Polarity Proteins: Shaping Dendritic Spines and Memory

Abstract

The morphogenesis and plasticity of dendritic spines are associated with synaptic strength, learning, and memory. Dendritic spines are highly compartmentalized structures, which makes proteins involved in cellular polarization and membrane compartmentalization likely candidates regulating their formation and maintenance. Indeed, recent studies suggest polarity proteins help form and maintain dendritic spines by compartmentalizing the spine neck and head. Here, we review emerging evidence that polarity proteins regulate dendritic spine plasticity and stability through the cytoskeleton, scaffolding molecules, and signaling molecules. We specifically analyze various polarity complexes known to contribute to different forms of cell polarization processes and examine the essential conceptual context linking these groups of polarity proteins to dendritic spine morphogenesis, plasticity, and cognitive functions.

Introduction

Dendritic spines are dynamic, small protrusions on neurons serving as the main sites for excitatory glutamatergic postsynaptic input (Figure 1A). These actin-rich protrusions compartmentalize biochemical and electrical signaling^1–7 (Figure 1B). Activity-dependent plasticity of dendritic spines is associated with learning^8,9 and is accompanied by remodeling of the cytoskeleton, scaffolding molecules, and glutamate receptors^10,11. Spine plasticity can result in long-term potentiation (LTP) or depression (LTD), which are associated with dendritic spine enlargement or shrinkage, respectively^12–14. Stable dendritic spines can persist for years^8,15 and are thought to encode long-term memories¹⁶. Considering that each dendritic spine has the ability to be independently altered¹², mechanisms must be in place to establish, remodel, and maintain the compartmentalization of these tiny structures to allow plasticity during learning and maintain stability for long-term memories.

Figure 1:

Compartmentalization of subcellular domains in neurons. (A) Neurons are polarized between the somatodendritic compartment (green), which contain dendritic spines, and the axon (blue). (B) The formation of dendritic spines leads to further compartmentalization between the spine head, spine neck, and the main dendritic shaft. The postsynaptic density (PSD), where neurotransmitter receptors, adhesion molecules, scaffolding and signaling proteins are clustered, establishes another subcompartment that is distinct from the rest of the spine head. Finally, nanocolumns (orange shading) form between the presynaptic terminal and the PSD, which align clusters of presynaptic vesicles and post synaptic receptors to facilitate synaptic transmission.

The compartmentalization of specific subcellular domains is often mediated by cell polarity proteins. Cell polarity is a fundamental feature found in nearly all cells from simple organisms, such as the budding yeast, to humans. In epithelial cells, these proteins contribute to the specification of apical versus basolateral membrane domains¹⁷. Polarity is also manifested at the tissue level where planar cell polarity (PCP) proteins coordinate the orientation of a sheet of epithelium¹⁸. Various protein complexes contributing to apical-basal and/or planar polarity have been implicated in establishing and maintaining dendritic spine structure and function. These polarity proteins also regulate cytoskeletal dynamics, which are important for the structural plasticity of dendritic spines. In addition, these proteins function to control membrane compartmentalization, regulating the clustering of receptor, signaling, and scaffolding molecules within the postsynaptic density (PSD) of the spine head¹⁹. In this perspective, we describe recent advances of the role of polarity proteins in dendritic spine morphogenesis and cognitive functions. In addition, we discuss future avenues of research in this field that will be critical for our understanding of brain functions under both physiological and pathological conditions.

Polarity proteins in dendritic spine morphogenesis, plasticity, and cognitive functions

The yeast cdc mutants

Many of the early studies on the cell polarity machinery used the budding yeast Saccharomyces cerevisiae, which forms a highly polarized bud during asymmetric cell divisions²⁰. Screens for cell division cycle (cdc) mutants led to the identification of several proteins important for cell polarity, including Septins (Cdc3, Cdc10, Cdc11, Cdc12), and Cdc42. Septins are a class of GTP-binding proteins that assemble into filaments to form arc or ring structures typically associated with the membrane, F-actin, and microtubules^21,22. In yeast, Septins form a ring in the bud neck to create a diffusion barrier preventing membrane protein movement between the bud and mother cell^23,24. Cdc12 deficient yeast displays elongated buds²³. Similarly in neurons, downregulation of Sept7 results in elongated dendritic spines²⁵. Additionally, knockdown of Sept2 and Sept6 decrease dendritic spine number, while knockdown of Sept2, Sept5, Sept6, or Sept7 all decrease dendrite complexity. Overexpression of Sept2, Sept6 or Sept7 increases dendritic spines and dendrite complexity^26,27. In proteomic and biochemical analyses, Septins are found associated with PSD fractions^26,28–30. Consistent with this, Sept7 also associates with and stabilizes postsynaptic density protein 95 (PSD-95) in a phosphorylation-dependent manner, which may serve to compartmentalize calcium signaling and stabilize the dendritic spine³¹.

Similar to their localization at the bud neck in yeast, Septins are enriched at the curvature between the dendrite shaft and dendritic spine neck^25,26,32,33. Sept6 creates ring or crescent shapes at dendritic branch points and at the base of filopodia and dendritic spines²⁷. Sept6 is found in between the microtubules (MTs) in the dendrite and the actin in the dendritic spine and is associated with MTs but not actin³⁴. Similarly, Sept7 forms a saddle-like structure underneath the plasma membrane to regulate the trafficking of dendritic spine membrane proteins, such as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPAR) subunit GluA2; however, it does not restrict cytoplasmic flow³³. Sept7 also forms a complex with Extracellular signal-regulated kinase 3 (ERK3) and Mitogen-activated protein kinase-activated protein (MAPKAP) kinase 5 (MK5), which together with kalirin-7 (Kal7) regulates dendritic spine formation³⁵. However, it remains elusive how Septins balance their roles between the formation of a diffusion barrier at spine neck and stabilizing the PSD at the spine head.

Apart from Septins, Cdc42 is another key regulator of cell polarity in the budding yeast. Cdc42 is a highly conserved Rho family small GTPase that is 80% identical between yeast and humans. In the budding yeast, activated, GTP-bound Cdc42 is highly concentrated in the daughter bud³⁶. Similarly in neurons, synaptic stimulation activates Cdc42, and activated Cdc42 is highly restricted to the stimulated dendritic spines³⁷. Also, forebrain specific conditional knockout of Cdc42 shows impaired structural and functional synaptic plasticity and remote memory recall³⁸. Although the exact mechanism of how Cdc42 regulates dendritic spines and memory remains elusive, Cdc42 is an upstream regulator of actin dynamics, suggesting a possible cytoskeleton regulation mechanism. In addition, Cdc42 is upstream of the Par polarity complex, which will be discussed in more detail below.

The apical Par complex

Partitioning defective (Par) proteins were identified in a C. elegans screen for zygotes exhibiting partitioning defects during their first asymmetric cell division³⁹. Par proteins (with the exception of Par2) are conserved from C. elegans to mammals and are essential for many cellular polarization processes including asymmetric cell division, epithelial polarization, and directional migration¹⁷. Par3, a PDZ containing scaffolding molecule, and Par6, an adaptor protein, interact with atypical protein kinase C (aPKC) to form the Par complex. However, Par6 and aPKC cycle between interacting with Par3 and the aforementioned Cdc42⁴⁰. In epithelial cells, the Par complex is essential for the specification of the apical membrane¹⁷. However, in neurons the Par complex regulates dendritic spine morphogenesis. In mature (Days in vitro 14) primary hippocampal neurons, knockdown of Par3 increases filopodia- and lamellipodia-like immature dendritic spines⁴¹, suggesting the Par complex regulates mature dendritic spine stability. Par3 binds to TIAM1, a guanine nucleotide exchange factor (GEF) for Rac GTPases, to restrict Rac activity to the dendritic spine. This spatially restricted localization of Rac activity is essential for the formation of mature dendritic spines⁴¹. During dendritic spine plasticity, TIAM1 forms a complex with Calcium/calmodulin-dependent protein kinase II (CaMKII). The TIAM1/CaMKII interaction maintains CaMKII and Rac activity, a necessary step for structural LTP and memory⁴². In addition to containing Rac activity to the dendritic spine via TIAM1 localization, the Par complex is upstream of LIMK2⁴³ and the p190 RhoGAP-RhoA pathway⁴⁴, both of which regulate actin dynamics. Thus, the Par complex regulates dendritic spine morphogenesis through several pathways ultimately converging on actin regulation, serving as a probable mechanism in the actin-rich dendritic spines.

Beyond actin regulation, the Par complex is implicated in long-term memory and late LTP maintenance through aPKC. PKCι/λ, a full length aPKC isozyme, traffics AMPARs to the dendritic spine during late LTP maintenance⁴⁵. PKMζ, a truncated constitutively active isozyme of aPKC and a proposed memory molecule, is necessary for late LTP maintenance and memory^46–48. However, PKMζ deficient mice display no dendritic spine or cognitive defects due to PKCι/λ compensation^49–51. Double conditional knockout (dcKO) of PKMζ and PKCι/λ results in decreased glutamatergic synapses and behavioral phenotypes including memory and social interaction defects⁵². Together, these data not only implicate the Par complex in dendritic spine morphogenesis and plasticity but also in cognitive functions.

How is the Par complex targeted to dendritic spines? Studies by the Tolias lab show that Par3 and TIAM1 directly interact with brain-specific angiogenesis inhibitor 1 (BAI1), a synaptic adhesion GPCR⁵³, resulting in their recruitment to the dendritic spine. A more recent study by the Penzes group shows Par3 directly interacts with contactin-associated protein-like 2 (CNTNAP2), also known as Caspr2, a synaptic cell adhesion molecule strongly associated with various neurodevelopmental disorders including autism spectrum disorders (ASD), intellectual disability, and schizophrenia⁵⁴. This association is particularly interesting considering that Par3 has been genetically linked to ASD⁵⁵ and schizophrenia⁵⁶ and Par6 is associated with bipolar disorders⁵⁷. Thus, it would be interesting to explore the interaction between the Par complex and CNTNAP2 in synaptic dysfunction in neurodevelopmental disorders.

The basolateral Scribble/Lgl/Dlg complex and Par1

In epithelia cells, the apical Par complex antagonizes the functions of basolateral protein complexes including the Scribble/Lethal giant larvae (Lgl)/Discs large (Dlg) complex and the Ser/Thr kinase Par1¹⁷. Interestingly, these basolateral polarity complexes are also emerging as important regulators in synaptic plasticity and cognitive functions. Scribble (Scrib) regulates the synaptic vesicle dynamics⁵⁸ in the presynaptic terminal through interaction with β-catenin⁵⁹. On the postsynaptic side, Scrib recruits nitric oxide synthase 1 adaptor protein (NOS1AP) to regulate Rac activity, which ultimately influences cytoskeletal dynamics and dendritic spine morphogenesis^60–62. Scrib deficient mice display increased number of enlarged spines, overactive Rac1, and aberrations in actin dynamics. Scrib deficient mice also exhibit impaired social behavior and enhanced learning and memory⁶³, linking the role of Scrib in dendritic spine morphogenesis to both social and cognitive functions.

A recent study from the Zou group highlights the role of Lgl in glutamatergic synapse formation and cognitive functions. Conditional knockout (cKO) of Lgl1 in pyramidal neurons leads to increased density of dendritic spines and glutamatergic synapses. In addition, a reduction in AMPAR to N-methyl-D-aspartate receptor (NMDAR) ratio and an impairment of synaptic plasticity was observed in these mice. Lgl cKO mice display increased locomotion, impaired novel object recognition and social interactions. Interestingly, knockout of Lgl1 rescues glutamatergic synapse number and cognitive deficits in the aPKC dcKO mice, indicating that antagonism between apical and basal polarity complexes may balance the control of synapse formation and cognition⁵².

Another basolateral polarity protein implicated in dendritic spine morphogenesis is the Ser/Thr kinase Par1, also known as microtubule affinity regulating kinase (MARK). In hippocampal neurons, Par1 is activated downstream of NMDAR to phosphorylate PSD-95⁶⁴. This PSD-95 Ser561 phosphorylation induces a conformational shift reducing the interaction between PSD-95 and its interacting partners to increase PSD-95 dynamics, a necessary step for both LTP and LTD⁶⁵. Activated Par1 also phosphorylates microtubule associated proteins (MAPs) to increase the dynamics of microtubules⁶⁶, which is important for dendritic spine plasticity⁶⁷. Taken together, these studies point to apical-basal polarity proteins, such as Par proteins, as key regulators of dendritic spine morphogenesis and cognition. However, much remains unknown of their exact mechanisms of action during dendritic spine plasticity, which begs further studies.

The Vangl/Celsr/Prickle PCP complex

In contrast to apical-basal or anterior-posterior polarity that mainly depend on cell intrinsic mechanisms, planar cell polarity (PCP) requires coordinated signaling across a group of cells. Key to this coordinated signaling is the asymmetric positioning of transmembrane receptors including Frizzled (Fz) and Van Gogh (Vang, or Vangl in vertebrates), which form intercellular interactions with each other. In planer polarized epithelial tissue such as the Drosophila wing epithelium, Fz localizes to the distal cell junctions, whereas Vangl localizes to the proximal junctions along with the cytoplasmic protein Prickle. Both Fz and Vangl laterally associate with Celsr (also called Flamingo), which is an atypical cadherin that forms trans-homophilic adhesions¹⁸. In neurons, Vangl2 interacts directly with PSD-95 through PDZ domains⁶⁸ and regulates of N-cadherin internalization⁶⁹. Several recent studies from the Zou lab elegantly demonstrate a key role for the Vangl/Celsr/Prickle PCP signaling complex in excitatory synapse formation and plasticity. Using postnatal day 7 (P7) conditional knockout (cKO) of Vangl2 and Celsr3, Zou and colleagues show Vangl2 and Celsr3 play opposing roles in glutamatergic synapse formation. Celsr3 promotes the formation of excitatory synapses while having no effects on inhibitory synapses⁷⁰. Celsr3 cKO mice also display altered LTP, hyperactivity, and learning and memory deficits⁷¹. On the other hand, Vangl2 inhibits the formation of excitatory synapses in vivo. This is in contrast with earlier studies using primary neuronal cultures derived from germline knockout and loop tail mutations of Vangl2, which show Vangl2 depletion decreases dendritic branching, dendritic spine density^72–73, and synapse formation^69,73. However, the Vangl2 mouse lines used in these earlier studies also develop alterations in dendrite complexity, neural tube formation, and axon projection^74–76, suggesting the effects of Vangl2 on early brain development may have complicated the interpretation of its role in synapse formation. Interestingly, Vangl2 is significantly decreased in the aforementioned Lgl cKO mice, which show increased excitatory synapse formation⁵². This indicates crosstalk between the apical-basal polarity pathway and PCP pathway providing further evidence for Vangl2 as a negative regulator of dendritic spine and excitatory synapse formation.

Prickle2, which binds the cytoplasmic tail of Vangl, interacts with PSD-95 and NMDARs in the CA1 region of the hippocampus⁷⁷. Prickle2 inhibits the N-cadherin-Vangl2 interaction and is required for normal synapse formation⁶⁹. Loss of both Prickle1 and Prickle2 during postnatal development at P7 causes up to an 80% loss of PSD-95 positive glutamatergic synapses and a decrease in overall dendritic spine density⁷⁸. A point mutation of Prickle2, E8Q, which is found in autistic patients⁷⁹, also leads to decreases in PSD-95 containing synapses, dendritic spines, and spatial learning speed. Prickle2 antagonizes Vangl2 and stabilizes the Celsr3/Frizzled3-Celsr3 intercellular complex to promote synapse formation while the E8Q mutation prevents Prickle2 from stabilizing the complex⁷⁸.

Finally, beyond synaptogenesis during development, Vangl2, Celsr, and Prickle regulate synapse maintenance in adulthood. In adult mice, knocking out Celsr2 and Celsr3 reduces dendritic spine and synaptic density in the hippocampus⁸⁰. Similarly, loss of Prickle1 and Prickle2 in adulthood causes a drastic reduction in excitatory synapses⁷⁸. By contrast, knocking out Vangl2 increases the number of synapses⁸⁰. Interestingly, oligomeric β-amyloid (Aβ) binds Celsr3, which disrupts the Fz/Celsr3 signaling and enhances Vangl2 function providing a potential mechanism for synaptic loss during Alzheimer’s disease pathogenesis⁸⁰. This suggests Celsr, Prickle, and Vangl2 play similar roles in regulating dendritic spines in both development and adulthood. Together, these studies reveal key roles for PCP signaling in dendritic spine morphogenesis as well as cognitive and social behavior.

The Wnt/Frizzled/Dishevelled complex

In the Drosophila wing epithelium, the Fz receptors are restricted to the distal edge of the apical membrane along with the cytoplasmic scaffolding protein Dishevelled (Dvl). This Fz/Dvl signaling complex is regulated by Wingless-type (Wnt). The canonical Wnt pathway regulates gene transcription through β-catenin. Noncanonical Wnt signaling includes the Wnt/PCP pathway and the Wnt/Ca²⁺ pathway⁸¹. Several studies have addressed the role of the Wnt/Fz/Dvl signaling complex in dendritic spines; however, they have focused on downstream mechanisms not directly related to PCP signaling. In neurons, both Dvl and Fz are highly localized to dendritic spines⁸² suggesting a dendritic spine specific mechanism. Wnt2, Wnt5a, and Wnt7a increase dendritic spine density^83,85 while Wnt5a also increases clustering of PSD-95⁸⁶ and synaptic transmission⁸⁴. Wnt7a functions through Dvl1 and CaMKII to increase excitatory synaptic transmission⁸⁵. In addition, Wnt7a acts through Dvl1 signaling to promote multiple innervated spines (MIS)⁸⁷. Thus, the Wnt/Fz/Dvl pathway may play a role in memory since MIS facilitates long-term memory^88–90. Lastly, Wnt7a/b acts through Fz7 to recruit AMPARs to the synapse to increase synaptic strength and spine growth during LTP⁹¹. Together, these studies demonstrate that the Wnt/Fz/Dvl signaling complex plays a vital role in dendritic spine morphogenesis and plasticity.

Interestingly, the positive effects of Wnt5a on synaptic transmission happen within 30 minutes, which is believed to be due to the activation of the Wnt/Ca²⁺ pathway⁸⁴. However, long-term treatment of Wnt5a (>12h) leads to a reduction in glutamatergic synapses^70,80,92, which may be due to the activation of the Wnt/PCP pathway. Given that all three downstream pathways of Wnt may potentially be involved in synapse formation and plasticity, it will be interesting to further dissect the complex role of Wnt in the PCP pathway versus pathways unrelated to PCP signaling.

Concluding Remarks and Perspectives

Dendritic spines and their associated synapses provide the most basic computational unit within the brain. Understanding the mechanisms leading to the proper morphogenesis and plasticity of these structures is key to deciphering the neural codes for cognitive processes, such as memory and social behaviors. Here, we have highlighted the role of cell polarity regulators in dendritic spine morphogenesis and plasticity. Yet, many key questions remain. For example, one key feature of synapse formation is its high level of specificity. How do neurons properly identify their synaptic partners to form trillions of highly specific connections? This specificity would require coordinated signaling across groups of neurons. In epithelia, PCP proteins coordinate cell-to-cell interactions to create tissue polarity spanning a group of cells. In neurons, PCP proteins regulate both pre- and postsynaptic sites. It would be interesting to speculate that the regulation of PCP signaling is conserved in neurons to confer specificity in neurocircuitry formation.

Another important feature of dendritic spines is their ability to be individually regulated. How do neurons compartmentalize dendritic spines a few micrometers apart? How does a dendritic spine remain stable while the neighboring spine undergoes plasticity? In epithelial cells, the Par proteins, Septins, and PCP proteins all function to compartmentalize different membrane domains. Recent studies described here suggest similar mechanisms compartmentalize the dendritic spine from the dendritic shaft. However, further details are necessary to elucidate the mechanisms underlying the apparent heterogeneity of dendritic spines and their varied levels of diffusional isolation⁶.

Within the PSD of dendritic spines, AMPARs, NMDARs, and scaffolding proteins including PSD-95 and Shank form nanoclusters around 50-80 nm in size^93–95. AMPAR nanodomains are closely aligned with presynaptic release sites to form nanocolumns across the synaptic cleft (Figure 1B), which likely plays a role in modulating synaptic efficiency. The addition of new nanodomains are associated with plasticity⁹⁶. However, the mechanisms responsible for the formation and regulation of these nanodomains remain unclear. It will be interesting to explore whether polarity proteins play a role in establishing and maintaining these nanodomain structures within the synapse.

Finally, what are the synaptic mechanisms underlying the unique features of the human brain? Humans have a higher density of dendritic spines which are longer, resulting in increased morphological complexity⁹⁷. Additionally, these synapses are developed throughout a delayed maturation period compared to rodents. What are the molecular mechanisms driving these differences? To determine possible human-specific mechanisms, scientists have turned to experimental models, including induced pluripotent stem cells (iPSCs) and organoids both derived from patients and humanized rodent models. For example, humanized Slit-Robo Rho-GTPase Activating Protein 2 (SRGAP2C) mice were developed by replacing the mouse gene, SRGAP2A, with human SRGAP2C. Introducing the human gene SRGAP2C mimicked increased dendritic spine density, delayed maturation, and lengthened spine neck observed in the human brain⁹⁸. Since human dendritic spines are more compartmentalized than rodent spines, do polarity proteins function to enhance this compartmentalization? It would be interesting to examine the role of polarity proteins in human specific synaptic mechanisms using iPSC or humanized animal models.

In summary, studies over the last ten to fifteen years have revealed key roles for polarity proteins in the morphogenesis and plasticity of dendritic spines, as well as in cognitive and social functions of the brain. Much research remains to be done, and it will be exciting to move the field forward to further elucidate the mechanisms of synaptic regulation by these polarity determinants, especially in relation to various cognitive processes. These studies will not only shed light on how neural circuits modulate behavior but also advance the understanding of neurodevelopmental disorders and neurodegenerative diseases with aberrations in dendritic spine morphology and function.

Highlights.

• Polarity proteins regulate the compartmentalization of dendritic spines.

• Many polarity proteins regulate synaptic and cognitive functions.

• Neurodevelopmental disorders have been linked to polarity proteins.