ReviewNeuroscience

Emerging roles for angiomotin in the nervous system

See allHide authors and affiliations

Science Signaling  27 Oct 2020:
Vol. 13, Issue 655, eabc0635
DOI: 10.1126/scisignal.abc0635

Abstract

Angiomotins are a family of molecular scaffolding proteins that function to organize contact points (called tight junctions in vertebrates) between adjacent cells. Some angiomotin isoforms bind to the actin cytoskeleton and are part of signaling pathways that influence cell morphology and migration. Others cooperate with components of the Hippo signaling pathway and the associated networks to control organ growth. The 130-kDa isoform, AMOT-p130, has critical roles in neural stem cell differentiation, dendritic patterning, and synaptic maturation—attributes that are essential for normal brain development and are consistent with its association with autism. Here, we review and discuss the evidence that supports a role for AMOT-p130 in neuronal development in the central nervous system.

INTRODUCTION

Cells are held together by cell-cell contacts that play unique roles by preventing the random distribution of proteins between intracellular compartments and the movement of solutes across cell surfaces. The formation of cell contacts can be transient, such as the interactions between immune system cells, or more stable, as between mature epithelial tissues composed of tightly connected cell monolayers. Similar to the adhesive contacts that bind cells together in monolayers, chemical synapses function as adhesive sites (1). Through this adhesive function, synapses are important for stabilizing dendritic branches during development, which promotes the patterning of neural circuits (2, 3). As neuronal circuits develop, neurons acquire a polarized morphology in that synaptic inputs along dendrites are separated from the output in axons. Epithelial cell polarity, likewise, separates membrane sheets specialized for interactions with the outside environment from the inside lumen, which is essential to their biological functions. In both neurons and in nonneural systems, the connections between cells require a large number of adhesion proteins that form homophilic or heterophilic interactions across cell membranes (4). Adhesion proteins, in turn, bind to networks of cytoplasmic scaffolds that are linked to the cytoskeleton, which establishes and maintains cell organization. Failure of the regulatory factors that control these steps leads to a disruption of the developmental programs that operate to organize cells into complex tissues.

The Angiomotin (herein referred to as AMOT) family is a group of scaffolding proteins that are associated with adhesive contacts between vertebrate epithelial cells called tight junctions. The AMOT family consists of the AMOT, AMOTL1, and AMOTL2, genes that encode different splice variants (described below), which in mammals have largely redundant roles. Much of our understanding of AMOT function comes from studies in epithelial cells and more recent work in model organisms including mouse knockout studies. In epithelial cells, AMOTs promote tight junction integrity by recruiting and binding a variety of different proteins including enzymes, filamentous (F)–actin, signaling effectors, and Rho guanosine triphosphatases (GTPases) and their regulators. Depletion of AMOTs in zebrafish leads to defects in cell migration and morphology, suppresses angiogenesis, and impairs the self-organizing ability of the mouse embryo (57). In addition, cells that lack some AMOT members lose their ability to polarize and show a down-regulation of the Hippo signal transduction pathway, a conserved cell signaling cascade that, under normal conditions, controls tissue growth by maintaining a balance between proliferation and apoptosis (7, 8). The Hippo pathway was initially discovered in Drosophila but has since been shown to operate in a similar way in vertebrates. It is activated by changes in cell density, cell morphology, and mechanical forces. These signals trigger a chain of biochemical events that involve AMOT members and leads to the inactivation of the transcription coactivators yes-associated protein (YAP) and its paralog transcriptional coactivator with PDZ (PSD-95/Dlg-1/ZO-1)–binding motif (TAZ). Until recently, the AMOTs have been examined almost exclusively in immortalized cell lines with most information coming from cancer studies. However, work from a number of different laboratories has now provided new clues as to how one specific isoform of AMOT, AMOT-p130, operates in neuronal progenitors and primary neurons (911). This has not only revealed some common features that are related to its functions in various cell lines but has also uncovered unique roles in the nervous system that will open new avenues for investigation. For conciseness, we will only cover the signaling pathways of the AMOT members that we believe are relevant to the discoveries recently made in the nervous system.

Common functions of the AMOT proteins

The founding member of the AMOT family was discovered in a placenta yeast two-hybrid screen and encoded a protein with a molecular weight of 80 kDa that was named AMOT (AMOT-p80) (12, 13). In many contexts, AMOT-p80 has been associated with oncogenic transformation due to its involvement in cell migration (14). Soon after, a second longer 130-kDa isoform (AMOT-p130), produced by alternative splicing of the AMOT-p80–encoding gene, was uncovered (15). This was shortly followed by the identification of two additional AMOT members, AMOT-like 1 (AMOTL1/JEAP) and AMOT-like 2 (AMOTL2/MASCOT), which operate as adaptors or scaffolding components (13, 16). While several annotated splice isoforms have been identified for the AMOTL1 and AMOTL2 proteins, only a shorter 60-kDa splice-site variant of AMOTL2 (p60 AMOTL2) has been functionally characterized (see below) (17). All members share a C-terminal PDZ-binding motif, a proximal coiled-coil domain and, with the exception for AMOT-p80 and p60 AMOTL2, an N-terminal extension containing an F-actin–binding region and multiple leucine/proline-proline-x-tyrosine (L/PPxY) motifs important for interactions with various factors including Hippo pathway proteins (Fig. 1) (13, 15, 18). The AMOTs are conserved in vertebrates but seem to be missing in Drosophila (19). The protein 4.1, ezrin, radixin, moesin (FERM) domain protein Expanded (Ex) has identical roles in fly to those described for mammalian AMOTs, which has raised the idea that Ex represents an analog in the arthropod lineage, although this has yet to be tested. Moreover, all AMOT members have been reported to localize to tight junctions (20), specialized cell-cell contacts in vertebrates that are absent in flies (21), indicating that their distribution is functionally preserved.

Fig. 1 Domain structure of the AMOTs and their interactions.

Schematic of the structural domains and protein interaction motifs within the splice variants encoded by the AMOT, AMOTL1, and AMOTL2 genes. Noted above are various proteins known to interact with Amot-p130, Amot-L1, and Amot-L2, the interaction regions indicated by the color-matched horizontal lines. Also noted in the F-actin–binding domain of these family members is the conserved serine residue and LATS phosphorylation site.

CREDIT: A. KITTERMAN/SCIENCE SIGNALING

It is unclear exactly how the AMOTs are recruited to tight junctions, but associations via PDZ domain-motif linkages to other proteins situated at the apical/lateral boundary, including PATJ/PALS1, PAR3, and the Rho GTPase-activating protein (GAP), ARHGAP17 (RICH1/NADRIN), contribute to this localization (2224). However, AMOT-p130 does not seem to be constitutively coupled to tight junctions, in certain cell types. For instance, in MCF-7 human breast cancer cells, AMOT-p130 associates with a complex at the apical surface consisting of bone morphogenetic protein (BMP) receptor 2 (BMPR2) and its effector SMAD1 (25). Stimulation with BMP6, a ligand for BMPR2, induces SMAD1 phosphorylation, which also results in dephosphorylation of AMOT-p130. This somehow triggers internalization of AMOT-p130 in endosomes and defines apical identity. How AMOT-p130 is dephosphorylated in MCF-7 cells is not clear, but consequent trafficking of AMOT-p130 at the apical surface is driving BMP-SMAD signaling. Although vesicle trafficking has been shown to control apical/basal polarity in a number of systems (26), whether the same mechanism pertains to AMOTs in other cell types is not known.

Loss of apical/basal polarity as a result of hypoxia is also a critical step in the progression of many cancers (27). In epithelial tumors, expression of p60 AMOTL2 closely correlates with disruption of normal protein distribution and increased invasiveness (17). Low oxygen levels activate the stress-induced transcription factor c-Fos, which regulate p60 AMOTL2 that accumulates in vesicles. The vesicle linkage permits p60 AMOTL2 to associate with members of transport protein particle II complex (TRAPPII), a multi-subunit guanine-exchange factor (GEF) complex that is known to mediate endosomal membrane trafficking (28). Through a mechanism that is not completely clear, p60 AMOTL2 then disables TRAPPII and restricts delivery of vesicles with cargo containing the polarity proteins CRB3 and PAR3 that are normally destined to the apical/basolateral surfaces. A critical consequence of the spatial antagonism by p60 AMOTL2 is that it causes defects in cell polarity, which can translate into enhanced invasiveness. Gaining insight into how p60 AMOTL2 interferes with the TRAPPII complex function will therefore be key to understanding how transformation of some epithelial cancers can be countered.

An important, but less investigated, feature of all the members is their ability to self-interact, through their coiled-coil domain (20). In cells expressing mutants that lack the coiled-coil fragment, AMOTs are unable to localize to destination sites at cell-cell contacts and fail to establish functional interactions (22, 2931). Thus, apart from providing critical spatial information, the coiled-coil domain may offer diversity in targeting different signaling components. It is also possible that transitions between discrete oligomer conformations account for differences in isoform performances in the same cell or tissue.

Posttranslational modifications of the AMOTs

The stability of the AMOTs is regulated by various posttranslational modifications (PTMs), which influences cell fate in different contexts. Examples of PTMs include phosphorylation, ubiquitination, and ribosylation (20), which, in some cases, cooperatively regulate their functions.

AMOT phosphorylation

Early proteomic screens first linked the AMOT family to kinase networks involved in cell division, but the functional relevance of these associations is unknown (32, 33). More recently, kinases with contextual roles in cellular energy levels [AMPK (adenosine 5′-monophosphate–activated protein kinase)], protein synthesis [mTOR (mammalian target of rapamycin)], and cell proliferation [protein kinase C (PKC), AKT/PKB, and SRC] have been found (5, 3437). Perhaps the most extensively studied examples of kinases with links to the protein family come from the mammalian Hippo pathway.

Hippo signal transduction pathway

Central to the signaling cascade are the serine/threonine kinases LATS1/2, which are activated in response to various stimuli, including changes in cell polarity and mechanical tension (8). The key effectors of LATS1/2 are the transcription coactivators YAP and TAZ (also known as WWTR1) (38). Both proteins are structurally similar and drive proliferation when present in the nucleus by regulating a number of transcription factors including the transcriptional enhanced associate domain (TEAD) 1-4 family to control organ growth. When inactivated by cytoplasmic retention, YAP and TAZ are targeted for destruction through a mechanism involving phosphorylation, followed degradation (Fig. 2). The association between AMOTs and LATS1/2 depends on both the central coiled-coil domain and the L/PPxY motifs within the N-terminal region (31, 3942). The N-terminal region contains, in addition, a short motif that binds directly to F-actin (39), which can alter the affinity of the AMOTs for other interaction partners. Upon binding, LATS1/2 phosphorylates Ser175 (AMOT-p130), Ser262 (AMOTL1), and Ser159 (AMOTL2), which triggers the release of the AMOTs from cortical F-actin into the cytoplasm (18, 31, 41, 42). Moreover, binding appears to also directly stimulate LATS1/2 activity and, in case of AMOT-p130, is mediated by recruiting activators that stimulate LATS1/2 autophosphorylation (43). What happens then is not entirely clear, but several lines of evidence indicate that AMOT phosphorylation is a cue for the inactivation of YAP/TAZ at cell-cell contacts (4446). Phosphorylation might provide additional targeting functions for AMOT-p130 that promote the scaffolding of YAP/TAZ, although the details that define these steps in different systems are probably distinct. For example, expression of a phosphomimetic mutant of AMOT-p130 also binds to F-actin (47). Thus, in this case, phosphorylation might increase binding avidity for the actin cytoskeleton, which, in turn, may influence function. Either way, these results clearly indicate that additional, as yet unidentified inputs control complex formation.

Fig. 2 Involvement of AMOT-p130 in the Hippo pathway.

The Hippo pathway is activated by various stimuli that include extracellular cues, cell shape, and mechanical forces mediated by the actin cytoskeleton. These stimuli initiate a core kinase cascade (in brackets), promoted by the adaptor proteins SAV1 and MOB1, that leads to phosphorylation of the cotranscription activators YAP and TAZ, thereby sequestering them to the cytoplasm. Amot-p130 interaction with YAP—possibly phosphorylated—can sequester YAP/TAZ to tight junctions. The phosphorylated status of YAP/TAZ influences the balance of transcriptional activation of proliferation versus apoptosis within the cells of a tissue.

CREDIT: A. KITTERMAN/SCIENCE SIGNALING

AMOT ubiquitination

AMOT-p130 expression levels are low across different cell types (44, 48), whereas serum withdrawal can enhance AMOT-p130 expression (40). This suggests that biological control mechanisms exist, which restrains AMOT levels under normal conditions, which otherwise could consequent tissue growth decisions. The ubiquitin-proteasome system (UPS) provides an additional regulatory framework in this context in mature cells and tissues by controlling the stability of the different AMOT members. A number of ligases from the homologous to the E6-AP carboxyl terminus (HECT) family including NEDD4-1, NEDD4-2, Itch/AIP4, and HECW2, as well as the RING domain ligase, RNF146, have all been implicated in AMOT ubiquitination (40, 4952). The consensus sites for these ubiquitin ligases reside in the coiled-coil domain (53), and transfer of ubiquitin linkages on a lysine residue within this domain causes degradation for the most part, but important exceptions exist. For instance, Itch/AIP4 ubiquitination of some AMOT members can either trigger degradation or enhance their stability (40, 49), an outcome that is likely context and cell type dependent. Moreover, mono- and K63-linked poly-ubiquitin transfers are known to induce alterations in protein trafficking and have been reported accordingly to modulate the stability and signaling pattern of AMOTL1 and AMOTL2 rather than triggering their degradation (52, 53). Likewise, in human cells, grown in vitro, AMOTs are targeted but not degraded by NEDD4L/NEDD4-2 to facilitate assembly and release of HIV-1 (54). Thus, under some conditions, AMOTs appear to be protected from degradation, but in other contexts, they are degraded through the same ligases, suggesting that different cues control their stability (49). In some instances, the target protein of a ubiquitin ligase is distinguished from numerous other proteins by the addition of PTMs that are centered in a short consensus motif (55). This event maximizes the interaction of the substrate and the ligase by priming either the target or the enzyme for the subsequent ubiquitin transfer step. As an example, the addition of adenosine 5′-diphosphate–ribose moieties on AMOT-p130, AMOTL1, and AMOTL2 is catalyzed by the polymerases tankyrase 1 and 2 (TNKS1/2); this then creates binding sites for the E3 ligase RNF146 to mediate ubiquitin conjugation, which is followed by proteasomal degradation (51, 56, 57). PTMs can, in this way, provide interaction platforms that enable additional associations to occur between ubiquitin ligases and their binding partners. Like other PTMs, ubiquitylation is also reversibly modified by deubiquitylating enzymes that can cleave ubiquitin from substrate proteins (58). USP9X is a ubiquitin-specific protease that is a pivotal component of the machineries that control intracellular endocytic trafficking, replication stress, and spine development (5961), and at least two AMOT members are targets of USP9x (53, 61, 62). USP9x-mediated removal of ubiquitin linkages on AMOTL2 and AMOT-p130 inhibits YAP/TAZ transcriptional activity and limits cell proliferation. Thus, failure of the controls on the AMOTs and their corresponding pathways can prevent cell growth but, in other instances, may facilitate cell invasiveness (63). An important area of future research, therefore, will be to understand how different PTMs influence AMOT functions in normal development and in disease.

AMOT interactions converge on actin filaments

The actin cytoskeletons, together with microtubules, are routes for protein trafficking and determines the shape and motility of cells (64). All members of the AMOT family accumulate at tight junctions, whereas their cellular distribution appears to be contextual and involve associations with both the cytoskeleton and Rho GTPases. The small Rho GTPases Rac1, Cdc42, and RhoA act as binary switches that cycle between on (GTP-bound) and off [guanosine diphosphate (GDP)–bound] states. Their functions are controlled by GEFs and GAPs, which catalyze the transfer of GDP for GTP and the hydrolyzation of GTP to GDP, respectively. The Rho family plays fundamental roles across diverse tissues by stimulating assembly and disassembly of F-actin either directly or through their effectors. Their activity can spark widespread changes in the corresponding signaling networks with profound effects on cellular behavior.

During development, the ezrin-radixin-moesin–like protein NF2/Merlin binds AMOTs and plays a role in linking membrane components to the underlying actin cytoskeleton (31, 65, 66). AMOT binding causes a release of NF2/Merlin autoinhibitory conformation, which stimulates LATS1/2 and Hippo pathway activity (67). In other instances, their binding triggers membrane release of a RhoGAP, ARHGAP17, which acts to reduce Rac1-GTP levels and affect Ras–mitogen-activated protein kinase (MAPK) signaling (66). Whether cytoskeletal elements are involved in these events are not known, but it is likely that formation of the NF2/Merlin-AMOT complex triggers changes in actin dynamics or effectors bound to F-actin that alter signaling.

Through a different mechanism involving mechanotransduction, NF2/Merlin can coordinate the migration of epithelial monolayers via a junctional complex that includes AMOT (68). Cell movement is known to involve reciprocal activation of the Rho GTPases, which induces contractile tension in the actin cytoskeleton that is propagated through cell contacts (69). During this process, NF2/Merlin relocalizes to the cytoplasm, enabling AMOT to bind and negatively regulate ARHGAP17, which relieves the inhibition of Rac1 at the membrane. This triggers the polarized distribution of Rac1-GTP to the leading edge in migrating cells that give rise to the coordinated movement of cell sheets.

Another binding partner of the AMOTs with links to both NF2/Merlin and the actin cytoskeleton is the WW domain–containing protein kidney and brain expressed protein (KIBRA) (70). In a mouse model of mammary tumor growth, overexpression of KIBRA causes defects in the formation of RhoA-mediated stress fibers and nuclear YAP/TAZ assembly (71). Consistent with earlier findings, this study also supports the idea that KIBRA is a sensor for tension that can modulate Hippo activity through mechanical force (7173). The AMOTs have been described to have actin-bundling activity (9, 15, 74, 75), and an interesting speculation is that they cooperate together with KIBRA to facilitate mechanotransduction. Additional studies will likely resolve this question in the near future. A second mechanism by which KIBRA can affect the actin cytoskeleton is by association with synaptopodin, a regulator of actin-dependent cell morphogenesis and dendritic spine and podocyte motility (76, 77). This type of indirect actin modulation probably also operates in Drosophila where the fly orthologs KIBRA and Merlin cooperate to control Hippo signaling and tissue growth (78).

Last, the AMOT-NF2/Merlin-KIBRA signaling axis has been shown to coordinate cell differentiation in mouse preimplantation embryos by separating cell polarity from Hippo pathway activity (31). During the initial fate decision, cells are segregated into inner and outer cell populations that form two distinct cell lineages through a process that involves RhoA (79). In the outer cell mass, RhoA-GTP competes with Merlin for binding, causing AMOT-p130 to remain bound to the actin cytoskeleton. In the inner cell mass, low RhoA activity somehow induces phosphorylation of the serine at residue 176 (same as Ser175 in humans), which decreases the affinity of AMOT-p130 for F-actin and displaces it from the plasma membrane. Release into the cytoplasm allows AMOT-p130 to bind and inactivate YAP/TAZ, which enable proper inner cell mass differentiation. Thus, through a complicated set of interactions, AMOT functions converge on actin dynamics to provide cues that break symmetry and establish cell polarity or trigger movement.

Although the AMOT family has been studied for more than a decade, our current knowledge of their functions, with the exception for a few early reports, has mainly been derived from immortalized cell systems (15, 22). The systematic sampling of their roles in the nervous system is still largely missing, but three recent reports begin to unravel a role for AMOT-p130 in neuronal development and maturation (10, 11, 80). Further evidence has implicated the AMOT-p130 isoform in autism (81), raising the importance of AMOT as an integral signaling molecule in nervous system function. Examining how the AMOTs operate during neuronal development will provide us with a better understanding of how they respond to changes in the brain and also identify the molecular networks in which they interact. From here, we summarize the functions of AMOT-p130 that impinge on neural stem cell (NSC) development, dendritic morphology formation, and synaptic plasticity.

Emerging roles of AMOT-p130 in cells of the nervous system

Neurons in the brain are derived from stem and progenitor cells that undergo asymmetric cell division, giving rise to progeny in each division with distinct cell fates. Stem cells self-renew and proliferate, whereas daughter cells destined to become neurons migrate to their final destination in the nervous system where they rapidly start to polarize and extend neurites. In the mammalian nervous system, NSCs and lineage-specific progenitors generate new neurons and glia throughout life (8284). Early reports identified the subventricular zone and the dentate gyrus of the hippocampus in the adult brain as regions with increased neurogenesis (85, 86). Later work linked changes in these areas to early neurodegenerative symptoms, raising the possibility of using stem cells as a potential therapy for brain disorders (87).

An important fate decision occurs when NSCs switch from a proliferative to a differentiated state in the developing brain. The inhibition of cell cycle progression is the main determinant for this decision in the embryo, whereas at least in adults, it is a reversible process (88). Symmetric cell division by contrast gives rise to a pool of NSCs with the capacity to self-renew. The decision therefore whether or not to remain in a proliferative state is a major step that affects the type and number of the progeny generated by NSCs. Because of their pervading roles as transcriptional regulators, YAP/TAZ has emerged as major determinants for sustained NSC proliferation (8992). For example, in the vertebrate, neural tube cell proliferation is determined by the activity of nuclear YAP/TAZ, which ensures the expansion of the neural progenitor/NSC pool. Nuclear YAP/TAZ stimulates transcription of genes that promote cell cycle development, which suppresses neuronal differentiation (9398). By contrast, pharmacological inhibition of actin dynamics retains YAP/TAZ in the cytoplasm and activates the developmental programs that drive neuronal differentiation (Fig. 3).

Fig. 3 Functions for AMOT-p130 in stem cell specification and dendritic arborization.

Left: The transcription coactivator YAP has a critical role in the expansion of undifferentiated stem cells and progenitor cells. During self-renewal of human pluripotent stem cells (hPSCs) and neural stem cells (NSCs), YAP is retained in the cytoplasm by AMOT-p130 through a mechanism involving actin cytoskeleton rearrangement. The diminished proliferative capacity of hPSCs increases the pool of NSCs with limited ability to proliferate, which subsequently differentiate to generate neurons. Right: Intermediate NSCs polarize into neurons by growing extensions that become an axon and multiple dendrites. During dendritic development, AMOT-p130 forms a complex with YAP in arbors and recruits the p70S6K1 kinase, which phosphorylates its effector rpS6. The resulting scaffold stimulates the overall growth and patterning of dendritic arbors, which promote the formation of circuitry in the CNS.

CREDIT: A. KITTERMAN/SCIENCE SIGNALING

A similar switch in embryonic stem cell identity can be induced when cells are grown on different substrates. Stiff substrates maintain YAP/TAZ in the nucleus and promote continued cell proliferation, whereas flexible ones trigger their cytoplasmic retention and induce differentiation (99). How changes in substrate alter the localization of YAP/TAZ is not entirely clear, but the mechanism seems to involve a number of actin-binding proteins that modulate cytoskeletal organization (100). Moreover, YAP/TAZ cellular distribution is independent of LATS1/2 activity because kinase depletion does not reverse YAP/TAZ inactivation when cells are grown on more flexible substrates (101). Together, these studies reveal the importance of YAP/TAZ signaling in stem cell proliferation and point to the role of mechanical force in this process.

A key question has concerned the fate of YAP/TAZ when stem cell differentiation is initiated, and new data have implicated AMOT-p130 in this process. Work from Zaltsman et al. (10) recently addressed this question. The authors showed that when human pluripotent stem cells (hPSCs) undergo neural specification, YAP becomes enriched in the cytoplasm. This enrichment enhances the association between YAP and AMOT-p130, which triggers the degradation of YAP through a proteasome-mediated pathway. As cultured hPSCs develop into neurons, AMOT-p130 levels that are low in undifferentiated cells gradually increase (10). The subsequent exclusion of nuclear YAP and its enhanced association with AMOT-p130 seems to be independent of Hippo kinase activity but is supported by changes in the actin cytoskeleton (Fig. 3, left). A similar mechanism appears to operate also in NSCs to prevent YAP from entering the nucleus during fate determination (102). In this case, however, both the phosphorylation status of AMOT-p130 and cell contractility play key roles for the outcome. During growth of NSCs on soft substrates, the phosphorylated AMOT-p130 is detached from the actin cytoskeleton, which potentiates its binding to YAP. The resulting protein complex support NSC differentiation by stimulating the activity of β-catenin, a known driver of both adult neurogenesis and mechanosensing pathways (103, 104). Cell tension, in contrast, induced by increased substrate stiffness, triggers a switch in the subcellular distribution of YAP, which, as discussed previously, favors its nuclear accumulation. This causes AMOT-p130 as well to localize in the nucleus, although exactly how this is mediated remains uncertain—neither a nuclear localization import nor an export sequence motif has been described for AMOT-p130 to date. The subsequent response in NSC contractility to substrate rigidity triggers cytoskeletal changes that, similar to other systems, seems to be driven by the RhoA-ROCK pathway. Together, this mechanism defines the conversion of NSCs into neural cells and points to a role for AMOT-p130 in this program. This model might also explain how actin remodeling and mechanosensitive responses control the AMOT-p130/AMOTL2/YAP axis during cell fate specification in the mouse blastocyst (105, 106).

As discussed above, both YAP and AMOT-p130 levels are regulated by the proteasome in hPSCs, which indicate that they both undergo degradation at some point of development. The UPS is critical for the timely removal of proteins involved in self-renewal, pluripotency, and specification of NSC identity (107). In summary, these results suggest that YAP nuclear exclusion is maintained and perhaps also established by AMOT-p130, which forces hPSCs to exit the cell cycle and implicate AMOT-p130 as a key player in neuronal differentiation (10). Is this link between YAP and AMOT-p130 exclusive to mitotic cells in the nervous system? Studies on differentiating neurons in the cerebellum do indicate that it is not.

Dendritic morphogenesis

At the onset of differentiation, NSCs begin to develop a polarized morphology by forming a single axon and multiple dendrites. As the axon start to extend, a specialized structure called the growth cone senses chemical cues in the environment, which determines the direction of axon growth. As neurons mature, dendritic branches are constantly remodeled to allow for the integration of a range of synaptic input unique to each cell type in the nervous system (2).

A number of proteins have been identified to be important for the development of dendritic circuits and synapse formation with two new studies pointing to the involvement of AMOT-p130 in both of these processes. Using a floxed AMOT mouse, Rojek et al. (11) demonstrated that AMOT-p130 plays a role in modulating dendritic arbors in cerebellar Purkinje cells (PCs) and hippocampal neurons through a mechanism involving YAP. In postnatal PCs and in hippocampal neurons, Cre-mediated excision of AMOT-p130 leads to a reduction in the arborization and complexity of the dendrites. In the cerebellum, the lack of complexity is linked to reduced cerebellar size and leads to behavioral defects in gait, consistent with deficits in PC circuitry. Because YAP has been shown to associate with AMOT-p130 in cell lines, Rojek et al. generated a YAP conditional mutant mouse. Cre-mediated excision of YAP produces a similar reduction in dendritic complexity to that of AMOT-p130, implicating YAP as an effector of AMOT-p130 in dendritic arborization. How do AMOT-p130 and YAP stimulate dendritic morphology? To address this question, the authors focused on two previously described effectors of YAP/TAZ: the p70S6K1 kinase (S6K1) and its target the ribosomal protein S6 (rpS6) (108). S6K1 has received much attention due to its ability to regulate protein synthesis by phosphorylation of rpS6, and both factors have been implicated in dendritic growth (109, 110). In the cerebellum and in cultured hippocampal neurons, depletion of AMOT and YAP reduces the levels of S6K1 to influence phosphorylation of rpS6 and dendritic complexity. Conversely, expression of a constitutively active S6K1 rescues dendritic morphology, suggesting that S6K1/rpS6 acts downstream of the AMOT/YAP complex (Fig. 3, right). The molecular details of how the AMOT-p130/YAP interaction affects dendrite growth and what the contribution is of AMOT80 depletion in this context remain unclear, however. S6K1 and rpS6 appears to be involved, although exactly how they operate is uncertain, as well. Both components have been linked previously to dendritic arborization through mTOR, a multiprotein complex that regulates a slew of factors involved in protein synthesis, many with the potential to engage in a wide range of cellular behaviors (110112). Also, there are indications that mTOR1 can directly regulate microtubule dynamics (113), raising the possibility that the effects on dendrite growth through S6K1 and rpS6 are indirect. On the basis of current data, it is also not clear what induces YAP in mature neurons given that YAP expression, as discussed above, is reduced upon cell cycle exit. One possibility is that YAP levels oscillate in the nervous system, which has been reported for some proteins involved in cell cycle regulation (114). However, no such link has been implicated yet for YAP/TAZ in the brain, but examination in neural cells might resolve this issue in the future. Nevertheless, it is clear from this study that AMOT-p130 and, potentially, also other AMOT members play a role in the growth and patterning of dendrites through YAP and S6K regulation. It will be of interest to determine how neural activity controls AMOT-p130 function and whether the protein is sensitive to activity-induced dendritic patterning.

Dendritic spine development

During early development of the nervous system, the ability of dendritic spines to form synaptic contacts is critical for the growth and loss of dendritic arbors. Later in development, spines control the number of excitatory contacts that a neuron can make, thereby integrating synaptic input with signaling output (2, 3). The growth of dendritic arbors and spines is driven by signaling pathways that converge on the cellular cytoskeleton. Microtubules make up the core structural backbone in dendrites, while actin localized near dendritic membranes generates protrusions that give rise to dendritic spines and their precursors, filopodia. Through a mechanism that is not fully understood, immature filopodia develop into mature, synapse-bearing spines, with the actin cytoskeleton providing critical support throughout this process (115, 116).

Evidence for the involvement of AMOT-p130 in the cellular machinery that controls the development of spines was shown by our group (9). In rat hippocampal neurons, AMOT-p130 accumulates in the postsynaptic density (PSD) where it can associate with the actin cytoskeleton and the multiple PDZ domain protein (MUPP1), which, in turn, binds PSD-95, a major structural component of the postsynapse. Silencing of AMOT-p130, which inhibits spine maturation, enhances the occurrence of filopodia along dendrites and leads to a reduction in PSD-95 expression levels. These effects are likely derived from a structural loss in cytoskeleton dynamics because fluorescent recovery after photobleaching experiments shows that green fluorescent protein (GFP)–actin mobility is reduced when coexpressed together with wild-type AMOT-p130. Notably, a mutant of AMOT-p130 that is unable to bind PSD components can instead increase the mobility of GFP-actin. Considering that AMOT-p130 is stably anchored in the synapse, this indicates that its role, at least in part, is to act as a bridge between PSD components and the actin cytoskeleton. What then regulates AMOT-p130 in the synapse?

In nonneuronal cells, LATS1 directly phosphorylates AMOT-p130 on a conserved serine residue (Ser175), which restricts the association of AMOT-p130 with actin (18, 31, 41, 42). LATS1 can similarly phosphorylate AMOT-p130 in the brain; however, phosphorylation occurs only during the first week of postnatal development and decreases sharply for the ensuing time points (Fig. 4, A and B). The same phosphorylation site on AMOT-p130 can be stimulated—in mature hippocampal neurons—with reagents that disrupt F-actin stability, which causes AMOT-p130 to dissociate from actin and affects spine morphogenesis. Expression of a loss-of-function mutant of AMOT-p130 that cannot efficiently bind actin produces filopodia-like protrusions, similar to that of silenced AMOT-p130. The consequent spine morphology defect suggests that AMOT-p130 suppresses cytoskeleton dynamics, whereas LATS1-mediated phosphorylation attenuates this process by excluding AMOT-p130 from actin. Later in development, LATS1 activity somehow ceases, which enables AMOT-p130 to fine-tune actin dynamics and promote spine morphogenesis.

Fig. 4 Regulation of AMOT-p130 during early neuronal development.

(A) In the developing and early postnatal brain, AMOT-p130 is chronically phosphorylated by the Hippo kinase LATS1 restricting its localization from protrusions along dendrites. Through an important but yet unknown mechanism, LATS1 kinase activity diminishes, which increases the pool of unphosphorylated AMOT-p130, which can associate with F-actin. This allows AMOT-p130 to diffuse into developing spines. (B) During maturation of synapses, dephosphorylated AMOT-p130 enriches in the spine head where it interacts with structural scaffolds linked to the PSD including PSD-95 and MUPP1, stabilizing synaptic F-actin. The link between F-actin and AMOT-p130 points to a number of interactions, which has been described in epithelial cells that may potentially bind in spines as indicated and remain to be discovered (dashed arrows). LATS1 accumulate in the synapse but is functionally separated from AMOT-p130 in the mature brain and neurons.

CREDIT: A. KITTERMAN/SCIENCE SIGNALING

Applying an antagonist of protein phosphatase 2 (PP2A) reportedly enhances LATS1/2 activity (117) and induces AMOT-p130 Ser175 phosphorylation in mature neurons in culture (9). The subsequent fate of AMOT-p130 in neurons is not currently known, but for a clue, we might look to similar mechanisms in other model systems. One speculation is that LATS1/2 phosphorylation early in development acts as a stability-altering cue that triggers AMOT degradation. What prevents LATS1 from phosphorylating AMOT-p130 throughout the neuron is also not currently known. Both proteins are endogenously expressed in the nervous system and localize to the same synaptic compartments upon neuron maturation. One possibility is that LATS1 and AMOT-p130 are functionally separated in the molecularly crowded synapse (118). Alternatively, the conserved phosphorylation site in AMOT-p130 may be masked by binding to the actin cytoskeleton and other synaptic components. A mass spectrometry screen revealed that PP2A forms a complex with AMOT in other systems (22). If such a complex exists also in the hippocampus, which remains to be shown, then it would explain the rapid phosphorylation of AMOT-p130 in response to PP2A inhibition in neurons. Thus, separation of AMOTs into discrete cellular compartments is sustained by actin cytoskeleton binding and phosphorylation-dephosphorylation cycles that are mutually exclusive. Together, these data raise the notion that LATS proteins function to restrict AMOT-p130 during synapse formation and that once the postsynaptic region develops, LATS kinase activity is reduced, enabling AMOT-p130 to bridge filamentous actin with the PSD elements to enhance synapse stability.

Concluding thoughts

Understanding the differentiation, organization, and plasticity of neurons will support and expand the development of new therapies aimed at preventing and reversing neurological disorders. Uncovering the associations involved in these processes will be essential for the evaluation of how synapses integrate and transmit information and how neuronal circuits are built. Relatively recent discoveries have identified a slew of molecules with key functions in cell fate specification in different organisms, and many more are being discovered. How these interactions are controlled and in which context-dependent settings these signaling complexes operate are still being investigated. A combination of data from different laboratories and different experimental methodologies has now provided evidence of a role for the AMOT-p130 isoform in the nervous system. During stem cell differentiation and neuronal maturation, AMOT-p130 is involved in a wide range of processes from self-renewal to the patterning of dendrites and control of spine development. The molecular networks surrounding AMOT-p130 in these contexts include F-actin, cell adhesion molecules, and components of the Hippo pathway or their targets, and more associations will likely be added to this interaction map in the near future. For instance, it seems likely that AMOT-p130 can interact with KIBRA (which has been linked to memory defects) and Merlin, as loss of either protein produces a similar synapse development defect (119, 120) to that of silenced AMOT-p130. Moreover, with essential roles in all aspects of actin dynamics, one might expect that Rho GTPases are coupled as well to AMOT-p130 in dendritic spines in some way. Elaboration of these associations is also important in the context of understanding whether AMOT-p130 functions in human central nervous system (CNS) disorders, given the described link between AMOT-p130 and autism (81). In addition, there are several isoforms of the AMOTs that presumably have distinct functions in the brain. For example, AMOTL1 and AMOTL2 are both expressed in various CNS regions (http://mouse.brain-map.org/) (20, 121), but their functions in these CNS regions remain unknown. As AMOTs can associate with each other, it will also be crucial to assess whether homo- and hetero-interactions between members exist in the CNS and how such complexes might operate in the nervous system. Many other questions remain as well. For example, in all cell types studied, AMOTs are modified by ubiquitin-mediated degradation. Does the same type of control also functions in neurons? What other types of PTMs control the functions of AMOTs? We know that phosphorylation plays a role in the spatial distribution of AMOT-p130 in the embryo, but almost nothing is known about other types of modifications in the mature nervous system. Last, how do AMOT-p130 expression levels affect dendritic patterning? The possibility that all AMOT members play subtle roles in all of these processes either directly or indirectly remains to be discovered.

REFERENCES AND NOTES

Acknowledgments: Funding: Work in the Fawcett laboratory was supported by funding from the Canadian Institutes of Health Research (CIHR–MOP 84366 and PJT 159738) and a National Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant (RGPIN/05146). Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article