Hippo Tips the TGF-β Scale in Favor of Pluripotency

Abstract

How TGF-β signaling switches from enforcing pluripotency to promoting mesendodermal differentiation remains an open question. Recently in Cell Reports, Beyer et al. demonstrated that Hippo signaling components recruit the NuRD complex to repress expression of key genes targeted by TGF-β and thus determine whether TGF-β signaling will favor pluripotency or differentiation.

TGF-β signaling influences numerous cells types, directing cell-type-specific responses throughout developmental and disease processes. In human embryonic stem cells (hESCs), TGF-β signaling is required to maintain pluripotency, and yet it is also the key pathway that induces differentiation of hESCs into mesendoderm (Oshimori and Fuchs, 2012). In a recent paper in Cell Reports, Beyer et al (2013) describe a new mechanism underlying these divergent responses to TGF-β signaling.

The canonical TGF-β signaling pathway is mediated through activation of the transcription factors SMAD2 and SMAD3 (SMAD2/3). When these proteins are activated by phosphorylation they form a complex with SMAD4 and are retained in the nucleus where they cooperate with other transcription factors to form a stable complex on DNA (Oshimori and Fuchs, 2012). Through context-dependent partnering with different transcription factors, SMAD2/3 tend to occupy unique enhancers in different cell types (Mullen et al., 2011).

In ESCs, SMAD2/3 occupy the genome with the ESC master transcription factors OCT4 and NANOG (Brown et al., 2011; Mullen et al., 2011). With removal of serum and fibroblast growth factor from hESC media, SMAD2/3 activation directs mesendodermal differentiation (D'Amour et al., 2005). It has remained unclear how activation of SMAD2/3 transitions from reinforcing to antagonizing hESC pluripotency. Beyer et al (2013) now show that components of the Hippo signaling pathway play a key role in determining how hESCs respond to activation of SMAD2/3.

The authors first took a proteomics approach to identify proteins from hESC nuclear extracts that bound to a 400 bp region of the NANOG promoter. The NANOG gene is bound by OCT4, NANOG and SMAD2/3 in hESCs (Brown et al., 2011; Mullen et al., 2011), and its expression is induced during early mesendodermal differentiation (Greber et al., 2008). Using mass spectrometry, the authors identified components of the Hippo signaling pathway that were associated with the NANOG promoter. These factors included TAZ, YAP and TEAD family members. The authors showed through biochemical approaches that OCT4, SMAD2/3 and TAZ/YAP/TEAD are part of a physical complex on DNA and found that depletion of TAZ and YAP together or depletion of the four TEAD family members was sufficient to induce activation of key mesendodermal genes. In addition, loss of these factors also resulted in further induction of OCT4 and NANOG expression, which is associated with early mesendodermal differentiation (Greber et al., 2008). These findings suggest that TAZ/YAP/TEAD act to repress gene expression in hESCs.

To understand how components of the Hippo signaling pathway repress mesendodermal genes, the authors returned to their proteomic analysis and identified several components of the nucleosome remodeling and deacetylase (NuRD) complex that were also associated with the NANOG promoter. The NuRD complex has been implicated in regulating ESC fate (Hu and Wade, 2012), and depletion of multiple components of the NuRD complex each produced a phenotype similar to loss of TAZ and YAP or TEAD family members. Taken together, these results suggest that the repressive effects of the TAZ/YAP/TEAD are mediated through recruitment of the NuRD complex.

SMAD2/3 and the mesendodermal transcription factor FOXH1 bind to many of the same genes in hESCs (Kim et al., 2011). When the authors compared the genome-wide binding sites for TEAD, SMAD2/3 and OCT4 (TSO) with the binding sites for FOXH1, they found that 20% of TSO sites were also occupied by FOXH1. They confirmed co-occupancy at key mesendodermal genes by chromatin immunoprecipitation (ChIP) and then tested the effect of loss of FOXH1 and disruption of the TSO complex on mesendodermal gene activation. As expected, hESCs deficient in FOXH1 retained normal expression of pluripotency genes suggesting that FOXH1 expression does not affect the pluripotent state of hESCs. In addition, hESCs deficient in FOXH1 were not able to activate mesendodermal markers under differentiation conditions, consistent with a block in differentiation. Significantly, when TAZ, YAP and FOXH1 were depleted, hESCs were no longer able to activate mesendodermal genes in response differentiation signals. Thus, TAZ and YAP appear to recruit the NuRD complex to genes co-occupied by SMAD2/3, OCT4 and FOXH1. Loss of TAZ and YAP, loss of TEAD or loss of the NURD complex all result in the induction of these co-occupied genes in a process that requires FOXH1. Therefore, in hESCs, components of the Hippo pathway appear to mediate repression of TGF-β signaling by binding a set of key mesendodermal genes. The loss of binding of these Hippo factors that occurs with differentiation acts as a switch to allow rapid induction of mesendodermal genes bound by OCT4, SMAD2/3 and FOXH1 (Figure 1).

Figure 1. Loss of TAZ/YAP and TEAD results in activation of mesendodermal genes bound by SMAD2/3.

A set of key mesendodermal genes are bound by OCT4, SMAD2/3, SMAD4, FOXH1, TAZ/YAP and TEAD family members in hESCs. These genes are repressed in hESCs where TAZ/YAP and TEAD act to recruit the NuRD complex (left). With mesendodermal differentiation, TAZ/YAP and TEAD no longer bind key mesendodermal genes. In their absence the NuRD complex is released and mesendodermal genes are activated in a process that is dependent on SMAD2/3 and FOXH1 (right).

The Hippo signaling pathway is conserved from yeast to mammals and has been shown to affect ESC pluripotency and to interact with TGF-β signaling (Barry and Camargo, 2013). However, this study provides the first evidence that components of the Hippo pathway are directly involved in repressing gene expression in ESCs and that the presence or absence of these Hippo factors controls how developmental genes respond to SMAD2/3 binding. These findings raise the question: What controls TAZ/YAP and TEAD binding during mesendodermal differentiation? The authors quantified expression of CTGF, a gene that is occupied by TAZ/YAP/TEAD but not OCT4 or SMAD2/3. CTGF expression was unchanged upon mesendodermal differentiation suggesting that Hippo signaling may not have been altered. However, TAZ/YAP have been shown to be involved in regulating hESC pluripotency, and YAP expression decreases with differentiation of mESCs (Barry and Camargo, 2013). Further understanding of the expression, subcellular localization and genomic occupancy of TAZ/YAP and TEAD family members during hESC differentiation should shed further light on how this pathway regulates transcriptional responses to SMAD2/3 activation.

Integrating the concept of TAZ/YAP/TEAD as transcriptional repressors into recent discoveries should continue to advance our understanding of how cells respond to TGF-β signaling. For example, cell cycle has recently been shown to affect how hESCs respond to SMAD2/3 activation with hESCs in early G1 being the most responsive to mesendodermal differentiation signals (Pauklin and Vallier, 2013). Hippo signaling regulates proliferation in many different cell types (Barry and Camargo, 2013) where nuclear localization of YAP or its homologs are associated with increased rates of cell division and presumably a shorter duration of G1. In hESCs loss of TAZ/YAP binding is associated with induction of mesendodermal genes by SMAD2/3 raising the question of whether loss of TAZ/YAP binding could be associated with longer duration of G1 and contribute to activation of mesendodermal genes. Further investigation of the interaction between Hippo and TGF-β signaling will increase our understanding of how they modulate hESC state and provide insight into the coordination of these two pathways in development and disease.