with fibrotic scar tissue, and the functional output of the heart is reduced. In contrast, the neonatal mouse heart can regenerate after partial resection or ischemia within the first week after birth (7). In these animals, mature cardiomyocytes, rather than distinct cardiac stem or progenitor cells, undergo proliferation to promote regeneration (188). In neonatal mice, conditional deletion of Yap1 in cardiomyocytes, using αMHC-Cre, impairs heart regeneration after ischemia (189). In contrast, cardiomyocyte-specific overexpression of constitutively active (S112A) murine Yap (homologous to human YAP S127A) reduces fibrosis and promotes cardiac regeneration beyond postnatal day 7 (Fig. 4B) (189). This is consistent with the observation that overexpression of active Yap promotes cardiomyocyte proliferation both in vivo and in cultured cardiomyocytes (183, 184, 190). In response to myocardial infarction, Yap can be found in the nuclei of cardiomyocytes surrounding the site of injury (190), suggesting increased activation of Yap at the border of the infarcted area. Moreover, adult mice with conditional Yap1 knockout using αMHC-Cre display greater injury, increased apoptosis, and reduced proliferation of cardiomyocytes after myocardial infarction (190). However, it is unknown whether Yap is involved in cardiac remodeling and myocardial regeneration after myocardial infarction, and if so, which cells express Yap1, how are they activated, and how do they contribute to the functional recovery of the heart. Conditional deletion of Sav1, or Lats1 and Lats2, in cardiomyocytes using Nkx2.5-Cre stimulates proliferation in uninjured hearts and increases proliferation, reduces fibrosis, and improves cardiac function in response to partial heart resection at postnatal day 8 in mice, as well as myocardial infarction in adult mice (191). Thus, inhibition of Hippo signaling promotes heart regeneration through increased cardiomyocyte proliferation.

Whether Hippo signaling also plays a role in cardiac hypertrophy is less clear. Conditional deletion of Yap1 in cardiomyocytes in Tnnt2-Cre mice during development does not affect the size of cardiomyocytes (184). Likewise, postnatal retro-orbital injection (192) of adenovirus encoding GFP and Tnnt2-Cre into mice with heterozygous floxed alleles of Yap1 results in targeted deletion of Yap1 in a small percentage of cardiomyocytes. GFP-positive (Yap1-deleted) cardiomyocytes have normal size at baseline and after pressure overload stress (184). Likewise, αMHC promoter–driven YAP S127A or S112A in the heart of postnatal mice does not alter cardiomyocyte size (184, 189). In contrast, conditional overexpression of wild-type Yap in neonatal cardiomyocytes increases size and induces expression of genetic markers of the hypertrophic fetal cardiomyocytes, including genes that encode atrial natriuretic factor, brain natriuretic peptide, and β-myosin heavy chain (190). Exposing cultured cardiomyocytes to the GPCR agonist phenylephrine induces hypertrophy and increases expression of atrial natriuretic factor mRNA, and expression of short hairpin RNAs targeting endogenous Yap1 attenuates the effects of phenylephrine in other cell types (69), suggesting that phenylephrine may inhibit Hippo signaling in these cells. Moreover, adult mice with conditional deletion of Yap1 with αMHC-Cre have reduced cardiomyocyte hypertrophy in response to myocardial ischemia (190). Thus, Yap can either promote or inhibit cardiomyocyte hypertrophy, a discrepancy that may be explained by the method of gene targeting, the timing or duration of Yap depletion, the type of myocardial stress, or the mutational status of Yap1. Nevertheless, these studies reveal that Yap, and potentially Hippo signaling, can influence cardiomyocyte hypertrophy.

Mst1 is activated by oxidative stress and promotes cardiomyocyte death (193, 194). Overexpression of Mst1 promotes apoptosis of cultured cardiomyocytes. In contrast, the inhibition of endogenous Mst1 by expression of a kinase-inactive (K59R) variant, which functions as a dominant negative, reduces cardiomyocyte apoptosis induced by pharmacological inhibition of protein kinase C or protein phosphatases (195). Furthermore, cardiomyocyte-specific overexpression of Mst1 causes a dose-dependent increase in cardiomyocyte apoptosis: mice with low amounts of Mst1 overexpression have a modest increase in basal apoptosis, whereas mice with higher expression have a robust increase in apoptosis and rapidly progress to dilated cardiomyopathy, heart failure, and premature death (195). The SARAH domain of Rassf1a binds to Mst1 and promotes its activation in response to pressure overload in mouse hearts (196). Cardiomyocyte-specific overexpression of Rassf1a increases Mst1 activation and exacerbates cardiac dysfunction induced by pressure overload. In contrast, overexpression of Rassf1a with a mutation in the SARAH domain (L308P), which cannot bind Mst1, acts as a dominant negative and protects against cardiac injury in this context. Furthermore, cardiomyocyte-specific deletion of Rassf1A attenuates Mst1 activation and is protective against injury and heart failure due to pressure overload. Knockout of Rassf1A in mice does not protect against fibrosis and cardiac hypertrophy induced by tumor necrosis factor–α (196), indicating cell type–specific effects of Rassf1A in heart.

Yap is also important for heart homeostasis in adult mice. Homozygous deletion of Yap1 in cardiomyocytes of postnatal mice using αMHC-Cre results in a rapidly developing dilated cardiomyopathy, and these mice die of heart failure by 12 weeks of age (190). Conditional knockout of Yap1 in the heart leads to robust increases in cardiomyocyte apoptosis and fibrosis with suppressed cardiac function, possibly due to suppression of the prosurvival kinase Akt. Activation of Hippo signaling may contribute to arrhythmogenic cardiomyopathy. Neurofibromin 2 (Nf2), a cytoskeletal protein that modulates Yap activity through physical interaction with the Hippo pathway members Ww45 and Kibra (197), and Mst1 are activated in a mouse model of arrhythmogenic cardiomyopathy and in human patients (198). The inactivation of Yap in these contexts may contribute to increased adipogenesis, which is a hallmark and contributing factor in arrhythmogenic cardiomyopathy. Thus, Yap is likely a mediator of cardiomyocyte survival, proliferation, and signal transduction in the adult mammalian heart.

The myocardin family of coactivators bind to the transcription factor Srf (serum response factor) and may affect both Yap and Srf signaling in muscle cells (199). Myocardins inhibit skeletal muscle differentiation (200) and drive the expression of contractile genes in smooth muscle and cardiac cells (201). Yap physically interacts with myocardin and inhibits its ability to elicit expression of contractile genes through interaction with Srf (199). The presence of a PPxY motif in myocardin suggests that it could bind to the WW domains of Yap, although this has not been empirically tested.

The Hippo signal transduction network as a drug target

Given the importance of Hippo signaling in skeletal and cardiac muscle biology, drugs that target this pathway may be relevant for therapeutic purposes. Verteporfin, which is used as a photosensitizer in photodynamic therapy in the eye (202), can disrupt the interaction between Yap and Tead in human embryonic kidney (HEK) 293 cells, inhibit activation of Tead target genes, and reverse Yap-induced hepatomegaly in vivo (203). Therefore, verteporfin may be useful in treating hyperproliferative muscle disorders, such as rhabdomyosarcoma. In addition, drugs that target GPCRs, such as β-blockers (69, 204206), or the mevalonate pathway, such as statins (166, 167), potentially could be used to target the Hippo pathway in muscle disease or to promote regeneration. Statins and GPCR-targeting drugs are among the most widely prescribed drugs, and it has been estimated that 30 to 50% of all medications exert their effect via GPCRs (207). Although not all statins and GPCR-targeted drugs necessarily affect Hippo signaling, this suggests that the effects and side effects of these drugs should be considered in light of potential impact on Hippo signaling. The availability of approved drugs that may target the Hippo pathway provides an exciting opportunity to test whether they can be used to treat skeletal muscle or heart diseases.

Conclusions and outlook

Since the late 1980s, proteins in the Hippo signal transduction network have been identified as regulators of skeletal and cardiac muscle gene expression, development, organ growth, stem cell function, regeneration, and disease. Additionally, in nonmuscle cells, proteins in the Hippo pathway interact with proteins involved in skeletal and cardiac muscle biology, suggesting that these mechanisms may exist in muscle and the heart. However, our knowledge of the Hippo signal transduction network in skeletal and cardiac muscle is modest compared to that of other signal transduction pathways in these tissues. Thus, there are still many unanswered questions in this field, for example:

• Can information from the ENCODE project (208) and other genome-wide analyses that have fundamentally changed our understanding of chromatin and gene regulation lead to better understanding of the regulation of gene expression by the Hippo pathway? Genome-wide chromatin immunoprecipitation analysis shows that Yorkie DNA binding correlates with the activating chromatin mark, trimethylation of Lys4 of histone H3. The same study also shows that Yorkie directly binds to chromatin-remodeling complexes (209). Thus, do Yap and Taz have similar functions to Yorkie? How do the Hippo transcriptional regulators interact with chromatin and chromatin-remodeling proteins throughout the mammalian genome, especially in skeletal muscle and heart cells, and during tumorigenesis in rhabdomyosarcoma?

• Given that Yap and Taz typically promote cell proliferation, which is associated with progressive shortening of telomeres, is there a relationship between the Hippo pathway and aging in skeletal muscle and the heart?

• Is the Hippo pathway involved in regulating the skeletal and heart muscle adaptation to endurance and resistance exercise training or in mediating other forms of muscle plasticity?

• Given that several approved drugs target Hippo signaling, can these drugs be used to treat skeletal and heart muscle diseases, including muscular dystrophy, cellular damage after myocardial infarction, and rhabdomyosarcoma? Are there other molecules that can target Hippo signaling specifically through Hippo pathway proteins that are only present in muscle?

REFERENCES AND NOTES

Acknowledgments: We would like to thank A. Farooq, B. Link, J. Miesfeld, W. J. Hong, A. Pobbati, S. Veeraraghavan, T. Heallen, and J. F. Martin for sending high-resolution figures of their work. Funding: Research in the Aberdeen lab has been supported by a Medical Research Council project grant (99477; H.W., P. S. Zammit, and C. De Bari), Sarcoma UK grant (H.W., G. Murray, R. Urcia, and M. Jaspars), Friends of Anchor grant (H.W. and C. De Bari), and a grant by Tenovus Scotland (G11/05; C. De Bari and H.W.). Research in the Sudol lab was supported by the Mechanobiology Institute of the National University of Singapore. Research in the Sadoshima lab has been supported by NIH grants HL112330, HL102738 (J.S.), and HL122669 (D.P.D.R.) and American Heart Association Scientist Development grant 11SDG7240066 (D.P.D.R.). Competing interests: The authors declare that they have no competing financial interests.
View Abstract

Navigate This Article