The Hippo Signaling Pathway in Development and Disease

Abstract

The Hippo signaling pathway regulates diverse physiological processes, and its dysfunction has been implicated in an increasing number of human diseases, including cancer. Here, we provide an updated review of the Hippo pathway; discuss its roles in development, homeostasis, regeneration, and diseases; and highlight outstanding questions for future investigation and opportunities for Hippo-targeted therapies.

Introduction

Growth and patterning are fundamental and intertwined developmental processes that sculpt the final size and shape of an organ. Decades of developmental genetics, epitomized by saturation mutagenesis screens for patterning defects in Drosophila embryos, have led to the discovery of a handful of evolutionarily conserved signaling pathways that control pattern formation in animal development (Barolo and Posakony, 2002). The revelation that the same developmental morphogens are used reiteratively to specify cell fates in diverse organs and organisms provides a unifying theme for modern developmental biology. Yet, for much of the post-molecular era of developmental biology, our knowledge of how growth is regulated in a developing organ had been limited (Hariharan, 2015). Indeed, as developmental morphogens often influence both growth and patterning, whether morphogen signaling alone is sufficent to explain growth control has been debated (Schwank and Basler, 2010).

Against this backdrop, the Hippo signaling pathway was discovered around the turn of the 21^st century as a potent mechanism that restricts the growth of Drosophila tissues. The first four tumor suppressors linked to this pathway, warts (wts) (Justice et al., 1995; Xu et al., 1995), salvador (sav) (Kango-Singh et al., 2002; Tapon et al., 2002), hippo (hpo) (Harvey et al., 2003; Jia et al., 2003; Pantalacci et al., 2003; Udan et al., 2003; Wu et al., 2003), and mob as tumor suppressor (mats) (Lai et al., 2005), were isolated based on the massive overgrowth of homozygous mutant clones in genetic mosaics. Interestingly, two of these tumor suppressor genes, wts and hpo, encode kinases of the Nuclear Dbf2-related (NDR) family and the Sterile-20 family, respectively. Initial biochemical studies demonstrated that these tumor suppressors constitute a kinase cascade in which Hpo phosphorylates and activates Wts, and Wts in turn represses the transcription of target genes such as diap1, presumably through an unknown transcriptional regulator (Wu et al., 2003). These revelations eventually led to the discovery of the transcriptional coactivator Yorkie (Yki) as a direct and physiological substrate of Wts (Huang et al., 2005), followed by the identification of Scalloped (Sd) as the DNA-binding transcription factor partner of Yki in regulating target gene transcription (Zhang et al., 2008; Wu et al., 2008). Together, these early studies in Drosophila have established the core kinase cascade of the Hippo pathway leading from the kinase Hpo to the nuclear Yki-Sd transcriptional complex.

The elucidation of the Hippo kinase cascade and the evolutionary conservation of its constituents have sparked much interest in understanding the physiological function and molecular regulation of this pathway in diverse animal phyla. The existence of an evolutionarily conserved kinase cascade culminating on the phosphorylation of YAP and TAZ, the mammalian counterpart of Yki, was soon demonstrated in mammalian cells (Dong et al., 2007; Lei et al., 2008; Zhao et al., 2007). In fact, these components are conserved even in the closest unicellular relatives of metazoans, suggesting that Hippo signaling represents an ancient mechanism predating the emergence of multicellularity (Sebé-Pedrós et al., 2012). A decade of intense research has expanded the Hippo kinase cascade into a complex signaling network, linking the core kinase cascade to diverse signals such as cell adhesion and polarity, mechanical forces, soluble factors, and various stress signals. Recent studies have further implicated the Hippo pathway in diverse physiological and pathological processes beyond developmental size control, such as cell fate determination, stem cell regulation, regeneration, immunity, and cancer. This review is intended to provide an updated view of the Hippo signaling network in normal physiology and disease, with a focus on recent advances in both Drosophila and mammals.

The Kinase Cascade of the Hippo Pathway

In the canonical Hippo kinase cascade, the Hpo-Sav complex (MST1/2-SAV1 in mammals) phosphorylates and activates the Wts-Mats complex (LATS1/2-MOB1A/B in mammals). The activated Wts-Mats complex then phosphorylates and inactivates Yki (YAP/TAZ in mammals). Recent studies have not only elucidated detailed molecular mechanisms in the canonical Hippo kinase cascade but also identified additional kinases and phosphatase converging onto the cascade (Figure 1). Since an evolutionarily conserved protein may have different names in Drosophila and mammals, we often describe conserved molecular interactions in this review using the names of the corresponding homologues separated by slashes.

Figure 1. Kinase Activation Mechanisms in the Hippo Kinase Cascade.

Hpo/MST is activated by Tao-1/TAOK-mediated phosphorylation or trans-autophosphorylation of its activation loop site (blue circles). Sav/SAV1 forms a heterotetramer with Hpo/MST to facilitate Hpo/MST activation and localization to the plasma membrane. Activated Hpo/MST then phosphorylates multiple sites (yellow circles) in its linker region. Binding of these phosphorylation sites by Mats/MOB1 helps recruit Wts/LATS to Hpo/MST. Hpo/MST then phosphorylates the HM of Wts/LATS to promote Wts/LATS autophosphorylation and activation. MAP4Ks function redundantly with Hpo/MST to phosphorylate the HM of Wts/LATS leading to its activation. Conversely, the linker phosphorylation sites of Hpo/MST recruit the STRIPAK PP2A phosphatase complex to dephosphorylate and inactivate Hpo/MST, therefore creating a negative feedback to restrict Hpo/MST activity. Upstream regulators such as KIBRA and Mer/NF2 facilitate the kinase cascade by recruiting Wts/LATS to the plasma membrane for its activation by Hpo/MST.

Kinase Activation Mechanisms in the Canonical Kinase Cascade

The activation of Hpo or its mammalian counterpart MST1/2 requires phosphorylation of a key residue within the activation loop (Thr195 for Hpo and T183/T180 for MST1/2). Tao-1 (TAOK1/2/3 in mammals) was identified as an upstream kinase catalyzing this event, although loss of Tao-1 leads to much weaker tissue overgrowth than the hpo mutation, indicating additional mechanisms of Hpo activation (Boggiano et al., 2011; Poon et al., 2011). Besides Tao-1/TAOK-mediated activation, Hpo/MST intrinsically forms homodimers through its C-terminal Sav-Rassf-Hpo (SARAH) domain, and in such homodimers each Hpo/MST subunit can activate the other subunit by trans-phosphorylating the same activation loop site (Ni et al., 2013; Praskova et al., 2004). Interestingly, the dimerization and hence autoactivation of Hpo/MST is modulated by two other SARAH-domain-containing proteins, Sav/SAV1 and RASSF; whereas Sav/SAV1 promotes Hpo/MST autoactivation by forming heterotetramers comprised of two subunits of each protein (Bae et al., 2017), the RASSF family proteins preclude autoactivation by forming RASSF-Hpo/MST heterodimers (Ni et al., 2013; Praskova et al., 2004). Despite these insights, it remains unclear whether the Hpo/MST homodimers, the Hpo/MST-RASSF heterodimers, and the Hpo/MST-Sav/SAV1 tetramers mediate different upstream signals to the Hippo kinase cascade.

Mechanistic studies in both Drosophila and mammalian cells have established that activation of the effector kinase Wts or its mammalian counterpart LATS1/2 requires two sequential phosphorylation events: Hpo/MST-mediated phosphorylation at the C-terminal hydrophobic motif (HM) of Wts/LATS followed by autophosphorylation at its T-loop motif (reviewed in Hergovich, 2016). The exact roles of the adaptor protein Sav/SAV1 in this process are less certain. Sav/SAV1 has been suggested to function either downstream of Hpo/MST by facilitating Hpo/ MST-Wts/LATS association (Harvey et al., 2003) or upstream of Hpo/MST by recruiting Hpo/MST to plasma membrane (Su et al., 2017; Yin et al., 2013), or antagonizing a Hpo/MST-inhibitory phosphatase complex (Bae et al., 2017) called striatin-interacting phosphatase and kinase (STRIPAK) (Goudreault et al., 2009; Ribeiro et al., 2010). Resolving this uncertainty will require sensitive genetic epistasis and structure-function analyses with more quantitative readout.

The other adaptor protein in the core kinase cascade, Mats/MOB1, is best known to promote Hippo signaling by facilitating the intrinsic kinase activity of Wts/LATS toward Yki/YAP/TAZ (Lai et al., 2005). Interestingly, Mats/MOB1 itself is phosphorylated by Hpo/MST at two conserved residues (Thr12 and Thr35), although the physiological role of this phosphorylation remains uncertain. While some studies suggested that this phosphorylation enhances Mats/MOB1-Wts/LATS interactions (Ni et al., 2015; Praskova et al., 2008), others reported a negligible effect of this phosphorylation event on Wts-Mats interaction (Vrabioiu and Struhl, 2015) or YAP phosphorylation (Plouffe et al., 2016; Zhou et al., 2009). Besides these mechanisms, Mats/MOB1 may further enhance Hippo signaling by binding to the autophosphorylated linker in Hpo/MST (Ni et al., 2015; Figure 1), although the physiological importance of this binding was questioned by the ability of mutant forms of Mats/MOB1 defective in Hpo/MST binding to rescue mats null flies (Kulaberoglu et al., 2017). Clearly, more investigation is needed to better understand the activation mechanism of the Hippo kinase cascade, especially the roles of Mats/MOB1 and Sav/SAV1.

Negative Regulation of Hpo/MST Activity by the STRIPAK Phosphatase Complex

The intrinsic ability of Hpo/MST to homodimerize and autoactivate suggests the existence of mechanisms that prevent its excessive activation. Recent studies revealed a built-in negative feedback mechanism whereby autophosphorylation of an unstructured linker in Hpo/MST creates docking sites for SLMAP, which in turn recruits the STRIPAK phosphatase complex to dephosphorylate and inactivate Hpo/MST (Bae et al., 2017; Zheng et al., 2017; Figure 1). Significantly, regulation of the STRIPAK complex also provides a molecular conduit to couple certain upstream physiological signals to Hippo signaling. For example, bacterial infection activates Hippo signaling in Drosophila fat body cells to promote antimicrobial defense through Toll-receptor-mediated phosphorylation and degradation of a STRIPAK subunit called Cka (Liu et al., 2016). Whether the STRIPAK complex also mediates other regulatory inputs into the Hippo pathway remains to be determined.

Alternative Kinase Cascade: Resolving the Puzzle of Hpo/MST-Independent Pathway Activation

The first evidence for Hpo/MST-independent activation of Wts/LATS came from the unexpected finding that Mst1/2 null mouse embryonic fibroblasts display normal LATS1/2 and YAP phosphorylation (Zhou et al., 2009). It was soon followed by additional reports showing that YAP/TAZ phosphorylation stimulated by F-actin perturbation was unaffected in Mst1/2 null cells (Yu et al., 2012; Zhao et al., 2012). These puzzling findings were recently resolved by the identification of the MAP4K subfamily of kinases, including Happyhour (Hppy) and Misshapen (Msn) in Drosophila and MAP4K1/2/3/4/5/6/7 in mammals, as alternative Hpo-like kinases that function redundantly with Hpo/MST to directly phosphorylate the HM of Wts/LATS (Li et al., 2014; Meng et al., 2015; Zheng et al., 2015). Besides the MAP4K subfamily, TAOK1 and TAOK3 were also reported to phosphorylate the HM of LATS1/2 in mammalian cells (Plouffe et al., 2016), although this activity is unlikely conserved for their Drosophila counterpart Tao-1 (Boggiano et al., 2011; Poon et al., 2011). In cultured mammalian cells, these alternative Hpo-like kinases act in parallel to MST1/2 to activate LATS1/2 in response to serum starvation, contact inhibition, and F-actin disruption (Meng et al., 2015; Plouffe et al., 2016; Zheng et al., 2015).

It is worth noting that much of the characterization of the Hpo/MST-independent mechanisms has been based on cultured cells. The physiological contexts in which these Hpo-like kinases regulate Wts/LATS activation in vivo are largely undefined. An exception is the Drosophila midgut, where Wts was reported to be activated by Msn in enteroblasts but by Hpo in intestine stem cells and enterocytes (Li et al., 2014). Another potential context to examine the functional contribution of the Hpo-like kinases is mouse epidermis. Epidermis-specific knock out of Mst1/2 in mouse has no discernable phenotype (Schlegelmilch et al., 2011), but deletion of the Lats1/2 cofactors Mob1a/b in the same tissue results in tumorigenesis, suggesting the operation of a potential Mst1/2-independent mechanism of Lats1/2 activation in this context (Nishio et al., 2012). Clearly, more studies are required to understand the division of labor between the canonical and the alternative kinase cascade in vivo.

The Transcriptional Machinery of the Hippo Pathway

The Hippo kinase cascade converges on its nuclear effector Yki/YAP/TAZ to regulate gene expression programs. Phosphorylation of Yki/YAP/TAZ by Hippo signaling inactivates these transcriptional coactivators by excluding them from the nucleus (Dong et al., 2007; Lei et al., 2008; Zhao et al., 2007), and additionally for YAP/TAZ, by promoting their degradation (Liu et al., 2010; Zhao et al., 2010). When Hippo signaling is low, Yki/YAP/TAZ enter the nucleus to drive gene expression (Figure 2). Understanding the function and regulation of the transcriptional machinery of the Hippo pathway is critical for interrogating Hippo signaling in physiology and disease.

Figure 2. The Transcriptional Machinery of the Hippo Signaling Pathway.

When Hippo signaling is low, Yki/YAP/TAZ enter the nucleus, where they use their N-terminal homology (NH) domain to interact with the C-terminal domain of the DNA-binding transcriptional factor Sd/TEAD. The DNA-binding TEA domain of Sd/TEAD targets the Yki/YAP/TAZ-Sd/TEAD complex to Hippo Responsive Elements (HREs) of target genes, where Yki/YAP/TAZ activate target gene transcription by recruiting multiple transcriptional complexes, including the Trr (MLL3/4 in mammals) H3K4 methyltranferase complex (through direct binding of its NcoA6 subunit), the ATP-dependent SWI/SNF chromatin remodeling complex, and the Mediator complex. When Hippo signaling is high, Yki/YAP/TAZ are sequestered in the cytoplasm and Sd/TEAD instead interacts with Tgi/VGLL4 and other unknown corepressors to repress target gene transcription.

The TEF/TEAD Family of DNA-Binding Transcription Factors as Obligatory Partners of Yki/YAP/TAZ in Transcriptional Regulation

Although other DNA-binding transcription factors have been reported to interact with Yki/YAP/TAZ in various contexts, the TEF/TEAD family transcription factors, Sd in Drosophila and TEAD1/2/3/4 in mammals, are the primary and physiologically relevant DNA-binding partners for the Yki/YAP/TAZ coactivators. Not only were they identified as Yki/YAP-binding proteins in multiple unbiased protein-protein interaction screens in both Drosophila and mammals (Vassilev et al., 2001; Wu et al., 2008), they were also shown to bind to various Hippo target genes such as diap1 (Koontz et al., 2013; Wu et al., 2008; Zhang et al., 2008) and CTGF in mammals (Zhao et al., 2008). The physiological importance of Sd/TEAD-Yki/YAP interactions was further supported by the discovery of a disease-causing point mutation in human TEAD1 (TEAD1^Y421H underlying Sveinsson chorioretinal atrophy) (Kitagawa, 2007) and the unbiased recovery of a missense mutant allele in Drosophila Yki (Yki^P88L) that specifically disrupts this interaction (Wu et al., 2008), as well as structural studies of TEAD-YAP co-crystals that independently pinpoint these residues in the protein binding interface (Chen et al., 2010b; Tian et al., 2010; Li et al., 2010). Indeed, multiple genome-wide chromatin occupancy studies revealed significant overlap between YAP- and TEAD-binding sites (Zanconato et al., 2015; Zhao et al., 2008). These observations support Sd/TEAD as the obligatory DNA-binding partners of Yki/YAP in Hippo signaling. A direct implication of this revelation is that any Yki/YAP-induced changes in the expression of Hippo target genes must be genetically dependent on endogenous Sd/TEAD, a premise that has been exploited as a robust genetic assay to validate bona fide Hippo pathway components and target genes in Drosophila (Yu and Pan, 2018).

Default Repression by Sd/TEAD

As expected of a critical Yki partner in Hippo signaling, loss of sd fully rescues tissue overgrowth and elevated target gene transcription caused by Yki overexpression or inactivation of Hippo pathway tumor suppressors (Wu et al., 2008; Yu and Pan, 2018). Strikingly, loss of sd also fully rescues tissue undergrowth and decreased Hippo target gene transcription in yki loss-of-function mutants. These unique genetic properties of sd led to a default repression model (Koontz et al., 2013), which posits that in the absence of Yki, Sd engages transcriptional corepressors such as Tondu-domain-containing growth inhibitor (Tgi) to actively repress the transcription of Hippo pathway target genes, and the physiological function of Yki is to derepress the same target genes by competing and/or neutralizing Sd’s corepressors (Figure 2). This model explains why growth is severely compromised by the loss of function of yki but not that of sd, as the loss of yki represses while loss the of sd derepresses growth-promoting genes. The potent activity of VGLL4, the mammalian counterpart of Tgi, to suppress YAP-driven hepatomegaly and tumorigenesis in transgenic mice supports the conservation of default repression in mammals (Koontz et al., 2013), which provides a potential explanation for the increased expression of YAP target gene Ctgf in Tead1/2 double knockout mouse embryos (Ota and Sasaki, 2008) or Tead4 knockdown C2C12 cells (Benhaddou et al., 2012).

A defining feature of Tgi/VGLL4 is the presence of two Tondu domains, both of which are required for growth suppressive activity. Although the Tondu domains of VGLL4 do not resemble the TEAD-binding domain of YAP in the primary sequence, both adopt a similar 3D structure in co-crystals with TEAD (Jiao et al., 2014), and indeed the corepressor Tgi/VGLL4 and the coactivator Yki/YAP competitively bind to the same C-terminal domain of Sd/TEAD (Koontz et al., 2013). Besides Tgi, Sd also dimerizes with Vestigial (Vg), which contains a single Tondu domain (Halder et al., 1998). Unlike Tgi/VGLL4, which is expressed in multiple tissues and functions as a transcriptional corepressor for Sd/TEAD in Hippo signaling, Vg and its mammalian counterpart VGLL1/2/3 are expressed in a tissue-specific manner, function as transcriptional coactivators for Sd/TEAD, and regulate a distinct set of target genes, even in the same tissue (Simon et al., 2016). What determines the target gene selectivity of Vg versus Tgi and Yki, despite engaging the same DNA-binding transcription factor, remains to be elucidated. It is also unclear what molecular features dictate the repressor function of Tgi versus the activator function of Vg and Yki. Furthermore, as Tgi only partially accounts for the default repressor activity of Sd (Koontz et al., 2013), additional Sd corepressors likely exist. Indeed, a recent study reported the identification of Nerfin-1 and its mammalian homolog INSM1 as Sd/TEAD-binding corepressors, although whether these zinc finger proteins are involved in Sd/TEAD-mediated default repression remains unclear (Guo et al., 2019).

Posttranslational Modification of the Yki/YAP/TAZ-Sd/TEAD Complex as a Nexus for Signaling Crosstalk

Besides Wts/LATS-mediated phosphorylation of Yki/YAP/TAZ by Hippo signaling, the Yki/YAP/TAZ-Sd/TEAD complex is regulated by additional posttranslational modifications. For example, YAP can be phosphorylated by other kinases including the cell cycle regulator CDK1 (Yang et al., 2013), the cellular energy sensor AMPK (Mo et al., 2015; Wang et al., 2015b), the integrin and interleukin-activated SRC family kinases (Li et al., 2016; Taniguchi et al., 2017), the osmotic stress sensitive Nemo-like kinase (NLK) (Hong et al., 2017; Moon et al., 2017), the viral infection-activated kinase IKKε (Wang et al., 2017a), and the nuclear kinase PRP4K (Cho et al., 2018). YAP activity is also regulated by methylation (Oudhoff et al., 2013), O-GlcNAcylation (Peng et al., 2017), and K63-linked polyubiquitination (Yao et al., 2018). Finally, palmitoylation of the TEAD transcription factors is required for their proper folding, stability (Noland et al., 2016), and YAP/TAZ binding (Chan et al., 2016). Together, these posttranslational modifications provide mechanisms for potential crosstalk between Hippo and other signaling pathways.

Hippo Target Genes and Mechanisms of Transcriptional Activation by Yki/YAP/TAZ

Early studies in Drosophila revealed that the Hippo pathway coordinately regulates cell proliferation and cell survival and have thus focused on target genes involved in these processes, such as the cell cycle regulators cyclin E, the cell death inhibitor diap1, and the pro-proliferative and anti-apoptotic microRNA bantam. Recent genome-wide expression profiling and chromatin immunoprecipitation analysis confirmed that genes related to cell proliferation and cell death are enriched among all Hippo target genes in both Drosophila and mammals (Kapoor et al., 2014; Oh et al., 2013; Zanconato et al., 2015). These studies also significantly expanded the repertoire of Hippo target genes to those involved in cell migration, extracellular matrix (ECM) organization, and cytoskeleton organization. Of note, the oncogene c-Myc is an evolutionally conserved Hippo pathway target in both Drosophila and mammals, suggesting that Yki/YAP/TAZ may induce other oncogenic transcriptional factors to further enhance their oncogenic activity (Cai et al., 2018; Neto-Silva et al., 2010; Zanconato et al., 2015). Conversely, another class of Hippo target genes encode upstream negative regulators of Yki/YAP/TAZ, such as Expanded (Ex) (Hamaratoglu et al., 2006), Kibra (Genevet et al., 2010), Crumbs (Crb) (Hamaratoglu et al., 2009), four-jointed (Fj) (Cho et al., 2006), and Wts (Jukam et al., 2013) in Drosophila and Lats2, Kibra, and Nf2 in mice (Chen et al., 2015b; Moroishi et al., 2015). Yki/YAP/TAZ-mediated regulation of these tumor suppressors constitutes a negative feedback that maintains signaling homeostasis and minimizes the detrimental effect of aberrant pathway activation by stress or oncogenic mutations.

The genome-wide analyses further revealed that the majority of YAP/TAZ-binding sites are located at distal enhancers (Galli et al., 2015; Zanconato et al., 2015) and that the Sd-Yki (or TEAD-YAP) complex cooperates with other transcriptional factors at these enhancers, such as dE2F1 in Drosophila and AP-1 and Myc in mammals, to regulate gene expression. (Croci et al., 2017; Nicolay et al., 2011; Zanconato et al., 2015). Recent studies have also shed light on the mechanisms by which the coactivator Yki/YAP/TAZ activates target gene transcription. In Drosophila, Yki has been reported to interact with multiple general transcriptional regulators including the Trithorax-related (Trr) Histone H3 Lysine 4 (H3K4) methyltransferase complex (through direct binding between Yki and the Trr subunit NcoA6) (Oh et al., 2014; Qing et al., 2014), the SWI/SNF-related Brahma chromatin remodeling complex (Jin et al., 2013; Oh et al., 2013), GAGA factors (Oh et al., 2013), and the Mediator complex (Oh et al., 2013). In mammals, YAP has been reported to recruit the Mediator complex and the CDK9 kinase to regulate transcriptional elongation (Galli et al., 2015), and TAZ has been reported to interact with the SWI/SNF chromatin remodeling complex (Skibinski et al., 2014). The relative contribution of different general transcriptional regulators and whether each regulator is required for the expression of all or selected Hippo targets remain to be determined.

The Upstream Regulators of the Hippo Pathway

Unlike other developmental pathways such as the Notch, Wnt, Hedgehog, and TGFb pathways, which are activated by relatively specific ligands, studies in the past decade have linked Hippo signaling to a diverse array of upstream signals, including cell polarity, adherens junctions (AJs), cytoskeleton, mechanical forces, GPCR ligands, and some stress signals (Figure 3).

Figure 3. Upstream Regulators of the Hippo Pathway in Drosophila and Mammals.

The signaling diagram illustrates the core kinase cascade (shown within the dashed-line box) and the upstream regulators of the Hippo pathway in both Drosophila and mammals. The upstream regulators link diverse signals such as cell polarity (the Crb/CRB3 complex, the aPKC-PAR complex, and the SCRIB group proteins Scrib, Lgl, and Dlg), cell junctions (adherens junctions in Drosophila and mammals, tight junctions, and focal adhesions in mammals), cytoskeleton (F-actin and spectrin), mechanical forces, GPCR ligands, and some stress signals to the core kinase cascade. The core components of the Hippo kinase cascade are highlighted in red, while the upstream regulators that promote or antagonize Hippo signaling are indicated by purple or green, respectively. Arrowed or blunted ends indicate activation or inhibition, respectively.

Cell Polarities

The tumor suppressors Merlin (Mer, NF2 in mammals) and Ex were the first upstream regulators genetically linked to the Hippo kinase cascade in Drosophila (Hamaratoglu et al., 2006). Together with two additional tumor suppressors, Kibra and Pez, these tumor suppressors physically interact with each other and function cooperatively in the apical domain of epithelial cells to regulate Hippo signaling in diverse developmental contexts (Baumgartner et al., 2010; Genevet et al., 2010; Poernbacher et al., 2012; Yu et al., 2010). These findings implicate the apical domain as a “subcellular niche” for Hippo pathway activation.

The apical and basal domains of epithelial cells are defined by distinct polarity protein complexes and are separated from each other by cell-cell junctions. These complexes include two apical complexes, the Crb complex and the aPKC-PAR complex, as well as the basal-laterally localized SCRIB group proteins including Scribble (Scrib), Lethal (2) giant larvae (Lgl), and Discs large (Dlg). In Drosophila, the apical domain determinant Crb directly recruits Ex to the apical domain to promote Hippo signaling (Chen et al., 2010a; Grzeschik et al., 2010; Ling et al., 2010; Robinson et al., 2010). Although mammals lack a clear Ex ortholog, the involvement of Crb in Hippo signaling appears to be conserved as CRB3 acts with angiomotin (AMOT)-like proteins to promote Hippo signaling (Varelas et al., 2010). The basal-lateral SCRIB group proteins have also been linked to Hippo signaling, although the underlying mechanisms are less defined (Grzeschik et al., 2010).

Besides apical-basal polarity, Hippo signaling in Drosophila is further regulated by two planar cell polarity (PCP) regulators, the protocadherins Fat (Ft) and Dachsous (Ds). Much of the Ft signaling is mediated by the atypical myosin Dachs (D), whose asymmetrical localization at the apical membrane is regulated by Ft in a process that requires the Casein kinase 1ε discs overgrown (Dco), the DHHC palmitoyltransferase approximated (App), the F-box protein Fbxl7, and the SH3-domain-containing protein dLish/Vamana (reviewed in Blair and McNeill, 2018). D, in turn, influences the protein level of Wts through poorly understood mechanisms (Cho et al., 2006). Besides the Wts protein level, Ft signaling has also been shown to regulate Hippo signaling through additional mechanisms, such as the protein conformation of Wts (Vrabioiu and Struhl, 2015) and Dlish-mediated Ex degradation (Wang et al., 2019).

Structure-function analysis of the intracellular domain (ICD) of Ft revealed distinct motifs that mediate its function in Hippo versus PCP signaling (reviewed in Blair and McNeill, 2018). Of note, the Ft ICD motifs required for Hippo signaling are not conserved in FAT4, the mammalian ortholog of Ft. Indeed, most genetic analyses of Fat4 knockout mice revealed mutant phenotypes indicative of PCP but not Hippo signaling (Saburi et al., 2008), with the exception of the heart where Fat4 deletion was reported to cause increased cardiomyocyte size and proliferation in a YAP-dependent manner (Ragni et al., 2017). Besides Fat4, mammals have three more distantly related Ft family members. Among them, Fat1 has been implicated as a tumor suppressor regulating Hippo signaling in head and neck squamous cell carcinoma (Martin et al., 2018) and breast cancer (Li et al., 2018).

Adherens Junctions

Cell-cell junction proteins, especially components of AJs, have been implicated as regulators of Hippo signaling. E-cadherin (E-cad) was reported to mediate contact inhibition of proliferation in cultured mammalian cells through a pathway involving NF2, KIBRA and LATS1/2 (Kim et al., 2011). Another AJ component, α-catenin (α-cat), was also reported to restrict YAP activity, either through direct binding to YAP (Schlegelmilch et al., 2011) or by inhibiting integrin-mediated activation of SRC tyrosine kinase (Li et al., 2016). The requirement for α-cat is likely cell-type specific because YAP activity is still sensitive to contact inhibition in α-cat null 293A cells (Plouffe et al., 2016). In contrast to the reported inhibitory activity in mammalian cells, E-cad and α-cat appear to promote Yki activity in Drosophila (Rauskolb et al., 2014; Yang et al., 2015). Instead, another cell-cell adhesion protein in Drosophila AJs, Echinoid (Ed), was shown to restrict Yki activity by promoting Sav membrane association and stability (Yue et al., 2012).

Besides AJs, tight junctions (TJs) and focal adhesions (FAs) have also been implicated in Hippo signaling in mammals. TJ localization of the Hippo pathway components YAP (Oka et al., 2010), CRB3 (Fogg et al., 2005), and AMOT (Yi et al., 2011; Zhao et al.,2011) has been reported. FAs have been shown to regulate the Hippo pathway through several mechanisms, including ILK-mediated suppression of NF2 (Serrano et al., 2013), FAK-Src-PI3K-mediated inhibition of LATS1/2 (Kim and Gumbiner, 2015), Gα13-mediated suppression of RhoA (Wang et al., 2016b), and mechanical forces (Elosegui-Artola et al., 2017).

Mechanical Cues and Cytoskeleton

It has long been appreciated that cell proliferation is influenced by both chemical ligands such as growth factors as well as mechanical signals such as cell-cell contact and stiffness of the ECM (Discher et al., 2009). The mechanisms that convert mechanical cues into growth-regulatory signaling have attracted much attention by cell biologists. Several lines of evidence have implicated Yki/YAP/TAZ as key mediators of this growth-regulatory mechanotransduction pathway. The subcellular localization of YAP is sensitive to cell-cell contact and cell density, displaying nuclear localization in sparsely cultured cells and cytoplasmic localization in confluent cultures (Zhao et al., 2007). Further analyses revealed that cell-cell contact influences YAP activity by altering cell geometry; sparsely plated cells adopt a flat and spread-out morphology, while cells grown at a high density become round and compact (Aragona et al.,2013). Indeed, manipulating cell geometry by other means result in similar changes in YAP activity (Dupont et al., 2011; Wada et al., 2011). For example, cells grown on stiff ECM or spread across a large surface are also flat and show nuclear localization of YAP, while cells grown on soft ECM, on a small surface, or detached from a culture plate are round and show cytoplasmic YAP localization (Dupont et al., 2011; Wada et al., 2011; Zhao et al.,2012). Further supporting its role in mechanotransduction, YAP/TAZ activity is sensitive to diverse mechanical perturbation such as direct stretching, shear stress from fluid flow, or cytoskeleton tension (Aragona et al., 2013; Benham-Pyle et al., 2015; Codelia et al., 2014; Deng et al., 2015; Elosegui-Artola et al., 2017; Fletcher et al., 2015; Nakajima et al., 2017; Rauskolb et al., 2014; Wang et al., 2016b).

Regardless of the nature of the mechanical cues investigated, many appear to influence Yki/YAP/TAZ activity through changes in the integrity and dynamics of the contractile actomyosin cytoskeleton (Aragona et al., 2013; Dupont et al., 2011; Fernández et al., 2011; Rauskolb et al., 2014; Sansores-Garcia et al., 2011; Wada et al., 2011; Zhao et al., 2012). A key regulator of the actomyosin cytoskeleton is the Rho-ROCK (Rho-associated protein kinase)-MLC (non-muscle myosin II light chain) pathway, and indeed direct manipulation of Rho, ROCK or MLC leads to similar changes in Yki/YAP/TAZ activity in Drosophila and mammalian cells (Cai et al., 2018; Dupont et al., 2011; Rauskolb et al., 2014; Zhao et al., 2012). A recent study reported that matrix stiffness acts through the small GTPase RAP2 to activate LATS1/2 via the Rho inhibitor ARHGAP29 and the Hpo-like kinases MAP4K4/6/7, suggesting the involvement of both Rho-dependent and -independent mechanisms in mechano-regulated Hippo signaling (Meng et al., 2018). Another cytoskeleton system, the spectrin-based membrane skeleton (SBMS) also regulates Yki/YAP activity but in an opposite manner to the actomyosin cytoskeleton (Deng et al., 2015; Fletcher et al., 2015). Mechanistically, this function was attributed to spectrin’s role in antagonizing non-muscle myosin II activity (Deng et al., 2015) or binding to canonical upstream regulators such as Crb and Ex (Fletcher et al., 2015). Regardless of the mechanisms involved, it will be interesting to examine whether the SBMS plays a direct role in mechanotransduction, as well as the relationship between the SBMS and the actomyosin cytoskeleton in mechanotransduction.

Cytoskeleton-mediated regulation of Yki/YAP/TAZ activity has been reported to involve both Wts/LATS-dependent (Fletcher et al., 2018; Meng et al., 2018; Plouffe et al., 2016; Rauskolb et al., 2014; Zhao et al., 2012; Zheng et al., 2015) and -independent mechanisms (Aragona et al., 2013; Chang et al., 2018; Dupont et al., 2011; Elosegui-Artola et al., 2017). Interestingly, cytoskeletal tension appears to influence the activity of Wts/LATS, but less of Hpo/MST (Plouffe et al., 2016; Rauskolb et al., 2014; Zhao et al., 2012; Zheng et al., 2015). The FERM domain protein Merlin, whose physical interaction with Wts is sensitive to F-actin status in the cells (Yin et al., 2013), and the LIM family protein Ajuba, which sequesters Wts/LATS at AJs in a cytoskeletal tension-dependent manner (Rauskolb et al., 2014), have been implicated in coupling cytoskeletal tension to Wts/LATS. The LATS-independent regulation of YAP by cytoskeletal tension may involve force-dependent alteration of nuclear pore size that modulates the nuclear-cytoplasmic shuttling of YAP (Elosegui-Artola et al., 2017) or mechano-regulated interactions between YAP/TAZ and the SWI/SNF transcription complex (Chang et al., 2018).

Soluble Factors

Studies in cultured mammalian cells have identified G-protein-coupled receptors (GPCRs) and their agonists as regulators of Hippo signaling (Yu et al., 2012). Interestingly, different G proteins have opposing effects on Hippo signaling, with Gα₁₁, Gα₁₂, Gα₁₃, Gα_i, Gα_o, and Gα_q activating and Gα_s inhibiting YAP/TAZ (Yu et al., 2012). Gα_12/13 or Gα_q/11-coupled GPCRs activate the GTPase Rho, which triggers actin polymerization to downregulate LATS1/2 activity (Yu et al., 2012), while Gα_s-coupled GPCRs activate PKA, which subsequently inactivates Rho to upregulate LATS1/2 activity (Kim et al., 2013; Yu et al., 2013). Besides the GPCR agonists, other soluble factors have also been reported to modulate Hippo signaling, often through poorly characterized molecular mechanisms. These include factors that can activate Yki/YAP/TAZ, such as Wnt (Azzolin et al., 2014; Park et al.,2015), epidermal growth factor (EGF) (Fan et al., 2013), and insulin (Straßburger et al., 2012), as well as factors that can inactivate Yki/YAP/TAZ, such as hyaluronan (Ooki et al., 2019) and FGF15 (Ji et al., 2019).

Stress Signals

Given its pivotal role in normal development and physiology, it is not surprising that the Hippo pathway responds to internal and environmental stresses that perturb these normal processes, such as energy deficit (DeRan et al., 2014; Mo et al., 2015; Wang et al., 2015b), hypoxia (Ma et al., 2015), oxidative stress (Lehtinen et al., 2006), endoplasmic reticulum stress (Wu et al., 2015), cytokinesis failure (Ganem et al., 2014), osmotic stress (Hong et al., 2017; Lin et al., 2017b), and pathogens (Geng et al., 2015; Liu et al., 2016; Meng et al., 2016; Wang et al., 2017a). Cellular energy stress leads to an increased AMP:ATP ratio, which activates the energy sensor AMP-activated protein kinase (AMPK) to restore energy homeostasis. Activated AMPK phosphorylates AMOTL1 at Ser793 to activate LATS1/2 and also directly phosphorylates YAP at multiple sites to impair YAP-TEAD interaction (DeRan et al., 2014; Mo et al., 2015; Wang et al., 2015b). In Drosophila, the LKB1-AMPK pathway regulates Yki activity in the central nervous system, and the AMPK-related kinases salt-induced kinase 2 and 3 can phosphorylate Sav (Gailite et al., 2015; Wehr et al., 2013). In addition, the nutrient and energy sensor Target of Rapamycin (TOR) has been shown to promote Yki/YAP activity in Drosophila and mammalian cells (Liang et al., 2014; Parker and Struhl, 2015). Pathogens represent another class of environmental stress that activates the Hippo pathway. Bacterial infection activates Hippo signaling in both Drosophila and mammals (Geng et al., 2015; Liu et al., 2016), while MST1/2, LATS1/2, and YAP/TAZ are implicated in anti-viral immune response in mammalian cells (Meng et al., 2016; Wang et al., 2017a).

A Synthesis: The Hippo Pathway Couples Tissue Architecture to Growth Control

The complex regulation of Hippo signaling is perhaps not surprising given the paramount importance of precise growth control in development, homeostasis, and regeneration. A common theme emerging from the regulation of Hippo signaling by diverse upstream inputs suggests that the Hippo pathway may have evolved as a key mechanism that couples tissue architect to growth control. As a consequence, this pathway is exquisitely sensitive to perturbation of normal tissue and cellular integrity, such as changes in cell polarity, cell junctions, cell tension, cell density, as well as the ECM and growth factor milieu. This ensures that any deviation from the normal tissue and cellular architecture in development, homeostasis, or regeneration can be effectively restored by compensatory cell proliferation mediated by the Hippo pathway effector Yki/YAP/TAZ.

Hippo Signaling in Development, Tissue Homeostasis, and Regeneration

While the Hippo pathway was first appreciated for its critical role in organ size control in Drosophila, research in the past decade has extended its physiological function to additional processes such as lineage specification in early embryogenesis, cell differentiation in organ development, tissue-resident stem cell regulation, and tissue repair. The widespread functions of Hippo signaling in diverse physiological processes implicate it as a major regulator of tissue biology in development, homeostasis, and regeneration.

Hippo Signaling in Blastocyst Development and Embryonic Stem Cells

The first cell fate specification in mouse embryogenesis is the specification of trophectoderm (TE) versus inner cell mass (ICM). At the 32-cell stage, differential position and polarity cues lead to a Hippo-high state in the inner cells and a Hippo-low state in the outer cells (reviewed in Rossant, 2016). Active Hippo signaling in the inner cells leads to cytoplasmic YAP localization that promotes ICM lineage specification, while inactive Hippo signaling in outer cells results in nuclear YAP that drives TEAD4-dependent genes such as Cdx2 to specify the TE lineage (Nishioka et al., 2009). The Hippo pathway regulator AMOT appears to be a key player in conferring position-dependent activation of Hippo signaling in this process; AMOT is localized at AJs to activate Hippo signaling in the inner cells, but in the outer cells, AMOT is redistributed from the AJs to the apical domain, resulting in the suppression of Hippo signaling (Hirate et al., 2013; Leung and Zernicka-Goetz, 2013).

While embryonic stem cells (ESCs) are derived from the ICM, ESCs paradoxically show nuclear accumulation of YAP (Lian et al., 2010). Early studies suggested that YAP/TAZ are required for ESCs to maintain pluripotency and self-renewal, with YAP activation preventing ESC differentiation even in differentiation medium (Lian et al., 2010; Varelas et al., 2008). However, it was reported more recently that YAP/TAZ are dispensable for self-renewal but rather required for differentiation of ESCs (Azzolin et al., 2014). The reasons for these discrepancies are unknown but may be related to the different culture conditions, especially given that YAP/TAZ activity is sensitive to many physical and chemical cues.

Hippo Signaling in Organ Size Control in Development and Homeostasis

Echoing the activity of its Drosophila counterpart, liver-specific overexpression of YAP (Camargo et al., 2007; Dong et al., 2007) or inactivation of Sav1 (Lee et al., 2010; Lu et al., 2010), Nf2 (Zhang et al., 2010), Kibra (Hermann et al., 2018), Mst1/2 (Lu et al., 2010; Song et al., 2010; Zhou et al., 2009), Mob1a/b (Nishio et al.,2016), or Lats1/2 (Chen et al., 2015b; Yi et al., 2016) results in hepatomegaly and liver cancers. Despite these similarities, these models may differ in the specific liver cancer types. For example, while mice with liver-specific Mst1/2 knockout primarily develop hepatocellular carcinomas (Zhou et al., 2009), Nf2- or Lats1/2-deficient livers primarily develop cholangiocarcinoma (Chen et al., 2015b; Lee et al., 2016; Yi et al., 2016; Zhang et al., 2010). Whether these differences are due to different levels of YAP activation remains to be determined.

Similar to the liver, targeted inactivation of Sav1, Mst1/2 or Lats2, or overexpression of YAP in the developing heart, increased cardiomyocyte proliferation and heart size (Heallen et al., 2011; von Gise et al., 2012; Xin et al., 2011). Conversely, deletion of Yap in embryonic or fetal cardiac progenitors reduced cardiomyocyte proliferation (von Gise et al., 2012; Xin et al., 2011). These findings suggest that the Hippo pathway restricts embryonic cardiomyocyte proliferation and heart size through its effectors YAP and TAZ.

Besides the liver and the heart, a systematic survey examining the effects of postnatal Mst1/2 inactivation in mice revealed that the size of the stomach and the spleen was also increased (Song et al., 2010). Deletion of Lats1/2 or overexpression of YAP in gastrointestinal mesenchyme both led to enlargement of the stomach (Cotton et al., 2017). In contrast to the aforementioned tissues, loss of Hippo signaling in the lung, kidney, limb, intestine, mammary gland, and pancreas does not increase overall organ size (Cai et al., 2010; Chen et al., 2014a; Gao et al., 2013; George et al., 2012; Song et al., 2010). For example, pancreatic-specific Mst1/2 knockout mice have a smaller pancreas because of metaplasia (Gao et al., 2013; George et al., 2012), and renal-specific Lats1/2 knockout mice show reduced kidney size due to impaired cell differentiation (McNeill and Reginensi, 2017). Thus, the impact of loss of Hippo signaling on organ size is context dependent and likely dictated by other factors such as the cellular composition of a given tissue. For example, the number of embryonic progenitor cells imposes a limit on the size of the pancreas but not the liver, which may explain why loss of Hippo signaling can increase the size of the liver but not the pancreas in adult mice (Stanger et al., 2007).

Hippo Signaling in Cell Differentiation

Soon after uncovering the role of Hippo signaling in growth control, studies in Drosophila also linked this pathway to cellular differentiation, most notably in specifying rhodopsin expression in pupal photoreceptors (Mikeladze-Dvali et al., 2005) and maintaining dendritic morphology in larval sensory neurons (Emoto et al., 2006). The dual control of growth and differentiation by Hippo is recapitulated in mammals, where Hippo signaling controls the size of some organs but plays a more widespread role in cell differentiation. Here, we provide a synopsis of genetic analyses in mice that have implicated Hippo signaling in diverse contexts of cellular differentiation.

Genetic studies in the mouse liver, where YAP activation has a profound impact on organ size, revealed that YAP is preferentially required in biliary epithelial cells. Loss of Yap during mouse liver development resulted in hypoplastic biliary ducts that were gradually lost over time (Zhang et al., 2010). Conversely, mice with liver-specific deletion of Nf2 or Lats1/2 displayed a significant increase of biliary epithelial cells at the expense of hepatocytes (Chen et al., 2015b; Yi et al., 2016; Zhang et al., 2010). These phenotypes mirror the observation that YAP activation can cause trans-differentiation of hepatocytes into biliary epithelium (Yimlamai et al., 2014).

Pancreatic-specific Mst1/2 knockout mice were born with normal pancreata but developed ductal metaplasia postnatally due to the dedifferentiation of acinar cells to ductal cells (Gao et al., 2013; George et al., 2012). This phenotype is mediated by hyperactivation of YAP, as it was suppressed by halving the dosage of the Yap gene (Gao et al., 2013). Consistent with these findings, transient YAP overexpression in mouse pancreata from E13.5 to E17.5 resulted in expansion of ductal cells and impaired differentiation of acinar cells (George et al., 2012). In contrast to the exocrine lineage, loss of Mst1/2 had little effect on pancreatic endocrine cells (Gao et al., 2013; George et al., 2012). Indeed, Yap expression is confined to exocrine cells and is gradually lost in endocrine cells during pancreatic development, which explains the normal development of endocrine cells in the pancreatic-specific Mst1/2 knock-out mice (Gao et al., 2013; George et al., 2012).

During lung development, YAP functions at the transition zone between the airway and the distal lung compartment to control Sox2 expression and ultimately the generation of airway epithelia (Mahoney et al., 2014), and Crb3-mediated Hippo signaling restricts YAP activity to promote proximal airway differentiation (Szymaniak et al., 2015). YAP is also required for branching morphogenesis by controlling mechanical force production and epithelial cell proliferation during lung development (Lin et al., 2017a). In the adult lungs, YAP is required for the maintenance of airway basal stem cells, with the loss of Yap leading to unrestrained differentiation and loss of these stem cells (Zhao et al., 2014).

In the endodermal epithelia of the gastrointestinal tract, YAP activation promotes the expansion of progenitor cells accompanied by the loss of differentiated cell types such as the secretory cells (Cai et al., 2010; Camargo et al., 2007). In addition, YAP and TAZ are required to coordinate growth and patterning in the gut mesenchyme, as their activation leads to expansion of primitive mesenchymal progenitors and inhibition of the induction of smooth muscle lineage (Cotton et al., 2017).

The Hippo pathway has been implicated in multiple aspects of kidney development and homeostasis, including embryonic glomerular and lower urinary tract development, maintenance of podocyte homeostasis and glomerular filtration barrier, renal tubular cyst growth, and renal fibrogenesis (reviewed in Wong et al., 2016). Although YAP and TAZ often function redundantly in other contexts, they play distinct roles in cap mesenchymal (CM) cells in kidney development, with Yap deletion impairing nephron induction and morphogenesis (Reginensi et al., 2013) and Taz deletion causing renal cyst formation (Makita et al., 2008). In contrast, YAP/TAZ activation resulting from Lats1/2 deletion causes a dramatically reduced kidney size due to differentiation of nephron progenitor cells into myofibroblastic cells (McNeill and Reginensi, 2017).

In the skin, YAP expression is high in the basal epidermal progenitors and declines in the suprabasal differentiated cells (Schlegelmilch et al., 2011). Epidermis-specific Yap knockout led to thinner, fragile skin and a lack of epidermal tissues in the distal part of the limbs (Schlegelmilch et al., 2011), while its overexpression increases the number of basal epidermal progenitors and inhibits terminal differentiation (Schlegelmilch et al., 2011; Zhang et al., 2011). These results indicate that YAP plays a critical role in epidermal stem cell proliferation and tissue expansion. Indeed, keratinocyte-specific knock out of Mob1a/b by K14-Cre leads to hyperplastic keratinocyte progenitors and defective keratinocyte terminal differentiation (Nishio et al., 2012).

In the nervous system, YAP is required to maintain neural progenitor cell number in the embryonic neural tube (Cao et al., 2008). YAP further controls ependymal integrity by sustaining proper proliferation and apical attachment of ependymal precursor cells (Park et al., 2016). In the peripheral nervous system, YAP/TAZ regulate myelination and expression of laminin receptors in Schwann cells (Poitelon et al., 2016). Besides the aforementioned tissues, the involvement of Hippo signaling in cell differentiation has also been reported in salivary gland (Enger et al., 2013; Szymaniak et al., 2017), mammary gland (Chen et al., 2014a), blood vessel (Nakajima et al., 2017; Wang et al., 2017b), and heart (Lai et al., 2018; Xiao et al., 2018).

Hippo Signaling in Tissue Regeneration

The growth-regulatory function of Hippo signaling in normal development has prompted investigations into its potential roles in regenerative growth. Early studies in both Drosophila and mice revealed an essential role of Yki/YAP in intestinal regeneration (Cai et al., 2010; Karpowicz et al., 2010; Ren et al., 2010; Shaw et al., 2010). In mouse intestines, YAP/TAZ are highly expressed at the base of the crypts, and their inactivation has no discernable effects on normal development but severely compromises regeneration upon intestinal injury (Azzolin et al., 2014; Cai et al., 2010). Interestingly, upon injury YAP is acutely induced post-transcriptionally to elicit a regenerative program by promoting the expansion of transient amplifying cells (Cai et al., 2010). YAP is also acutely activated in regenerating hepatocytes after partial hepatectomy (Grijalva et al., 2014) and in the basal cell layer of the epidermis after wounding (Elbediwy et al., 2016). The transient activation of YAP/TAZ and its functional requirement in multiple models of tissue regeneration suggest that artificial activation of YAP/TAZ may be exploited in regenerative medicine. Indeed, a small molecule inhibitor of MST1/2 (XMU-MP-1) has been shown to facilitate intestinal and hepatic repair upon injury (Fan et al., 2016). Along the same line, expression of an active form of YAP in in vitro cultured human cadaver islets robustly increased proliferation of insulin-producing β-cells without a negative influence on differentiation or functional status (George et al., 2015), suggesting that YAP activation may be useful in β-cell expansion.

The mammalian heart represents a particularly promising context to harness the power of YAP/TAZ in tissue regeneration. Although the adult heart has largely lost regenerative capacity, the fetal and early neonatal heart in mammals can regenerate following cardiac injury (Porrello and Olson, 2014). Interestingly, deletion of Yap abolished the regenerative capacity of the early neonatal mouse heart, and conversely, cardiac-specific YAP overexpression prolonged the regenerative capacity beyond the neonatal period (von Gise et al., 2012; Xin et al., 2013). Besides small molecule activators of YAP, gene therapies delivering virus-encoded short hairpin RNA against Hippo pathway tumor suppressors may provide an alternative approach to enhancing heart regeneration and improving heart function after myocardial infarction (Leach et al., 2017).

Hippo Signaling in Disease

Given its widespread roles in normal development and homeostasis, it is not surprising that the Hippo pathway has now been linked to diverse human diseases, including cancer, autoimmunity, and developmental anomalies (Figure 4). Thus, a mechanistic understanding of Hippo signaling holds great promise for developing effective therapeutics for Hippo-related human diseases.

Figure 4. Hippo Pathway Mutations in Human Diseases.

Mutations of Hippo pathway components have been identified in many human diseases, including cancer (highlighted in blue dialogue balloons) and other non-cancer diseases (highlighted in yellow dialogue balloons). The dialogue balloons also point to the specific Hippo pathway components whose mutations are associated with the indicated diseases.

Hippo Signaling in Cancer

Echoing the growth-regulatory function of Hippo signaling in Drosophila, mouse models with targeted deletion of Mst1/2 (Lu et al., 2010; Song et al., 2010; Zhou et al., 2009), Sav1 (Cai et al., 2010; Lee et al., 2010), Nf2 (Giovannini et al., 2000), Kibra (Hermann et al., 2018), Lats1/2 (Chen et al., 2015b; Yi et al., 2016), or Mob1a/b (Nishio et al., 2016) or overexpression of YAP (Camargo et al., 2007; Dong et al., 2007) were reported to show hyperplasia and tumorigenesis in various tissues. Accordingly, the Cancer Genomic Atlas identified the Hippo pathway as one of the eight major signaling pathways commonly altered in human cancers (Sanchez-Vega et al., 2018).

Loss-of-function mutations of Hippo pathway tumor suppressors are well documented in several tumor types. Inactivation of NF2 causes neurofibromatosis type 2, which is characterized by nervous system tumors including schwannomas, meningiomas, and ependymomas (Rouleau et al., 1993). Loss-of-function NF2 mutations are also observed in 19%–50% of malignant mesothelioma (Murakami et al., 2011; Sekido, 2011), 18%–22% of renal cell carcinoma (RCC) (Mehra et al., 2016), as well as in cervical squamous cell carcinoma (CSCC) (Wang et al., 2018b). Genetic studies in both Drosophila and mouse models have established NF2 as an essential regulator of the Hippo pathway, and further implicated YAP as a dosage-sensitive target for potential therapeutic intervention (Hamaratoglu et al., 2006; Liu-Chittenden et al., 2012; Zhang et al., 2010). Besides NF2, LATS1 and LATS2 mutations are observed in 11% of malignant mesothelioma (Miyanaga et al., 2015; Murakami et al., 2011), LATS2 mutations have been identified in CSCC (Wang et al., 2018b), and SAV1 is frequently mutated in RCC and CSCC (Wang et al., 2018b). Mutations in PTPN14, the human homolog of Drosophila Pez, are reported in 31% of mucinous tubular and spindle cell carcinoma, a relatively rare subtype of RCC (Mehra et al., 2016) and 23% of basal cell carcinoma, a common malignant neoplasm of the skin (Bonilla et al., 2016).

In comparison to mutations of upstream tumor suppressors of the Hippo pathway, activation of YAP/TAZ, as indicated by elevated protein level and/or nuclear accumulation, is even more prevalent in human cancers. Several mechanisms may contribute to YAP/TAZ activation in human cancers. First, amplification of the YAP gene locus occurs in various cancers including CSCC, hepatocellular carcinoma, medulloblastoma, and oral squamous cell carcinoma (Fernandez-L et al., 2009; Overholtzer et al., 2006; Wang et al., 2018b; Zender et al., 2006). Second, chromosomal translocations leading to the production of gain-of-function fusion proteins involving YAP or TAZ have been found in epithelioid hemangioendothelioma (Antonescu et al., 2013; Tanas et al., 2011), luminal breast cancer (Li et al., 2013), and ependymoma (Pajtler et al., 2015; Parker et al., 2014). Third, a missense mutation in YAP (R331W) has been reported as an activation mutation that predisposes carriers to lung adenocarcinoma (Chen et al., 2015a). Lastly, YAP/TAZ may be activated in human cancers due to the extensive crosstalk between Hippo and other oncogenic pathways. For example, uveal melanomas show prevalent activation of YAP/TAZ due to gain-of-function mutations in GNAQ/GNA11 (Feng et al., 2014; Yu et al., 2014). YAP/TAZ activation is observed in most human colorectal cancers, which frequently carry mutations in the adenomatous polyposis coli (APC) gene (Azzolin et al., 2014; Cai et al., 2015). Oncogenic mutations of KRAS in pancreatic ductal adenocarcinoma (Kapoor et al., 2014) and lung cancer (Shao et al., 2014) also lead to YAP activation. Inactivation of liver kinase B1 (LKB1) leads to YAP/TAZ activation in lung adenocarcinoma (Mohseni et al., 2014). Additionally, Kaposi sarcoma-associated herpesvirus (Liu et al., 2015) and Merkel cell polyomavirus (Nguyen et al., 2014), two oncogenic viruses, have been shown to activate YAP/TAZ in sarcoma and Merkel cell carcinoma, respectively. However, it should be noted that despite the predominant oncogenic role for YAP/TAZ in human cancers, loss of YAP was reported in 10% of multiple myeloma patients and low YAP expression is associated with poor prognosis in acute myeloid leukemia (Cottini et al., 2014), suggesting that YAP may be a context-dependent tumor suppressor in certain cancer types.

Recent studies have also shed light on how YAP/TAZ activation confers cancer-associated features in human malignancies, some of which are unique to mammalian tumor biology. Besides an evolutionarily conserved role in promoting cell proliferation and survival, YAP/TAZ are required and sufficient to confer cancer stem cell (CSC) traits (Cordenonsi et al., 2011). YAP/TAZ have also been shown to promote metastasis by inducing epithelial-to-mesenchymal transition (EMT) (Lamar et al., 2012), as well as the expression of growth factors (Zhang et al., 2009), matricellular proteins (Calvo et al., 2013), and cytokines (Guo et al., 2017; Wang et al., 2016a) that modify the tumor microenvironment. In addition, YAP/TAZ have been shown to promote tumor immune evasion through regulatory T cells (Ni et al., 2018) or PD-L1 (Janse van Rensburg et al., 2018). Lastly, YAP/TAZ activation has been shown to make cancer cells resistant to various chemotherapies including taxol, doxorubicin, 5-fluorouracil, and cisplatin as well as targeted therapies against RAF and MEK (reviewed in Zhao and Yang, 2015). Indeed, YAP amplification was identified as a central driver of tumor relapse of KRAS-driven lung and pancreatic cancers upon inhibition of KRAS signaling in the established tumors (Kapoor et al., 2014; Shao et al., 2014). Thus, besides being a pathway-specific drug target, YAP/TAZ may provide a general target to sensitize tumor cells to diverse therapeutic reagents.

Hippo Signaling in Other Diseases

Besides cancer, mutations in Hippo pathway components have been implicated in other human diseases. YAP activation in NF2 patients frequently leads to cataract formation in the eyes (reviewed in Kresak and Walsh, 2016). Conversely, loss of YAP activity is also pathological in the eyes, as a point mutation in TEAD1 (Y421H) that affects a key residue mediating YAP-TEAD1 interaction underlies Sveinsson chorioretinal atrophy (SCRA), an autosomal dominant eye disease (Fossdal et al., 2004), and heterozygous mutations in YAP cause a type of congenital malformation of the eye called coloboma (Williamson et al., 2014). In addition, MST1 mutations are associated with autosomal recessive primary immunodeficiency (Nehme et al., 2012).

Besides the disease-associated gene mutations, altered expression, phosphorylation, or subcellular localization of YAP/TAZ protein has been reported in several non-cancerous diseases. Pathological activation of Hippo signaling is implicated in arrhythmogenic cardiomyopathy (AC), a myocardial disease characterized by the replacement of cardiac myocytes with fibro-adipocytes, cardiac dysfunction and arrhythmia (Chen et al., 2014b), and decreased YAP protein level is associated with Stanford type A aortic dissection (STAAD) (Jiang et al., 2016). Conversely, elevated YAP protein level or decreased YAP phosphorylation level was observed in pulmonary hypertension, atherosclerosis, and Alexander disease (Bertero et al., 2016; Wang et al., 2016b; Wang et al., 2018a). Nuclear accumulation of YAP was also observed in the cystic epithelia of polycystic kidney disease where it drives cyst growth (Cai et al., 2018; Happé et al., 2011). Elevated TAZ protein levels in nonalcoholic steatohepatitis (NASH) promote liver inflammation and fibrosis (Wang et al., 2016c), while mislocalization of TAZ from intercellular junctions to the nucleus has been reported in patient tissues with Sjogren syndrome (SS), a complex autoimmune disease that primarily affects salivary and lacrimal glands (Enger et al., 2013). Besides altered YAP/TAZ activity, increased MST1 activity was observed in the motor neurons of sporadic amyotrophic lateral sclerosis (sALS) patients (Lee et al., 2013), and decreased expression of LATS2, TEAD2, and TEAD4 were observed in keratoconus corneas (Kabza et al., 2017). It should be noted, however, in most of these contexts, the exact molecular mechanisms underlying the aberrant Hippo pathway activity and their contribution to disease etiology remain to be determined.

Drugging the Hippo Pathway

Given its critical roles in diverse physiological and pathological processes, therapeutic targeting of Hippo signaling may offer an important avenue of intervention in cancer and regenerative medicine. As shown for the MST1/2 inhibitor XMU-MP-1 (Fan et al., 2016), development of kinase inhibitors against MST1/2 or LATS1/2 may yield potential YAP/TAZ activators to facilitate tissue repair in regenerative medicine. Conversely, YAP/TAZ inhibitors may be exploited as potential cancer therapies. While much of the efforts have focused on disrupting the YAP-TEAD complex (Liu-Chittenden et al., 2012), recent mechanistic insights into the transcriptional machinery of the Hippo pathway have suggested new approaches for targeting the YAP oncoprotein. For example, a polypeptide that mimics the Sd/TEAD corepressor VGLL4 can inhibit the growth of YAP-driven gastric cancer xenografts (Jiao et al., 2014). Targeting the transcriptional machinery downstream of Yki/YAP, such as the Trr/MLL H3K4 methyltransferase complex (Oh et al., 2014; Qing et al., 2014) and the Mediator complex (Galli et al., 2015), may provide additional points of intervention. An alternative to these Yki/YAP-centric approaches is to activate Hippo signaling by targeting upstream regulators of the Hippo pathway. Promising examples include Tankyrase inhibitors, which suppress YAP through the AMOT family of proteins (Wang et al., 2015a); statins, which suppress YAP by inhibiting the mevalonate pathway required for membrane targeting of Rho GTPase, and other inhibitors targeting Rho GTPase or its effector ROCK (Sorrentino et al., 2014; Wang et al., 2014); phosphodiesterase inhibitors rolipram and ibudilast, which suppress YAP by activating cAMP-PKA signaling (Yu et al., 2013); and metformin, which suppresses YAP by activating AMPK (Mo et al., 2015; Wang et al., 2015b). Anecdotally, statin use was associated with a 47 percent reduction in the risk of colorectal cancer (Poynter et al., 2005), although it remains unclear whether any part of this effect is mediated by YAP inhibition. A caveat with this approach is that these chemicals also impact other cellular functions besides Hippo signaling and may therefore present unwanted side effects. Irrespective of the approaches being explored, any systemic application of YAP inhibitors or activators should take into consideration the normal physiological functions of Hippo signaling in diverse tissues to maximize the therapeutic window of a particular intervention and minimize unwanted side effects.

Concluding Remarks

Starting from the simple genetic model Drosophila, the Hippo pathway has emerged as an evolutionarily conserved signaling cascade in diverse metazoans and their closest unicellular relatives. A mechanistic understanding of its core kinase cascade and nuclear effectors provides a reliable foundation for investigating this pathway in many physiological, pathological, and evolutionary contexts. Less certain, however, is how the Hippo kinase cascade responds versatilely to signals as diverse as cell adhesion, polarity, mechanical forces, soluble factors, and cellular stresses, especially in intact tissues. Indeed, despite the growing list of proteins and signals that have been implicated as upstream regulators of Hippo signaling, how these entities are molecularly linked to the Hippo kinase cascade is not well understood. For example, the actin cytoskeleton is believed to play a critical role in relaying mechanical forces to Hippo signaling, but the exact biochemical mechanisms by which the actin cytoskeleton impacts the core kinase cascade remain poorly understood. In addition, as many of these upstream signals have been inferred from studies in cultured cells, it is equally important to define whether and how these upstream signals regulate Hippo signaling in vivo using animal models such as mice and Drosophila. For example, to what extent do mechanical forces control the final organ size through the Hippo pathway? How does the mechanosensitive Hippo pathway cooperate with other developmental pathways to stop organ growth at the appropriate time? Addressing these outstanding questions will likely require the development of novel tools to monitor and manipulate mechanical forces in vivo, as well as live reporters to monitor the spatiotemporal regulation of Hippo signaling in intact tissues. A deeper understanding of Hippo pathway regulation will not only shed light on the longstanding question of organ size control but also guide the ongoing efforts to harness this pathway for disease intervention and regenerative medicine.