Abbreviations: β-cat, β-catenin; Δβ-cat, truncated fragment of β-catenin; Scrib, scribble.
*Address correspondence to: Nancy Y. Ip, Division of Life Science, State Key Laboratory of Molecular Neuroscience and Molecular Neuroscience Center, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China. Tel.: +852-2358-7267. Fax: +852-2358-1552. E-mail: email@example.com
Human cognitive functions depend on the precise neural network formed by neurons in the brain. Neurons communicate through neurotransmission at the synapse, which comprises a presynaptic terminal that contains the active zone for neurotransmitter release, a postsynaptic apparatus that harbors a high density of receptors for relaying the signals to the target cell, and surrounding glial cells that modulate the efficacy of neurotransmission across the synapse. The process of synapse formation is initiated when the presynaptic terminal (i.e., the axon) navigates and finds its postsynaptic target. Once formed, the protein composition of the synapse, including receptors, scaffold proteins, and kinases, changes in order to maintain optimal connectivity between presynaptic and postsynaptic neurons. Synapse dysfunction is associated with a number of neurological disorders. For example, synaptic failure is an early feature of neurodegenerative diseases including Alzheimer's disease and Parkinson's disease; moreover, the hyperactivity or hypoactivity of specific types of synapses is linked to psychiatric disorders such as schizophrenia and depression . Therefore, the precise organization of synaptic proteins and the spatial and temporal coordination of synaptic signaling events are central research topics in neurobiology. Although extensively studied, the molecular basis of synapse development and plasticity is not fully understood. This review summarizes the emerging evidence suggesting Axin, a scaffold protein, plays critical roles in the orchestration of synaptic protein complexes and regulation of synaptic functions in the central nervous system.
Axin and its Binding Partners in Synapse Development
Axin, a multidomain protein encoded by the mouse Fused gene, is named for its role in the inhibition of vertebrate axis formation during embryonic development. Axin is highly conserved across different species, sharing 87% amino acid identity between human and murine and 66% between chicken and mouse . Axin mRNA is ubiquitously expressed in different tissues of postimplantation embryos (E7.5–E9.5) and is enriched in ventricular zone of the brain from E11 to E16 [2, 3]. Axin protein expression in the brain is reduced from E12 to P5, and the protein is concentrated in the cortical plate and upper intermediate zone from E15 to P2 . Loss of Axin in mice leads to early embryonic lethality along with abnormalities in structures such as the forebrain and anterior–posterior axis. This indicates that Axin plays an essential role in embryonic development . A point mutation of Axin in zebrafish leads to deficits in the fate determination of the telencephalon and eyes and abolishes the establishment and elaboration of asymmetries in the central nervous system [6, 7]. Misexpression of Axin in Drosophila impairs cortical cell fate determination and visual system development . These studies strongly suggest that Axin plays critical roles in neural development.
Axin forms a large complex with various proteins via a number of interacting domains, most of which are critical players in canonical Wnt signaling. For example, Wnt coreceptor, low-density lipoprotein-related protein receptor (LRP), interacts with the Axin C-terminal region. Adenomatous polyposis coli (APC) binds to the N-terminal regulators of the G-protein signaling domain (RGS), whereas Dishevelled (Dvl) binds to the C-terminal Dishevelled/Axin homologous domain (DIX). The glycogen synthase kinase-3β(GSK-3β)-binding domain is located in the region spanning amino acids 353–437 of Axin, and an adjacent region (amino acids 437–506) is responsible for the association with β-catenin, enabling the formation of a ternary complex. In canonical Wnt signaling, Axin acts as the backbone for the destruction complex comprising Dvl, APC, casein kinase I (CKI), GSK-3β, and E3-ligase β-transducin repeat-containing protein (β-TrCP), which facilitates the phosphorylation, ubiquitination, and degradation of β-catenin. Wnt stimulation releases β-TrCP from the destruction complex, increasing cytosolic unphosphorylated β-catenin; β-catenin subsequently accumulates in the nucleus, triggering the expression of Wnt target genes . Besides this essential role in Wnt signaling, Axin exerts its biological functions through binding to other signaling partners including cyclin-dependent kinase 5 (Cdk5), growth factor receptor-bound protein 4 (Grb4), and synaptic scaffolding molecules (S-SCAM) [4, 10, 11]. Although Axin has been well studied in embryonic development and carcinogenesis, its role in the central nervous system development is only beginning to be elucidated. It is interesting that a number of Axin-binding partners are important components in central synapses and neuromuscular junctions, suggesting that Axin plays an active role in the scaffolding of synaptic architecture. This review focuses on the role of Axin and its binding partners in central synapses.
Axin has been implicated in early neuronal differentiation by its regulation of neurite outgrowth . A recent finding from our laboratory revealed an unexpected role of Axin in the regulation of axon dynamics in the early stage of synapse development . In developing neurons, Axin is targeted to the nascent axons and concentrated at growing axon tips. Loss of Axin function blocks the initiation of axons, leading to severe deficits in axon formation in the developing cortex. The action of Axin depends on the Cdk5-mediated phosphorylation at Thr485, a site close to the GSK-3β-binding domain, which enhances Axin's association with GSK-3β and inhibits GSK-3β activity. This in turn stabilizes the microtubules of the extending axons by reducing the GSK-3β-mediated collapsin response mediator protein-2 phosphorylation via a mechanism independent of the canonical Wnt signaling. Interestingly, Axin interacts with the Src-homology-3 domain of the adaptor protein, Grb4, which couples ephrin-B signaling through the adjacent SH2 domain . Ephrin-Bs are the ligands for Eph receptors, a unique family of receptor tyrosine kinases capable of transducing signals bidirectionally into either the receptor- or ligand-expressing cells. The forward and reverse signaling of Eph/ephrin are important for various aspects of synapse development, including synapse formation, maturation, maintenance, and plasticity . Once reverse signaling is activated, Ephrin-Bs recruit Grb4 to their phosphorylated tyrosine residues and relay the exogenous stimulus to the cytoskeleton via small Rho GTPases (e.g., Rac and Cdc42) to control the pruning of the axonal growth cone and presynaptic differentiation . Thus, it would be intriguing to further investigate whether Axin present at the tips of the growth cones is involved in the pathfinding of axons to their destination or the initial recognition of presynaptic and postsynaptic sites between two neurons.
In addition to its role in axon development, accumulating evidence suggests that Axin is an important scaffold protein for anchoring postsynaptic proteins. Recent results from our laboratory revealed the accumulation of Axin at the postsynaptic density (PSD) fractions enriched with PSD-95 (N. Y. Ip, unpublished data). Axin–Grb4 interactions can also occur at the postsynaptic sites, where Grb4 coclusters with G-protein-coupled receptor kinase-interacting protein 1 (GIT1) upon the activation of synaptic ephrin-Bs. The disruption of this signaling cascade abolishes dendritic spine morphogenesis as well as presynaptic and postsynaptic marker clustering . These findings suggest that Axin acts as a local stabilizer of the ephrin-B/Grb4/GIT1 complex during synapse development. Axin is likely to be the key scaffold protein that orchestrates ephrin-B and its signaling complex at both presynaptic and postsynaptic sites; however, the precise molecular mechanism is unclear. Another key postsynaptic interacting partner of Axin is S-SCAM, a multidomain protein that modulates AMPA receptor-mediated synaptic transmission by recruiting the transmembrane AMPA receptor-regulating protein, stargazin [16, 17]. As the S-SCAM-binding domain overlaps with the GSK-3β-binding region, S-SCAM competes with GSK-3β to bind to Axin, and inhibits β-catenin phosphorylation and degradation . Further study is required to elucidate the synaptic localization and regulation of potential ternary S-SCAM/stargazing/Axin complex to confirm whether the interaction between components serves as a mechanism for protecting synaptic β-catenin and maintaining synaptic structural stability.
GSK-3β is well known for its function in the establishment of the polarized structures of dendrites and axons. During this process, phosphorylated and inactivated GSK-3β is specifically localized at the tip of the nascent axon to maintain the axon's fate . The precise regulation of GSK-3β activity is critical for the processes of synapse development described below.
When an axon is approaching its target, the Wnt-7a-induced inactivation of GSK-3β is a key step in stabilizing the microtubules during growth cone remodeling . In addition, growth cone collapse induced by semaphorin3A also requires the formation of GSK-3β/Axin/β-catenin complex . Inhibition of GSK-3b mimics the Wnt-7a-mediated signaling by inducing the clustering of synaptic proteins, suggesting that GSK-3β plays an important role in the maturation process of synapses . Although GSK-3 activation is associated with the suppression of presynaptic glutamate release, there is still a lack of direct evidence demonstrating the role of GSK-3 in neurotransmitter release . However, GSK-3 does in fact play critical roles during activity-dependent bulk endocytosis on elevated neuronal activity but not clathrin-mediated endocytosis on low-intensity stimulation . In activity-dependent bulk endocytosis, the calcium-dependent phosphatase, calcineurin, dephosphorylates a group of proteins called dephosphins (e.g., dynamin I); this process is essential for the initiation of activity-dependent bulk endocytosis . GSK-3 activity is initially acutely inhibited by strong depolarizing stimulation via the Akt-mediated phosphorylation, thus ensuring the dephosphorylation of dynamin I and maintaining dynamin I–syndapin I interaction to trigger endocytosis . The subsequent rephosphorylation of dephosphins is indispensable for the completion of endocytosis, during which GSK-3 phosphorylates dynamin I and other dephosphins with a Cdk5-mediated priming step at adjacent serine/threonine residues . Therefore, GSK-3 is essential for maintaining the pool of synaptic vesicles for the next round of neurotransmitter release at the presynaptic terminals.
At postsynaptic sites, GSK-3 is a key kinase that coordinates different signaling cascades. Highly enriched in synaptosomal fractions, GSK-3 interacts with AMPA receptor subunits GluA1 and GluA2, suggesting it has a key role in synaptic function regulation . Long-term potentiation (LTP) and long-term depression (LTD) are two major forms of plasticity believed to be associated with cognitive processes such as learning and memory. On the induction of NMDA receptor-dependent LTP, GSK-3β activity is inhibited by PI3K and Akt-mediated phosphorylation, preventing the synapse from undergoing LTD . Moreover, in the late phase of LTP, which depends on protein synthesis, synaptic activity abolishes the inhibitory effect of GSK-3 on mammalian target of rapamycin, which is the key kinase that triggers protein translational machinery; this enables protein synthesis for prolonged LTP expression . Interestingly, the pharmacological inhibition of GSK-3 activity in neurons blocks the induction of NMDA receptor-dependent LTD, indicating that this process requires GSK-3. The LTD-inducing signal activates GSK-3β probably via the PP1-mediated dephosphorylation at GSK-3β Ser9 and Akt inhibition . These findings suggest that GSK-3 has a critical role in the regulation of metaplasticity, in which the previous GSK-3 activity of the synapse affects the current status of the synapse to determine if LTP or LTD can be induced. Importantly, aberrant GSK-3 activity is believed to influence the effect of synaptic stimulation during learning and memory processes, possibly implicating GSK-3 in neurodegenerative diseases .
Among the substrates of GSK-3β, tau is a key microtubule-binding protein particularly important for the regulation of presynaptic and postsynaptic functions. The ability of tau to bind microtubules is tightly regulated via phosphorylation by GSK-3β. GSK-3β dysfunction leads to the hyperphosphorylation and mislocalization of tau, which are associated with neurodegeneration . Interestingly, by interacting with and sequestering GSK-3β, Axin reduces tau phosphorylation and maintains its association with the microtubule cytoskeleton . Thus, it would be intriguing to investigate the expression and localization of Axin in neurodegenerative diseases to determine whether manipulating Axin during the disease pathogenesis exerts any protective effect.
Cadherins are cell-surface adhesion molecules present on both presynaptic and postsynaptic membranes that form homophilic complexes across synapses. β-Catenin interacts with the C-terminus of cadherins; the resultant cadherin adhesive complex functions as central organizer that regulates the strength of synaptic contacts . As a key interacting partner of Axin, β-catenin not only acts as an adaptor protein to maintain the structure stability of synaptic compartments but also translocates into the nucleus to control the expressions of synapse-related genes, thus regulating synapse development. Although Axin interacts with both β-catenin and N-cadherin, its role in stabilizing this β-catenin/N-cadherin adhesion complex is not well understood .
β-Catenin–cadherin interaction is critical for synaptic vesicle clustering at the presynaptic terminal. Increasing β-catenin phosphorylation at Tyr654 by brain-derived neurotrophic factor (BDNF) or manipulating the cytoplasmic tyrosine kinase, Fer, attenuates the affinity between β-catenin and cadherin and enhances synaptic vesicle mobility [33, 34]. In addition, β-catenin recruits scribble protein to the adhesion complex together with a Rac/Cdc42 guanine nucleotide exchange factor called β-pix [35, 36]. The interaction between β-catenin and β-pix is essential for the synaptic localization of β-pix. Blockade of β-pix function destabilizes the actin cytoskeleton, leading to deficits in the localization and recycling of synaptic vesicles . Therefore, β-catenin is critical for regulating the dynamic changes of presynaptic structures during synapse formation and remodeling.
Postsynaptic β-catenin also shapes synaptic structure and function. Neuronal activity induces the translocation of β-catenin from the dendritic shaft to spines concomitant with attenuated Tyr654 phosphorylation and increased affinity for cadherin. Inhibiting β-catenin phosphorylation leads to enhanced clustering of postsynaptic PSD-95 and presynaptic synapsin I . β-Catenin overexpression reduces miniature excitatory postsynaptic current (mEPSC) amplitude and surface AMPA receptor clusters. The phenotype of β-catenin overexpression in neurons mimics the effect of chronically elevated neuronal activity with respect to AMPA receptor downscaling and enhanced dendritic growth and branching. This suggests that β-catenin is a key regulator of the homeostatic environment that prevents neuron overexcitation . On the other hand, β-catenin deficiency in postsynaptic neurons reduces the number of mushroom-shaped mature spines and increases immature spines without markedly changing the total number of spines. Consistently, β-catenin loss-of-function neurons exhibit reduced mEPSC amplitude, which depends on the AMPA receptor abundance within the spine, but not mEPSC frequency, reflecting functional presynaptic release probability . Instead, the basal release probability is believed to be set by postsynaptic cadherin rather than β-catenin. However, blockade of postsynaptic β-catenin but not N-cadherin function abolishes the homeostatic upregulation of release probability; this upregulation is triggered by the chronic inhibition of neuronal activity in a retrograde manner, which may involve gene transcription . Thus, there are different mechanisms that regulate synaptic efficacy at the basal level and during homeostatic adaptation.
Although β-catenin has been studied extensively as a downstream transcriptional regulator of Wnt signaling in embryonic development, its roles in gene regulation in synapse formation and plasticity are not well understood. LAPSER1 is a tumor suppressor that binds to β-catenin at the postsynaptic density. NMDA receptor activation results in the translocation of both LAPSER1 and β-catenin into the nucleus, activating gene transcription (e.g., Tcfe2a and c-Myc) . Elucidating the precise roles of these target genes in synapses requires further investigation. Interestingly, β-catenin is cleaved at its N-terminus by NMDA receptor-dependent calpain activity, generating a truncated fragment resistant to GSK-3β phosphorylation and degradation. The translocation of the truncated fragment into the nucleus induces an immediate early gene, Fosl1, whose expression is associated with improved performance during the learning process [42, 43]. The upregulation of calcium channel subunit Cav3.1 by β-catenin further suggests that β-catenin-dependent gene expression plays an important role in synapse development .
APC was initially identified as a tumor-suppressor gene in colon cancer. Its roles in controlling the progression of cell cycling, migration, and differentiation have been well studied. It is highly expressed in the central nervous system during the embryonic and early postnatal stages. Moreover, it is required for the clustering of neurotransmitter receptors and other proteins at the presynaptic and postsynaptic sites, probably because of its ability to regulate microtubule bundling and actin nucleation [45-47].
At the presynaptic nerve terminal, APC associates with the Axin/β-catenin complex. Wnt abolishes the APC–β-catenin interaction and in turn facilitates the formation of APC/α7-nAChR (a nicotinic acetylcholine receptor) complex with a concomitant increase in the number and size of APC/α7-nAChR clusters at the synaptic sites . At the postsynaptic membrane of chick ciliary ganglion neurons, APC associates with α3-nAChR and colocalizes with PSD-93. The interaction between APC, PSD-93, and the microtubule plus-end-binding protein, EB1, is essential for α3-nAChR surface expression and clustering . APC also interacts with PSD-95 in glutamatergic synapses; blockade of their interaction in postsynaptic neurons reduces PSD-95/GluA1 clustering and AMPA receptor-mediated synaptic current, suggesting that APC plays a key role in hippocampal synaptogenesis [50, 51].
Interestingly, nicotinic synapses require postsynaptic APC for the membrane enrichment of S-SCAM and neuroligin. In turn, interfering APC function reduces the clusters of neurexin on the opposing presynaptic terminals along with decreased clustering of SV2 and other active-zone proteins as well as defects in presynaptic vesicle recycling. These findings suggest that APC mediates retrograde signaling that directs presynaptic specification .
Dvl is a key downstream player of Wnt. At the presynaptic terminal, Dvl inhibits GSK-3β activity, thus destabilizing APC at the microtubule plus-end. Instead of transcriptional regulation in canonical Wnt signaling, this pathway directly controls the growth and directionality of microtubules during axon remodeling . Dvl also colocalizes with the presynaptic active zone marker, Bassoon, and mediates the retrograde signaling to promote presynaptic differentiation, synaptic vesicle recycling, and neurotransmitter release . At the postsynaptic sites, Dvl is concentrated at postsynaptic density and mediates the Wnt-dependent local activation of Ca2+/calmodulin-dependent protein kinase II, which is required for spine maturation and synaptic transmission at the excitatory synapse .
As a multidomain scaffold protein, Axin provides an ideal platform for the docking of different synaptic components. This review summarizes some recent evidence indicating that Axin and its interacting partners play important roles in synapses (Fig. 1). It should be noted that molecules such as protein phosphatase 2a, CKI, and other Wnt signaling components not discussed above because of space limitations are also involved in different aspects of synapse development. However, direct evidence demonstrating Axin function at synapses remains limited. Several key questions should be addressed in future studies. For example, how the expression, localization, and post-translational modification of Axin as well as Axin's interactions with its binding partners are regulated during synapse development have not been systematically investigated. It remains unclear whether and how manipulating endogenous Axin level affects synapse formation and remodeling. It would also be interesting to know the differential mechanisms by which Axin organizes the signaling complex in Wnt-dependent and Wnt-independent synaptic events. As synaptic dysfunction occurs at the early stage of neurodegenerative diseases and neuropsychiatric disorders, it would be intriguing to investigate how Axin is involved in disease progression and if Axin is a key biomarker of such diseases.
The authors apologize to the many authors whose excellent works were not cited in this review because of space limitation. The authors thank Dr. Kwok-On Lai for his critical comments on the manuscript, and Ka-Chun Lok for his help in preparing the figure. The studies by N. Y. Ip were supported in part by the Research Grants Council of Hong Kong (HKUST 660110 and 660810), the National Key Basic Research Program of China (2013CB530900), the Shenzhen Peacock Plan, the Theme-based Research Scheme of the University Grants Committee (T13-607/12-R), and the Innovation and Technology Fund for State Key Laboratory (ITCPT/17-9).