Anchored kinase and phosphatase signaling networks in amyloid β-induced synaptic dysfunction

Abstract

Impaired excitatory synaptic function and synapse loss are early hallmarks of Alzheimer’s disease (AD). There is strong biochemical, genetic, physiological, and anatomical evidence that the accumulation of soluble amyloid beta (Aβ) oligomers in the brain leads to AD-related synapse dysfunction and cognitive impairment. Long-term potentiation (LTP), a key form of synaptic plasticity for learning and memory, is disrupted in several mouse models harboring familial early-onset AD–linked mutations that lead to Aβ accumulation, and even acute applications of Aβ oligomers block LTP within minutes and promote synapse loss within days. How does Aβ cause such profound synaptic dysfunction? It is increasingly appreciated that Aβ hijacks normal synaptic signaling pathways involved in plasticity, biasing them toward long-term depression (LTD) and eventual synapse elimination. LTP and LTD in the hippocampus are ultimately driven by insertion and removal of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid–type glutamate receptors (AMPARs) from synapses, respectively, in response to Ca²⁺ signals generated by N-methyl-d-aspartate–type glutamate receptors or L-type voltage-gated Ca²⁺ channels. Central to the signaling pathways regulating AMPAR trafficking during LTP and LTD is a postsynaptic Ser/Thr kinase/phosphatase signaling network that is coordinated by the scaffold protein A-kinase anchoring protein (AKAP) 79/150 (human79/rodent150; AKAP5/Akap5 gene). This AKAP-organized signalosome includes N-methyl-d-aspartate–type glutamate receptors, AMPARs, L-type voltage-gated Ca²⁺ channels, G-protein–coupled receptors, adenylyl cyclase, the cyclic adenosine mono-phosphate (cAMP)-dependent protein kinase, and the Ca²⁺-calmodulin–dependent protein phosphatase 2B/calcineurin. This minireview will highlight recent findings that Gαs-coupled β₂-adrenergic and Gαq-coupled group 1 metabotropic glutamate (mGlu1/5) receptors signal through AKAP79/150-anchored cAMP-dependent protein kinase and protein phosphatase 2B/calcineurin to mediate multiple aspects of Aβ synaptotoxicity.

Significance Statement

Recent studies reveal that amyloid beta engages A-kinase anchoring protein–scaffolded G-protein–coupled receptor signaling pathways to disrupt synaptic plasticity and promote synapse loss. These pathways contain several potential therapeutic targets involved in both local cAMP-dependent protein kinase and calcineurin/protein phosphatase 2B (CaN) signaling that regulates L-type voltage-gated calcium channels, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors, and N-methyl-d-aspartate receptors to impair synaptic plasticity as well as distal CaN signaling to the nucleus that regulates gene expression to drive synapse loss.

1. Introduction

Regulation of N-methyl-d-aspartate–type (NMDAR) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid–type (AMPAR) glutamate receptors in the hippocampus is required during long-term potentiation (LTP) and long-term depression (LTD) synaptic plasticity to increase and decrease synaptic strength, respectively, to support learning and memory.1, 2, 3 Under basal conditions, AMPARs regulate the majority of fast excitatory synaptic transmission. However, during plasticity induction, NMDARs become active in response to coincident detection of presynaptic glutamate release and postsynaptic depolarization that removes a Mg²⁺ ion blocking the channel pore to allow Ca²⁺ influx through the receptor.⁴ This NMDAR Ca²⁺ influx activates downstream signaling pathways including protein kinases, such as Ca²⁺-calmodulin–dependent protein kinase II (CaMKII) and the cAMP-dependent protein kinase (PKA), and protein phosphatases, such as protein phosphatases 1 and 2B (PP2B/calcineurin [CaN]). These protein kinases and phosphatases in turn regulate AMPAR activity by modulating channel properties and receptor localization to the postsynaptic density both directly through receptor phosphorylation and indirectly through phosphorylation of AMPAR-associated proteins and postsynaptic structural proteins.^5,6 To this end, these kinase and phosphatase pathways also trigger corresponding changes in postsynaptic dendritic spine structure, with LTP promoting spine formation and enlargement and LTD favoring spine shrinkage and elimination.^7,8 Importantly, alterations in these LTP and LTD mechanisms impacting dendritic spine structure are implicated in nervous system disorders, including schizophrenia, Down syndrome, autism, and Alzheimer’s disease (AD).^9,10

NMDAR-induced changes in postsynaptic structure and function during LTP and LTD depend on the activation of what are inherently broad specificity kinases and phosphatases (ie, CaMKII, PKA, and CaN) that have diverse cellular functions both inside and outside of the nervous system. Thus, it is important to understand how the signaling functions of these multifunctional kinases and phosphatases are specialized to control synaptic signaling events. Accordingly, central to LTP and LTD are mechanisms by which postsynaptic scaffold proteins target kinases, phosphatases, and other signaling proteins to multiprotein complexes positioned near both upstream activators and downstream targets at the synapse. This scaffolding promotes precise, spatiotemporally restricted control of substrate phosphorylation in response to highly localized postsynaptic signals.¹¹ This review will focus on the organization of postsynaptic cAMP and Ca²⁺ second messenger–activated kinase/phosphatase signaling networks by A-kinase anchoring protein (AKAP) family scaffold proteins in neurons,^12,13 and how these local signaling networks are engaged by pathogenic amyloid beta (Aβ) oligomers to promote synaptic dysfunction in AD.

2. Aβ-mediated synaptic dysfunction in AD

AD is the most common neurodegenerative disease and cause of dementia in the elderly population worldwide. AD postmortem human brains exhibit amyloid plaques containing insoluble aggregates of 40- to 42-amino acid Aβ peptides that are derived from proteolytic processing of the amyloid precursor protein (APP), neurofibrillary tangles containing hyperphosphorylated forms of the microtubule-associated Tau protein, and atrophy of regions important for learning and memory, including most prominently the hippocampus and cortex.¹⁴ The majority of AD cases are sporadic and increase in prevalence with aging; however, there are a small number of AD cases (<5%) caused by rare, familial mutations (ie, familial early-onset AD [FAD]) in either APP or presenilins 1 and 2, which are components of the γ-secretase APP proteolytic processing machinery. These FAD mutations promote Aβ overproduction leading to early AD onset (<60 years of age). In addition, individuals with Down syndrome (also known as trisomy 21) develop early-onset AD pathology due to Aβ overproduction related to triplication of the APP gene, which is located on human chromosome 21.¹⁴

Two of the earliest manifestations of AD that precede neuronal cell death and strongly correlate with initial cognitive impairment are loss of excitatory synapses and the disruption of synaptic plasticity in the hippocampus and cortex.15, 16, 17, 18, 19, 20, 21, 22, 23, 24 Importantly, both humans with AD and rodent FAD models that overproduce Aβ exhibit impairments in spatial learning and formation of new declarative memories that depend on the hippocampus.25, 26, 27, 28 Considerable evidence indicates that this synaptic dysfunction is triggered by soluble, low-molecular-weight oligomers of Aβ.29, 30, 31, 32 For example, acute exposure to even nanomolar concentrations of synthetic or patient-derived Aβ₄₂ oligomers blocks fundamental forms of synaptic plasticity at hippocampal CA1 synapses, including LTP, within minutes and triggers chronic synaptic depression (ie, LTD) and elimination of excitatory synapses over the course of days to weeks.^{10,18,19,21,22,24,}33, 34, 35 Numerous studies have demonstrated that this Aβ synaptotoxicity is mediated at least in part by disrupting the normal regulation of LTP and LTD by NMDARs, which supply Ca²⁺ and other signals required for plasticity induction.^{18,19,21,23,36} As mentioned above, NMDAR-regulated insertion and removal of AMPARs mediate increases and decreases in excitatory synaptic strength that support the expression of LTP and LTD, respectively.1, 2, 3 Thus, Aβ oligomers may impair learning and memory through altering LTP/LTD balance via disrupted AMPAR trafficking, but the signaling pathways linking Aβ to these plasticity deficits remain incompletely characterized.

3. AMPAR regulation by PKA/CaN signaling during plasticity and Aβ synaptotoxicity

AMPARs are glutamate-gated cation channels composed of tetrameric assemblies of GluA1-4 subunits. Synapses in the adult hippocampus nearly exclusively harbor AMPARs containing the GluA2 subunit, which renders them Ca²⁺-impermeable (ie, CI-AMPARs).37, 38, 39 However, it is now appreciated that transient synaptic incorporation of a small pool of Ca²⁺-permeable AMPARs (CP-AMPARs), lacking the GluA2 subunit (likely GluA1 homotetramers), plays a major role in diverse plasticity mechanisms, including LTP, LTD, and homeostatic plasticity40, 41, 42, 43, 44, 45, 46, 47, 48 (but see the papers by Adesnik and Nicoll,⁴⁹ Gray et al,⁵⁰ Ancona Esselmann et al,⁵¹ and Gainey et al⁵²). Incorporation of CP-AMPARs is thought to induce a “metaplastic” state by providing a new source of Ca²⁺, which can engage signaling pathways to alter subsequent plasticity induction. In many cases, across different brain regions, this metaplastic state primes synapses for subsequent LTD mediated either by NMDAR or metabotropic glutamate receptor (mGluR) activation.^36,45,53, 54, 55 In this regard, CP-AMPARs are mobilized to hippocampal CA1 synapses within minutes following acute Aβ application, where they impair LTP by hijacking signaling pathways favoring LTD.^36,56 In addition, blocking CP-AMPARs prevents Aβ-mediated dendritic spine loss in hippocampal neurons overexpressing human APP containing FAD mutations.⁵⁷ Thus CP-AMPARs are recruited to synapses after both acute and prolonged Aβ exposure and are responsible for at least some key aspects of Aβ synaptic pathology. However, until recently, little consideration had been given to involvement of CP-AMPARs in Aβ synaptotoxicity.

Accumulating evidence indicates that PKA and CaN play key roles regulating CP-AMPARs during normal synaptic plasticity, in part through controlling GluA1-Ser845 phosphorylation.^42,43,45,58, 59, 60, 61, 62, 63 In addition, several prior studies have implicated CaN-LTD signaling as contributing to LTP inhibition and cognitive impairments after exposure to Aβ.^19,21,64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74 In the case of PKA signaling, the conclusions of previous studies are mixed, with some implicating PKA in Aβ-mediated synaptic dysfunction, possibly through CP-AMPAR recruitment, and others finding that strong, exogenous stimulation of PKA signaling may instead override the deleterious impacts of Aβ.^56,75, 76, 77, 78, 79 However, these previous studies of PKA and CaN signaling in Aβ-mediated synaptic dysfunction primarily relied on pharmacological manipulation of cAMP levels or kinase and phosphatase activities that would globally impact all PKA or CaN signaling, including both presynaptic and postsynaptic. Thus, a major question not answered by these prior studies is how Aβ specifically engages PKA and CaN signaling at synapses to impact AMPAR trafficking.

4. AKAP79/150-anchored PKA and CaN regulation of CP-AMPARs in LTP/LTD balance

To this end, studies over the last 20 years have found a specific requirement for local synaptic pools of PKA and CaN, anchored at the synapse via the postsynaptic scaffold protein AKAP79/150, in coordinating CP-AMPAR incorporation and removal at hippocampal synapses (reviewed in the article by Purkey and Dell'Acqua⁵). AKAP79/150 is targeted to the excitatory PSD⁸⁰ in dendritic spines by a polybasic N-terminal membrane-targeting domain that is palmitoylated81, 82, 83, 84 and binds phosphoinositide lipids,⁸⁵ F-actin,⁸⁶ and cadherin transsynaptic cell adhesion molecules.⁸⁷ Once at the postsynaptic membrane, AKAP79/150 nucleates multiprotein G-protein–coupled receptors (GPCR) signaling complexes containing the β₂-adrenergic receptor (β₂AR),^88,89 Gαs,⁸⁹ adenylyl cyclase (AC),90, 91, 92, 93 PKA,^80,89,94,95 CaN,96, 97, 98, 99 and protein kinase C.100, 101, 102 This signaling complex is linked to Ca_V1.2 L-type voltage-gated calcium channels (LTCCs) via direct protein-protein interactions.^98,103 and indirectly to NMDARs and AMPARs by the PSD-95 and synapse-associated protein 97 scaffold proteins.104, 105, 106, 107 Interestingly, AKAP79/150 localization to dendritic spines is increased after LTP induction and decreased after LTD induction in coordination with AMPAR synaptic insertion and removal, respectively.^81,86,108, 109, 110 Several studies have implicated AKAP-PKA/CaN signaling in CP-AMPAR trafficking regulation during LTP and LTD by using mice with complete Akap5 gene knockout, and more importantly, Akap5 knock-in mutations that selectively disrupt synaptic anchoring (but not overall cellular levels) of PKA (150ΔPKA, 150D36; deletions in the PKA-RII binding amphipathic α-helical motif) or of CaN (150ΔPIX; deletion of the CaNA docking PxIxIT motif). In particular, these studies found that AKAP79/150-anchored PKA and CaN signaling oppose each other at the synapse to regulate GluA1 phosphorylation and dephosphorylation and CP-AMPAR insertion and removal, respectively, to control LTP, LTD, and the balance between them.^43,45,46,62,111, 112, 113 However, although genetic disruption of AKAP79/150-PKA anchoring has pronounced impacts on both LTP and LTD, basal GluA1-S845 phosphorylation is surprisingly only partially reduced, thus indicating potential roles for other postsynaptic AKAPs, such as AKAP12/Gravin,¹¹⁴ and/or cGMP-dependent protein kinase signaling¹¹⁵ in maintaining GluA1 phosphorylation.

5. Recent studies on GPCR-AKAP signaling complexes in Aβ synaptoxicity

Several recent discoveries indicate that Aβ actively engages the local, postsynaptic PKA/CaN signaling pathways that are organized by AKAP79/150 to impair LTP and favor LTD. As discussed above, AKAP-PKA signaling is required for CP-AMPAR synaptic recruitment that can facilitate either LTP or LTD under normal conditions^36,43,45 depending on how long the newly inserted receptors remain in the synapse after induction, with CP-AMPARs persisting for >10 minutes after LTP induction^36,41 but only briefly visiting the synapse during LTD induction before being removed by AKAP-CaN signaling⁴⁵ (Fig. 1, top). However, in the presence of Aβ, AKAP-PKA signaling recruits CP-AMPARs even in the absence of any LTP or LTD plasticity-inducing stimulus, and these receptors remain in the synapse for at least 30 minutes.⁴⁵ This persistent Aβ-mediated CP-AMPAR recruitment prevents additional AMPAR recruitment and instead primes synapses for AKAP-anchored CaN-LTD signaling, aided by CP-AMPAR Ca²⁺ influx, that removes AMPARs to prevent LTP induction³⁶ (Fig. 1, bottom). Of note, disruption of either AKAP-PKA anchoring in 150ΔPKA mice or AKAP-CaN anchoring in 150ΔPIX mice significantly prevents acute Aβ-mediated LTP impairment.³⁶ Thus, although CP-AMPARs can positively contribute to LTP, they also paradoxically render synapses vulnerable to Aβ-mediated synaptotoxicity that hijacks LTD mechanisms to inhibit LTP.

Fig. 1.

Model for Aβ-mediated disruption of LTP/LTD balance through CP-AMPAR trafficking regulated by β₂ARs, Ca_V1.2 LTCCs, AKAP150, PKA, and CaN. AKAP-PKA signaling can promote synaptic recruitment of both CI- and CP-AMPARs downstream of NMDAR and LTCC Ca²⁺ influx to mediate either LTP or LTD (top). However, prior activation of β₂AR signaling by Aβ promotes AKAP-PKA–mediated CP-AMPAR recruitment to prevent subsequent LTP induction and to prime synapses for LTD mediated by AKAP-CaN signaling that removes both CI and CP-AMPARs (bottom). Aβ, amyloid beta; AKAP, A-kinase anchoring protein; β₂AR, β₂-adrenergic receptor; cAMP, cyclic adenosine monophosphate; CaN, calcineurin/protein phosphatase 2B; CI, Ca²⁺ impermeable; CP-AMPAR, Ca²⁺-permeable-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; LTCC, L-type voltage-gated calcium channel; LTD, long-term depression; LTP, long-term potentiation; NMDAR, N-methyl-d-aspartate receptor; PKA, cAMP-dependent protein kinase.

Recent studies indicate that AMPARs containing GluA3 subunits, which can form heterotetrameric CI-AMPARs with GluA2 and heterotetrameric CP-AMPARs with GluA1, are regulated by PKA during synaptic plasticity and are required for Aβ to inhibit LTP and promote dendritic spine loss.116, 117, 118 However, in the case of GluA3 regulation, the cAMP effector Epac2 appears to play a central role in addition to PKA.^119,120 AKAP79/150 may also recruit p85 ribosomal S6 kinase (p85S6K) to synaptic GluA1 to maintain Ser845 phosphorylation, and this synaptic localization of p85S6K is reduced in AD mouse models and human postmortem AD tissue.¹²¹ Thus, disruption of postsynaptic AKAP and cAMP signaling by Aβ in AD is likely more widespread than impacts on just GluA1 and PKA.

Although it has been appreciated for over 20 years that Aβ associates with excitatory synapses122, 123, 124, 125 and can directly interact with several different cell surface receptors,^10,79,126, 127, 128, 129, 130, 131, 132, 133, 134 the links between Aβ receptors and signaling pathways leading to impaired LTP and synapse loss remain tenuous. Thus, a long-standing and still unresolved issue in the field is which Aβ receptor(s) mediates its synaptotoxicity? Two candidate GPCR types previously implicated in Aβ-mediated neuronal dysfunction that also couple to AC-cAMP-PKA and Ca²⁺-CaN signaling in neurons are β₂ARs^78,79 (via Gαs stimulation of AC and LTCC Ca²⁺ influx) and group I mGluRs (ie, mGlu1/5)^126,128, 129, 130^,133, 134, 135, 136, 137, 138, 139, 140, 141 (via Gαq-phospholipase C [PLC]-Ca²⁺ release from the endoplasmic reticulum [ER] and Ca²⁺-CaM stimulation of AC). In the case of mGlu1/5, cellular prion protein is thought to play a role as an Aβ coreceptor.^{128,129,133,134,140} Importantly, AKAP150 scaffolds both Gαs-stimulated AC5^90,91,93 and Ca²⁺-CaM–stimulated AC8^84,91,92 isoforms and is known to control signaling between plasma membrane channels and ER Ca²⁺ stores at ER-PM junctions,^92,142, 143, 144, 145, 146 where mGlu1/5 also congregate via linkage to Homer scaffolds.^147,148 In addition, AKAP150 forms signaling complexes with β₂ARs, AMPARs, and Ca_V1.2 LTCCs that coordinate PKA phosphorylation of Ca_V1.2-Ser1928 to increase channel opening and GluA1-Ser845 to increase CP-AMPAR synaptic insertion.^45,89,142,149, 150, 151, 152, 153 Indeed, Aβ activation of β₂AR signaling has been recently implicated in synaptic CP-AMPAR recruitment and LTP impairment through enhancement of Ca_V1.2 LTCC activity by PKA phosphorylation of Ser1928¹⁵⁴ (Fig. 1, bottom). However, it remains to be elucidated how Aβ-mediated enhancement of Ca_V1.2 Ca²⁺ influx is driving CP-AMPAR synaptic insertion. One possibility is that the increased LTCC Ca²⁺ influx is driving additional stimulation of cAMP production by AC1/8-scaffolded AKAP79/150 to further drive PKA phosphorylation of GluA1-Ser845, which promotes CP-AMPAR recruitment. Alternatively, enhanced Ca_V1.2 Ca²⁺ influx could be activating CaMKII to drive CP-AMPAR synaptic insertion.⁴¹ Consistent with this possibility, CaMKII activation has also been implicated in Aβ-mediated impairment of LTP.^155,156

In the future, it will be important to determine whether Aβ is signaling only through activation of β₂ARs located on the plasma membrane cell surface or also somehow engaging β₂ARs located on intracellular organelles, such as endosomes, which have recently been shown to regulate PKA phosphorylation of GluA1-Ser845 and CP-AMPAR trafficking during LTP.¹⁵⁷ Accordingly, AKAP79/150 is localized to recycling endosomes via palmitoylation of 2 conserved Cys residues in its N-terminal targeting domain, and palmitoylation-deficient AKAP79/150CS mutants interfere with synaptic recruitment of CP-AMPARs during LTP.81, 82, 83 Thus, it will be important to determine whether AKAP79/150 targeting of PKA and CaN to endosomes is also involved in Aβ signaling via β₂ARs to disrupt LTP.

Although recent findings indicate that the β₂AR appears to be at least one of the receptors mediating Aβ′s impacts on synapses, studies discussed above also implicate mGlu1/5 as Aβ receptors/coreceptors. In addition, a recent study using super-resolution fluorescence imaging revealed that Aβ preferentially binds to a subset of dendritic spines where it localizes immediately adjacent to the PSD (Fig. 2).¹⁵⁸ Intriguingly, this perisynaptic Aβ distribution closely resembles the localization of mGlu1/5 receptors159, 160, 161 and AKAP150 relative to the PSD^83,162 (Fig. 2). Consistent with Aβ binding to spines activating highly localized receptor signaling pathways, it was also demonstrated that Aβ inhibits LTP-associated dendritic spine enlargement induced by optical uncaging of glutamate (ie, structural LTP) only at spines bound to Aβ but not at adjacent spines lacking Aβ.¹⁵⁸ Moreover, recent findings using AKAP150ΔPIX mice indicate that Aβ locally engages AKAP-CaN signaling downstream of mGlu1 activation to decrease NMDAR Ca²⁺ influx imaged in single dendritic spines¹⁶³ (Fig. 3). Such persistent, low-level Ca²⁺ influx through NMDARs resulting from this mGlu1-AKAP-CaN signaling, combined with Ca²⁺ contributions from CP-AMPARs recruited by β₂AR-AKAP-PKA signaling (as discussed above), would together sustain downstream AKAP-CaN signaling that subsequently removes both CP and CI-AMPARs, resulting in synaptic depression. In addition, depression of Ca²⁺ influx through NMDARs may also favor nonionotropic, conformational signaling through p38 mitogen-activated protein kinase that can drive LTD, spine shrinkage, and spine loss, including downstream of Aβ.^57,164, 165, 166

Fig. 2.

Super-resolution imaging reveals that Aβ oligomers bind to perisynaptic regions adjacent to the PSD where AKAP150 is also enriched. (A) Stochastic optical reconstruction microscopy (ie, STORM) imaging reveals that Aβ oligomers (Aβ_o) (HiLyte647 labeled) bind to dendritic spines in nanoclusters located in perisynaptic regions adjacent to the PSD, which is defined by high PSD-95 (mouse anti-PSD-95-CF568 labeled) localization density (gray objects in the middle and right panels). (B) Nanoclusters of AKAP150 (rabbit ant-AKAP150-Alexa647 labeled) localizations are also prominently found adjacent to the PSD defined by PSD-95. Panel (A).¹⁴⁴ Reproduced under the open access Creative Commons Attribution 4.0 International license. Panel (B).¹⁴⁸ Reproduced under the open access Creative Commons Attribution License (CC BY). Aβ, amyloid beta; AKAP, A-kinase anchoring protein; PSD-95, postsynaptic density 95.

Fig. 3.

Model for Aβ-mediated inhibition of NMDAR Ca²⁺ influx and promotion of dendritic spine loss through AKAP150-CaN signaling downstream of mGlu1/5 and CP-AMPARs. Acute Aβ exposure activates mGlu1/5-AKAP-CaN signaling to drive GluN2B-Ser1166 dephosphorylation to promote depression of postsynaptic NMDAR Ca²⁺ influx in coordination with SSHL1 phosphatase dephosphorylation and activation to drive downstream cofilin-Ser3 dephosphorylation that activates its F-actin severing function and destabilize dendritic spines. Aβ recruitment of CP-AMPARs (see Fig. 1) subsequently promotes AKAP-CaN–mediated dephosphorylation of NFAT, which translocates to the nucleus to activate a transcriptional program upregulating Mdm2 expression to drive chronic dendritic spine loss. Aβ, amyloid beta; AKAP, A-kinase anchoring protein; CaN, calcineurin/protein phosphatase 2B; CP-AMPAR, Ca²⁺-permeable-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; ER, endoplasmic reticulum; Mdm2, murine double-minute 2; mGlu1/5, metabotropic glutamate 1/5; NFAT, nuclear factor of activated T cells; NMDAR, N-methyl-d-aspartate receptor; PKA, cAMP-dependent protein kinase; PSD-95, postsynaptic density 95; SSHL1, Slingshot 1L.

Interestingly, NMDAR synaptic Ca²⁺ influx is positively regulated by PKA phosphorylation of Ser1166 on the NMDAR GluN2B subunit,¹⁶⁷ and previous studies established that GluN2B subunit-selective antagonism mimics and occludes the negative impacts of Aβ on NMDAR Ca²⁺ influx.¹⁶⁸ Accordingly, recent studies using neurons from wild-type versus AKAP150ΔPIX mice found that Aβ reduces phosphorylation of GluN2B-Ser1166 in an AKAP-CaN-dependent manner that is also inhibited by antagonism of mGlu1, but not mGlu5, receptors¹⁶³ (Fig. 3). However, despite both proteins being enriched around the periphery of the PSD (Fig. 2), it is not known whether AKAP79/150 forms a direct complex with mGlu1 (or mGlu5). Also, it is not clear whether AKAP150-anchored PKA is solely responsible for controlling GluN2B-S1166 phosphorylation or whether another NMDAR-linked AKAP, such as Yotiao/AKAP9 that coanchors protein phosphatase 1,¹⁶⁹ is also involved. Both questions warrant future investigation.

It is well established that chronic Aβ-mediated activation of CaN signaling eventually leads to even more complete depression of both AMPAR and NMDAR transmission by trafficking these receptors away from synapses and ultimately results in loss of dendritic spine synapses.^{10,19,21,33,64,70,74,131} CaN-dependent dendritic spine loss with chronic Aβ exposure has been linked not only to LTD-like dephosphorylation of GluA1-Ser845 to promote AMPAR endocytosis^64,70,74,170, 171, 172 but also to activation of nuclear factor of activated T-cell (NFAT)–dependent transcription,^70,170,171 a process that is also coordinated by AKAP-PKA/CaN signaling.^{45,62,98,99,153,173,174} Accordingly, recent discoveries using neurons from AKAP150ΔPIX mice found that Aβ-mediated spine loss involves AKAP-CaN signaling, not only to the spine cytoskeleton via the F-actin severing protein cofilin (Fig. 3)¹⁶³ but also to the neuronal nucleus via NFAT transcription factors⁵⁷ (Fig. 3). Notably, double RNAi knockdown of NFATc3 and NFATc4, the 2 major NFAT isoforms expressed in hippocampal neurons, prevented, whereas NFATc3 overexpression mimicked and occluded, FAD mutant APP/Aβ-mediated dendritic spine loss.⁵⁷ In addition, Aβ exposure increased NFATc3 translocation to the nucleus, and CP-AMPAR antagonism reversed this increase and also prevented downstream dendritic spine loss.⁵⁷ Regarding AKAP-CaN activation of the F-actin severing activity of cofilin, Aβ promoted cofilin-S3 dephosphorylation, which is mediated by the phosphatase Slingshot 1L downstream of its dephosphorylation by CaN, and cofilin-S3 dephosphorylation and downstream dendritic spine loss were both prevented by either disruption of AKAP-CaN anchoring in AKAP150ΔPIX mice or antagonism of mGlu1, and to a lesser extent mGlu5.¹⁶³ Past studies also found that APP/Aβ-driven synapse loss could be prevented by either CaN inhibition or mGlu1/5 antagonism.¹⁹

In the case of Aβ activation of NFAT-regulated gene programs, the E3-ubiquitin ligase murine double-minute 2 (Mdm2), which regulates levels of PSD-95 at the synapse and of p53 transcription factors in the nucleus, was identified as an NFAT-induced target gene whose enzymatic activity was required for Aβ-mediated spine loss in hippocampal neurons expressing FAD mutant APP.⁵⁷ Interestingly, mAKAP/AKAP6 has also been implicated in PKA and CaN regulation of transcription factors, including NFAT, myocyte enhancer factor 2, and cAMP response element binding protein, in cardiac myocytes to control genes regulating cardiac hypertrophy and in neurons to control cell survival.175, 176, 177, 178, 179 Thus, it may be worth investigating whether AKAP6-CaN signaling is also involved in regulating neuronal gene expression in Aβ-mediated synapse loss and neurodegeneration.

6. Conclusions and potential relevance for AD therapies

Most approved AD therapies have historically focused on mitigating cognitive impairments by using cholinesterase inhibitors to help compensate for cholinergic neurodegeneration and memantine to block extrasynaptic NMDAR hyperactivity that may promote glutamatergic neurodegeneration. However, these therapies provide modest, temporary benefits as they do not fundamentally alter AD progression. Although more recent antiamyloid antibody therapies that are targeted at clearing Aβ from the brain show modest improvements in cognitive function and in slowing decline, they are associated with serious side effects, including intracranial hemorrhages in some partients.^180,181 Thus, investigation of signaling mechanisms downstream of Aβ that cause synaptic dysfunction, as reviewed here, is also likely to be important to inform the development of novel disease-modifying AD therapies.

In prior studies, inhibition of CaN phosphatase activity with the immunosuppressant drugs cyclosporin A or tacrolimus (FK506) prevented impacts of Aβ on LTP and spines in vitro in neuronal cultures and ex vivo in rodent brain slices.^{19,21,66,72,74,172} Retrospective studies of patients taking tacrolimus, which readily crosses the blood-brain barrier (BBB), and cyclosporin A, which does not cross the BBB as readily, for immunosuppression revealed reduced incidence of AD only in patients taking tacrolimus.^69,182 In addition, tacrolimus improves learning and memory in AD mouse models.^65,73,183 However, there are side effects of tacrolimus, in addition to immunosuppression, that would limit use to treat AD, including renal toxicity, cardiovascular complications, and impaired blood glucose regulation.^183,184

Because of the essential roles that PKA plays in all cells, PKA inhibitors have no clinical utility and systemic stimulation of cAMP-PKA signaling with β₂AR agonists or PDE inhibitors, such as rolipram, to favor strong LTP signaling that may override the LTD-promoting actions of Aβ-mediated by CaN75, 76, 77^,185 also causes undesirable systemic effects including increased heart rate and blood pressure, nausea, and vomiting.^186,187 Furthermore, consistent with β₂ARs and PKA signaling instead mediating some of Aβ′s deleterious impacts, as discussed above, retrospective studies of patients taking BBB-permeable βAR blockers, such as propranolol, to treat cardiovascular issues have actually found delayed progression and/or reduced AD incidence. However, these β-blockers are not β₂AR specific and could also be acting through enhancing cerebrospinal fluid flow to promote Aβ clearance.188, 189, 190 Likewise, CP-AMPAR antagonists also have drawbacks not only due to the normal functions these receptors play during plasticity in principle neurons but also because they are basally expressed at some excitatory synapses in GABAergic interneurons. In addition, at higher doses, CP-AMPAR antagonists block NMDARs and kainate receptors.⁴

Increased LTCC expression, phosphorylation, and activity have been linked to altered synaptic plasticity and cognitive impairments associated with aging, and these mechanisms may be further upregulated in AD.191, 192, 193, 194 Similarly, BBB-permeable dihydropyridine LTCC blockers, such as nimodipine and isradipine, can improve cognition in AD mouse models and in humans with AD in some studies.^191,192,195 However, dihydropyridines are commonly used to treat hypertension due to the prominent role of Ca_V1.2 in regulating smooth muscle contraction to mediate vasoconstriction; thus, it can be difficult to deconvolve the impacts of dihydropyridines on cognition due to antagonizing LTCCs in cerebral vasculature versus in neurons. Yet, AD does have a prominent vascular component, where both basal blood flow and neurovascular coupling to increase blood flow (ie, functional hyperemia) are impaired early in disease progression,196, 197, 198, 199, 200, 201, 202, 203, 204 and it has been hypothesized^205,206 that these perturbations may cause hypoxia and energy deprivation that compromises neuronal health and cognitive function. Interestingly, functional hyperemia is impaired in the 5xFAD mouse AD model, and capillary endothelial cells from these mice exhibit reduced function of Kir2.1 potassium channels that are required for neurovascular coupling via sensing increases in extracellular K⁺ resulting from neuronal activity and then relaying a hyperpolarizing signal along the capillary to smooth muscle cells in upstream arterioles to promote vasodilation.²⁰⁷ These impairments in Kir2.1 activity and functional hyperemia in 5xFAD mice may both be related to Aβ-mediated activation of PLC, leading to a depletion of phosphatidylinositol-4,5-bisphosphate, which is required for maintaining Kir2.1 function.208, 209, 210 Thus, it might be worth exploring whether AKAP-GPCR signaling complexes could also be involved in Aβ-PLC coupling that contributes to vascular dysfunction that in turn impacts neuronal function and cognition in AD. Interestingly, both AKAP79/150 and group 1 mGluRs are expressed in brain endothelial cells, where they participate in Gq-PLC-Ca²⁺-protein kinase C signaling pathways that promote vasodilation and are disrupted in hypertension.^211,212 However, AKAP79/150 is also expressed in vascular smooth muscle cells, where it instead promotes pathologic PKA-Ca_V1.2 and CaN-NFAT signaling that increases vasoconstriction in the development of diabetic hypertension.213, 214, 215, 216 Thus, AKAP-PKA/CaN plays important, yet complex and context-specific roles in vasculature signaling mechanisms that could be of therapeutic relevance. Likewise, group 1 mGluRs also have complex signaling roles across the brain in neurons, astrocytes, and vasculature, thus making their antagonism likely beneficial in some cases but perhaps deleterious in others, leading to safety and efficacy issues.^217,218

It follows that targeted AD therapies that more specifically interfere with specific, localized AKAP signaling events engaged by Aβ to promote synaptic dysfunction could hold great promise. Of note, AKAP150/AKAP5 expression has been reported to be increased in postmortem AD brains,¹²⁰ and mAKAP/AKAP6 expression has been reported to be decreased in olfactory neurosphere stem cells derived from patients with AD. Importantly, impairments in olfaction are some of the first symptoms that occur during AD progression.²¹⁹ It follows that a greater understanding of how key signaling events are controlled by AKAP-organized multiprotein complexes and how these pathways are coopted by Aβ to disrupt synaptic plasticity and promote synapse loss and neurodegeneration could provide new insights into the development of novel AD therapies. In this regard, identification of Mdm2 as an AKAP-CaN-NFAT-regulated target gene required for Aβ-mediated dendritic spine loss⁵⁷ suggests that Mdm2 inhibitors previously developed to treat cancer could potentially be repurposed to treat AD.^220,221

Conflict of interest

Mark L. Dell’Acqua reports financial support from the National Institutes of Health. He reports a relationship with National Institutes of Health that includes funding grants. Given his role as the Molecular Pharmacology Editorial Board member, he had no involvement in the peer review of this article and had no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to another journal editor. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.