**Preface**

The current Special Issue is directed to all the neurologists and other specialists involved in Alzheimer's Disease research and care. The papers aim to review old and new molecular pathways in Alzheimer's Disease that can be used as potential alternative targets to improve both diagnosis and treatment of this neurodegenerative disorder. In fact, although the effectiveness of anti-amyloid therapy is still controversial, it has been shown to improve the quality of life and reduce AD progression in mild or moderate forms of the disease. However, potential adverse effects include urinary tract infections, nervous system disorders, intracranial hemorrhage, and amyloid-related imaging abnormalities. These treatments seem to reduce the burden of brain amyloid, which is the final waste product of complex molecular pathways that lead to AD neurodegeneration.

This Special Issue first discusses the current and future molecular methods suggested to improve AD diagnosis. Patients who cannot be selected for specific treatments or studies due to mixed or atypical presentations should undergo cerebrospinal fluid (CSF) biomarker interpretation to differentiate AD from other forms of dementia, such as vascular forms. By integrating clinical, neuropsychological, and radiological data with the AT(N) biochemical profiling system (amyloid, tau pathology, and neural loss), researchers and physicians can refine AD diagnosis for both research and clinical purposes, even in atypical clinical presentations. Current molecular research also proposes novel serum plasma markers, such as plasma phospho-tau-181, which will be adopted to refine AD diagnosis and to predict its progression, without the need for an invasive lumbar puncture for CSF biomarker determination. However, this novel marker of disease deserves extensive validation, and the optimal method of determination should be standardized to determine its exact sensitivity or specificity values.

The second topic covered deals with the complex molecular pathways associated with AD pathophysiology. Identifying innovative molecular targets could lead to more effective treatments to reduce both the incidence and the progression of this neurodegenerative disease. An increased deterioration of cognitive function has been observed in patients showing a status of comorbidity, considering vascular risk factors, such as diabetes, hypertension, dyslipidemia, and cigarette smoking, which are associated with neuroinflammation, neurovascular unit dysfunction, and blood–brain barrier breakdown. Evidence suggests that AD and its associated comorbidities share molecular pathways leading to a faster cognitive decline. One of the most intriguing molecular overlaps between neurodegenerative and systemic diseases is symbolized by diabetes mellitus. Insulin resistance translates into a chronic signaling activation of the mechanistic target of rapamycin (mTOR), leading to blood–brain barrier dysfunction, tau hyperphosphorylation, and amyloid plaque formation. Thus, the insulin receptor could represent a potential target to improve neurogenesis. Several other genes involved in oxidative stress and mitochondrial dysfunction have been associated with mild cognitive impairment and AD, representing targets of interest for future investigations.

Another typical aspect of AD is characterized by neuroinflammation, especially in its later stages. Microglia seem to play a pivotal role in the neuroinflammatory component observed in AD neuropathology. Of note, melatonin and other similar molecules act on neuroinflammatory pathways and seem able to upregulate SIRT1-mediated brain-derived neurotrophic factor with regard to prolonged microglial exposure to Aǃ42. This could translate into a reduced expression of inflammatory markers, such as IL-1ǃ and (TNF)-΅, with a subsequent downregulation of the proinflammatory pathway, which is mediated by NF-NjB. Other molecules, such as curcumin, have been shown to be effective in reducing inflammation, oxidative stress, and the aggregation of amyloidogenic proteins.

Despite actual knowledge, more insights into the molecular mechanisms leading to the amyloid cascade are still needed to improve diagnostic methods and to explore novel therapeutic agents acting on different molecular targets of the neuropathogenetic cascade of Alzheimer's disease.

#### **Lorenzo Falsetti**

*Editor*

### *Editorial* **Molecular Research on Alzheimer's Disease**

**Lorenzo Falsetti**

Internal and Subintensive Medicine Department, Azienda Ospedaliero-Universitaria delle Marche, 60131 Ancona, Italy; lorenzo.falsetti@ospedaliriuniti.marche.it

Alzheimer's disease (AD) is the most common form of dementia worldwide. Despite its prevalence and incidence, there are few and limited specific treatments for this disabling and progressive disorder. Anti-amyloid therapy effectiveness is still controversial, but it seems to improve quality of life and reduce AD progression in mild or moderate forms, at the cost of potentially serious adverse effects, especially urinary tract infection, nervous system disorders, intracranial hemorrhage, and amyloid-related imaging abnormalities. However, this treatment seems able to reduce the burden of brain amyloid, which represents the final waste product of complex molecular pathways, leading to AD neurodegeneration. The first topic discussed in this Special Issue is related to current and future molecular methods suggested to improve AD diagnosis. Often, patients cannot be selected for specific treatments, studies, or enrolled in clinical trials due to mixed or atypical presentations: in these settings, AD diagnosis should be enriched with cerebrospinal fluid (CSF) biomarker interpretation, which still represent the cornerstone for differential diagnosis in the setting of neurodegenerative diseases, allowing one to differentiate AD from other forms of dementia, such as vascular forms [1]. Moreover, integrating clinical, neuropsychological, and radiological data with the AT(N) biochemical profiling system (amyloid, tau pathology, and neural loss) allows the researcher and the physician to refine AD diagnosis for both research and clinical purposes, allowing one correctly frame the patient, even in atypical clinical presentations. Current molecular research is also proposing novel serum plasma markers, such as plasma phospho-tau-181, that, in the near future, will be adopted to refine AD diagnosis and to predict its progression, without the need for an invasive lumbar puncture for CSF biomarker determination [2]. Still, this novel marker of disease deserves extensive validation: especially, the optimal method of determination should be standardized to determine its exact sensitivity or specificity values. The second topic covered in this issue of Biomedicines, "Molecular Research of Alzheimer's Disease", deals with the description of the complex molecular pathways associated with AD pathophysiology. Identifying innovative molecular targets could lead to more effective treatments to reduce both the incidence and the progression of this neurodegenerative disease. Most AD cases are not inherited and become clinically evident in elderly, multicomorbid subjects. In this setting, the pathophysiology of AD neurodegeneration seems complex and seems to be associated with the disruption of several molecular pathways, compromising the function of neuronal, glial, and neurovascular units. An increased deterioration of cognitive function has been observed in patients showing a status of comorbidity, considering—among a patient's associated disorders—vascular risk factors, such as diabetes, hypertension, dyslipidemia, and cigarette smoking, which are associated with neuroinflammation, neurovascular unit dysfunction, and blood–brain barrier breakdown [3]. Evidence suggests that AD and its associated comorbidities share molecular pathways, leading to a faster cognitive decline: one of the most intriguing molecular overlaps between neurodegenerative and systemic diseases is symbolized by diabetes mellitus. Alterations in the insulin signaling pathway and glucose resistance in AD subjects' brains are common and typical, and AD is commonly referred as to type 3 diabetes mellitus. Insulin resistance translates into a chronic signaling activation of the mechanistic target of rapamycin (mTOR), leading to blood–brain barrier dysfunction, tau hyperphosphorylation, and amyloid plaque formation. Thus,

**Citation:** Falsetti, L. Molecular Research on Alzheimer's Disease. *Biomedicines* **2023**, *11*, 1883. https:// doi.org/10.3390/ biomedicines11071883

Received: 20 June 2023 Accepted: 25 June 2023 Published: 3 July 2023

**Copyright:** © 2023 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the insulin receptor could represent a potential target to improve neurogenesis. Some molecules, such as amarogentin, seem able to interact with this receptor, representing potential candidates for future clinical studies [4]. Other classes of newer drugs, such as SGLT-2 inhibitors [5] and sestrins [6], could be considered to reduce mTOR activity, acting after the insulin receptor cascade and slowing neurodegeneration. Sestrins seem able to also act in other commonly disrupted pathways in AD, such as, for example, by improving antioxidation and adjusting autophagy. Several other genes involved in oxidative stress and mitochondrial dysfunction have been associated with mild cognitive impairment and AD [7], representing targets of interest for future investigations. Another typical aspect of AD is characterized by neuroinflammation, especially in AD's later stages [8]. As in most tau-dependent neurodegenerative diseases, the interplay between astrocytes, microglia, and neurons often shift from an early, neuroprotective, tau-clearing phenotype with an exacerbated autophagy-lysosomal pathway to a "loss of function" phenotype, leading to neuronal excitotoxicity, often associated with a neuroinflammatory phenotype, which is related to increased tau pathology, oxidative stress, and increased amyloid deposition [9]. Microglia seem to play a pivotal role in the neuroinflammatory component observed in AD neuropathology. Of note, melatonin and other similar molecules act on neuroinflammatory pathways and seem able to upregulate SIRT1-mediated brain-derived neurotrophic factor with regards to prolonged microglial exposure to Aβ42. This could translate into a reduced expression of inflammatory markers, such as IL-1β and (TNF)-α, with a subsequent downregulation of the proinflammatory pathway, which is mediated by NF-κB [10]. Other molecules, such as curcumin, have been shown to be effective in reducing inflammation, oxidative stress, and the aggregation of amyloidogenic proteins [11]. Albeit interesting, most of the published papers in this issue show evidence at a preclinical stage, and further clinical studies are required to validate and to extend the interesting results collected in this issue of Biomedicines. Despite actual knowledge, more insights into the molecular mechanisms, leading to the amyloid cascade, are still needed to improve diagnostic methods and to to explore novel therapeutic agents acting on different molecular targets of the neurodegenerative cascade of Alzheimer's disease.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

## *Review* **Curcumin as Scaffold for Drug Discovery against Neurodegenerative Diseases**

**Filippa Lo Cascio 1,2, Paola Marzullo 3, Rakez Kayed 1,2 and Antonio Palumbo Piccionello 3,\***


**Abstract:** Neurodegenerative diseases (NDs) are one of major public health problems and their impact is continuously growing. Curcumin has been proposed for the treatment of several of these pathologies, such as Alzheimer's disease (AD) and Parkinson's disease (PD) due to the ability of this molecule to reduce inflammation and aggregation of involved proteins. Nevertheless, the poor metabolic stability and bioavailability of curcumin reduce the possibilities of its practical use. For these reasons, many curcumin derivatives were synthetized in order to overcome some limitations. In this review will be highlighted recent results on modification of curcumin scaffold in the search of new effective therapeutic agents against NDs, with particular emphasis on AD.

**Keywords:** curcumin; Alzheimer's disease; amyloid; tau

#### **1. Introduction**

For last two centuries, natural occurring products have attracted the attention of many researchers due to their health benefits in the prevention and treatment of several diseases [1]. In 1815 Vogel isolated a yellow pigment, called curcumin, from the rhizome of Curcuma Longa, an East Indian plant [2]. Curcumin is the most abundant polyphenol and the most biologically active molecule found in the turmeric root; other minor components, known as curcuminoids, are demethoxycurcumin, bisdemethoxycurcumin, and cyclocurcumin [3]. Curcumin is one of the main elements of the Southeast Asian diet and it has been widely used for centuries as a traditional Indian and Asian medicine. After its first extraction, several studies showed that this polyphenolic molecule exhibits a broad spectrum of biological activities. Curcumin offers several health benefits, including anticancer [4], hypoglycemic activities [5], as well as the ability to be used as an analgesic, antiseptic or antimalarial [6]. In addition, curcumin has been shown to have anti-inflammatory [7], antioxidant [8,9], and antiamyloidogenic properties, which are relevant for the treatment of Alzheimer's disease (AD), and related diseases [10,11].

Worldwide, 50 million people have dementia. Unfortunately, this number is expected to increase exponentially, affecting 152 million people by 2050 [12]. AD is the most prevalent progressive neurodegenerative disease associated with age and the most common form of dementia [13], contributing to 60–70% of cases. AD is characterized clinically by progressive loss of memory, language problems, social withdrawal, deterioration of executive functions and eventually death [14,15]. Histopathologically, as Alzheimer's disease progresses, the brain shrinks dramatically and is characterized by cortex damage, and progressive degeneration of limbic and cortical brain structures, mainly in the temporal lobe [15]. This atrophy also affects the cortical association areas and the hippocampus, which is critical for the formation of new memories [16]. As a result of this pattern of

**Citation:** Lo Cascio, F.; Marzullo, P.; Kayed, R.; Palumbo Piccionello, A. Curcumin as Scaffold for Drug Discovery against Neurodegenerative Diseases. *Biomedicines* **2021**, *9*, 173. https://doi.org/10.3390/ biomedicines9020173

Academic Editor: Lorenzo Falsetti Received: 11 January 2021 Accepted: 8 February 2021 Published: 9 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

cortical thinning, it is also possible to observe an enlargement of ventricles and a functional alteration of Wernicke's and Broca's areas [17]. A common characteristic of age-related neurodegenerative diseases, including AD, is the pathological accumulation of unfolded and aggregation-prone proteins in the brain, which are considered the major cause of synaptic loss and progressive neuronal death observed in these disorders [18]. The two major systems involved in proteostasis maintenance are the autophagy-lysosomal system and the ubiquitin proteasome system [19]. However, these two systems have been found to be impaired in many neurodegenerative diseases, including AD. Therefore, the failure of these systems in maintaining proteostasis may also contribute to the pathological aggregation of proteins as well as formation of insoluble and fibrillar amyloid inclusions [20].

The major neuropathological features of AD are synaptic and neuronal degeneration and the presence of amyloid plaques and neurofibrillary tangles (NFTs).

Neuritic plaques are polymorphous aggregates made up of the amyloid Aβ peptide (Aβ) aggregates. The ≈4 kDa Aβ fragment originates from the transmembrane amyloid precursor protein (APP) by concerted proteolytic cleavage of β- and γ-secretase [21]. Monomeric Aβ1-40 (Aβ40) and Aβ1-42 (Aβ42) species can aggregate to form Aβ oligomers that can further aggregate and assembly into amyloid fibrils [22]. A growing body of evidence suggest that the oligomeric/prefibrillar Aβ peptide is the neurotoxic species that trigger the amyloid cascade, leading to the damage and eventual death of neurons associated with AD [23–25]. On the other hand, NFTs are intracellular inclusions of hyperphosphorylated tau, a microtubule associated protein. In its native state tau is a monomeric protein [26]. Tau is a natively unfolded protein involved in microtubule stabilization and axonal transport. However, under pathological conditions, tau can undergo abnormal post-translational modifications, including phosphorylation or acetylation [27,28]. As result of these modifications, tau detaches from the microtubules causing their disassembly, cytoskeletal instability, and axonal transport perturbation [29,30]. Unbound tau can self-aggregate forming soluble tau oligomers that assemble into paired helical filaments (PHFs) [31–33]. The PHFs mature into fibrils that constitute the intracellular NFTs, observed in the brain of AD patients [27]. Increasing evidence suggests that synaptic dysfunction and neuronal loss precede the formation of NFTs [34–39], indicating that the smaller and prefibrillar aggregates, tau oligomers, may be responsible for the toxic effects during the early stage of AD and other tauopathies [40,41]. Therefore, tau oligomers are considered to be highly toxic and to seed tau misfolding, thus propagating the pathology seen across different neurodegenerative diseases [38,42].

Despite the many efforts made to develop new treatments and therapeutic approaches to prevent the onset of the disease and to reverse the disease process, to date, there are no effective therapeutics. Nowadays, the therapeutic strategies available are only symptomatic treatments that counterbalance neurotransmitter disturbance, thus ameliorating a few of the clinical symptoms associated with the disease. The established treatments available are acetylcholinesterase inhibitors (e.g., Donepezil or Tacrine), antagonists of glutamate NMDA receptor (e.g., Memantine), agonist of nicotinic or muscarinic receptors, antioxidants and anti-inflammatory agents [43,44].

Growing evidence demonstrates a protective effect of curcumin against Aβ plaque formation; however, the mechanism of action is not yet fully clarified. Some studies have classified curcumin as an inhibitor of Aβ aggregation, others as disaggregating and destabilizing of amyloid fibrils [45]. In addition, curcumin has been shown to hamper Aβ oligomerization but not its fibrillization [46]. Recently, curcumin has been shown to attenuate amyloid-β aggregate-associated neurotoxicity by promoting the formation of "off-pathway" nontoxic soluble oligomers and prefibrillar proteins [47].

Curcumin has also been shown to exert a neuroprotective role by inhibiting tau aggregation. Indeed, curcumin has been shown to inhibit tau oligomerization, disintegrate preformed tau oligomers, inhibit β-sheet formation, and disaggregate tau filaments [48,49]. In addition, in vitro studies have shown that curcumin prevents the aggregation of other amyloidogenic protein, including α-synuclein (α-syn), which is a presynaptic protein

involved in PD. PD is a debilitating neurodegenerative disorder characterized by the gradual loss of dopaminergic neurons in the substantia nigra pars compacta and clinically characterized as movement disorder. α-syn accumulates abnormally and aggregates in the cytosol as Lewy bodies and in the neuronal processes as Lewy neurites [50]. Several studies have showed that curcumin inhibits α–syn aggregation and reduces α-syn-induced cytotoxicity [51,52].

The neuroprotective effect of curcumin is certainly due to its ability to modulate the aggregation pathways and toxicity of amyloidogenic proteins and mitigate inflammation and oxidative stress, known to be key factors in the progression of neurodegenerative disorders [53,54].

This review attempts to explore the protective role of curcumin and its related compounds in the treatment of neurodegenerative disorders as a potential modulator of pathogenic pathways associated with AD and related diseases.

#### **2. Physicochemical Characteristics of Curcumin**

Due to the relevant biological and health benefits of curcumin, several chemists proposed a potential structure of curcumin. In 1913 Lampe et al. synthesized curcumin for the first time [55]. A general procedure for the synthesis of curcumin with a higher yield was later reported by Pabon [56]. In this reaction scheme 2,4-diketones, such as acetyl acetone, reacts with conveniently substituted aromatic aldehydes, particularly vanillin aldehyde, to synthesize curcumin. To prevent a Knoevenagel condensation due to the high acidity of the α-methylene group, the reaction is carried out in the presence of boron oxide as a complexing agent for the dienolate group. In this way, the condensation reaction involves terminal alkyl groups of di-ketone and primary or secondary amines, usually *n*-butylamine, are used to deprotonate these groups. Alkyl borates act as drying agents to remove the water formed by condensation reaction between boron complex and aromatic aldehyde. In the final step, boron complex gives the final product in acidic conditions. The reaction is refluxed, using aprotic solvent such as ethyl acetate (Scheme 1).

**Scheme 1.** Synthesis of curcumin with Pabon's method.

Several research groups follow the general method proposed by Pabon with slight modifications. For example, boron oxide has been replaced with boric acid with a lower yield [57,58]. An alternative procedure, reported by Rao et al., replaced boron oxide with borontrifluoride to obtain curcuminoid difluoroboronites that can be then hydrolyzed using aqueous methanol at pH 5.8 to get curcuminoid compounds [59]. To synthesize polyhydroxy curcuminoids, it is necessary to protect the hydroxyls groups on the starting benzaldehyde. These groups were protected as ethers and deprotected using aluminum chloride [60]. Curcumin is a low molecular mass polyphenolic compound (368.38 g/moL) with a melting point of 183 ◦C [61]. The IUPAC name of curcumin is 1,7-bis(4-hydroxy-3-methoxy-phenyl)-1,6-heptadiene-3,5-dione and is also known as diferuloyl methane. Curcumin is a hydrophobic molecule with a log *p* value of 3.29. It is insoluble in water and soluble in polar organic solvent, like methanol, ethanol, dimethylsulfoxide, dimethylformamide, or ethyl acetate. It is partially soluble in hexane or cyclohexane [62].

Curcumin is a symmetrical molecule composed of two aromatic rings substituted with o-methoxy phenolic groups and a β-diketone moiety as a central linker. The heptadienone linkage exhibits keto-enol tautomerism (Scheme 1) that influences physicochemical and antioxidant properties of curcumin [63,64]. Curcumin is present in its bis-keto form in acidic and neutral pH conditions (pH 3–7) due to the presence of an acid proton linked to a highly activated carbon between the two aromatic rings. Conversely, under basic conditions (pH > 8), the enol form predominately and curcumin acts as an electron donor. Indeed, the antioxidant activity of curcumin is attributed to its enolic form [65]. X-ray crystallography studies confirmed that the enol form has a lower energy as compared to the diketone tautomer and it is the exclusive form in solution [66]. Moreover, keto-enol tautomer can exist as syn and anti isomers with the syn-enol form being more stable. In the syn form the two methoxy groups are on the same side with respect to keto-enol and hydroxy groups. Thus, it is possible to identify a polar surface with either a phenolic or enol group and a nonpolar area with methoxy groups [67].

Curcumin, as well as other polyphenolic compounds, displays a strong absorption in the visible region with a maximum absorption around 410–430 nm and another band with maximum absorption at 265 nm. In the presence of nonpolar solvents, including hexane or cyclohexane, a blue-shift of the absorption spectrum is observed. Conversely, in polar solvents, such as methanol or DMSO, the peak is shifted towards the lower frequencies [65]. These observations can be justified by the shift of the keto-enol tautomerism towards the enol form in a polar solvent or towards the bis-keto form in nonpolar solvent. The enol form exhibits a larger electronic delocalization and, therefore, a red-shifted absorption peak is observed [68].

#### **3. Curcumin Bioavailability: Metabolic Reactions and New Formulations**

Despite the relevant biological activities of curcumin, several studies have revealed a low oral bioavailability due to its poor solubility in water, low permeability and absorption, fast metabolism, and excretion in vivo. Oral administration of curcumin in rats (500 mg/kg) showed 1% of bioavailability in rat plasma [69]. In addition, several clinical studies have revealed extremely low serum levels following oral administration [70,71]. However, curcumin bioavailability improves once it is injected intravenously in rats [72].

The gastrointestinal tract represents the first physical barrier that limits the oral absorption due to the presence of the mucus layer and the tight junction proteins [48]. In addition, following oral administration, curcumin is rapidly metabolized by both conjugation and reduction pathways in the body, resulting in the formation of several pharmacologically inactive metabolites. Indeed, O-glucoronide or O-sulphate have been the principal curcuminoid metabolites found in the plasma following oral administration in rats. Furthermore, bioreduction products such as dihydrocurcumin or tetrahydrocurcumin and their conjugates formed by alcohol dehydrogenase were identified by HPLC and mass spectrometry analyses [73]. The α,β-unsatured β-diketo moiety of curcumin can be susceptible to degradation by hydrolysis at room temperature in neutral or alkaline conditions (pH ≥ 7). Several degradation products, including ferulic acid, ferulic aldehyde, vanillin, vanillic acid, and feruloylmethane, are also found in the serum; however, the amount of conjugated metabolites is more than the amount of reduction products. Probably in biofluids the β-keto function is not free but bound to proteins and, therefore, is not hydrolysable [62,74]. Curcumin is also photoreactive and undergoes photodegradation when exposed to sunlight, forming similar products to those found following hydrolytic degradation [75,76].

Several preclinical studies have suggested curcumin as a potential therapeutic approach for AD and related diseases; however, no clinical trials have been successful. The failure of these studies may be due to curcumin low brain bioavailability after oral administration and fast metabolism [77,78]. Consequently, alternative formulations and new drug delivery systems, including liposomes and nano-based approaches, have been developed to increase curcumin brain delivery. Most of the new delivery systems proposed are characterized by the presence of a central hydrophobic pocket in which curcumin binds

through hydrophobic interactions. These macromolecular systems preserve curcumin from degradation and enhance its absorption as well as its distribution [75]. Furthermore, liposomes modified with penetrating agents, including curcumin as well as other amyloidstargeting ligands, have been shown to facilitate the passage of the compound through the blood–brain barrier (BBB) and have been considered as suitable vehicle for the delivery of therapeutics in the central nervous system [79]. Recently, Giacomelli et al. developed a novel and advanced curcumin delivery system based on nanoparticles named lipid-core nano-capsules. As result of their studies, they found that curcumin loaded nano-capsules display a significantly higher neuroprotective effect against Aβ toxicity in a mouse model of AD as compared to free curcumin [80]. In addition, Yang et al. proposed a novel curcumin-loaded nanoparticle system made of chitosan and bovine serum albumin. Using this formulation, they observed an increased penetration of curcumin through the BBB and microglia activation with a subsequent increase of Aβ phagocytosis [81]. Cyclodextrins are also used as absorption enhancers in several pharmaceutical formulations. Li et al. demonstrated that α-cyclodextrin enhances intestinal absorption of curcumin via transcellular and paracellular mechanisms [82]. In recent years, the study and the analysis of several crystalline solid forms of curcumin have also been a focus of great interest due to the different physicochemical properties exhibited by these polymorphs [83,84]. In addition to these alternative formulations, Wang et al. suggested exosomes, membrane-bound extracellular vesicles, as potential delivery system of curcumin. Therefore, exosomes-derived from curcumin-treated cells were used as carrier to release selectively curcumin in the brain. This delivery system increases the percentage of curcumin crossing the BBB through receptor-mediated transcytosis. They observed a decreased phosphorylation of tau through inhibition of AKT/GSK pathway, following injection in an AD mice model [82,85].

#### **4. Relationship between Structural Properties and Biological Activity of Curcumin Derivatives**

The synthesis of novel curcumin derivatives represents an effective alternative to obtain curcumin analogs with a better solubility in biofluid in an effort to improve the pharmacokinetic profile of curcumin and its biological activity [48,86–88]. As mentioned above, curcumin acts as a neuroprotective agent blocking multiple mechanisms involved in neurodegeneration by interfering with the accumulation of misfolded aggregate proteins, including Aβ and tau, inflammation and oxidative stress. The modulation of each of these pathological pathways requires distinct structural feature of curcumin. Therefore, comprehensive structure–activity studies are extremely important to identify novel curcumin derivatives as potential therapeutic agents for neurodegenerative diseases.

In recent years, researchers have synthesized several compounds able to block Aβ fibrillogenesis. In this process, Aβ monomers aggregate to form oligomers, which then assemble to form insoluble aggregates [89]. This transformation is characterized by a structural transition from α-helix to β-sheet structure [90]. It is known that the short Aβ fragment, KLVFF (Aβ16-20) binds to full length Aβ and it is important for amyloid fibril formation [91]. A shared model hypothesizes that phenylalanine residue in the KLVFF sequence of Aβ peptides interact through Π-Π interactions during Aβ aggregation. Small molecules, such as curcumin, are able to block or break these interactions and could be valid candidates to revert amyloid formation [92].

Reinke and Gestwicki have created a library of compounds resembling curcumin structure to investigate and evaluate the effect of the three main features of curcumin on inhibition of amyloid aggregation [93]. The structural features contributing to the inhibitory potency of curcumin are the two aromatic rings, the substitution pattern of these phenyl groups, and the length and flexibility of the central linker region. To perform structural considerations and better understand which feature is critical for the inhibition of Aβ aggregation, Reinke and Gestwicki synthesized curcumin analogs by modifying only one structural feature at the time and retaining the other two. As a result of their studies, they found that compounds lacking one aromatic group are less active than curcumin, suggesting that both aromatic rings are essential to interact through hydrophobic interactions and hydrogen bonding with phenylalanine residue of Aβ monomers and inhibit amyloid formation [94]. Furthermore, hydroxyl substitution on the aromatic end group or other polar functional substituents are required for the inhibiting activity. In addition, Reinke and Gestwicki showed that both length and flexibility of the central linker region are key factors to take into consideration in the design of new Aβ aggregation inhibitors. Indeed, the inhibiting activity is negatively affected when the central linker region is too long, too short, or too flexible. The optimal length of the central linker is 8–16 Å and no more than one or two rotating sp3-hybridized carbons are required for an ideal flexibility.

It has been shown that the homeostasis of metal ions is critical for maintaining normal physiological functions. Some ions such as Al, Fe, Cu, and Zn have been observed in the brain of AD patients [95]. Their imbalance in the brain is closely related to the Aβ deposition and tau accumulation, suggesting that they play a role in the degenerative process of AD. The histidine residues of Aβ peptides (H13/H14) are good coordination sites for metal ions [96]. Curcumin can interact with metal ions forming strong complexes. Indeed, the α,β-unsaturated β-diketo moiety of curcumin has shown excellent chelating properties. Many studies reported the synthesis of stable metal-curcumin complex with a stoichiometry 2:1 (ligand:metal) [62]. Curcumin-metal complexes decrease Aβ plaques as well as suppress inflammatory processes by preventing metal induction of nuclear factor kappa B, NF-kB [97,98]; however, metal chelators can disrupt the normal brain homeostasis. Zhang et al. designed a novel curcumin derivative, named CRANAD-17, as a chelating agent to attenuate Aβ crosslinking induced by Cu [99].

Some curcumin derivatives exert their neuroprotective effects by promoting phagocytosis of Aβ fibrils. For example, it was demonstrated by Fiala et al. that bisdemethoxycurcumin (BDC) enhances macrophage-activated Aβ clearance and reduces the inflammation state [100]. Recently, Gagliardi et al. showed that the treatment of BDC derivatives in a human monocytic cell line, mimicking the peripheral blood mononuclear cells of AD, revealed overexpression of genes essential for macrophage function, including mannosylglycoprotein 4-beta-*N*-acetylglucosaminyltransferase and vitamin D receptor. BDC also showed a protective anti-inflammatory effect through downregulation of NF-kB and β-site APP cleaving enzyme 1 (BACE1) genes [101]. BACE 1 protein and γ-secretase enzyme cleaves APP to generate Aβ peptides, with the β-cleavage being the rate-limiting step of this sequential proteolytic pathway [102]. Inhibitors of BACE1 activity are, therefore, considered as a possible therapeutic approach and many BACE1 inhibitors have been synthesized. Two critical curcumin structural features associated with BACE 1 inhibitory activity are the phenolic rings and the unsaturated alkenyl linker between the aromatic rings. Derivatives with multiple hydroxy groups have been found to be more active than compounds with nonsubstituted phenyl groups or substituted with methoxy groups or halogen. When researchers replaced the phenol group with indole or a pyrrole ring, they produced more active compounds. These data suggest that inhibitor molecules interact with BACE 1 through hydrogen bonds [103]. Reduction products are not active against BACE 1. A planar structure is required to maintain the inhibitory activity and sp2 carbons give the optimal rigidity to the molecule. The substitution of 1,3-dicarbonyl moiety of curcumin by isosteric heterocycles, isoxazole, and pyrazole, resulted in the formation of potent inhibitors of γ-secretase enzyme [104].

Up to this point of our discussion, we have analyzed the structural aspects of curcuminoids that have been found to be important for their anti-Aβ aggregation activity. Curcumin and its derivatives also exert a protective role against misfolded tau aggregates. Unlike Aβ, tau lacks in hydrophobic residues and therefore its aggregation process does not lead to the formation of Π-Π interactions. However, under particular circumstances, the small degree of hydrophobicity, compared to other proteins, is sufficient to drive tau aggregation [105]. Tau aggregation inhibitors interact with tau by electrostatic interaction and hydrogen bonding. A recent study suggested two ruthenium-curcuminbipyridine/phenanthroline complexes as inhibitors of tau aggregation. In these complexes,

the metal ion is bound to the enol group of curcumin and to the nitrogen atoms of the ancillary ligands, bipyridine or phenanthroline. Curcumin inhibits aggregation at the nucleation stage while the positive charged ruthenium complex inhibits the elongation phase, reducing longer fibrils formation [106].

Oxidative damage plays an important role in neurodegeneration, therefore, to treat neurodegenerative disorders another pharmacological approach is to develop antioxidant compounds. The antioxidant property of curcumin is due to the abstractable phenolic hydrogen [107]. This group can reduce, for example, superoxide radicals to generate less reactive phenoxyl radicals that are resonance stabilized [108]. Ferrari et al. developed curcumin analogs substituted on the central carbon of the heptadienone linker and demonstrated that these complexes exhibit good metal chelating properties. However, these compounds showed less scavenging activity because of the presence of the substituent on central linker shifts the keto-enol tautomerism towards the di-keto form with less stabilization of phenoxyl radical [109].

#### **5. Curcumin Derivatives and Hybrids Molecules**

Given the large number of biologically active curcumin-like molecules that have been synthesized, we can divide them into different classes accordingly to the modified part of curcumin structure. Discussion on selected compounds is reported in the following subsections.

#### *5.1. Monocarbonyl Analogs of Curcumin (MACs)*

The monocarbonyl analogs of curcumin (MACs) belong to the group of compounds with a central core modification, as shown in Figure 1. Several researchers have proposed MACs as anticancer as well as anti-inflammatory agents. These compounds have been shown to exhibit a higher anticancer potency than curcumin in many cancer cell lines [110]. Recently, monocarbonyl derivatives have also been proposed for the treatment of AD.

**Figure 1.** Structures of monocarbonyl analogs of curcumin (MACs).

The removal of the keto-enol motif, susceptible to hydrolysis, enhances the stability of MACs compared to curcumin; however, the presence of the enone group is important for the anti Aβ aggregation activity [111]. MACs can be divided into two main groups: acyclic (**1**) and cyclic MAC compounds (**2**, **3** and **4**) (Figure 1). In the cyclic MACs, the carbonyl group could also be part of a 5 or 6 membered heterocyclic ring containing NR, O, S or SO2 groups [112]. In a general synthetic method reported by Ohori et al., MACs have been produced from aryl-aldehyde and acetone. To obtain cyclic MACs, acetone was substituted by cyclic ketone. The reaction is carried out in ethanol using sodium hydroxide and cetyltrimethylammonium bromide as catalysts [110].

Orlando et al. synthesized MAC derivatives with 5-carbon spacer between the aromatic rings. Compound **1** in Figure 1 showed a higher percentage of anti-Aβ aggregation activity compared to curcumin (IC50 0.8 μM vs. IC50 1.0 μM) [111]. In addition, newly synthesized monocarbonyl derivatives with a piperidone structure in the linker (**2**) are identified as potent inhibitor of Aβ fibrillation. Furthermore, substitution of a nitrogen atom within the piperidone ring with a carbon atom or *N*-methylation reduced the antiaggregation activity. The aliphatic linker gives the ideal flexibility, while the *N*-methylpiperazine groups on the two aromatic ends enhance hydrogen bond interactions with Aβ peptides. Compound **2** decreased the β-sheet structure and stabilized the α-helix content as assessed by circular dichroism spectroscopy (CD). In addition to the antiaggregating activity, compound **2** showed antioxidant and metal chelating properties [113]. MAC **3** was demonstrated to be more stable than curcumin at physiological pH. Furthermore, UV measurements of MAC **3** revealed that its spectra remained unchanged over time, while UV spectra of curcumin showed a decreased intensity and a shift towards the red. The protein hen egg white lysozyme (HEWL) was used as a model protein to study the inhibitory effect of MAC **3** on amyloid formation. MAC **3** also exhibits an optimal length of linker (8.84 Å) and it is more rigid than curcumin due to the absence of rotating sp3 carbon within the backbone. These structural features make it as an ideal inhibitor of Aβ aggregation. Fluorescence measurements using the molecular probe 8-anilinonaphatalene-sulphonate (ANS), showed decreased hydrophobic surface upon treatment with the curcumin analog MAC **3**. Docking studies demonstrated that the compound binds to the catalytic tryptophan (62 and 63) of HEWL with the carbonyl component pointing toward the hydrophilic residues of the active site [114]. The monocarbonyl-cyclohexanone derivative **4** was tested in vitro as an antioxidant agent and showed good scavenging activity for reactive oxygen species (ROS) [115]. MAC 4 exhibits antioxidant properties and protective effect against H2O2-induced cytotoxicity in PC12 cells. The compound was able to rescue the levels of glutathione (GSH) as well as the activity of superoxide dismutase and catalase. Moreover, compound **4** showed to increase mRNA expression of Nrf2, a key transcription factor of the antioxidant response [115].

#### *5.2. C4-Substituted Curcumin Derivatives*

Medicinal chemists have synthesized C4-substituted curcumin derivatives by modifying the central core of curcumin with the insertion of one or two alkyl substituents (Figure 2).

**Figure 2.** Structures of C4-substituted curcumin derivatives.

The synthetic pathway follows Pabon's route. In the first step, alkyl-substituents are inserted into an alkaline environment using acetyl-acetone and appropriate alkyl halides [116,117].

This modification as well as the synthesis of MACs, leads to the production of more stable compounds. Mono-substituted compounds have an acidic hydrogen within the linker and, therefore, retain the keto-enol tautomerism. The enolization of these derivatives is involved in the interaction with Aβ aggregates. Particularly, the anion of enol form has a reddish color while the other forms, including the neutral enol form and neutral or anion keto form, are yellow or colorless in solution. The same reddish color is observed when these molecules are incubated with Aβ aggregates, indicating an increase of the anion of the enol form when it is bound to Aβ aggregates. Notable, the interaction of the compound with Aβ monomer did not cause change in color. Therefore, this observation supports the involvement of the enol form in the binding to Aβ aggregates [118]. A decreased anti-Aβ inhibitory activity was observed for disubstituted compounds, including α,αdimethylcurcumin. This compound lacks enol tautomer and planarity that are essential to interact with Aβ aggregates.

Ferrari et al. synthesized curcumin analogs, known as **K2T**, by introducing t-butyl ester in the C4 position [109]. **K2T** derivatives exhibit metal chelating properties towards gallium and copper ions and inhibit Aβ aggregation. The presence of alkyl substituent shifts the tautomeric equilibrium towards the di-keto tautomer and limits the radical scavenging ability. **K2T21** (Figure 2) is the best compound of this series with a good metal chelating property. In addition, the Cu(II):**K2T21** complex maintains a scavenger activity [109]. The derivative **K2F21** (Figure 2), functionalized with a phthalimide group in the α-position and vanillin-like structure on the aromatic portion, showed a higher stability, depolymerization, antioxidant and antiapoptotic activities [116]. Molecular docking simulations showed a high probability of van der Waals interactions of **K2F21** with β2-site of Aβ1-40 fibrils. The β2-binding site is located within the residues 31–40, which is a region known to be involved in modulating Aβ aggregation through the action of methionine 35 (Met35) [116].

#### *5.3. Heterocyclic Derivatives*

Several research groups designed and synthesized curcumin derivatives by replacing 1,3-dicarbonyl moiety with isosteric pyrazole and isoxazole rings. Isoxazole derivatives were synthesized at reflux starting from curcumin and hydroxylamine hydrochloride using pyridine and ethanol as solvents. Curcumin was converted into pyrazole and *N*-substituted pyrazole by reaction with corresponding hydrazines (NH2NH2 or RNHNH2), using reaction conditions reported by Narlawar et al. Particularly, from their series, compound **5** and **6** (Figure 3) are the ones exhibiting the most interesting biological effects.

**Figure 3.** Structures of heterocyclic derivatives.

Both compounds interact with Aβ42 aggregates and can be used as an imaging agent for diagnostic purposes. In addition, compound **6**, at lower concentrations, demonstrated the ability to depolymerize tau aggregates, inhibit tau aggregation, and produce

γ-secretase activity [104]. **CNB001** (Figure 3) is a pyrazole analog of curcumin that has been extensively studied because of its potential as reliable therapeutic candidate for the treatment of AD. **CNB001** was shown to improve memory and long-term potentiation and mitigate motor impairments in rats [119,120]. In addition, the compound showed antiinflammatory activity through inhibition of proinflammatory mediators in LPS-induced microglia [121]. Another compound with curcumin-pyrazole structure is **GT863** (also named PE859, Figure 3). Evaluation of its anti-Aβ and tau aggregation activity revealed that substitution of the aromatic ring with a bicyclic system along with the protection of the phenolic group from metabolic reaction, may increase the inhibitory activity [122]. It was shown that **GT863** reduces the production of Aβ through alteration of nicastrin maturation, an essential glycoprotein component of the γ-secretase complex [123].

Heterocyclic derivatives of curcumin also include compounds with an oxadiazole ring instead of a β-diketone group. Our research group has synthesized a series of curcumin-like compounds with 1,2,4-oxadiazole or 1,3,4-oxadiazole motif. Compounds **7** and **8a** were tested as Aβ aggregation inhibitors using different biophysical techniques and both showed to affect the aggregation pattern of Aβ [124]. Induced Fit Docking (IFD) observations revealed that both compounds bind to Aβ in a saddle between Met35 and Val39 via hydrophobic interactions. In the presence of compound **7** was observed a perturbation of β-sheet content, while it was partially preserved in the presence of compound **8a**. The IFD results suggest that compound **8a** can interfere with Aβ aggregation by hampering the packing of oligomers along the fibril major axis. On the other hand, due to steric hindrance, compound **7** interferes with the formation of β-sheet structure, resulting in the formation of toxic off-pathway structures. Indeed, compound **7** showed to trigger less Aβ aggregation and enhance Aβ 1-40 toxicity, probably due to the higher presence of toxic oligomers in the medium [124]. Other newly synthesized curcumin derivatives demonstrated modulation of the aggregation pathway of preformed tau oligomers. Particularly, compound **8b** was found to convert toxic tau oligomers into more nontoxic tau aggregates and mitigated tau oligomer-associated toxicity in the human neuroblastoma cell line, SH-SY5Y, and primary neuronal cultures [87].

Novel pyrimidine **9**, pyrazine **10** and pyridazine **11** curcumin derivatives, with general structure reported in Figure 1, were efficiently synthesized and tested in vitro and in vivo as potential Aβ and tau imaging probes for the diagnosis of AD. These heterocyclic derivatives showed an excellent capacity to label Aβ as well as tau aggregates [125].

#### *5.4. Tetrahydrocurcumins (THCs)*

Hydrogenated curcuminoids attracted researcher's attention for their biological properties (Figure 4). These compounds lack the diketone bridge necessary to bind Aβ fibrils and to exert antioxidant activity.

**Figure 4.** Structures of tetrahydrocurcumins (THCs) derivatives.

Tetrahydrocurcumin (**THC**) is a more stable metabolite of curcumin. Phenolic rings and methoxy groups, but not double bonds of curcumin, mediate the anti-inflammatory effect. **THC** has been shown to reduce neuroinflammation through reduction of IL1β in the brain and inhibition of LPS inducing the release of iNOS; however, it does not display any amyloidogenic inhibitor activity [126]. Tetrahydrodemethoxycurcumin (**THDMC**) and tetrahydrobisdemethoxycurcumin (**THBDMC**) are reported as inhibitors of acetyl cholinesterase activity (AChE). Particularly, the absence of the double bond and the methoxy group increases their inhibitory activity. In addition, **THCs** were conjugated with a dihydropyrimidinone, a known AChE inhibitor, to create a series of **THCs-DHPM** compounds. Compound **THBDC-DHPM** produced by **THBDMC** with methoxy group on phenyl group exhibits the lowest value of IC50 against the enzymatic activity [127].

#### *5.5. Curcumin-Like (CL) Compounds*

To overcome the instability issues due to the β-di keto moiety, our research group synthesized a library of curcumin-like compounds without the α-carbon within the linker (Figure 5). In the synthetic method adopted, the diacetyl reacted in a double aldol condensation with the appropriate aromatic aldehyde. The resulting compounds **CL3** and **CL8** were found to modulate the aggregation state of recombinant toxic tau oligomers and disease-relevant tau oligomers [87,128]. Notably, **CL3** affected both the size and the surface hydrophobicity of brain-derived tau oligomers reducing their associated neurotoxicity [128].

**Figure 5.** Structures of curcumin-like (CL) compounds.

#### *5.6. Aromatic Ring Substitution: Methoxy and Hydroxy Groups*

Aromatic rings of curcumin and curcumin-like compounds interact with Aβ aggregates by hydrophobic and hydrogen bonds. Therefore, chemical structure of curcumin containing two aromatic rings is optimal for inhibition of amyloidogenesis. Reinke and Gestwicki synthesized a series of curcumin analogs without an aromatic ring and evaluated their inhibitor activity on Aβ aggregation. These compounds did not show any effect on Aβ aggregation, even at higher concentrations (500 μM). Conversely, curcumin inhibits Aβ aggregation with an IC50 value of 10 μM [93].

In the same study, they also evaluated the influence of aromatic substituent on the neuroprotective activity of curcumin and its derivatives. They showed that substituents capable of taking part in hydrogen bonding are essential for the activity.

Several structural–activity studies were performed to evaluate the effects of the position and number of methoxy and hydroxy group on the biological activity of the resulting derivatives (Figure 6).

**Figure 6.** Structures of compounds with methoxy/hydroxy substituted aromatic rings.

Compounds with an anti-Aβ aggregation activity were obtained by maintaining methoxy and hydroxy group in para- and meta- positions, while orto-substitution did not improve the activity. Compound **12** was obtained from curcumin through the substitution of the para-hydroxy group with a methoxy group. Researchers observed a decreased polarity and an increased permeability across the BBB. Moreover, this substitution preserves the conjugation reaction of -OH groups with sulphate or glucuronide groups [111]. In addition, the mono-carbonyl compound **13** methoxy substituted showed an improved inhibitory activity on Aβ aggregation [114]. Hitoshi et al. showed the importance of the catechol motif for anti-Aβ aggregation activity. Particularly, they demonstrated that the ophenol derivative **14** increases the water solubility as compared to para- and meta- phenol. Previous studies reported that an increased dihedral angle improved water solubility. In ophenol compound hydroxyl and β-di keto groups are close and can interact by participation of water. This interaction causes a torsion of the molecule, which increases the angle between the phenol group and central linker and, thus, increases water solubility [129]. Compound **14** is a potent inhibitor of BACE1 enzyme. Computational studies suggest that the inhibitor interacts with BACE1 in the P3 pocket. It did not interact with the aspartic acid residues of active site (Asp32 and Asp228); hydroxy groups and ketone motif are involved in hydrogen bonds with Glu230 and Glu339 [103].

#### *5.7. Aromatic Ring Substitution: Halogenated and Prenylated Derivatives*

Chemists have made several efforts to synthesize new curcumin-like compounds that can be used as diagnostic tools to detect amyloid formation in the brain. They designed the compound **15** by the introduction of tri-fluomethoxy groups on the aromatic rings (Figure 7).

**Figure 7.** Structures of compounds with halogenated or prenylated aromatic rings.

This derivative passes through the BBB and reaches the brain, where it can be detected using fluorine 19 by MRI. Moreover, the trifluoromethoxy group is important for anti-Aβ aggregation activity [118]. Several studies reported other halogenated curcumin derivatives as biologically active compounds for the treatment of AD, including the chlorosubstituted compound **16** [130]. In addition, the bromo-derivative **17** exhibits antioxidant properties [60]. Synthetic halogenated derivatives of curcumin have been identified as ligands for nuclear receptor of vitamin D as well as for nuclear receptor of retinoid (RXR and RAR) [131]. It was demonstrated that agonists of RXR stimulate Aβ clearance through induction of ApoE expression [132].

To increase the number of hydrogen bonding and electrostatic interactions between curcumin derivatives and Aβ peptide, a *N*-methylpiperazine group was introduced to the aromatic ring. Thus, a series of mono-carbonyl *N*-methylpiperazine substituted derivatives was synthesized and compound **2**, reported above (see Figure 1), showed the best inhibitory activity of Aβ aggregation [113]. Instead, prenylated curcumin analogs were prepared to enhance the hydrophobic contacts of curcumin with Aβ monomers. Compound **18** is the most active compound in term of antifibrillogenic and anti-inflammatory activities. It retains methoxy and hydroxy groups that are essential for H-bonds and the 4-prenyloxy group that is required to establish hydrophobic interactions with the nucleation core of Aβ peptides [133]. Indeed, the compound approaches to the self-recognition hydrophobic core, 16KLVFFA21, which is known to be a nucleation site for the pathogenic aggregation of Aβ. In particular, **18** assumes an extended geometry by interacting with six different amyloid segments in the same residues, Leu18 and Lys16. Specifically, they form hydrophobic interactions with six Leu18 residues, which contact both the phenyl ring as well as the alkyl chain. In addition, they form H-bonds with six Lys16 residues, which approach the β-keto-enol central core and the substituents on phenyl rings. When the complexity of simulated amyloid structure increases, the interactions between compound **18** and the fibril structures are governed by hydrophobic interactions [133].

#### *5.8. Other Aromatic Rings*

Among lateral changes made to the structure of curcumin there is the substitution of one aromatic group with an indole moiety. The asymmetric compound **19** showed an inhibitory activity against the BACE1 enzyme and can be considered as anti-AD drug (Figure 8) [103,134]. In this compound, the substitution of phenol group with the indole motif improves the H-bonds with BACE1 enzyme.

**Figure 8.** Structure of compound **19** with an indole aromatic ring.

The pyrazole curcumin derivative **GT863** (Figure 3) also presents an indole ring on the lateral part of the molecule. It inhibits both Aβ and tau aggregation. This compound is more stable than curcumin due to the presence of the pyrazole moiety and the protection of the phenolic hydroxyl group that is prone to metabolic reaction [122].

#### *5.9. Hemi-Curcuminoids*

Hemi-curcuminoid β-diketones are another class of asymmetric curcumin-like compounds that can be considered the start point for the development of new neuroprotective agents (Figure 9).

**Figure 9.** Structures of hemi-curcuminoids.

These derivatives were prepared following Pabon's method starting from substituted benzaldehydes and di-ketonic compounds. Compounds **20** and **21** display metal chelating properties and scavenger activity against ROS. The para-hydroxy group is considered a required structural feature to exert the antioxidant effect [135].

Claisen–Schmidt Aldol condensation of the aromatic aldehyde with acetone under basic conditions generated a series of hemi-curcuminoid compounds [87]. These compounds were tested as a modulator of the aggregation state of preformed tau oligomers. Compound **22** showed interaction with tau oligomers promoting the formation of larger, less hydrophobic and nontoxic tau aggregates as assessed in human SH-SY5Y neuroblastoma cells [87].

#### *5.10. Calebin A Derivatives*

Darrik and colleagues isolated Calebin A from Curcuma Longa and synthesized it for the first time [136]. After evaluating the protective role of Calebin A against Aβ associated toxicity in neuronal cells, several Calebin A derivatives have been synthesized (Figure 10).

**Figure 10.** Structures of Calebin A and its derivatives.

Compound **23** showed that the hydroxyl group is important to protect cells from Aβ25-35 associated toxicity [137]. Calebin A derivatives are usually obtained by protecting the aromatic substituents [138]. Compounds **24** and **25** were synthesized following an alternative synthetic procedure, by substitution reaction of a cinnamic acid derivative on iodoketone in basic conditions. Both compounds were shown to protect neuronal cultures from tau oligomer-associated neurotoxicity by promoting the formation of larger nontoxic tau aggregates [87].

#### *5.11. Hybrid Compounds*

The bibliography about curcumin and curcumin-like compounds reports several studies on the design and biological evaluation of hybrid molecules. These compounds comprise curcumin fused with other biological active entities to enhance its biological and pharmaceutical properties (Figure 11).

**Figure 11.** Curcumin hybrid molecules.

Curcumin-melatonin hybrids were discovered as neuroprotective agents and have the potential to become new therapeutic strategies to treat AD. Melatonin as well as curcumin exhibit antioxidant, anti-inflammatory, and anti-Aβ aggregation activity [139]. Moreover, clinical studies have revealed that AD patients experience circadian dysfunctions due to the decreased melatonin levels in cerebrospinal fluid [140].

The presence of the methoxy group and the acetamide moiety is believed to be crucial for the neuroprotective activity of melatonin and its hybrids. Curcumin-melatonin hybrids without the p-hydroxy group of curcumin exhibited reduced neuroprotection. Contrarily, the double bound and conjugation of the β-diketone group with a phenyl ring is not necessary for the activity. The compound **26** inhibited the formation of Aβ oligomers and showed neuroprotective effects on MC65 cells [141].

Bivalent compounds were created by binding curcumin to the steroidal compound diosgenin using linkers with different lengths. The steroidal part acts as an anchor to localize curcumin into membrane lipid raft where ROS and Aβ oligomers are produced. Compound **27** with steroid moiety attached to the α-carbon of curcumin showed better neuroprotective property than those with a different position attachment. In addition, it exhibited antioxidant and anti-Aβ aggregation inhibitor activities [142]. Steroidal compounds **28** and **29** increased acetylcholine in the brain through inhibition of AChE activity and overexpression of choline acetyl transferase levels. Moreover, both compounds exert an antioxidant effect due to the phenolic groups combined to the methylene group of β-diketone group. The steroid portion can be involved in an interaction with the estrogen receptors and modulate the expression of antioxidant enzymes via intracellular pathway. Additionally, the increased levels of GSH may have a decisive role in the antioxidant effects observed. Both compounds exhibit an antiapoptotic activity as a result of the overexpression of the antiapoptotic factor Bcl2 in the brain of AD mice model [143]. Therefore, these hybrid molecules can be considered as an alternative approach to reverse the oxidative stress in neurodegeneration [143]. Additionally, bioconjugates of curcumin with demethylenated piperic acid, valine or glutamic acid display a protective effect against GSH depletion in dopaminergic neuronal cells and, therefore, may exert their antioxidant activities in the treatment of Parkinson's disease [144]. To increase curcumin water solubility, a sugar moiety was attached to the phenol residues. The hydroxyl groups of sugar can act as β-sheet breakers through competitive hydrogen bonding with amyloids fibrils. Sugar-curcumin conjugate **30** was shown to inhibit Aβ and tau aggregation [145].

To date, acetylcholinesterase inhibitors, such as Tacrine or Donepezil, are used in the treatment of Alzheimer's disease. Recently, new curcumin-hybrids were created by fusion of curcumin with these known drugs. Donepezil-curcumin hybrids **31** and **32** have a potent AChE inhibitor effect and both compounds showed metal chelating properties for Cu2+. Compound **31** also showed antioxidant activity in neuronal SH-SY5H cells. The phenyl group of the compound **31** interacts by stacking type interactions with Trp86 and the nitrogen atom binds to Tyr337 through cation-Π interaction. The carbonyl group forms a hydrogen bond with the main chain NH group of Phe295. On the other hand, compound **32** interacts with both catalytic site (CAS) and anionic site (PAS) of AChE. The two phenyl groups and the hydroxyl group interact with the active site. The hydroxyl forms an H-bond with the carbonyl group of Ser286 and Arg289, while the phenyl ring on the piperidine part formed Π-Π interactions with Trp84. The piperidine ring is located on the aromatic pocket connecting PAS and CAS sites [146,147]. Tacrine-curcumin hybrid **33** has been evaluated as an AChE inhibitor and it showed a higher potency than Tacrine [148]. Indeed, docking simulations showed that the benzene ring binds to the catalytic site of AChE, while the Tacrine motif interacts with the anionic site. The carbonyl group forms an H-bond with the Tyr121 thus stabilizing the complex. Therefore, due to the ability to inhibit both the catalytic as well as the anionic sites, compound **33** exhibited the most potent activity. Moreover, the curcumin moiety is responsible for the antioxidant properties, while the β-diketone moiety confers remarkable ion-chelating ability [148].

Biological activities of compounds discussed in this paragraph are presented in Table 1.


**Table 1.** Summary of biological activities of compounds showed in Figures 1–11.

#### **Table 1.** *Cont.*


#### **6. Conclusions**

After more than two centuries from the first discovery of curcumin, its clinical applications are still under investigation. Many potential applications of this compound are envisioned but its administration and metabolic fate need to be still deeply investigated. To overcome some issues related to curcumin use as a drug candidate and address its low bioavailability, many curcumin derivatives were synthesized and tested. The field of neuroprotective compounds accounts for hundreds of different derivatives and many of them are promising drugs for the treatment of AD and related diseases. In the next years, the clinical development of these compounds will assess the real effectiveness of curcumin as a lead compound for the synthesis of novel neuroprotective drugs.

**Author Contributions:** All authors contributed to paper selection and discussion. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research and APC were funded by MIUR, within the "FIRB-Futuro in Ricerca 2012" Program-Grant Project RBFR12SIPT. Mitchell Center for Neurodegenerative Diseases, and National Institute of Health grants: R01AG054025, R01NS094557 (R.K.).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Critical Molecular and Cellular Contributors to Tau Pathology**

**Liqing Song, Evan A. Wells and Anne Skaja Robinson \***

Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA; liqing@cmu.edu (L.S.); eawells@andrew.cmu.edu (E.A.W.)

**\*** Correspondence: anne.robinson@cmu.edu; Tel.: (+1)-412-268-7673

**Abstract:** Tauopathies represent a group of neurodegenerative diseases including Alzheimer's disease (AD) that are characterized by the deposition of filamentous tau aggregates in the brain. The pathogenesis of tauopathies starts from the formation of toxic 'tau seeds' from hyperphosphorylated tau monomers. The presence of specific phosphorylation sites and heat shock protein 90 facilitates soluble tau protein aggregation. Transcellular propagation of pathogenic tau into synaptically connected neuronal cells or adjacent glial cells via receptor-mediated endocytosis facilitate disease spread through the brain. While neuroprotective effects of glial cells—including phagocytotic microglial and astroglial phenotypes—have been observed at the early stage of neurodegeneration, dysfunctional neuronal-glial cellular communication results in a series of further pathological consequences as the disease progresses, including abnormal axonal transport, synaptic degeneration, and neuronal loss, accompanied by a pro-inflammatory microenvironment. Additionally, the discovery of microtubule-associated protein tau (*MAPT*) gene mutations and the strongest genetic risk factor of tauopathies—an increase in the presence of the ε2 allele of apolipoprotein E (*ApoE*)—provide important clues to understanding tau pathology progression. In this review, we describe the crucial signaling pathways and diverse cellular contributors to the progression of tauopathies. A systematic understanding of disease pathogenesis provides novel insights into therapeutic targets within altered signaling pathways and is of great significance for discovering effective treatments for tauopathies.

**Keywords:** tauopathies; Alzheimer's disease; prion-like propagation; tau self-aggregation; endocytosis; neuron-glial communication; neuroinflammation; apolipoprotein E

#### **1. Introduction**

Intraneuronal accumulation of neurofibrillary tangles (NFT) made of abnormally hyperphosphorylated tau is centrally involved in the pathogenesis of primary tauopathies, such as supranuclear palsy (PSP), corticobasal degeneration (CBD), Pick's disease (PiD), and frontotemporal dementia with Parkinsonism linked to chromosome 17 (FTDP-17), and secondary tauopathies such as Alzheimer's disease (AD) [1]. The development of tau pathology has been postulated to follow spatiotemporal patterns, starting from the dissociation of phosphorylated tau from microtubules and followed by the formation of toxic tau species via self-aggregation [2]. Even though polyanionic molecules are normally required for inducing tau aggregation in vitro, modifications to tau, such as site-specific mutations and site-specific phosphorylation, have driven spontaneous seeding and selfaggregation of tau in vivo under pathological situation [3]. Physiologically, extracellular tau is present in brain interstitial fluid (ISF) and then passes into the cerebrospinal fluid (CSF) [4,5]; however, the elevated concentrations of tau found in the brain ISF of human P301S tau transgenic mice has suggested that cellular tau release may be a part of disease progression [6]. Additionally, soluble tau concentrations in brain homogenates decrease with the deposition of intracellular insoluble tau, suggesting that transcellular tau propagation requires cellular internalization of extracellular tau, which has also been found to mediate the progression of neurodegeneration [6–8]. The cellular pathways for internalizing tau species are regulated by both heparan sulfate proteoglycan (HSPG)-mediated

**Citation:** Song, L.; Wells, E.A.; Robinson, A.S. Critical Molecular and Cellular Contributors to Tau Pathology. *Biomedicines* **2021**, *9*, 190. https://doi.org/10.3390/ biomedicines9020190

Academic Editor: Lorenzo Falsetti

Received: 18 January 2021 Accepted: 11 February 2021 Published: 14 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

cellular uptake and specific receptor-mediated endocytosis, which are highly dependent on the isoform being internalized [8–10].

Extensive experimental data have demonstrated that transcellular propagation of soluble tau species occurs mainly through synaptic connections, leading to neuronal dysfunction characterized by the breakdown of cytoskeletal integrity, abnormal axonal transport, and synapse loss [9,10]. In particular, glial cells, activated microglia, and reactive astrocytes are also involved in the progression of tau pathology by directly affecting the homeostasis of the neuronal microenvironment or indirectly exerting inflammatory effects across multiple tauopathies [11,12]. For example, the degree of glial cell activation correlates with the severity of neurodegeneration in AD, in terms of the degeneration of synapses, neuronal loss, the formation of NFTs, or even cognitive impairment [13]. Alternatively, dysfunctional neuron-glial communication has been widely observed in AD patients and has recently developed in vitro tau pathology animal models [14,15]. Abnormal neuron-glial crosstalk strongly impairs neuronal homeostasis including neuronal metabolism, synaptogenesis, neurotransmission, and neuromodulation, contributing to the progression of neurodegeneration [14,16]. The investigation of critical molecular and cellular contributors to tau pathology provides a comprehensive understanding of tau pathogenesis that will accelerate the discovery of novel therapeutic targets and the development of drugs for treating tauopathies.

The purpose of this review is to summarize the factors that contribute to the formation of tau aggregates, tau cell-to-cell propagation, and glial contributions in tauopathies, by using the scientific evidence published in the last decade that bring promising insights into the therapeutic development for tau protein pathology. Keywords for this topic, such as tauopathies, Alzheimer's disease, prion-like propagation, tau self-aggregation, endocytosis, neuron-glial communication, neuroinflammation, and apolipoprotein E were first chosen, and searches conducted in PubMed, Google Scholar, and Web of Science. The results of these searches were then refined and categorized into cellular contributors at the early stage and later stage of neurodegeneration, based on the characterized Braak-like spatiotemporal staging scheme for tau pathology. Lastly, the combined keywords search strategy was used for searching for potential treatments for tauopathies such as using the affected signaling pathway and tau phosphorylation together. The pathological roles of phosphorylation sites, Hsp90 and site-specific mutations in tau aggregation, the roles of CX3CR1/fractalkine signaling in microglia and neurons, the roles of the glutamate-glutamine cycle between astrocyte and neurons in the progression of tau pathologies, and the possible therapeutic role of NLR3 inflammasome in the treatment of tauopathies are the major focus of this review. A list of the abbreviations used in this review is provided in Table 1.

**Table 1.** Table of abbreviations used in this review.



**Table 1.** *Cont.*

#### **2. Factors Involved in the Formation of Tau Seeds**

The formation of NFTs from soluble tau is a multistep process. This process begins with the dimerization of two conformationally altered monomers and is followed by the formation of intermediate soluble oligomers with varying higher-order conformations and degrees of phosphorylation. Tau oligomers have been implicated as toxic 'tau seeds' capable of seeding new aggregates by recruiting normal monomers. Despite evidence of tau trimers being the minimal unit of spontaneous cellular uptake and intracellular fibrillary structure formation in vivo [17], the folding potency of monomer could be much more critical in initiating the early nucleation process of tau aggregation (Figure 1).

**Figure 1.** The molecular mechanisms involved in tau aggregation. Molecular factors such as site-specific phosphorylation, site-specific mutations on MAPT, and specific chaperones (Hsp90) are associated with tau aggregation.

#### *2.1. Site-Specific Phosphorylation-Mediated Tau Self-Aggregation*

Previous work details that although tau itself is intrinsically disordered, proteins in solution possess a 'paperclip-like' conformation where the N- and C-terminal ends of tau fold over in proximity to the center of the repeat domains [18]. Site-specific phosphorylation directly influences the conformation of monomeric tau and affects the stability of a folded conformation, contributing to the propensity for tau to aggregate [19]. Two hexapeptides, known as PHF6s, 275VQIINK280, and 306VQIVYK311, are located at the beginning of the second and third repeat domains of the MBDs, and appear to drive β-sheet structure formation during the tau aggregation process. The accessibility of residues in the two PHF6s defines the structural differences between inert (Mi) and seed-competent (Ms) tau monomer, meaning that the inert (Mi) tau monomer has less inter-chain accessibility to these residues compared with that in the seed-competent (Ms) monomer [2]. Phosphorylation outside of, but proximal to, these regions is relevant to the formation of

NFTs. A previous study systematically investigated the effects of different phosphorylation sites on tau self-aggregation, using a series of in vitro pseudo-phosphorylated tau proteins [20]. Phosphorylation sites T175/T176/T181 within N-terminal, recognized by AT270 antibody, mainly suppress tau aggregation [21]. In addition, phosphorylation at three sites, S202/T205/S208, within the proline-rich region (PRR) is enough to induce tau self-aggregation without any exogenous aggregation inducer [22]. The monoclonal antibody AT8 that specifically recognizes tau phosphorylation at the S202/T205 site has been established as a valid biochemical marker for identifying abnormally phosphorylated tau as well as the paired helical filament form. Moreover, phosphorylation sites near the C-terminus have been found to preferentially promote tau self-aggregation. For example, pseudo-phosphorylated S396, specifically recognized by PHF-1 antibody [21], has led to increased tau aggregation in the presence of metal ion inducer. In particular, the strong effect on aggregation has been seen in pS422 tau protein, which showed increased aggregation in the presence of both metal ions and heparin inducers [20], which may be related to the conversion of tau monomer from inert to seed-competent form, due to increased accessibility of these residues [3], as shown in Figure 1. By performing a comprehensive electrochemiluminescence ELISA assay, Ercan-Herbst et al. [23] found that specific phosphorylation events (pS198, pS199, and pS416) correlated with increased oligomerization in all brain regions, which implies that phospho-sites regulate tau aggregation during the progression of AD neurodegeneration. Collectively, phosphorylation plays a major role in tau self-aggregation by altering the charge and conformations of physiological tau.

#### *2.2. Hsp90-Mediated Tau Aggregation*

Tau phosphorylation and aggregation that lead to conformational changes could involve molecular chaperones, which regulate protein folding, degradation, and accumulation. The protective effect of Hsp70 and Hsp104 in tauopathies has been described in previous studies [24]. Hsp70 inhibits the aggregation of tau protein by forming a complex with tau oligomer or fibril tau, preventing toxic effects or further seeding of tau aggregation [25,26]. Despite the recognition of its disaggregase activity for many aggregates, a distinct mechanism of Hsp104 in preventing tau aggregation is related to its holdase activity on soluble amyloid tau through the small subdomain of nucleotide-binding domain 2 (ssNBD2) [27].

In contrast to the preventative functions of Hsp70 and Hsp104, heat shock protein 90 (Hsp90), one of the major tau-binding chaperones, has been found to drive the aggregation of tau species [28]. Although Hsp90 is normally thought to act as cellular protection during stress, Hsp90 binding to tau at the VQIVYK motif facilitates a conformational change that results in its phosphorylation by glycogen synthase kinase 3, which further promotes tau aggregation [28]. Additionally, a recent study found that Hsp90 binding to tau uncovered the repeat domains by conformationally opening the 'paper-clip' structure of tau, suggesting that the formation of tau oligomers was caused by the conversion of tau monomers from inert to aggregation-prone forms [29].

#### *2.3. Site-Specific Mutations and Tau Aggregation*

Abnormal tau mutants related to FTDP-17 possess distinct structures leading to a differential aggregation propensity [30–32]. Recently, Strang and coworkers demonstrated that the susceptibility of FTDP-17-associated mutants to aggregate with seeded, exogenous fibrillar tau depended highly on site-specific mutations and their surrounding amino acid sequences [33]. Robust aggregation with exogenous tau fibril seeds, both homotypic and heterotypic, has been seen in FTDP-17 mutations at sites P301 and S320. In particular, the unique property of the P301L variant in regulating the aggregation propensity of tau has been demonstrated by mutating individual proline residues into leucine residues within conserved PGGG motifs in each of the four MTBDs in tau [33]. Only P301L showed a propensity to aggregate when seeded with exogenous fibrillar tau. In contrast, other FTDP-17-associated variants near the PHF6 site showed no propensity to aggregate when

seeded. Double mutants at P301L/S320F and P301S/S320F have been shown to facilitate aggregation. For these P301L/S320F and P301S/S320F tau protein variants, robust aggregation was observed in vivo without exogenous fibrillar tau seeding [33]. A possible underlying cause of this enhanced aggregation propensity is altered conformation with higher accessibility to PHF6, that converts the inert monomer into aggregation-prone monomer; alternatively, more frequent interactions with chaperones may be required to stabilize a folded conformation for these variants. In either case, further investigation is needed to identify the mechanism.

#### **3. Molecular Mechanisms of Tau Cellular Uptake**

Transcellular tau propagation has been implicated in tauopathies following a 'prionlike' transmission pattern [34], suggesting that the internalization of extracellular tau by recipient cells is mediated mainly by endocytosis. Recent studies showing distinct features of prion-like propagation of tau species under diverse cell and animal models are summarized in Table 2. Endocytosis can be divided generally into clathrin-dependent and -independent internalization, of which the latter can be further divided into caveolindependent, -independent endocytosis, and actin-dependent macropinocytosis. Previous studies highlighted cellular internalization pathways associated with tau including bulk endocytosis [35], heparin sulfate proteoglycan (HSPG)-associated macropinocytosis [8], and clathrin-mediated endocytosis [35].

The majority of extracellular tau consists of soluble oligomers and monomers, while a minority of tau species exist in truncated forms cleaved by various proinflammatory cytokines in AD brains [6]. The size and conformation of tau species determine the cellular mechanisms for extracellular tau uptake, which may not be restricted to one particular pathway [8,36]. For instance, smaller sized tau aggregates enter neurons in a dynamin-dependent endocytosis pathway that is independent of actin polymerization [35]. For larger tau aggregates, actin-dependent macropinocytosis has been identified as the main pathway for internalization by neuronal cells [37]. However, the cellular entry pathways of monomeric tau are highly dependent on the specific conformation and isoform. A recent study demonstrated that monomeric tau could enter human neurons via both the dynamin-dependent endocytosis process and through actin-dependent macropinocytosis, which could be regulated by HSPGs [35].

#### *3.1. The Effects of HSPGs on the Cellular Uptake of Tau Seeds*

HSPGs are highly expressed on the cell surface and have been identified as critical cell-surface endocytosis receptors for tau internalization in various studies. Most recent research has focused on understanding the interaction of heparan sulfate (HS) with tau protein at the structural level, which would provide a mechanistic understanding of how tau-HS interaction regulates tau internalization during the progression of tau pathologies. HS-tau interactions appear to be driven mainly by electrostatic forces between negatively charged sulfo groups on HS and positively charged lysines or arginines on tau protein [38]. Even though electrostatic interactions between tau and HS are relatively nonspecific, a few studies have also identified the importance of specific HS sulfation patterns on the tau-HS interaction. Prior works demonstrated the crucial role of the 6-O-sulfation of HSPGs in the tau-HS interaction by performing an SPR competition assay [39]. Moreover, 6-O-desulfated heparin showed the weakest competitive effect on tau binding to heparin immobilized on a chip among a variety of HS derivatives tested, including N-desulfated and 2-O-desulfated HS derivatives. NMR mapping showed that HS derivatives bound the second repeat motif (R2) in tau. Consistently, a knockout of 6-O-sulfotransferase also significantly reduced tau uptake by HEK293 cells [40]. Reduced intracellular tau uptake and tau cell surface binding in a 3-O-sulfotransferase knockout cell line compared with the wild-type cells suggest that tau protein is capable of recognizing the less common 3-O-sulfation site of HS [41]. The importance of sulfation was further validated in competition assays performed by Zhao and coworkers—3-O-sulfated low molecular weight HS (LMWHS) oligosaccharides

had higher inhibitory effects on the tau-HS interaction compared with those without sulfation in an SPR competition assay, further validating the specific role of 3-O-sulfation in tau-heparin interactions. Furthermore, 3-O-sulfation is rare and minor sulfation is found on HS chains, which is likely not responsible for any charge effects in HS chains. More likely, HS interacts with tau via a specific 3-O-sulfation of HS recognized by both the PRR2 and R2 regions of tau instead of non-specific electrostatic interactions. The tau-heparin interaction has also been found to be chain size-dependent due to enhanced electrostatic interactions [40]. Knockouts of extension enzymes of the HSPG biosynthetic pathway, such as extension enzymes exostosin 1 (*EXT1*), exostosin 2 (*EXT2*), and exostosin-like 3 (*EXTL3*) in HEK293T cells significantly reduced the uptake of tau oligomers [36].

HSPGs can be considered as the natural receptors for the uptake of macromolecules, such as larger tau fibrils, through the micropinocytosis pathway; nevertheless, the exact role of HSPGs in the uptake of tau species dominated by the clathrin-mediated pathway needs further investigation within specific systems. Key questions include whether HSPGs are part of a multi-receptor complex or merely an initial attachment site during tau uptake. Moreover, HS-modifying enzyme expression patterns show cell-type-specific patterns, resulting in enormous HS diversity because of the many different cell types in the brain. Because of the heterogeneity of HS-expression, the specific role of HSPGs in tau internalization should be investigated on a cell type-specific basis.

*Biomedicines* **2021**, *9*, 190


**Table2.**Asummaryofrecenttautransmissionmodels.

#### *3.2. Receptor-Mediated Endocytosis of Tau*

Apart from HSPG-dependent uptake, cellular internalization of tau is also regulated by specific receptor-mediated endocytosis, as suggested by several previous studies [35,44]. Rapid dynamin-dependent endocytosis of tau species would typically require one or more receptors, the identities of which are still under investigation. Muscarinic receptors M1/M3 have been found to regulate monomeric tau internalization by neurons [43]. Glial cells including microglia and astrocytes also take up tau efficiently. CX3CR1 has been demonstrated to mediate monomeric tau uptake in microglia [45]. For astrocytes, monomeric tau was internalized in a non-HSPG dependent pathway [46]; further study is still needed to identify specific receptors responsible for rapid dynamin-dependent endocytosis of monomeric tau in astrocytes (Figure 2). Low-density lipoprotein receptorrelated protein-1 (LRP1) represents a promiscuous endocytosis receptor for macromolecular ligands, including ApoE and Aβ, and delivers these ligands to the endosomal/lysosomal compartments. Knockdown of LRP1 abolished uptake of various forms of tau, including monomers, oligomers, and fibrils in H4 neuroglioma cells, suggesting that it may serve as a master regulator of tau uptake [44]. Additionally, knocking down LRP1 also prevented tau transmission within human tau transgenic mice. Once associated with specific ligands, LRP1 is also involved in the activation of signaling pathways including MAPK, by assisting the assembly of the intracellular protein complex [47]. LRP1 is also abundantly expressed by radial glia, microglia, and astrocytes, and involved in the clearance of Aβ [47,48]. Further studies are still needed to identify whether and how LRP1 is involved in tau endocytosis by glial cells, and whether the tau-LRP1 interaction alters the immune response of glial cells.

**Figure 2.** Receptor-mediated endocytosis of tau species is facilitated by several receptors. Central nervous system cells actively internalize monomeric tau via receptor-mediated endocytosis in addition to the HSPG-dependent pathway. Monomeric tau is internalized by muscarinic receptors M1 and M3 in neurons [43]. CX3CR1 mediates monomeric tau uptake in microglia [45]. For astrocytes, monomeric tau can be internalized in a non-HSPG dependent pathway [46]. Further work should be focused on identifying specific receptors of tau endocytosis. Additionally, LRP1 has recently been identified as a major regulator of tau spread in the brain [44]; LRP1 is abundantly expressed by microglia, astrocytes, and neuronal cells [47,48].

#### **4. Cellular Contributors to Tau Pathology**

In 1991, the work of Braak proposed the sequence of progression of Alzheimer's disease neuropathology, demonstrating that soluble hyperphosphorylated tau first appears in the locus coeruleus (LC) neurons and subsequently appears along LC axons to their terminals in the entorhinal cortex (EC) [49,50]. Transgenic mice models that display human tau pathology have been established to recapitulate the development of neurodegeneration and diverse pathological phenotypes, including gliosis, synaptic loss, tangles, and neuronal loss (Figure 3A). These models also demonstrate the involvement of diverse cellular contributors, including neuronal cells, microglia, and astrocytes, to the progression and spread of tauopathies (Figure 3B).

**Figure 3.** The interplay between different cell types including neurons, microglia, and astrocytes, in tauopathy progression. (**A**) A representative human tau pathology model: PS19 transgenic mice express mutant human *MAPT* with a P301S mutation and display a series of pathological features of tauopathies, such as gliosis, synaptic loss, tangles, and neuronal loss over their lifetimes. Adapted from [51,52]. Note that in this model, no plaques were found; LTP, long-term potentiation; LTD, long-term depression. (**B**) Both microgliosis and astrogliosis are involved in the progression of tauopathies and abnormal neuronal activities, as indicated by phenotypic characterization of human tau pathology models [53]. Copyright © 2021, John Wiley and Sons.

As described in detail in the following sections, the development of tau-related pathologies has been postulated to follow spatiotemporal patterns and is characterized by multiple progressive stages, each with pathological features in the form of differential cellular behaviors and distinguished phenotypes (Figure 4). At the earliest stage, tau seeds formed by phosphorylated tau dissociate from microtubules spread along a transsynaptic pathway, involving the release of tau species in the synaptic cleft, with subsequent internalization by post-synaptic neurons [54]. Glial cells, on the other hand, adapt a neuroprotective phenotype with microglia classically activated to engulf tau species in a CX3CR1-dependent way [55], and astrocytes actively involved in clearing tau species with an exacerbated autophagy-lysosomal pathway (ALP) [56]. As the disease progresses, astrocytes display a 'loss-of-function' phenotype by exhibiting a decreased level of glutamate transporters, leading to neuronal excitotoxicity and upregulated tau release [57]. Additionally, microglia develop an alternative pro-inflammatory phenotype after responding to diverse pro-inflammatory stimuli, including higher concentrations of tau protein and the presence of reactive oxygen species [11]. These activated microglia continue to produce proinflammatory cytokines, such as TNF-α and IL-1β, which are necessary and sufficient to convert inactive astrocytes into reactive astrocytes, resulting in further neuroinflammatory cytokine release [58].

**Figure 4.** Cellular contributors to tau-dependent degeneration. The development of tau-related pathogenesis has characteristic stages, starting from the formation of tau species consisting of phosphorylated tau dissociated from microtubules ➀. Abbreviation: pTau, phosphorylated tau. Hyperphosphorylation of tau incorrectly sorts tau into the somatodendritic compartment, which is linked to dysfunctional axonal transport ➁, one of the earliest pathological features of tauopathies. Glial cells adapt a more neuroprotective phenotype with microglia classically activated ➂ and astrocytes actively phagocytosing tau species ➃. As disease progresses, astrocytes transform into a loss-of-function phenotype via lower-level expression of astrocyte-specific transporters, leading to neuronal excitotoxicity and upregulated tau release. Additionally, alternatively activated microglia in a pro-inflammatory phenotype are necessary and sufficient to induce reactive astrocytes with the capability of releasing neuroinflammatory cytokines ➅. At the later stages of disease progression, the microglial-exosomal pathway acts as the essential tau propagation pathway ➇ as an alternative to transsynaptic transduction, due to extensive synaptic degeneration and neuronal death ➈. The formation of neuronal and glial tau plaques is the most important hallmark of tauopathies ➉. Red arrows indicate pathological consequences of change. Created in BioRender.com.

Activation of the NLRP3 inflammasome in microglia has been demonstrated to facilitate the progression of tau pathologies, mainly through intensifying neuronal tau hyperphosphorylation in an IL-1 receptor-dependent way [59]. At the late stage of disease progression, dysfunctional synaptic transmission caused by synaptic loss and neuronal death leads to microglia-exosomal tau transmission that takes precedence over transsynaptic tau transmission [60].Finally, neuronal and glial tau plaques are formed, which are the most important hallmarks of tauopathies [9].

Overexpression of the ε4 allele of apolipoprotein (*ApoE4*) in multiple cell types shows cell-type-specific effects; overexpression in neuronal cells upregulates neurotransmitter release while enhancing inflammatory signaling of microglia. For astrocytes, ApoE4 overexpression downregulates phagocytosis of pathogenic proteins and disrupts lipid transport and metabolism. Taken together, ApoE4 serves as a common genetic risk in AD and primary tauopathies, and can worsen tau pathology, indicating an overlap between ApoE4 and tau pathogenesis. LRP1, as a major receptor of tau species and ApoE, may play an intermediate role between ApoE and tau species, which could point to a therapeutic potential for treating tauopathies via LRP1 interaction.

#### *4.1. The Involvement of Neuronal Activity in the Spreading of Pathogenic Tau*

Under physiological conditions, tau is crucial for microtubule stabilization and is located mainly in axons [61]. Immunoblot analysis with phosphorylation-dependent antibodies revealed that phosphorylated tau is missorted into the somatodendritic compartment during the early stages of AD progression [62]. Missorted tau results in axonal transport deficits and loss of synaptic functions and is more prone to forming toxic tau oligomers if seeded [62].

The progression of tau pathology follows a defined hierarchical pattern, starting from the EC, then advancing into anatomically connected neurons downstream in the synaptic circuit, such as the dentate gyrus (DG), the hippocampus, and the neocortex, as demonstrated by tau transgenic animal models [54,63]. Despite the identification of the physiological role of neurons in regulating synaptic tau release and translocation [5], the specific neuronal activities resulting in the propagation of tau pathologies are still under investigation. Amyloid precursor protein (*APP*) transgenic mouse models show that endogenous tau in CSF increases during the progression of amyloid plaque formation, accompanied by hyperexcitable neurons [64,65]. A key question is whether the hyperexcitable neurons are essential for the release of pathogenic tau, independent of Aβ. Indeed, tau pathology mouse models combined with novel neuronal stimulation approaches showed that neuronal hyperexcitability and accelerated synaptic tau release are critically linked and independent of Aβ toxicity [66]. Using an optogenetic activation approach, the stimulated side of the hippocampus of the rTg4510 mice line tended to accumulate more human tau protein, along with increased evidence of neuronal atrophy [66]. Additionally, tau pathology spread from the stimulated-EC to the synaptically connected DG region, suggesting that the propagation of tau pathology accelerates through synaptic circuits.

Abnormal extracellular glutamate levels have been proposed as one of several mechanisms that account for an excitotoxic microenvironment in AD [67]. Notably, alterations in synaptic glutamate homeostasis caused by dysfunctional astrocytes can be deleterious to neuronal cells. To some extent, the activities of reactive astrocytes correlate with the reduction in astroglial glutamate transporters, which in turn elevates the extracellular glutamate level. Accumulation of excess glutamate contributes to neuronal excitability through activating NMDA (N-methyl-D-aspartate) receptors. NMDA receptors, present in glutamatergic neurons, respond to the glutamate levels via binding to their GluN2 subunit that activates increased calcium flux in the neurons [57]. Sequentially, activation of extrasynaptic NMDA receptors has been linked to tau-induced neuronal cell death mediated by calpain I and ERK/MAPK activation [68]. Therefore, alteration of astroglial glutamate transporters and overstimulation of extrasynaptic NMDA receptors of neuronal

cells may have an overlapping role in neuronal hyperexcitability, and these actions have been implicated in the progression of tau pathology along with synaptic connections [69].

#### *4.2. Glial Cells Are Involved in the Progression of Tau Pathology*

Even though tau is expressed primarily by neurons, most primary tauopathies are characterized by the presence of both neuronal and glial tau pathologies [12]. Glial cells adopt immune functions and closely interact with neuronal cells for maintaining brain development and homeostasis [70]. Most glial tau pathologies have been observed in astrocytes and oligodendrocytes, and in some cases, tau pathologies have also been seen in microglia. Moreover, both primary tauopathies and AD are characterized by microgliosis and astrogliosis, along with a significant increase in the pro-inflammatory cytokines [6,71]. Glial cell dysfunction has also been implicated in the progression of neurodegenerative diseases [51]. This part of the review aims to highlight the role of dysfunctional neuronalglial communication in the spreading and propagation of pathological tau during the progression of tauopathies.

#### 4.2.1. Microgliosis in Tauopathies

#### Neuroprotective Effects of CX3CL1/CX3CR1 Signaling

Microglia are the innate immune cells of the CNS and account for 5–20% of total neural cells in the functional tissue of the brain [72,73]. They have two main CNS functions: immune defense and maintenance and promoting programmed cell death during development [72,73]. Recently, microglia-induced neuroinflammation has been linked to tau hyperphosphorylation, suggesting that microglia play an important role in the progression of tau-related neuropathogenesis [74]. As discussed previously, extensive studies have demonstrated that tau pathology predominantly spreads along with synaptic connections. Physiologically, microglia control and regulate synaptic plasticity through pruning of inactive synapses via phagocytosis during CNS development [75]. Among the key factors emerging as potential regulators of neuronal-microglial interaction, chemokine ligand 1 (CX3CL1) secreted by neurons plays an essential role in regulating phagocytic capability of microglia by binding to CX3CR1 [76], a key receptor that maintains the normal synaptic pruning ability of microglia [77]. Altered CX3CL1/CX3CR1 signaling has been demonstrated to regulate the pathological changes in both animal models of tauopathies and AD patients [78,79]. Single-cell RNA-seq of microglia in AD-transgenic mouse brains shows that CX3CR1 is upregulated as part of the initial innate immune response [80], which facilitates the internalization of tau by microglia to enhance the clearance of extracellular Tau [55].

However, at the later stages of AD, CX3CR1, among many other genes, is downregulated [80]. The downregulation of CX3CR1 has also been observed in human brain tissue from AD patients, showing that CX3CR1 levels decrease as microglial phagocytic phenotypes are reduced [55]. Microglia have been found to phagocytose extracellular tau oligomers directly via the tau-CX3CR1 interaction, which is impaired by the loss of CX3CR1 at the later stages of AD. The deletion of CX3CR1 in models of tau pathology has accelerated tau phosphorylation and exacerbated neurodegeneration [55,81]. This CX3CR1 deficiency led to elevated levels of tau phosphorylation on the AT8 (pS202), AT180 (pT231), and PHF1 (pS396/S404) epitopes [58], which is mediated by neuronal IL-1 and TLR-4 receptors triggered by the microglial release of proinflammatory cytokines [82]. Indeed, the deletion of CX3CR1 in the hAPP-transgenic mice model exacerbates microglial inflammation and neurotoxicity by upregulating the secretion of proinflammatory cytokines [78]. Similarly, CX3CL1 overexpression in the human tau transgenic mouse model rTg4510 significantly reduced neurodegeneration and microglial activation [83]. Therefore, the investigation of CX3CR1-CX3CL1 signaling has provided novel insights for treating tauopathies.

The Role of NLRP3-ASC Inflammasome Activation in Tau Phosphorylation

Extracellular fibrillary Aβ-induced microgliosis has been linked to NOD-like receptor family, pyrin domain containing 3 (NLRP3) inflammasome activation, which further exacerbates Aβ pathology [84]. The role of microglia and NLRP3-caspase recruitment domain (ASC) inflammasome activation has been demonstrated recently in Aβ-independent tau pathology [59]. Phagocytosis of fibrillar Aβ induced the assembly of the NLRP3 inflammasome consisting of NLRP3, ASC, and pro-caspase 1, which led to the caspase 1-dependent release of pro-inflammatory cytokines such as IL-1β and IL-18 [85]. Stancu and colleagues [86] demonstrated that aggregated tau was capable of activating the NLRP3 inflammasome, which further exacerbated the tau aggregate seeding and increased the secretion of proinflammatory cytokines. The importance of NLRP3 on the progression of tau pathology also was demonstrated in tau transgenic mice models deficient for NLRP3 or ASC. A significantly lower level of tau phosphorylation was observed in the hippocampal samples of the transgenic mice deficient for NLRP3 or ASC compared with wild-type mice [86]. Additionally, templated seeding of tau pathology was reduced in tau transgenic mice with an ASC deficiency.

Reduced activities of GSK-3β and CaMKII-α, but not p38/MAPK and Cdk5, were correlated with the deficiency of NLRP3 or ASC, suggesting the potential role of the NLRP3 inflammasome in regulating tau kinases in neuronal cells [86]. To understand how the NLRP3 inflammasome regulates tau phosphorylation, conditioned medium collected from LPS-activated microglia induced an increased level of tau phosphorylation in neuronal cells, along with the activation of CaMKII-α [59]. However, once the neuronal IL-1 receptor was inhibited, the effects on CaMKII-α were abolished, suggesting that the activation of the NLRP3 inflammasome in microglia promotes neuronal tau hyperphosphorylation in an IL-1 receptor-dependent manner via the regulation of multiple tau kinases (Figure 5). Potential therapeutic interventions targeting the NLRP3 inflammasome have been attempted for treating AD in mouse models [87]. By increasing *ASC* and *NLRP3* gene expression in Tau22 transgenic mice, the formation of tau aggregates was attenuated, as determined by thioflavin T staining and reduced tau phosphorylation at serine 416, due to diminished CaMKIIα activity [87]. Pharmacological NLRP3 inhibition using the molecular inhibitor MCC50 also significantly decreased tau-seed induced tau aggregates, as determined by AT8 detection, in tau transgenic mice [86].

**Figure 5.** The role of the NLRP3-ASC inflammasome in tau pathogenesis. Either fibrillary Aβ or tau species in the form of monomers or oligomers are sufficient to induce the assembly of the NLRP3 inflammasome, consisting of NLRP3, ASC, and pro-caspase 1, which further leads to caspase 1-dependent release of the pro-inflammatory cytokines IL-1β and IL-18 [85]. The activation of the NLRP3 inflammasome in microglia has been demonstrated to promote neuronal tau hyperphosphorylation in an IL-1 receptor-dependent manner via the regulation of multiple tau kinases, like GSK-3β and CaMKII-α.

Even though these studies demonstrate the involvement of neuroinflammation and altered CX3CR1-CX3CL1 signaling in the spreading of tau pathology, further investigation is still needed to uncover the interplay between neuroinflammation induced by extracellular tau aggregates and disrupted phagocytosis caused by impaired CX3CR1-CX3CL1 signaling. Most likely, the relationship between inflammation and phagocytosis will demonstrate the crucial role of microglia in the development of tau pathology.

Non-Transsynaptic Tau Propagation-Microglial and Exosomal Spreading of Tau Species as an Alternative Pathway

In addition to the important role of CX3CR1-CX3CL1 signaling and the NLRP3 inflammasome in tau pathogenesis via the crosstalk between neurons and microglia, exosomes are another key mediator between glial-neuronal communication for both synaptic pruning in the healthy brain as well as neuroinflammation under pathological conditions [88,89]. Physiologically, neuronal exosomes stimulate microglial phagocytosis under selective elimination of synaptic connections. Microglia-derived exosomes play a major role in hierarchical tau transmission [60], despite pathogenic tau readily propagating from neuron to neuron in the form of free-floating fibrils [34] and interconnected neuronal contacts [90], as well as neuronal exosomes [4].

Recently, a tau rapid-propagation mouse model was created with adeno-associated viral (AAV)-tau injection into the EC [60]. This model exhibits rapid tau pathology, as demonstrated by the spreading of human tau from EC to the DG within 1 month, recapitulating the perforant diffusion pathway of AD progression in the human brain [91]. Moreover, inhibition of exosome synthesis or depletion of microglia in this AAV-GFP/tau injection mouse model led to a dramatic reduction of AT8+ tau detected in the DG without changing the tau expressed in the injection site, indicating the important role of microglia-derived exosomes in the spreading of tau pathology. Pharmacologic inhibition of exosome synthesis in microglia not only dramatically reduced secretion of the tau-containing exosomes but also decreased the capabilities of exosomes to deliver hTau, as observed in co-cultured primary neurons [60]. As the synaptic connection becomes less functional throughout disease progression, the microglial and exosomal transmission pathways become the primary means of tau propagation [9], suggesting exosomal transmission as a potential therapeutic target.

#### 4.2.2. Astrogliosis in Tauopathies

The concept of astroglial excitability—activation of membrane ion receptors in response to stimulation—facilitates the bidirectional communication between neurons and astrocytes mediated by a 'tripartite synapse' [92]. The close physical proximity between synapses and astrocytes and resulting efficient neurotransmission explain why astrocytes are key regulators in maintaining essential neuronal functions, including synaptic plasticity and neurodevelopment [57]. Besides their crucial role in supporting neuronal functions in the CNS, astrocytes represent the largest group of glial cells that interact closely with microglia for maintaining efficient immune surveillance of the CNS [92]. Like microglia, astrocytes also express genes involved in phagocytosis [93], and eliminate synaptic debris [94], and protein aggregates, as seen by the clearance of Aβ [95]. In recent years, the involvement of astrocytes in the progression of tau pathology has drawn much attention because of their widely demonstrated role in the progression of neurodegeneration in tauopathies [11,57]. For example, reactive astrocytes induced by microglial activation have been observed to precede tangle formation in P301S tau transgenic mice models (PS19) [51].

#### Reactive Astrocyte Phagocytosis Has a Neuroprotective Effect

Under pathological conditions, astrocytes develop more neurotoxic features by transforming into reactive astrocytes (A1 subtype), induced by activated microglia and neuroinflammation in various human neurodegenerative disorders [16]. The phagocytic ability of reactive astrocytes appears to be enhanced in tau transgenic mouse models [96]. Astrocyte activation is accompanied by upregulated expression of transcription factor EB (TFEB), the key regulator of the autophagy-lysosomal pathway (ALP). When compared

with wild-type counterparts, two widely used tau pathology mouse models, rTg4510 and PS19, showed increased expression of TFEB and lysosomal protein LAMP1 [56]. In particular, astrocytes of rTg4510 mice (transgenic mice expressing human P301L tau protein) showed much higher nuclear localization of TFEB in GFAP-expressing astrocytes compared with the wild-type mice [96]. However, overexpression of astrocytic TFEB in rTg4510 showed minimal effects on neuronal activities. In vitro, TFEB overexpression in primary astrocytes led to enhanced cellular uptake of tau fibrils by stimulating lysosomal biogenesis. In contrast, the TFEB-transduced PS19 tau pathology mouse model showed reduced tau pathogenesis and reduced tau transmission compared to the rTg4510 mouse model. These data demonstrated that the neuroprotective effects of astroglial activation took place primarily at the early stage of tauopathies by enhancing endocytosis and subsequently, triggering intensive lysosomal-mediated degradation of abnormal tau species. The effects of ALP on regulating phagocytic properties of reactive astrocytes may be one of the mechanisms that explains why tau protein enters astrocytes more efficiently than neurons, as observed in prior work [97] and has been implicated in the glial inclusions, as seen in most of the primary tauopathies, including PSP, CBD, and PiD [97].

Neuroinflammatory Microenvironment Induced by Reactive Astrocytes

Reactive microglia secrete inflammatory cytokines such as IL-1a, TNF-a, C1q, and IL-1β [16,97]. These cytokines themselves are necessary and sufficient to induce the A1 subtype of astrocytes, which further stimulates inactive astrocytes in proximity. Reactive astrocytes lose their ability to promote neuronal survival, synaptogenesis, and phagocytosis. Enhanced release of inflammatory cytokines from activated glial cells can induce active neuronal p38 MAPK by interacting with multiple receptors, such as TNFR1, and lead to enhanced phosphorylation and aggregation of tau, which precedes the progression of tau pathology [98]. Thus, reactive astrocytes also play a role in neuroinflammation-induced tau pathology.

#### Dysfunctional Neuronal-Astroglial Communication

Reactive astrocytes also exhibit neurotoxicity by impairing glutamate transport between neurons and astrocytes and disturbing the synaptic neurotransmitter balance via direct contact [57,99]. Astrocytes are key regulators for maintaining homeostasis of major neurotransmitters like glutamate (Glu) and γ-aminobutyric acid (GABA) via the glutamineglutamate/GABA cycle [99]. The rapid uptake of tau species by reactive astrocytes disrupts intracellular Ca2+ signaling, leading to a significant reduction in the release of gliotransmitters such as glutamate, glutamine, and serine, and formation of synaptic vesicles [97]. Moreover, it has also been reported that conditioned medium (CM) collected from primary astrocyte cultures isolated from P301S mice decreased the expression of synaptic neuronal markers in cultured cortical neurons, while CM from control astrocytes enhanced these markers in co-cultured neurons [100]. Taken together, reactive astrocytes appear to affect neuronal tau pathologies by impairing the neuronal synaptic transmission as well as synaptic plasticity.

The glutamine (Gln)/glutamate (Glu) cycle (GGC) is critical for maintaining homeostasis of the major neurotransmitters Glu and γ-aminobutyric acid (GABA), which is a key metabolic interaction between neurons and astrocytes [57,101]. Astrocytes uptake excess Glu released by glutamatergic neurons in the synaptic cleft via glutamate transporter-1 (GLT1) receptor (see Figure 4). Glu is then converted into Gln by astrocyte-specific enzymes, and released into the extracellular space. Subsequently, Gln is taken up by neurons and metabolized into Glu by neuron-related enzymes. Recent studies have reported that the dysfunctional neuronal-astroglial communication via the GGC may contribute to tau protein pathology [57]. For instance, reduced expression of astrocytic glutamate transporters, such as GLT1, has been found to coincide with tau inclusion pathology, as well as neuromuscular weakness in the spinal cord and the brainstem, as seen in both tau transgenic mouse models and CBD patients [99]. The reduction of glutamate transporters in astrocytes

also elevates extracellular glutamate levels that then further overstimulate glutamatergic receptors (NMDA receptors), causing increased calcium flux and more neuronal excitotoxicity [62]. Subsequently, activation of NMDA receptors has led to tau phosphorylation at specific sites, the most efficient being Ser-396, mediated by p38/MAPK activation [102]. The mechanistic effects of tau pathology on the downregulation of glutamate transporters and reduction of GLT1 in glial cells are unknown, but investigations on the involvement of astrocytes in the progression of tauopathies have provided novel insights for treating glial tau pathology.

#### *4.3. ApoE4 Plays a Cell-Type Dependent Role in Tau Pathology*

ApoE protein serves as a major cholesterol carrier in the brain, as well as helping to clear Aβ plaques. Among the three alleles for ApoE, the presence of ApoE4 is considered an important genetic risk factor for Alzheimer's disease, leading to tau hyperphosphorylation in an Aβ-dependent manner [103]. However, a key question is whether ApoE4 influences tau pathology in primary tauopathies, such as PSD, in which tau pathology is not accompanied by Aβ. Using a P301S transgenic mouse model, Shi et al. [104] have demonstrated ApoE4-induced tau pathology independent of Aβ pathology, suggesting the crucial role of cholesterol in the tau pathogenesis of primary tauopathies.

ApoE4 expressed by different cell types has been shown to affect tau pathologies in a cell type-dependent manner. ApoE4 toxicity has been observed in multiple cell types, including neurons, as demonstrated in human iPSCs-derived cell types [105] despite ApoE4 being primarily produced by microglia and astrocytes in the CNS. ApoE4-expressing neurons exhibited tau hyperphosphorylation, while ApoE4 glial cells had reduced capacity for neuronal homeostasis and thus contributed to the pathogenesis of tau pathology. For instance, Wang and colleagues [106] showed that ApoE4-expressing neuronal cultures derived from human iPSCs expressed higher levels of the synaptic proteins SYN1 and PSD95, alongside an elevated release of neurotransmitters, compared with ApoE3-expressing neurons. Additionally, ApoE4 cerebral organoids exhibited an elevated level of phosphorylated tau (p-S202/T205) compared with ApoE3 organoids. ApoE4-expressing astrocytes had impaired lipid metabolism/transport and phagocytosis, while ApoE4-expressing microglia exhibited intensive immune reactivities upon LPS stimulation compared with wild-type microglia.

Neurons co-cultured with ApoE knockout glial cells displayed the greatest neuronal viability with the lowest level of TNF-α secretion [104]. Analogously, co-culturing P301S tau-expressing neurons with ApoE4-expressing microglia resulted in markedly reduced neuronal viability and a significantly high level of TNF-α secretion. Similarly, Friedberg and colleagues have demonstrated that inflammatory profiles of AD-associated microglia that regulate tau pathologies are highly dependent on the presence of ApoE4 [107]. These data have suggested that ApoE4 plays a crucial intermediate role between microglia inflammatory signaling and tau pathology. Furthermore, LRP1, a major receptor for ApoE and tau, has been shown to mediate the inflammatory responses of microglia via the regulation of the JNK and NF-κB signaling pathways [47]. ApoE may facilitate the assembly of the tau-LRP1 complex to exacerbate the pro-inflammatory signaling pathways on microglia.

#### **5. Therapeutic Approaches Targeting Molecular/Cellular Signaling Pathways**

A comprehensive understanding of cellular and molecular contributors to tau pathogenesis provides novel insights for discovering therapeutics for human tauopathies, including AD. Extensive investigations have demonstrated that HSPGs play a crucial role in the transcellular spreading of tau pathogenesis; therefore, HS-based therapeutics hold great potential for treating tau pathologies. Small molecules or anti-HS peptides interfering with HS-tau interaction are of therapeutic potential for the treatment of tauopathies, which have been reviewed previously [108]. Glycan-based compounds targeting 3-O-sulfated motifs on HS recognized by tau seeds represent a novel therapy for tauopathies [41].

NLRP3 inflammasome activation has been linked to the development of multiple inflammatory diseases, such as atherosclerosis, type II diabetes, and Alzheimer's disease, as well as various cancers [109]. The inhibition of NLRP3 inflammasome activity has been demonstrated to decrease tau phosphorylation and aggregation via attenuated neuronal GSK-3β and CaMKIIα activities. Antagonizing purinoreceptor (P2 × 7R) to prevent the assembly of an active NLRP3 inflammasome in microglia has been suggested as one of the best approaches to control neuroinflammation caused by microglial activation and has therapeutic potential for treating tauopathies [109,110].

Hsp90 directly binds to tau species [110], and Hsp90 inhibitors have been considered as promising therapeutics for treating tauopathies. However, disappointing clinical results due to poor blood–brain barrier permeability and toxicity of all tested drugs have led researchers to alternatives to Hsp90, such as Hsp90 co-chaperones, including ATPase homolog 1 (Aha1), a small 38-kDa cochaperone that binds to the N-terminal and middle domains of Hsp90. The role of Aha1 in tau pathogenesis via interactions with Hsp90 has been demonstrated in a transgenic tau mouse model, rTg4510 [111]. Overexpression of Aha1 led to an increased level of sarkosyl-insoluble tau, as well as the tau with T22 reactivity (anti-oligomer antibody). Treatment with KU-177, which binds specifically to Aha1, reduced the accumulation of insoluble P301L tau in cultured cells, suggesting that Aha1 may be a promising therapeutic target for tauopathies by directly reducing tau aggregation [111].

The autophagy-lysosomal pathway (ALP) shows beneficial effects on tau clearance in reactive astrocytes during the early stages of tau pathogenesis. Because of this, activation of TFEB, a key regulator of this pathway, could be considered as a promising treatment for tauopathies [96]. The therapeutic role of a novel TFEB activator named curcumin analog C1 has been studied using three AD animal models [56]. Treatment with curcumin analog C1 has significantly reduced the levels of Aβ42/Aβ40 in brain lysates from 5×FAD mice models, and phospho-tau epitopes (AT8+ and PHF1+) in a P301S mice model. In addition, curcumin analog C1 attenuated both APP and tau pathology in a 3×Tg AD mice model, accompanied by TFEB activation, increased autophagy, and lysosomal activity.

Neuroprotective effects of CX3CR1-CX3CL1 signaling in tau clearance through microglia phagocytosis has been suggested at an early stage of tau pathogenesis, revealing that the enhancement of this signaling at an early stage of disease progression could be beneficial for disease treatment. Indeed, soluble CX3CL1 overexpression by adenoviral transformation in the Tg4510 mouse model has rescued tau pathology by regulating microglial activation [83]. An alternative approach by Fan et al. [112] via neuronal CX3CL1 overexpression reduced neuronal loss and improved cognitive function in a P19 tauopathy model by enhancing neurogenesis through the CX3CL1–TGF-β2/3–Smad2 pathway. Taken together, CX3CL1 overexpression could be considered a key therapeutic target for treating AD by either promoting neurogenesis for neuronal loss recovery or attenuating microglia-induced neuroinflammation.

Given the overlapping effects on neuronal excitotoxicity by overstimulation of NMDA receptors and decreased expression of astrocyte-specific glutamate receptors observed in multiple neurodegenerative disorders, both NMDA receptors and reactive astrocytes have been implicated as therapeutic targets for treating tau pathology including AD. NMDA receptor-dependent excitotoxicity has been shown to depend on the extrasynaptic GluN2B-containing NMDA receptors rather than synaptic GluN2A-containing NMDA receptors. Antagonists selectively inhibiting extrasynaptic NMDA receptors may have neuroprotective effects [113]. Recently, a new NMDA receptor blocker, RL-208, has been tested on a mouse model of late-onset AD, showing cognitive function improvement in terms of increased synaptic protein density, increased phosphorylation of NMDA2B, reduced protein-related apoptosis, as well as decreased phosphorylated tau levels [114]. This study points out that this novel neuroprotective drug may be valuable for treating AD.

Ameliorating dysfunctional neuronal-astrocyte communication via reducing reactive astrocytes may pose an additional therapeutic target for neurodegenerative disorders. Decreased levels of astrocyte-specific glutamate transporters have been associated with the pathogenesis of tau pathology [115]. For instance, small molecule LDN/OSU-0212320 has been shown to upregulate the expression of EAAT2, a glutamate transporter, in astrocytes via translational activation. Thus, LDN/OSU-0212320 treatment has attenuated glutamate-induced cytotoxicity in neuron and astrocyte coculture, as indicated by the greatest neuronal survival compared to untreated cells. Furthermore, significantly ameliorated symptoms and prolonged lifespan upon the treatment with LDN/OSU-0212320 have been demonstrated in an ALS transgenic mice model [116]. Similar studies are still warranted to determine the drug efficacy for AD models.

*ApoE4* is the most recognized genetic risk factor of AD and is thus considered an important therapeutic target for treating AD. Despite the incomplete understanding of the mechanisms underlying the effects of *ApoE4*, the conversion of *ApoE4* to less-toxic isoforms of *ApoE*—either *ApoE3* or *ApoE2*—may hold promising therapeutic potential. For instance, gene editing to convert *ApoE4* to *ApoE3* or the addition of a 'structural corrector' on *ApoE4* expressing culture to refold *ApoE4* into more *ApoE3*-like conformation has rescued *ApoE4* neurons from AD pathology [106]. Additionally, an AAV-mediated *ApoE2* expression vector targeting the *ApoE4* gene of AD patients to transform the ApoE4 homozygote to an *ApoE2-ApoE4* heterozygote for treating AD is currently in clinical trials [117]. A summary of therapeutic approaches targeting altered molecular and cellular signaling pathways in tauopathies is presented in Table 3.


