2.1. Targeting Plasma Membrane Receptors and Cytosolic Ca²⁺

Proper intracellular Ca²⁺ homeostasis is crucial for many neuronal functions. Disruption of this homeostasis might be one of the main mechanisms via which Aβ and tau exert their neurotoxicity. The main plasma membrane channels involved in neuronal Ca²⁺ influx from the extracellular space are voltage-gated Ca²⁺ channels (VGCCs), which allow Ca²⁺ influx following neuronal depolarization, and receptor-operated Ca²⁺ channels (ROCs), which open upon specific binding of the agonist, with NMDARs among the most important examples.

Toxic Aβ increases cytosolic Ca²⁺, which may affect a variety of enzymes (such as proteases or phosphatases), promote cytoskeletal modifications, cause the generation of free radicals, or trigger neuronal apoptosis [29]. It has been proposed that Aβ can overactivate channels and/or form pores in the cytosolic plasma membrane, allowing massive influx of Ca²⁺ from the extracellular space and increasing the overall Ca²⁺ levels in the cytosol, severely limiting normal cellular function ^[30][31][30,31]. Aβ potentiates Ca²⁺ influx through VGCCs, particularly L-type VGCCs. Excessive Ca²⁺ influx through these channels has been observed in cultured neurons following Aβ exposure and was shown to be blocked by the L-type VGCC inhibitor nimodipine [32]. This phenomenon, however, was not observed in brain slices from AD mouse models [33]. In addition, AD patients taking L-type VGCCs inhibitors such as nilvadipine (NILVAD multicenter trial) showed reduced Aβ levels but no improvement in cognitive decline ^[34][35][34,35]. Acute application of tau aggregates has also been observed to increase cytosolic Ca²⁺ and elevate reactive oxygen species (ROS) production via nicotinamide adenine dinucleotide phosphate (NADPH), an effect that can be prevented by nifedipine and verapamil, both L-type VGCC inhibitors [36]. This suggests that tau fibrils could also incorporate into the cell membrane to activate VGCCs and lead to neuronal dysfunction [36].

CALHM1 (Ca²⁺ homeostasis modulator protein 1) is a Ca²⁺ channel highly expressed in neurons in the hippocampus that allows cytosolic Ca²⁺ influx in response to decreases in extracellular Ca²⁺ [37]. Its activation triggers different kinase signaling cascades in neurons. The CALMH1 polymorphism P86L has been proposed as a risk factor for late-onset SAD ^[9][37][9,37], an argument that has been challenged by other groups ^[38][39][38,39]. Nevertheless, increased levels of Aβ have been observed in transfected cells expressing the P86L polymorphism, suggesting a role for CALMH1 in AD [37]. Additionally, the P86L polymorphism alters the channel permeability to Ca²⁺ [37]. A recent study demonstrated that CALHM1 deficiency in mice leads to cognitive and neuronal deficits, which manifest memory impairment and hippocampal long-term potentiation (LTP) ^[40] [40], pointing to CALHM1 as a potential treatment target in AD.

NMDARs are a subfamily of ionotropic glutamate receptors involved in the excitatory synaptic transmission and synaptic plasticity of the brain. Specific types of NMDARs are much more permeable to Ca²⁺ than other ionotropic glutamate receptors and are often implicated in neuronal pathophysiology. NMDARs are mainly composed of GluN2A and GluN2B in the brain areas most affected in AD [41]. Extra-synaptic GluN2B-containing NMDARs have been associated with excitotoxicity (the excessive neuronal death induced by cellular Ca²⁺ overload due to excessive stimulation of glutamate receptors) and the toxic effect of Aβ oligomers in AD ^[42][43][42,43]. For this reason, selective GluN2B subunit antagonists may be a strategy to prevent synaptic dysfunction in AD. Aβ₄₂ peptides interact with NMDARs, potentiating their activity and leading to increased Ca²⁺ influx, thus contributing to the synapse loss observed in AD ^[44][45][44,45]. Additionally, as demonstrated in mouse models of AD, glutamate-induced excitotoxicity is inhibited by tau reduction ^[46] [46] and exacerbated by tau overexpression ^[47][48][47,48]. In turn, glutamate-induced excitotoxicity increases tau expression ^[49] [49] and phosphorylation [50], while activation of extra-synaptic NMDAR leads to tau overexpression, neuronal degeneration, and cell loss [51]. Memantine—a weak NMDAR antagonist—is one of the two FDA-approved drugs to treat AD patients and the only NMDAR antagonist [23]. It provides modest improvements to memory and cognitive performance in moderate to severe AD patients ^[52][53][52,53]. Memantine restricts excessive Ca²⁺ influx, thus reducing neuronal excitotoxicity, and, due to its low activity, the basal NMDAR function is preserved. Memantine has also shown neuroprotective effects against oxidative stress, neuroinflammation, and tau phosphorylation ^[54][55][54,55].

Ionotropic neuronal nicotinic acetylcholine receptors (nAChRs) respond to the neurotransmitter acetylcholine (ACh) and to drugs such as the agonist nicotine. They are permeable to Na⁺, K⁺, and Ca²⁺. The nAChRs expressing the α7 subunit have the highest conductance for Ca²⁺ and are found in brain regions most susceptible to AD [56]. In the basal forebrain, cholinergic neuronal loss and decreased levels of ACh mediate cholinergic impairment, which eventually leads to short-term memory loss ^[57][58][59][57,58,59]. The loss of cholinergic innervation in early AD led to the “cholinergic hypothesis of AD” [3]. Galantamine and rivastigmine (for use in mild to moderate AD) and donepezil (in mild to severe AD) are the cholinesterase inhibitors FDA-approved to treat AD [23]. These drugs act by increasing ACh levels, which delay the progression of AD through Ca²⁺-dependent mechanisms. Furthermore, supplemented with memantine, it has been proposed that this combination could provide greater benefits on behavior, cognition, and global outcomes in AD [60].

Exposure of hippocampal and cortical neurons to tau also increased intracellular Ca²⁺ levels through muscarinic receptors [61]. Interestingly, Ca²⁺ activates many kinases, including those responsible for tau phosphorylation—such as glycogen synthase kinase 3β (GSK3β)—and, therefore, Ca²⁺ dyshomeostasis may increase tau phosphorylation and NFT formation ^[62] [62]. Given that tau pathology correlates better with cognitive impairments than Aβ deposition, tau targeting is expected to be more effective once clinical symptoms emerge [63]. It has long been known that tau-expressing cells secrete normal and pathological tau [64], which can be taken up by other cells, seeding and spreading tau pathology ^{[4][65][66][67]}[4,65,66,67]. Led by immunotherapy approaches, the efforts to target tau with therapeutics focus on reducing tau pathology by limiting the spread of extracellular tau across brain regions [68]. Anti-tau immunotherapy has shown potential in numerous clinical studies. Both active and passive tau immunization seem to offer a promising option by reducing tau pathology [69]. Active tau immunization, however, seems to elicit a risk of adverse immune reactions from targeting the normal protein. Other tested approaches involve reducing tau expression (with small interfering RNA or antisense oligonucleotides; siRNA and ASOs, respectively), targeting tau modifications, reducing tau aggregation, and stabilizing microtubules. Preventing or reducing pathologic tau has been shown to improve cognitive and motor impairments in animal models with neurofibrillary pathology, and several tau antibodies and vaccines have been tested in preclinical studies in the last years. Immunotherapy is currently at the stage of drug development (recently reviewed in ^[68][69][70][68,69,70]), and, as of today, eight humanized tau antibodies and two tau vaccines are under clinical trial for AD or frontotemporal dementia ^[71] [71] (www.alzforum.org).

The use of intravital imaging and transgenic mouse models of AD have allowed for direct observation of cytosolic Ca²⁺ dysregulation. In vivo, neuronal cytosolic Ca²⁺ dyshomeostasis is more likely to be observed in the vicinity of amyloid β plaques, but is detectable in neurons throughout the cortex [15]. Higher Ca²⁺ levels were observed in neurons close to amyloid plaques in a commonly used mouse model of cerebral amyloidosis (APP_SwexPS1ΔE9, APP/PS1) [15], but only after plaque deposition and not before. Cytosolic Ca²⁺ overload was absent in mice harboring only the PS1 mutation (typically lacking plaque deposition). The mechanisms of Ca²⁺ dysregulation involved activation of calcineurin (CaN), a Ca²⁺/calmodulin-dependent protein phosphatase sensitive to subtle rises in intracellular Ca²⁺ levels, and whose activation induces long-term depression (LTD). Ca²⁺ dysregulation in neurites was linked to neurodegeneration (neuritic blebbing and beading), which can be partially prevented by inhibiting CaN [15]. Elevated Ca²⁺ levels in the neurites impair synapses, by increasing the frequency of spontaneous synaptic potentials and reducing plasticity. In addition, pathological increases in neuronal network activity—observed as increased frequencies of somatic Ca²⁺ transients—potentiate Aβ release into the extracellular space ^[72][73][72,73]. In the APP23/PS45 mouse model of AD (overexpressing mutant APP_Swe and mutant PS1_G348A), neuronal hyperactivity was observed around amyloid plaques in the cortex, only after plaque deposition [14]. Hyperactive neurons, however, were found in the CA1 region of the hippocampus in pre-depositing animals [13]. Direct application of soluble Aβ onto the wild-type (Wt) naïve brain increased cytosolic Ca²⁺ levels ^[12] [12] and induced neuronal hyperactivity [13]. Acute treatment with the γ-secretase inhibitor LY-411575, which reduces soluble Aβ levels, normalized the frequency of Ca²⁺ transients prior to plaque deposition [13].

Interestingly, AD patients are more prone to developing epileptic seizures [74]. Blocking network hyperactivity with the antiepileptic drug levetiracetam improves learning and memory, reverses behavioral abnormalities, and reverts synaptic deficits in the hippocampus in an AD mouse model [75]. In the same way, it has been observed that tau is implicated in neuronal circuit deficits in mouse models of AD expressing both Aβ and tau. Tau effects dominate those of Aβ and are mostly dependent on the presence of soluble tau [16]. According to the authors, this dramatic effect could suggest a possible cellular explanation contributing to disappointing results of anti-Aβ therapeutic trials. This abnormal network activity and its resultant AD-related cognitive deficits in mice point to neuronal hyperactivity as a promising therapeutic target in AD.

Aducanumab is a high-affinity, fully human immunoglobulin G1 (IgG1) monoclonal antibody that selectively binds to aggregated Aβ fibrils and soluble oligomers (and not monomers) in the brain parenchyma [25]. It was shown that it could ameliorate Ca²⁺ dysregulation in AD. Using multiphoton microscopy and a Ca²⁺ reporter, it was observed that a single topical application of the antibody onto the brain surface of mice depositing amyloid plaques (Tg2576 AD model) led to a reduction in existing amyloid deposits [76]. Peripheral administration of the antibody over a period of 6 months rescued Ca²⁺ overload in transgenic neurites, restoring them to control levels within 2 weeks. The authors suggested that aducanumab exerted its function by targeting amyloid deposits, including soluble oligomeric Aβ [76]. In March 2019, the termination of all aducanumab clinical trials was announced after an interim analysis of EMERGE and ENGAGE trials predicted the phase III placebo-controlled studies would not meet their primary end points. However, in a subsequent analysis of a larger dataset from the EMERGE trial, aducanumab met the primary end point, and the FDA accepted the aducanumab application for review [77]. If the case is approved, aducanumab would be the first drug to combat the root causes of AD.

2.2. Targeting ER Ca²⁺ and SOCE

The ER is an important subcellular organelle involved in protein synthesis, modification, and folding. Additionally, it is a dominant Ca²⁺ reservoir in the cell, critical for maintaining intracellular Ca²⁺ levels [27]. Ca²⁺ is released from the ER after activation of either inositol 1,4,5-trisphosphate receptors (IP₃Rs) or ryanodine receptors (RyRs). Ca²⁺ efflux from the ER modulates a range of neuronal processes, including regulation of axodendritic growth and morphology or synaptic vesicle release [78]. The sarco-endoplasmic reticulum ATPase (SERCA) pump, which actively consumes ATP, is important for extruding Ca²⁺ into the ER lumen, where it is sequestered by binding to proteins such as calsequestrin and calretinin, priming this organelle as a critical component of Ca²⁺ buffering.

Impaired IP₃R signaling in the ER was an early discovery in AD. It was shown that human cells from FAD patients exhibited enhanced Ca²⁺ release in response to IP₃R-generating stimuli [79]. Fibroblasts from asymptomatic members of AD families [80], as well as PS1 knock-in mice and other presymptomatic AD mouse models, showed the same enhancement [81]. These observations suggested that FAD mutations contribute to Ca²⁺ dysregulation, even before pathology deposition or cognitive impairments were evident. A reduction in IP₃R expression can normalize Ca²⁺ homeostasis and restore hippocampal LTP in mouse models of AD [82]. PSs are transmembrane proteins found in the ER membranes and form the catalytic core of the γ-secretase complex that processes APP and other type 1 transmembrane proteins, such as Notch ^[83][84][83,84]. PSs are essential for learning and memory, as well as neuronal survival during aging in the murine cerebral cortex ^[85][86][85,86]. Mutations in PSs have been shown to affect APP processing, leading to increased production of the more hydrophobic neurotoxic form Aβ₄₂ ^[87][88][89] [87,88,89] or increasing the Aβ42/40 production ratio [90]. It has also been proposed that PSEN mutations cause a loss of presenilin function in the brain, triggering neurodegeneration and dementia in FAD [91]. Mutations in PS1 and PS2 might stimulate IP₃Rs, leading to exaggerated Ca²⁺ release through these channels ^[79][80][81][79,80,81]. An alternative hypothesis suggested that PSs function as passive low-conductance leak channels in the ER membrane. AD-associated PS mutations might impair this leak function, resulting in ER Ca²⁺ overload [92] and leading to exaggerated increases in cytosolic Ca²⁺ upon stimulation of Ca²⁺ release. However, these observations have not been supported by other groups ^[93][94][93,94], and, despite extensive research, this subject is still a matter of controversy. Recently, it was proposed that the ER-based transmembrane and coiled-coil domain TMCO1 could be responsible for the ER Ca²⁺ leak [95].

RyR Ca²⁺ dysregulation was also observed before the histopathology and cognitive decline in AD. Both human brain tissue from AD patients and AD mouse models have shown increased expression of RyR (particularly RyR₂) in affected brain regions in AD [96]. Exaggerated Ca²⁺ release from RyR has been related to impaired neurophysiology and synaptic signaling events, contributing to memory impairment in AD ^[97][98][33,97,98]. The FAD PS mutations also exaggerate Ca²⁺ release through the RyR, as a result of either increased expression of RyRs or sensitization of the channel activity ^[99][100][99,100]. Furthermore, RyR-mediated Ca²⁺ release upregulates secretases, increasing APP cleavage, Aβ fragments, and plaque deposition, and its blockage leads to Aβ reduction and improved memory impairment [101]. RyRs can also themselves be activated by Ca²⁺, which amplify IP₃R activity via Ca²⁺-induced Ca²⁺ release mechanisms [102]. Additionally, Aβ aggregates themselves trigger ER Ca²⁺ release through IP₃Rs and RyRs ^[103][104][103,104]. Recently, stabilization of RyR₂ macromolecular complex by S107 (Rycal)—a benzothiazepine that prevents the dissociation calstabin2 from the RyR₂ complex—showed therapeutic potential in vitro and in mouse models of AD in vivo. Application or administration of S107 reversed ER Ca²⁺ leak, reduced APP cleavage and Aβ production, and restored synaptic plasticity and cognitive deficits ^[105][106][105,106].

It has been found that, in cells lacking PS1, PS2, or PS1/2, or cells expressing either PS2 or FAD-PS2, SERCA activity is diminished, resulting in increased cytosolic Ca²⁺ [107]. Conversely, other studies have shown that mutations in PS influence SERCA by accelerating Ca²⁺ sequestration via ATPase [108], leading to an overfilled ER. In any case, these data suggest that normal PSs are required for normal SERCA functioning and suggest that PSs are a candidate target for development of therapeutics, independent of their role in APP processing.

Increased Ca²⁺ release from intracellular ER Ca²⁺ stores might exacerbate disease-mediated pathology. Accordingly, dantrolene, a negative allosteric modulator of RyR—and its central nervous system (CNS)-penetrant version Ryanodex—has been shown to reduce amyloid pathology, normalize ER Ca²⁺ homeostasis, restore synaptic structure and density, normalize synaptic plasticity, and improve behavioral performance in mouse models of AD ^{[101][109][110]}[101,109,110]. This builds on RyR as a therapeutic target for AD, and further emphasizes the role of dysregulated ER Ca²⁺ as a key component in the AD pathogenesis.

ER Ca²⁺ depletion triggers a sustained extracellular Ca²⁺ influx to the cytosol through the store-operated Ca²⁺ entry (SOCE) pathway by activating STIM (stromal-interacting molecule) protein—which senses low Ca²⁺ concentration upon depletion of the ER stores—and plasma membrane channels Orai and TRPC (transient receptor potential canonical) [111]. Two forms of STIM are expressed in the brain (STIM1, predominantly in the cerebellum, and STIM2 in the hippocampus and cortex) [112]. SOCE refills the ER, keeping it ready for the next ER Ca²⁺ signal [113]. Disrupted SOCE has been observed in AD. SOCE is decreased by diverse FAD PS mutations ^[114][115][114,115] and in the presence of soluble Aβ [116]. It has also been proposed that SOCE deficits may be due to the decreased expression of STIM1 and/or STIM2 in FAD-linked PS1 mutations [117]. Related to this, overexpression of the dominant negative PS1 variant potentiates SOCE [118]. It has also been proposed that SOCE deficits might result from overfilled ER Ca²⁺ stores [114]. These findings, however, are inconsistent, as other groups have observed no differences or decreased ER Ca²⁺ concentration in mutant PS expressing cells ^{[11][93][107]}[11,93,107]. Recently, it has been proposed that neuronal SOCE is required for maintaining the morphology of mushroom spines, modulating Aβ production and promoting memory functions ^[117][119][117,119]. Ca²⁺ entry via SOCE activates Ca²⁺/CaM-dependent kinase II (CamKII), which is upstream of gene transcription for maintenance of mature spines. Attenuated SOCE-mediated Ca²⁺ influx might reduce CaMKII activity while inducing destabilization of mushroom spines. This can reduce LTP-mediated memory formation [117]. Attenuated SOCE may also lead to inadequate ER refill, which might induce neuronal cell death via apoptosis ^[120][121][120,121]. STIM2 overexpression in AD models restores spine morphology, implicating SOCE in AD ^[122] [122] and suggesting that targeting SOCE in AD may avoid or restore dendritic spine loss. Additionally, it was recently found that expression of TRPC1, a subfamily of TRPCs, is decreased in AD cells and mouse models. While deletion of TRPC1 did not impair cognitive function or lead to cell death in physiological conditions, it did exacerbate memory deficits and increase neuronal apoptosis induced by Aβ. On the contrary, overexpression of TRPC1 inhibited Aβ production and decreased apoptosis [123]. Together, these studies suggest another mechanistic target for therapeutic development within the Ca²⁺ hypothesis of AD.

2.3. Targeting Mitochondrial Ca²⁺

Mitochondria are crucial organelles that provide energy to the cell in the form of adenosine triphosphate (ATP) via the process of oxidative phosphorylation. Mitochondria form a dynamic tubular network that extends throughout the cytosol, undergoing fusion and fission, which regulates the morphology and structure of the mitochondrial network [124]. Neurons rely strictly on mitochondria to produce ATP, with mitochondria being recruited in areas like synapses, where high energy is required. Mitochondria also buffer Ca²⁺ and shape its signal [125], which is involved in neurotransmission and maintenance of the membrane potential along the axon. At the synaptic level, mitochondria regulate the Ca²⁺ levels necessary for synaptic functions [126]. Mitochondrial Ca²⁺ uptake activates some dehydrogenases at the electron transport chain (ETC), activating mitochondrial respiration and ATP production [127]. The electrochemical gradient created by the ETC allows mitochondria to take up Ca²⁺. This mitochondrial Ca²⁺ participates in signal transduction and the production of energy. Mitochondria contain two major membranes, the outer mitochondrial membrane (OMM), which contains voltage-dependent anion-selective channel protein (VDAC), permeable to most molecules, and the inner mitochondrial membrane (IMM), which is impermeable to molecules and ions, unless they contain specific channels or transporters.

Ca²⁺ is taken up into the mitochondrial matrix through the mitochondrial Ca²⁺ uniporter (MCU) complex, a highly Ca²⁺-sensitive ion conductance channel ^[128][129][128,129]. The MCU is a macromolecular complex of proteins, which includes the pore and several regulatory subunits. It is ubiquitously expressed among organisms and defines the pore domain of the complex [128]. Two other proteins participate in the Ca²⁺ permeant pore: MCUb, whose expression is restricted to most vertebrates ^[130][131][130,131], and the essential MCU regulator (EMRE) [132]. The response of the MCU to extramitochondrial Ca²⁺ is regulated by the mitochondrial Ca²⁺ uptake (MICU) family of proteins, which are in the intermembrane space. MICU1 and MICU2 act as Ca²⁺ sensors, each with two Ca²⁺-binding EF-hand motifs that confer sensitivity to Ca²⁺ [133]. MICU1 and MICU2 also act as gatekeepers of MCU [134], with MICU1 getting involved when the extramitochondrial Ca²⁺ concentration is high, activating the channel open state. At low concentrations, the main player seems to be MICU2, leading to minimal accumulation of Ca²⁺ within mitochondria ^[135][136][135,136], thus preventing mitochondrial Ca²⁺ overload at resting conditions. MICU3, a paralog of MICU1 and MICU2, is mainly expressed in the CNS [137], and has been proposed to enhance mitochondrial Ca²⁺ uptake in neurons [138]. In regulating the MCU pore, the mitochondrial Ca²⁺ uniporter regulator 1 (MCUR1) also plays a role [139]. It has been suggested as a necessary player in MCU-mediated mitochondrial Ca²⁺ uptake. The small Ca²⁺-binding mitochondrial carrier protein (SCaMC, also known as SLC25A23) ^[140] [140] seems to also participate in the mitochondrial Ca²⁺ uptake by interacting with MCU and M1CU1.

Mitochondrial Ca²⁺ efflux occurs via the Na⁺/Ca²⁺ exchanger (NCLX) ^[141] [141] and leucine zipper- and EF hand-containing transmembrane protein 1 (Letm1), located at the IMM [142]. Excessive Ca²⁺ in the mitochondrial matrix induces the activation of the mitochondrial permeability transition pore (mPTP) and allows the release of Ca²⁺ ions and small molecules such as cytochrome c [143]. Mitochondrial Ca²⁺ levels are tightly regulated since excessive levels of Ca²⁺ within mitochondria, i.e., mitochondrial Ca²⁺ overload, result in the impairment of mitochondrial function, suppression of ATP production, increase in reactive oxygen species (ROS) production, and mPTP opening. This can lead to caspase activation and cell death via apoptosis [144].

Mitochondrial function has long been considered one of the intracellular processes compromised at the early stages in AD and likely in other neurodegenerative diseases. Moreover, the “mitochondrial cascade hypothesis” was proposed to explain the onset of SAD [145], which posits that mitochondrial dysfunction is the primary process to trigger the cascade of events that lead to late-onset AD. Even though the validity of this hypothesis has yet to be demonstrated, numerous mitochondrial functions are disrupted in AD [146], including mitochondrial morphology and number [147], oxidative phosphorylation, mitochondrial membrane potential, ROS production [148], mitochondrial DNA (mtDNA) oxidation and mutation [149], mitochondrial–ER contacts [150], and mitochondrial dynamics, including mitochondrial transport along the axon and mitophagy [151]. Additionally, both Aβ and tau have been found in mitochondria. Aβ is imported to mitochondrial matrix via translocase of the outer membrane (TOM) [152], and a fraction of intracellular tau has been found within the inner mitochondrial space [153]. Once in mitochondria, they interact with specific intramitochondrial targets, leading to the dysfunction of the organelle. Furthermore, tau accumulation in mitochondrial synaptosomes has been proposed to correlate with synaptic loss in AD brains [154].

Mitochondrial Ca²⁺ dysregulation is considered a fingerprint of AD. Mitochondrial Ca²⁺ overload can be a result of three different processes: (i) increased mitochondrial Ca²⁺ influx (following Ca²⁺ influx from extracellular space or Ca²⁺ transfer from ER), (ii) decreased mitochondrial Ca²⁺ efflux through NCLX, or (iii) reduced mitochondrial Ca²⁺ buffering. Neurotoxic Aβ can lead to mitochondrial Ca²⁺ overload, as shown in in vitro and in vivo models ^{[155][156][157]} [155,156,157]. Primary neurons in culture exposed to Aβ oligomers triggered mitochondrial Ca²⁺ overload, leading to mPTP opening, release of cytochrome c, and cell death via mitochondrial-mediated apoptosis ^[156] [156]. Additionally, studies in mouse neuroblastoma N2a cells co-transfected with the Swedish mutant APP and Δ9 deleted PS1 showed similar mitochondrial impairment, evidenced by the increased mitochondrial apoptotic pathway and caspase-3 activity [158]. Furthermore, Aβ can interact with cyclophilin D—a regulator of mPTP—and promote the release of cytochrome c through the opening of mPTP [159]. This causes neuronal injury and decline of cognitive functions, as shown in a mouse model of AD. Genetic deletion of CypD in Tg AD mice rescues mitochondrial impairment and improves learning and memory [160], suggesting that CypD could represent a potential therapeutic target in AD.

Recently, we showed mitochondrial Ca²⁺ overload in a mouse model of cerebral amyloidosis (APP/PS1). Using in vivo multiphoton imaging and a ratiometric Ca²⁺ reporter, we demonstrated increased levels of mitochondrial Ca²⁺ following Aβ deposition, which preceded neuronal cell death. Moreover, naturally secreted soluble oligomers applied to the healthy brain of Wt mice also increased mitochondrial Ca²⁺ levels, a process that could be prevented by MCU inhibition with the specific channel blocker Ru360 [157]. We also showed, for the first time, that the expression of mitochondrial Ca²⁺ transport-related genes in brain tissue from AD patients was impaired compared to control cases. In particular, genes involved in mitochondrial Ca²⁺ uptake (MCU complex) were downregulated, whereas the only one encoding for Ca²⁺ efflux (NCLX) was upregulated, suggesting a compensatory response to prevent mitochondrial Ca²⁺ overload [157]. However, others reported that different techniques used for evaluating expression showed conflicting results [161]. Another mechanism proposed for mitochondrial Ca²⁺ overload in AD is impairment of mitochondrial Ca²⁺ efflux. Loss of NCLX expression and functionality has also been suggested in AD, whereas genetic rescue of NCLX expression in neurons restored cognitive decline and cellular impairment in transgenic mouse models of AD [161]. Additionally, the more general cytosolic Ca²⁺ overload observed in vivo (as previously cited) may contribute to the observed mitochondrial Ca²⁺ overload. These observations suggest that restoring mitochondrial Ca²⁺ levels in AD could be a promising new therapeutic target against AD.

It has been previously proposed that nonsteroidal anti-inflammatory drugs (NSAIDs) may help in preventing the cognitive decline associated with aging [162]. Unfortunately, results from several clinical trials have given rather pessimistic results ^{[163][164][165]}[163,164,165], partly due to inadequate CNS drug penetration of existing NSAIDs, suboptimal doses, unknown molecular targets (and, therefore, unknown pharmacodynamics), and toxicities. Nevertheless, in vitro studies have shown that NSAIDs such as salicylate and the enantiomer (R)-Flurbiprofen lacking anti-inflammatory activity, at low concentrations, are able to depolarize mitochondria and inhibit the driving force for mitochondrial Ca²⁺ uptake ^[166][167][166,167]. They act as mild mitochondrial uncouplers without altering cytosolic Ca²⁺ levels. This mild mitochondrial depolarization was able to prevent NMDA- and Aβ-induced mitochondrial Ca²⁺ uptake and cell death ^{[155][156][168]}[155,156,168]. These results point to mitochondrial Ca²⁺ as a key player in Aβ-driven neurotoxicity and suggest a new mechanism of neuroprotection by NSAIDs independent of their anti-inflammatory activity. Another compound, TG-2112x, has been recently suggested as neuroprotective and proposed as a new therapeutic opportunity. Tg-2112x partially inhibits mitochondrial Ca²⁺ uptake without affecting the mitochondrial membrane potential or mitochondrial bioenergetics, protecting neurons against glutamate excitotoxicity [169].

Abnormal tau hyperphosphorylation also influences mitochondrial transport along the neuronal axon, which leads to a reduction in and impairment of mitochondria at the presynaptic terminal with detrimental consequences and eventual cell death ^[170][171][170,171]. In vitro and in vivo studies have shown that tau dysregulates Ca²⁺ homeostasis in mitochondria. Mitochondrial Ca²⁺ buffering and homeostasis are disrupted in cells overexpressing tau and those exposed to extracellular tau aggregates ^[61][172][61,172]. Additionally, basal mitochondrial Ca²⁺ levels have been shown to be elevated in patient-derived human induced pluripotent stem cell (iPSC) neurons expressing a tau mutation, likely due to the inhibition of NCLX by tau [173]. Elevation in mitochondrial Ca²⁺ levels by tau also increased the vulnerability to Ca²⁺-induced cell death [173]. Phosphorylated tau has also been found to interact with VDAC in AD brains, leading to mitochondrial dysfunction [174].

Mitochondria and ER membranes are juxtaposed and establish contact points known as mitochondrial-associated membranes (MAMs). They are dynamic lipid rafts enriched in cholesterol and sphingomyelin, as well as in proteins associated with Ca²⁺ dynamics ^[175][176][175,176]. MAMs allow for communication between ER and mitochondria, including metabolic pathways and Ca²⁺ transfer from ER to mitochondria [177]. Increased contacts between ER and mitochondria have been found in human fibroblast cells derived from FAD patients, human brain tissue, and AD mouse models ^[178][179][178,179]. An increased association between the ER and mitochondria has also been observed in a Tg mouse model of tauopathy [180]. Increased contact promotes mitochondrial bioenergetics, but excessive Ca²⁺ transfer can contribute to mitochondrial Ca²⁺ overload and suppression of normal mitochondrial functions, and Aβ oligomers have been found to induce massive Ca²⁺ transfer from ER to mitochondria ^{[116][181][182][183]}[116,181,182,183].

Mitochondria-targeted protective compounds that prevent or minimize mitochondrial dysfunction could represent potential therapeutic strategies in the prevention or treatment of AD. However, several compounds targeting mitochondrial function have been tested in AD without a favorable outcome [184]. Nevertheless, the idea of AD as a multifactorial disease is widespread, and mitochondria as a therapeutic target combined with other medications is emerging as a valid therapy for AD. The list of pharmacologic approaches that directly target mitochondria includes antioxidants (such as vitamin E and C, coenzyme Q10, mitoQ, and melatonin) and phenylpropanoids (such as resveratrol, quercetin, or curcumin) [185]. Antioxidants are generally used to decrease oxidative stress and slow the progression of symptoms that generally accompany AD. Antioxidants such as coenzyme Q10 and mitoquinone mesylate (MitoQ) are antioxidants that directly target mitochondria [186]. Currently, there is a small clinical trial testing MitoQ on cerebrovascular blood flow in AD [187]. The Szeto-Schiller (SS) tetrapeptides, an alternative type of antioxidants that target mitochondria, are small molecules that can reach the mitochondrial matrix and act as antioxidants [188]. Specifically, SS31 (also known as elamipretide) selectively binds to cardiolipin and promotes electron transport while optimizing mitochondrial ATP synthesis [189]. In addition, SS31 inhibits mitochondria swelling and oxidative cell death. In mouse models of cerebral amyloidosis, it was shown that SS31 reduces Aβ production and mitochondrial dysfunction, and enhances mitochondrial biogenesis and synaptic activity [190]. Recently, SS31 combined with the mitochondrial division inhibitor 1 (Mdivi1) was tested in vitro with a positive outcome, suggesting this combination as a possible type of mitochondria-targeted antioxidant in AD [191]. Ongoing clinical trials regarding mitochondria in AD are reviewed in ^[187][192][187,192] and at www.clinicaltrials.gov.

2.4. Targeting Lysosomal Ca²⁺

Lysosomes are acidic organelles that participate in the endolysosomal system. They are important for autophagy and intracellular Ca²⁺ storage (with comparable Ca²⁺ levels to those of the ER) [193]. The Ca²⁺ transport in and out of the lysosomal lumen provides signals that modulate the fusion of autophagosomes and lysosomes. In order to maintain lysosomal Ca²⁺ homeostasis, lysosomes contain P/Q type VGCCs expressed in the lysosomal membrane that provide Ca²⁺ to the cytosol. Dysregulation in lysosomal Ca²⁺ release via VGCCs leads to defective autophagic fusion and flux [194]. The vacuolar-type H⁺-ATPase (V-ATPase) and Ca²⁺/H⁺ exchanger are in charge of lysosomal Ca²⁺ refilling [195]. It has been suggested that this refilling is largely dependent on ER Ca²⁺ [196]. V-ATPase activity predominantly maintains lysosomal pH; however, other ion channels localized to the lysosomal membrane participate in pH regulation during lysosomal proteolysis, including the chloride channel CLC7 [197] and the Ca²⁺ channel TRPML1 (mucolipin) ^[198][199][198,199]. Additionally, Ca²⁺ microdomains generated at the mouth of these channels have been suggested to take part in the regulation of autophagy [200].

Lysosomal Ca²⁺ efflux has been linked to changes in lysosomal pH. Recent reports suggested that decreased lysosomal Ca²⁺ in AD-linked mutations or PS1 knockout (KO) cells is a consequence of elevated lysosomal pH [201]. Raising lysosomal pH leads to autophagy defects and lysosomal Ca²⁺ efflux. PS1 mutant cells exhibit these defects. PS1 KO cells show deficiencies in lysosomal V-ATPase content and function, defective autophagy, and abnormal Ca²⁺ efflux. ^[201] [201]. Reversal of lysosomal pH abnormalities in PS1 KO cells, but not Ca²⁺ efflux deficits, was sufficient to rescue these same deficits [201]. These data suggest that lysosomal Ca²⁺ defects are secondary to lysosomal pH elevation, and that lysosomal Ca²⁺ dyshomeostasis contributes significantly to the overall Ca²⁺ dysregulation observed in PS1-deficient cells. However, other studies do not support these arguments, citing that, although the autophagosome and lysosome accumulation was apparent in PS1 or PS2 cells, defective lysosome acidification was not found [202]. In addition, defects in lysosome acidification or Ca²⁺ homeostasis have not been observed in FAD-PS2 models [203]. On the contrary, other studies have shown both reduced cytosolic Ca²⁺ signal and lower ER content in FAD-PS2 models. In particular, it was proposed that FAD-PS2 decreases ER and cis–medial Golgi Ca²⁺ levels by reducing SERCA activity, which could lead to defective autophagosome–lysosome fusion ^[203] [203]. Further studies are necessary to confirm these observations and demonstrate whether or not lysosomal Ca²⁺ or pH could be potential therapeutic targets for AD.

Autophagy is a lysosomal degradative pathway responsible for the recycling of different cellular constituents. Especially important under conditions of metabolic stress, this pathway aids in the cellular turnover of damaged or obsolete organelles in order to eliminate misfolded and aggregated proteins left behind by the ubiquitin-proteasome system [204]. Materials are engulfed within double-membrane vesicles (autophagosomes) and targeted to lysosomes for degradation of molecular components. Disruption of autophagy results in accumulation of autophagic vacuoles within swollen dystrophic neurites of affected neurons [205]. Lysosomal Ca²⁺ has been proposed to trigger transcriptional activation of autophagic proteins [200]. Impairment of the autophagy–lysosomal pathway has been described as a hallmark of AD related to lysosomal Ca²⁺ dyshomeostasis. This dysregulation impacts clearance of Aβ and hyperphosphorylated tau, and contributes to their accumulation in the brain ^[205][206] [206,207].