Alzheimer’s disease: phenotypic approaches using disease models and the targeting of tau protein

Abstract

Introduction:

Hyperphosphorylated and aggregated tau protein is the main hallmark of a class of neurodegenerative disorders known as tauopathies. Tau is a microtubule binding protein which is important for microtubule assembly and stabilization, for proper axonal transport and overall neuronal integrity. However, in tauopathies, tau undergoes aberrant post-translational modifications that fundamentally affect its normal function. The etiology of these devastating diseases is unclear and there is no treatment for these disorders.

Areas covered:

This review examines the neurobiology of tau, tau post-translational modifications and tau pathophysiology. Progress regarding the effort to identify and assess novel tau-targeted therapeutic strategies in preclinical studies is also discussed. We performed a search on PubMed of the relevant literature published between 1995 and 2020.

Expert opinion:

Tau diversity and the lack of clinically available test to diagnose and identify tauopathies are major obstacles; they represent a possible reason for the lack of success of clinical trials. However, given the encouraging advances in PET tau imaging and tau neurobiology, we believe that a more personalized approach could be on the horizon and that this will be key to addressing the heterogeneity of tau pathology.

1.0. Introduction

Tau is a microtubule-associated protein (MAP) predominantly expressed in neurons and located in several cellular compartments, including nucleus, cytoplasm, cell membrane, axon, dendrite and synapse [1]. Tau normally promotes polymerization and stability of microtubules and is also important for axonal transport of mitochondria, lipids, synaptic vesicles, proteins, and other organelles [1,2]. Tau exists as a naturally unfolded protein and is subjected to many post-translational modifications which modulate its function and cellular localization. This particularly flexible state allows tau to adopt different conformations required for binding to numerous partners and suggests its involvement in many signaling pathways [3]. However, under pathological conditions, aberrant post-translational modifications interfere with normal tau function and promote polymerization, fibrillization and ultimately its aggregation resulting in neurodegeneration [1].

Intracellular aggregates of pathological tau is the hallmark of a group of neurodegenerative disorders called tauopathies which includes Alzheimer’s disease (AD), the most common form of dementia in elderly adults, Pick’s disease (PiD), Frontotemporal dementia with parkinsonism 17 (FTDP-17), progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD) [1]. With the exception of a few mutations identified in the MAPT gene that encodes tau (especially in FTDP-17), the mechanisms leading to tau aggregation and neurodegeneration remain unclear and no therapy is available to cure or halt the progression of these disorders. Given the increasing prevalence of these diseases with aging, today AD and related tauopathies represent a significant economic burden for the US healthcare system. Since the extent of tau pathology strongly correlates with dementia and memory loss [4], the search for an effective tau modifying therapy seems to be the best approach to treat these neurological diseases. In this review, we will summarize current knowledge on the neurobiology of tau, tau post-translational modifications, tau pathophysiology and we discuss the progress on the effort to identify and assess novel tau-targeted therapeutic strategies in preclinical studies and their challenges. To reach this goal we performed a literature search on PubMed of the relevant papers published between 1995 and 2020.

2.0. Tau protein

In human, tau is encoded by MAPT gene located on chromosome 17. Alternative splicing of exons 2, 3, and 10 generates 6 brain isoforms (see Figure 1). Inclusion or exclusion of exon 2 and 3 generates tau isoforms which differ by the presence of zero, one or two amino-terminal inserts (0N, 1N and 2N). Exon 10 instead, encodes the second tau microtubule-binding domain, thus, inclusion or exclusion of exon 10 results in the presence of either 4 or 3 microtubule-binding repeats (called 4R-tau and 3R-tau). Recent studies have shown that, 4R-tau isoforms are more efficient at promoting microtubule assembly and have greater microtubule-binding affinity than the 3R-tau isoforms but, they also are more likely to aggregate [5]. The 3R-tau and 4R-tau are present in equal amount in the adult human brain. However, an imbalance in the 3R/4R ratio has been observed in various tauopathies [2, 6].

Figure 1.

Tau alternative splicing. In human, tau is encoded by MAPT gene located on chromosome 17. Alternative splicing of exons 2, 3, and 10 generates 6 brain isoforms. Inclusion or exclusion of exon 2 and 3 generates tau isoforms which differ by the presence of zero, one or two amino-terminal inserts (0N, 1N and 2N). Exon 10 instead, encodes the second tau microtubule-binding domain, thus, inclusion or exclusion of exon 10 results in the presence of either 4 or 3 microtubule-binding repeats (called 4R-tau and 3R-tau).

As a microtubule-associated protein, tau is crucial for microtubule assembly, structural stability and dynamics. Moreover, by binding to microtubules and actin microfilaments, tau regulates axonal transport, synaptic integrity and activity [7]. In addition to its canonical functions, tau nuclear localization and recent reports have unveiled novel functions of tau like maintenance of DNA integrity, RNA and ribosome stability, regulation of neuronal activity and neurogenesis [8, 9].

In pathological conditions tau is also released into the extracellular space by healthy or dying neurons and is found at significant levels in the interstitial fluids of the CNS and in the cerebrospinal fluid (CSF) of AD patients [3]. The pathogenic tau transfer among neurons seems to occur mainly across synapses. For instance, tau spreading has been widely demonstrated in wild type and transgenic animal models of AD and related tauopathy. In these studies, intracranial injections of either synthetic tau fibrils or human brain-derived pathological tau reached neuro-anatomically connected brain regions way distant from the original site of injection [10]. Although there are strong data in support of the cell-to-cell tau transmission hypothesis, currently, the mechanisms responsible for tau spreading are not fully understood. It is believed that one of the possible mechanisms involves exosomes. These extracellular vesicles are released upon fusion with the plasma membrane [11] and several studies have reported tau secretion via this pathway in vitro and in vivo [12, 13]. Indeed, in patients with AD tau from the CSF [14] and from the blood [15] has been found associated with exososomes. Alternative proposed secretion mechanism involves chaperone complexes [16], or members of the Rab GTPases family (Rab7a and Rab1a) [17]. On the other hand, among the potential mechanisms for tau uptake clathrin-mediated endocytosis, micropinocytosis, or direct membrane fusion have also been considered [10].

3.0. Tau post-translational modifications

Under physiological conditions, tau undergoes several post-translational modifications that can significantly affect its function, metabolism and clearance.

3.1. Tau phosphorylation

The most common form of post-translational modification of tau protein is phosphorylation. Phosphorylation and de-phosphorylation are reversible processes that occur on specific sites of a protein surface. Phosphorylation refers to the addition of a phosphate group to Serine and Tyrosine (phosphate-accepting amino acid) by enzymes called kinases, while de-phosphorylation refers to the removal of that phosphate group by enzymes called phosphatases. Adenosine triphosphate (ATP) and guanosine triphosphate (GTP) are the common energy donors that supply the phosphate group. The net activity of these two processes determines the phosphorylation state and function of the substrate [18]. Tau is characterized by a very high content of proline and lysine residues with a total of 85 potential phosphorylation sites. In a normal state, there is an average of 2–3 moles of phosphate per mole of tau, while in tauopathies this ratio is increased up to 7–8 [19]. Several studies have shown that tau affinity for microtubules is affected by its phosphorylation status and in particular by phosphorylation at S262, S293, S324 and S356, respectively found in tau repeat regions [20, 21]. When phosphorylation is present to such a high degree, tau detaches from the microtubules and accumulates in the cytoplasm, increasing the chances of self-aggregation, polymerization, mis-sorting and altered transport from axon to terminals and dendrites and ultimately leading to neuronal toxicity [22]. However, not all tau phosphorylation sites are equally prone to aggregation, rather a specific pattern of phosphorylation could be necessary to induce tau self-assembly [23]. For instance, hyper-phosphorylation of tau induced by recombinant ERK kinase (15 sites) as well as phosphorylation of pSer202/pThr205 residues in absence of phosphorylation of the Ser262 does not affect tau susceptibility to in vitro aggregation when compared to wild type [24]. On the other hand, the combination of phosphorylation at Ser202/Thr205/Ser208 with the absence of phosphorylation at Ser262 residue, promotes formation of tau filaments [24].

Throughout the years we have identified several kinases involved in tau phosphorylation including cyclin-dependent kinase 5 (CDK5), CDK2, GSK3β, the mitogen-activated protein kinases p38 (p38 MAPK), ERK, JUN N-terminal kinase (JNK), Ca²⁺/calmodulin-dependent protein kinase II (CaMKII), AKT, PKA and PKC. On the other hand, tau protein phosphatases include protein phosphatase 2A (PP2A), PP2B, PP2C, PP3 and PP5 [25]. PP2A is considered the major phosphatase and it has been reported to be reduced by 50% in the AD brain compared with healthy controls [9].

3.2. Tau acetylation

Tau lysine residues have also been found highly acetylated in the brain of AD and tauopathy patients compared to normal brains [26]. The histone acetyltransferase p300 (p300 HAT) and the cAMP-responsive element-binding protein (CREB)-binding protein seem to be the major enzymes promoting tau acetylation. Sirtuin 1 (SIRT1) and histone deacetylase 6 (HDAC6) instead, deacetylate tau. The acetylation of lysine K174 and K280 is an early event in the AD pathological cascade and contributes to tau pathology by affecting tau clearance through inhibition of ubiquitin binding [27]. Therefore, dysregulation of this process has been linked to tau aggregation, tau mis-sorting and altered neuronal plasticity. Recently, acetylation at Lys174 (K174) has been also shown to play a crucial role in tau homeostasis and cognitive deficits in a mouse model of tauopathy [28, 29].

3.3. Tau proteolytic cleavage

Numerous studies have shown that tau can be also a substrate for different proteases including calpain, cathepsin and several caspases in AD and related tauopathies (i.e., CBD, Pick disease, FTD and PSP). Like phosphorylation, tau truncation inhibits its biological function and promotes tau conformational changes and subsequent aggregation. Supporting this concept, a study showed that there is an inverse correlation between the abundance of tau cleavage products and cognitive performance [30]. Specifically, tau can be cleaved at Asp421 by caspase 3 and 6, at Asp314 by caspase 2 and at Asp402 and Asp13 by caspase 6 [31–34]. Caspase-3 can also modulate tau phosphorylation indirectly by cleavage of the protein kinase B (Akt) which in turn regulates the activation of the GSK3β kinase pathway. This was demonstrated in vitro in neuro-2A neuroblastoma (N2A) neuronal cells stably expressing human APP carrying the K670 N, M671 L Swedish mutation (APPswe) [35] as well as in vivo, in the 3xTg transgenic mouse model of AD [33]. Additionally, proteolytic cleavage of tau at glutamic acid 391 (E391) was also observed in AD brain, as a component of NFTs [36]. To study the biological consequences of this truncation, McMillian et al. generated a mouse model expressing E391-truncated tau and found that indeed, the E391 peptide promotes tau cellular mis-localization, phosphorylation and tau conformational change confirming the relevance of tau truncated species in the development of these pathologies [36].

3.4. Tau O-GlcNAcylation and N-glycosylation

Compared with the post-translational modifications above reported, the exact role of O-GlcNAcylative and N-glycosylative modification is less investigated and not quite understood. However, it seems that O-GlcNAcylation is protective against tauopathy since it has been reported to be reduced in AD compared to age-matched healthy controls [37]. On the other hand, N-glycosylation may promote tau phosphorylation and aggregation and is increased in AD [38]. However, further studies are needed to validate these preliminary observations.

3.5. Tau clearance

Tau is degraded by both the proteasome and autophagy systems and the preferred mechanism adopted by cells seems to be dependent on the nature of tau post-translational modifications and level of tau aggregation [39]. Soluble, monomeric tau interacts with E3 ligase and with the carboxy-terminus of heat shock protein (CHIP) which promotes tau ubiquitination and subsequently targets tau to the proteasome. On the other hand, some phosphorylated forms of tau, mono-ubiquitylated, caspase-cleaved-tau and oligomeric tau are selectively degraded by the autophagy machinery. There is significant evidence that both of these degradative systems are extremely important in counteracting the progressive accumulation of pathological tau aggregates [40]. In fact, upregulation of CHIP or treatment with proteasome activators has been shown to reduce accumulation of tau protein in vivo and in vitro [41]. Similarly, stimulation of autophagy via genetic or pharmacological approach decreased total and phosphorylated tau in vitro and in several mouse models of the disease. [39, 41].

4.0. Tau in neurodegenerative diseases

Abnormal tau protein phosphorylation and subsequently deposition in the brain is the primary characteristic of a heterogeneous group of neurodegenerative diseases called tauopathies. These disorders differ in their phenotypic manifestations, in the isoforms and filament structure of aggregated tau, in the type of cells in which the pathology is found and in the brain areas involved [3]. Tauopathies can be classified as primary (FTDL, PSP, PiD and CBD) or secondary (AD, Down’s syndrome) depending on whether tau is the predominant lesion or, even if still central to the neurodegenerative process, is associated with other pathological features (i.e., Amyloid β). Tauopathies can also be grouped based on the ratio of 3R/4R tau in the filaments. Mixed 3R and 4R tauopathies are AD, Down’s syndrome and a subset of FTDP-17. PiD is a 3R tauopathy while CBD and PSP largely contain 4R tau. Although tauopathies are mostly sporadic, several tau mutations localized to the microtubule binding domains have been identified (Table 1) [19]. The majority of these mutations affect tau alternative splicing of exon 10, thus the 3R/4R ratio. Other missense mutations affect protein-protein interaction and tau ability to bind to the microtubules. Other evidence have shown that a wide range of mutants (i.e., R5L, K257T, I260V, G272V, ΔK280, P301L, P301S, G335V, Q336R, V337M, and R406W) in the presence of polymerization-inducing agents promote formation of tau filaments in vitro [42]. In pathological conditions, aberrant tau modifications not only interfere with tau physiological functions but also trigger its self-aggregation. For example, without the need of inducers, a truncated tau fragment (corresponding to residues 297–391 of full-length tau) has been shown to self-assemble into PHF-like fibrils in vitro [43].

Table 1.

Classification of tauopathies based on tau isoform, cell type, morphology of tau aggregates and associated MAPT mutations.

Disease	Tau isoform	Neurons	Astrocytes	Glia	Mutations associated with tau pathological subtype
Alzheimer’s disease	3R + 4R	NFTs Tau oligomers	NA	NA
Down syndrome	3R + 4R	NFTs	NA	NA
Pick disease	3R	Pick bodies	tufted astrocytes	glial tau deposits	Exon 9: K257T,L266V , G272V Exon 10: ΔK280 Exon 11: L315A, S320F, P332S Exon 12: Q336R, Q336H, K369I,  G342V  Exon 13: E372G, G389R, R406W Intron 9: IVS9–15 Intron 10: IVS10+4
Progressive supranuclear palsy	4R	NFTs	tufted astrocytes	Coiled bodies	Exon 1: R5L Exon 10: S305S, S303S, S285R, N279K Intron 10: IVS10+16
Corticobasal degeneration	4R	NFTs, Neuropil threads	Astrocytic plaques	Coiled bodies	Exon 10: S305S Intron 10: IVS10+16 Exon 13: N410H
Globular glial tauopathy	4R	NFTs, Neuropil threads	Globular inclusions	Globular inclusions	Exon 1: R5H Exon 10: P301L, N296H IVS10+16 Exon 11: K317N
Chronic traumatic encephalopathy	3R + 4R	NFTs, Neuropil threads	tufted astrocytes	NA
Argyrophilic grain disease	4R	spindle-shaped argyrophilic grains	NA	Coiled bodies

NA= not available

By contrast, tau phosphorylation does not seem to be required for tau aggregation in vitro. However, recently it was shown that this post-translational modification works as an aggregation inducer and modulates in vivo propagation [44, 45]. In fact, de-phosphorylation of soluble tau oligomers prepared from end-stage AD prior injection in the brain of hTau mice significantly diminished tau seeding [45].

Other toxic species which precede the formation of fibrils and NFTs are tau oligomers. These intermediate forms of tau can propagate and spread through synapses in the brain. Several studies regarding the role of tau in neurodegeneration have suggested that tau oligomers and not NFTs is the main toxic entity leading to neuronal degeneration and synaptic failure in tauopathies [46, 47].

In this complex group of neurological disorders, the structure of tau species and their regional distribution can also be quite different. Notably, in AD tau lesions are mainly composed of paired helical filaments (PHFs) and straight filaments (SFs) while in other tauopathies they are characterized by different molecular conformers [48] On this regard, recent reports further verified the existence of disease-specific folds in tau filaments which, in the future may lead to a better understanding of the different clinical manifestations, mechanisms of tau spreading and morphology of tau lesions [48,49] For example, although in AD and PiD tau adopts a similar secondary structure, in the latter filaments assume a novel fold independently from phosphorylation at Ser262/356, which by contrast is consistently observed in AD [48]. Moreover, in AD, tau pathology is restricted to neuronal cells where tau is found in its oligomeric form and as NFTs. PiD is characterized by the presence of rounded cytoplasmic neuronal inclusions, composed of tau straight filaments, called Pick bodies and globular oligodendroglial inclusions. PSP and CBD instead, present tau accumulation in neurons and glia cells as well as in astrocytes (tufted astrocytes and astrocytic plaque, respectively) (Table 1) [19].

5.0. Tau as therapeutic target

Being the major driving pathological factor in tauopathies, tau protein represents the target of choice for the treatment of these diseases. Moreover, in recent years we have seen many promising drug candidates targeting β-amyloid failing clinical trials leaving the need for an effective disease-modifying therapy also for AD patients unfulfilled [50]. This fact together with the observation that, in AD tau pathology shows a better correlation with impaired cognitive function than amyloid pathology has emphasized the therapeutic role of tau protein as a target of choice [4]. On this regard, just recently La Joie et al. reported that tau-PET, but not amyloid-β PET, can predict with more than 40 percent accuracy the location and extent of brain atrophy and degeneration on an average of 15 months in advance in AD patients [51]. This finding not only confirms the role of tau as major driver of these pathologies but further supports tau as therapeutic target to slow down the neurodegenerative process also in AD. Every stage of tau biology such as synthesis, post-translational modifications, self-aggregation and clearance represents a possible target for therapeutic intervention (Figure 2). In this field of research, animal models have been a resourceful tool for the understanding of tau physiology and elucidation of the complexity of the mechanisms involved in the generation of the diverse pathological tau species responsible for neuronal toxicity during neurodegeneration. Tau transgenic mice engineered to express wild type or mutated human tau have been used to validate numerous tau-related targets at the preclinical stage. Most of tau transgenic mice are carriers of either the P301L [52] or P301S MAPT mutation [53] (rTg-tauP301L-4510 and PS19 mice, respectively). These mutations are mainly found in FTDP-17 patients and recapitulate the progressive accumulation of glial and neurofibrillary tangles, observed in humans [42]. However, these models are less relevant for AD in terms of tauopathy where mutations of tau have not been discovered or associated yet.

Figure 2.

Summary of the different tau therapeutic strategies assessed in preclinical studies.

Recently, a novel mouse model expressing truncated tau protein (296–390) corresponding to the PHF core-tau fragment found in AD has been developed [54]. Interestingly, although this model displays a weaker tau phenotype compared to other models carrying pathogenic tau mutations, it has the advantage to present evidence of an age-dependent spread of tau pathology similar to AD Braak staging [54]. Another concern regarding these models is the difference between human and mouse tau. It is well known in fact that adult mice express exclusively the 4R tau isoforms while human brain has both [55]. To this end, the transgenic mouse model called hTau, which expresses all 6 isoforms of wild-type human tau in the absence of the endogenous mouse tau was generated to better resemble the development of the classic AD tau neuropathology [56]. More recently, it has been shown that injection of isolated tau fibrils from the brains of AD, CBD, or PSP patients into the hippocampus of wild-type mice results in the development of classical tau neuropathology. Interestingly, tau fibrils from CBD and PSP spread only affecting glial cells, while AD tau fibrils were confined into neurons [57]. Thus, this model appears to be extremely valuable for the understanding of how different tau species propagate in neurological diseases and can facilitate the design of appropriate therapeutic strategies that specifically target them.

In recent years, cell based high-throughput screenings have emerged as excellent means to test and identify novel molecules drugs. However, since this approach requires a large number of cells, traditionally it implements immortalized cells of neuronal lineage, or non-neuronal cells (i.e., HEK cells). On the other hand, the use of primary neurons although more physiological carries the same limitation regarding the scale and the additional fact that human and rodent cells may behave differently in vitro assays. The advent of the human induced pluripotent stem cells (iPSCs) technology has opened up new possibilities in the field since it can generate a large number of different types of neurons which are more relevant to the human conditions [58]. Many investigators have used iPSCs generated from the fibroblasts of AD patients harboring mutation either in APP or PS1/2 [59]. Previous studies demonstrated that neurons or neural progenitor cells derived from these iPSCs recapitulated AD pathogenic phenotypes, such as increases in the ratio of pathogenic Aβ42 to Aβ40 oligomer and hyper-phosphorylated tau [60, 61]. Though iPSC technology has huge potential for a better understanding of tau neurobiology and therapy, it faces many challenges and limitations. Among them, genetic variability between different patients or clones derived from the same patient, and the ability to model the late onset disease since differentiated cells are often functionally immature [62] and tau generated by stem-cell derived neurons remains in the fetal state (i.e., 3R0N). However, cells expressing 4R-containing tau mutations, such as P301L, exhibit increased tau phosphorylation and accumulation, and most importantly this phenotype can be modulated by a pharmacological approach [63].

5.1. Microtubule stabilizers

Microtubule instability and disruption has been observed in animal models of tauopathy and in the brain of tauopathy patients [64]. It is intuitive that as a crucial microtubule binding protein, tau dysfunction will lead to reduced microtubules integrity and axonal damage. Thus, microtubules stabilizers may prove beneficial and could be an effective therapeutic strategy for these diseases. Epothilone D, an antifungal agent, successfully reduced tau pathology, restored microtubules function and improved cognition in several tau models [65–66]. Similarly, the peptide Davunetide (NAD) by stabilizing microtubules ameliorated pathology, memory deficits and axonal transport in tau transgenic mice [67, 68].

5.2. Inhibiting tau phosphorylation

Although tauopathies are heterogeneous diseases, hyper-phosphorylation of tau which as we described earlier deeply affects tau function is one the most common event in all of them. For this reason, targeting this specific post-translational modification of tau seems to be the most obvious therapeutic approach. Modulation of tau phosphorylation can be achieved by either inhibition of kinases or activation of phosphatases like PP2A. Several GSK3β inhibitors have been discovered and tested in preclinical phase [64]. Tideglusib, a specific and irreversible GSK3β inhibitor, proved to reduce tau phosphorylation, memory deficits and astrocytosis in APPsw-tauvlw mice [69]. Also, 3xTg mice treated with lithium chloride, a drug FDA approved for bipolar disorder, showed reduced phospho-tau levels, Aβ and better memory [70].

Besides GSK3β, inhibition of CDK5, a member of the cyclin-dependent kinases (CDKs), is also known to reduce tau phosphorylation and apoptosis in cortical neurons [71]. Moreover, 5-lipoxygenase (5LO), an enzyme that produces potent pro-inflammatory mediators and which is elevated in the brain of AD patients, has been shown to effect tau phosphorylation by activation of the CDK-5 kinase pathway. Following this discovery, our group has extensively demonstrated that 5LO inhibition or knockout prevents CDK5-dependent phosphorylation of tau in vitro and in vivo in several models of AD (3xTg) and tauopathy (P301S and hTau mice). By contrast, we have also shown that, 5LO overexpression results in hyper-phosphorylation of tau upon increased levels and activity of the CDK5 kinase pathway. Taken together these data support the novel hypothesis that pharmacologic inhibition of 5LO by modulating CDK5 would reduce tau hyper-phosphorylation and therefore could represent a valid therapeutic strategy for the treatment of human tauopathies [72].

Another possible therapeutic target is PP2A, the main tau phosphatase which is known to be reduced in the brain of AD patients when compared with healthy controls. As phosphatase modifiers, sodium selenite was shown to increase PP2A activity by interaction with the regulatory B subunit of the enzyme and to decrease tau phosphorylation in tau transgenic mice [73]. In addition, memantine, an FDA approved drug for the treatment of AD, has also been reported to enhance PP2A activity via inhibition of the inhibitor protein of PP2A (I2PP2A) [70].

5.3. Inhibiting tau acetylation

Since recent studies have shown that tau acetylation is another important tau post-translational modification and an early event in the development of tauopathy, research effort has been focused on the identification and development of novel drugs that interfere with this aspect of tau pathophysiology. To this end, the molecule salsalate, a non-steroidal anti-inflammatory drug, that inhibits tau acetylation at Lys174 by inhibition of p300 HAT, was tested in tau transgenic mice. Salsalate treatment significantly reduced total tau levels, tau acetylation, NFTs formation, and ameliorated hippocampal neuronal loss and spatial memory deficits in the PS19 model [29]. As a cyclooxygenase inhibitor, it is possible that some of the salsalate therapeutic effects are also due to its anti-inflammatory properties rather than to its p300 HAT inhibition. This concept is important because although salsalate is very promising as a drug candidate, there is great concern regarding its potential inhibitory activity on p300, a crucial acetyltransferase. In fact, p300 inhibition could result in numerous cellular changes and negative side effects [28, 64].

5.4. Inhibition of tau aggregation

Hyper-phosphorylated and cleaved tau is more prone to its irreversible aggregation leading to microtubule destabilization and neuronal damage. Thus, another promising therapeutic avenue is the inhibition of tau aggregation processes. Several molecules can inhibit this pathological process. The most common is methylene blue (MB), a phenothiazine which acts as nitric oxide synthase and guanylate cyclase inhibitor [74]. MB is the chloride salt of Methylthioninium (MT) which exists in equilibrium between a reduced leucomethylthioninium (LMT) and oxidized (MT+) form depending on environmental conditions [75]. LMT and MB have been shown to prevent tau fibrillization and reduce tau aggregates in vitro and in vivo [76–79]. In preventative studies, MB was effective in reducing tau pathology and memory dysfunction in several tau transgenic mouse models [76–79]. Unfortunately, in clinical trials, MB showed modest effects probably secondary to its complex and limited absorption [75]. Hence, a new generation of tau protein aggregation inhibitors (i.e., LMTX) has been successfully developed which show improved brain uptake [80]. Clinical trials investigating LMTX have provided some promising results in terms of rate of disease progression and brain atrophy. Curcumin, a natural extract from the curcuma plant also can bind to β-sheets and counteract tau aggregation. Interestingly, this compound when administered to the 3xTg mice ameliorated tau pathology and memory impairments [81].

5.5. Clearance of tau

In the past few years, several studies have revealed the presence of significant autophagic defects in the brains of AD and related tauopathies. Accumulation of autophagosomes and lysosomes is evident in human brain sections and is associated with pathological tau accumulation [82]. Autophagy is the preferred mechanism by which hyper-phosphorylated, truncated and oligomeric tau are degraded [39]. In line with this evidence, enhancing the autophagy-lysosomal system efficiency could significantly improve the clearance of toxic forms of tau and result in better neuronal health and synaptic function. Pharmacological studies in vitro and in vivo have confirmed this hypothesis. In tauopathy animal models (3xTg and P301S) both autophagy inducers (trehalose) or mTOR inhibitor like rapamycin, reduced phosphorylated and insoluble tau accumulation, increased neuronal survival and improved cognitive function [83,84].

In this regard, we have recently shown that consumption of extra virgin olive oil ameliorates memory, cognitive impairments and tau pathology in the 3xTg mice as a consequence of the activation of cell autophagy. [85]. Additionally, in support of the important role played by the autophagy machinery in the clearance of tau, we have demonstrated that this cellular mechanism can also be stimulated by pharmacological inhibition of the 12/15Lipoxygenase enzyme [86]. Thus, pharmacological blockade or genetic downregulation of this enzymatic pathway is sufficient to affect total tau levels and phosphorylation in neuroblastoma N2A cells transiently transfected with human tau (hTau) and in vivo in the 3xTg mice. Conversely, overexpression of 12/15Lipoxygenase worsens the AD phenotype through downregulation of the autophagic flux [86, 87].

An additional appealing therapeutic strategy that aims to enhance tau clearance is targeting the ubiquitin-proteasome machinery. Before undergoing heavy post-translational modifications and accumulates in its oligomeric forms tau is subject to degradation by the proteasome system. CHIP, an ubiquitin ligase, together with the chaperone heat shock protein 90 (Hsp90), play an important role in tau degradation. In a mouse model overexpressing the human P301L 4R tau mutation, reduction of CHIP caused increase in the accumulation of toxic tau [40]. Hsp90 forms a complex with misfolded tau and its levels are inversely proportional to soluble and tau oliogomers [88]. Inhibitors of this chaperone help cells to clear toxic tau and represent a promising drug candidate for the treatment of tauopathies. Thus, a recent paper showed that a brain-permeable Hsp90 inhibitor (PU-DZ8) reduced tau expression and phosphorylation in the TauP301L transgenic mouse model [89].

5.6. Reducing tau expression

Roberson and colleagues were among the first to report that the simple reduction of endogenous tau protein in an APP mouse model of amyloid pathology was enough to promote improvement in memory function without effecting Aβ metabolism and deposition [90]. After this initial observation, several studies have confirmed that tau lowering approach is a viable therapy in animal models of AD. Tau suppression was also beneficial in a pure tauopathy mouse model, the Tg4510, expressing the 4R mutated p301L tau where memory and neuronal loss was recovered as result of this approach [91].

In recent years, antisense short single-stranded oligonucleotides (ASOs), have been developed to selectively target human tau mRNA for either degradation or inhibition of translation. ASO efficacy has been tested in the PS19 mouse model. In this model intra-cerebroventricular administration of ASO reduced human and phosphorylated AT8 tau, neuroinflammation, prevented neuronal loss, rescued memory and improved survival at 6 and 9 months of age [92]. However, the important observation about this study is that ASO proved to be able to halt the progression of the pathology not only a young age but also at a later time point after the degenerative process is established. This fact could have important implication in terms of therapeutic efficacy since the diagnosis of these disorders is usually done at a later stage of the disease. One main concern is how to translate these preclinical data to humans. These transgenic mice were developed to express high levels of tau protein, so it is possible that the observed effects are consequent to the suppression of the artificially high expression of tau which for example is not observed in AD and tauopathy patients not carrying MAPT mutations.

MicroRNAs (miRNA) are small non-coding single-strand RNA molecules which regulate gene expression at the post-transcriptional level. They are transcribed by RNA polymerases II and III and generated through a series of cleavages of a primary miRNA (pri-miRNA) to form a mature miRNA. These mature miRNAs exert their function by binding to the 3’UTR of a target mRNA thus promoting its silencing either through its cleavage and subsequent degradation, or by inhibition of its translation [93]

One member of this very large family of cell function regulators, the miRNA-132/212 could represent an important alternative approach to ASOs. The miR132/212 directly targets tau mRNA and reduces its expression. Interestingly, miR132/212 is downregulated in the brain of AD and in PSP patients [94]. Several studies have shown that deletion of miR132/212 promotes abnormal tau metabolism and pathological aggregation in vitro and 3xTg mice [95]. On the other hand, miR-132 upregulation reduces the levels of total and post-translationally modified forms of tau, its cleavage, and release, and enhances long-term potentiation in vitro and in the P301S tau transgenic mice [96].

5.7. Tau immunization

In recent years, tau immunotherapy has made incredible progresses and based on preclinical data, both active and passive immunization have shown promising results. Currently several clinical trials implementing this approach are being conducted for AD and related tauopathies [97]. Importantly, anti-tau monoclonal antibodies have been proved to enter neurons, likely through clathrin-mediated endocytosis, and to clear extra- and intracellular tau potentially blocking tau spreading [98]. Although there are some concerns about safety, active immunization successfully reduced tau pathology in different tauopathy models. Immunotherapy targeting phospho-tau reduces the extent of aggregated tau and halts the progression of cognitive impairments in P301L mice [99]. Similarly, active immunization with peptide containing phospho-Ser422 epitope reduces tau pathology and improves cognitive functions in the THYTau22 mouse model [100]. Vaccination with fibrillar PHF-tau decreases brain levels of sarkosyl-soluble tau and NFTs in old tau transgenic mice [101]. Finally, immunization with wild type human tau results in reduced tau pathology and neuroinflammation in rTg4510 tau mice [102].

Since there are safety concerns about the risk of immunological adverse effects, recent studies have been focusing on passive immunization which also offers greater specificity. A variety of antibodies against phospho-tau, wild type and mutated tau as well as aggregated tau has been developed and tested in multiple in vitro and in vivo tauopathy models [70, 103]. For instance, antibody (D), developed against the mid-region of tau (amino acids 235–250), effectively neutralized tau seeding in an aggregation cellular model of both AD and PSP [104]. Subsequently, the same antibody was also tested in vivo in a transgenic mouse model overexpressing human 1N4R tau mutant (P301S and G272V), the Tgtau30 mice, that develops AD-like NFTs [105]. Passive immunization with antibody D confirmed its effectiveness in blocking tau seeding and prevented the formation of NFts in these animals injected with human tau derived from AD brain [106]. Another promising tau oligomer-specific monoclonal antibody (TOMA) administered intravenously and/or intracerebroventricularly reversed locomotor and memory deficits in the JNPL3 mouse model (expressing human mutant tau transgene P301L) and induced rapid reduction of tau oligomers without effecting phosphorylated and monomeric tau [107]. A panel of seven antibodies against different regions of tau (N-terminal and mid-region) was instead tested by Nobuhara et al. revealing an epitope-dependent efficacy, with the antibodies raised against the mid-domain (6C5 and HT7) of tau being the most successful in blocking tau neuronal uptake in vitro [108]. Consistent with this study, tau quantification by mass spectrometry in the CSF of AD, PSP and LBD patients revealed abundance of 1N/3R among all tau isoforms and prevalence of peptides from the tau mid-domain when compared to peptides with the N- and C-terminus domains providing a rational for the findings above described [109,110].

Extracellular tau is present in both full-length [111] and truncated forms [112] and which species is responsible for tau seeding and spreading is still topic of debate. In the preclinical stage, several antibodies have shown their potential as therapeutic strategies, especially the one targeting the mid-region of tau. However, The existence of a large variety of tau species, the complex molecular structures of tau seeds and varying degree of their accumulation among different tauopathies makes the design of these antibodies, and the prediction of their efficacy, rather challenging.

6.0. Conclusions

In conclusion, currently there a several tau-directed therapeutic strategies under evaluation in preclinical and clinical phase for the treatment of AD and related tauopathies. A number of factors are involved in the efficacy and safety of those specific approaches when translated to human. In fact, quite often promising results obtained in animal models has not been reproduced in a clinical scenario. Although transgenic tau mice have been very valuable in the investigation and validation of candidate targets, there are concerns about the artificially high expression of human tau, and the introduction of tau mutations necessary to generate the classical neuropathology in these animals. In fact, with few exceptions, tauopathies are mainly sporadic disorders and for instance no tau mutations have been identified in AD yet. Moreover, the heterogeneous nature of this class of neurological disorders adds complexity to the already challenging search for a treatment able to prevent or stop the progression of the pathology. For instance, one approach may not be suitable for all tauopathies. Thus, in the future the use of combination therapies designed to specifically target 3R, 4R or an oligomeric form of tau depending on the patient and type of tauopathy might address best the pathophysiology of these diseases and ultimately results in more effective treatments against them.

Expert Opinion

The complexity of tau biology offers many potential candidate targets for the treatment of tauopathies, including inhibition of tau phosphorylation and aggregation, tau reduction and immunotherapy. Preclinical studies in several animal models of these diseases have consistently demonstrated the validity of these strategies in slowing down the progression of memory and cognitive impairments confirming the role of tau as a key neuropathological driver of these conditions. Recently, experimental studies have indicated tau immunotherapy as one of the most successful in counteracting disease phenotype even in presence of Aβ. However, when translated to human, tau immunization repeatedly failed to show any efficacy raising a lot of doubts. Although over the years our understanding of tau pathophysiology has considerably improved, there are many other crucial questions that need to be answered to better redirect our efforts in the design of more appropriate preclinical and clinical trials targeting tauopathies.

For instance, what’s the right stage of tauopathy to be targeted? We know that typically when symptoms appear the pathology is already established and this has an impact not only on the level of success of a specific therapeutic approach, but also on the type of approach that should be adopted. In other words, one approach might be more suitable in the earliest stages of the disease compared to a later stage or vice versa. This is a very important matter when it comes to the design of a clinical study. Thus, in our opinion, a significant obstacle to the achievement of an effective therapeutic approach is also the lack of a clinically available test to diagnose and specifically identify tauopathies relatively early and obtain a reliable measure of the stage of the disease in living patients. In fact, currently, these diseases can be diagnosed with certainty only after post-mortem examination. Ongoing research is trying to develop reliable and sensitive tests to detect abnormal forms of tau protein associated with AD, PSP and CBD.

Moreover, tauopathies are very heterogeneous diseases not only for the brain areas involved but also in terms of distinct tau epitopes, tau fragments and deposits which likely reflect a diversity in the pathogenic mechanisms. The lack of knowledge in this regard represents a major weakness in the field and a possible reason for the limited success of current clinical trials. If the structure of toxic tau and the nature of tau seeds are not known, we cannot be certain whether we are targeting the right epitope responsible for the disease. Indeed, recent studies are moving in this direction and are precisely looking at the diversity of intra- and extra-cellular tau among tauopathies. We anticipate that, guided by this new knowledge, more effective immunotherapies will be available. The future holds the promise of a treatment customized to each patient depending on his/her type and degree of tau pathophysiology.

We believe that improving the understanding of tau neurobiology is the key for the successful optimization of antibodies specificity and therapeutic window. Nevertheless, a key step forward that might even facilitate clinical trials comes from a recent small longitudinal study involving patients in early clinical stages of AD, which by implementing tau-PET imaging identified the exact areas of neurodegeneration a year in advance. This study is encouraging and could have important implications in terms of development of a personalized treatment that will enable us to specifically slow down or even halt the progression of the neurodegenerative process depending on the predicted brain atrophy pattern for each patient.

Table 2.

List of compounds that have been tested as anti-tau therapeutic agents.

Name	Mechanism	References
Epothilone	Microtubule stabilizer	B. Zhang et al. J Neurosci, 32,3601–3611. Barten DM et al. J Neurosci, 32,7137–7145.
Davunetide	Microtubule stabilizer	Matsuoka Y et al. J Pharmacol Exp Ther, 325,146–153. Matsuoka Y et al. J Mol Neurosci, 31, 165–170.
Tideglusib	GSK3β inhibitor	Serenó L et al. Neurobiol Dis. 35(3):359–367.
Lithium	GSK3β inhibitor	Erin E et al. Nat Rev Neurol. 14(7): 399–415.
Zileuton	5 lipoxygenase inhibitor	Lauretti, E et al. Journal of Alzheimer’s disease 64(s1), S481–S489.
Sodium Selenite	Phosphatase modifier	van Eersel J et al. Proc Natl Acad Sci USA. 107(31):13888–13893.
Memantine	Phosphatase modifier NMDA antagonist	Erin E et al. Nat Rev Neurol. 14(7): 399–415.
Salsalate	NSAD Acetylation inhibitor	Min SW et al. Neurosci. 38(15):3680–3688.
Methylene Blue/LMT	Anti-tau aggregant	Congdon EE et al. Autophagy. 2012;8:609–622. Hochgrafe K et al. Acta Neuropathol Commun. 2015;3:25 Al-Hilaly et al. J Mol Biol. 2018 Oct 19;430(21):4119–4131
LMTX	Anti-tau aggregant	Melis et al. Cell Mol Life Sci. 2015 Jun;72(11):2199–222. Baddeley TC et al. J Pharmacol Exp Ther. 2015;352(1):110–8.
Curcumin	Anti-tau aggregant	Tsuyoshi H et al. CNS Neurosci Ther. 2010; 16(5): 285–297.
Rapamycin	Autophagy inducer (mTOR inhibitor)	Smita M et al. PLoS One. 2011;6(9):e25416. doi: 10.1371/journal.pone.0025416. Ozcelik S et al. PLoS One. 2013 May 7;8(5):e62459.
PU-DZ8	HSp90 inhibitor	Luo W et al. Proc Natl Acad Sci U S A. 2007;104(22):9511–6.
ASO	Tau expression modulator	DeVos SL et al. Sci Transl Med. 9(374). pii: eaag0481.
miR132/212	Tau expression modulator	Luikart, BW et al. PLOS ONE 6:e19077. Smith PY et al. Hum Mol Genet. 24(23):6721–635.
Anti-tau antibodies	Tau immunotherapy	Asuni AA, J Neurosci. 27(34):9115–29. Ando K et al. J Alzheimer’s Dis. 40 Suppl 1:S135–45. Selenica MB et al. Journal of Neuroinflammation. 11: 152 Castillo-Carranza DL et al. J. Neurosci 34, 4260–4272. Colin M et al. Acta Neuropathol 139, 3–25. Courade JP et al. Acta Neuropathol 136:729–745 Ando K et al. Am J Pathol. 178(2):803–16. Albert M et al. Brain. 142(6):1736–1750. Nobuhara CK et al. 187(6):1399–1412.

Article Highlights.

• The term tauopathy describes a class of neurodegenerative diseases pathologically characterized by aberrant accumulation of abnormal tau protein and tau aggregates. Clinically, it is characterized by the presence of a range of symptoms such as cognitive impairment and personality changes that depend on the location and distribution of pathology.

• Tau is subjected to many post-translational modifications which modulate its function and cellular localization. In disease conditions, tau undergoes post-translational modifications that interfere with its normal function and drive tau aggregation and toxicity ultimately leading to neurodegeneration.

• Pathologically modified tau isoforms have emerged as potential therapeutic targets

• Therapeutic approaches targeting amyloid-β have repeatedly failed in Alzheimer’s disease clinical trials; tau shows a better correlation with impaired cognitive function than amyloid-β pathology

• Every stage of tau biology such as synthesis, post-translational modifications, self-aggregation and clearance represents a feasible target for a viable therapeutic intervention.

• The most promising therapeutic strategies are tau reduction and tau immunotherapy.