Introduction

Alzheimer’s disease (AD) is the most common form of dementia with no effective treatment available. It is the sixth leading cause of death in United States which mostly affected the people of age 65 and above. The number of patients is estimated to be over 46 million worldwide and it is predicted that the number will reach 131.5 million by 2050 (Prince et al. 2015). It is estimated to have cost the world $818 billion and become a trillion dollar disease by 2018 (Prince et al. 2015). The pathological hallmarks of AD are the presence of extracellular amyloid-β (Aβ) deposition and the formation of intracellular neurofibrillary tangles (NFTs) in the brain, along with the typical degeneration and loss of neurons characterized by memory loss and learning difficulty (Bloom 2014; Wang et al. 2017). Several studies have focused on Aβ targeting pathology as a potential treatment for AD; however, clinical evidence has shown that anti-amyloid treatment for AD has not been effective at slowing down the progression of disease. Since then, attention has been shifted towards the research on the pathology of tau proteins in AD (Giacobini and Gold 2013; Desikan et al. 2012).

Tau is a normal, unfolded, highly soluble protein which plays a critical role in tubulin assembly and stabilization of microtubules, thereby promoting normal function of neurons (Duan et al. 2017). In AD, several post-translational modifications especially tau hyperphosphorylation were believed to be a key factor that affects microtubule assembly and induces tau aggregation. In tau hyperphosphorylation, tau undergoes conformational changes in which the conversion of tau monomer to tau oligomer induces the aggregation of tau into pair helical filament, leading to the formation of NFTs (Chirita et al. 2005; Sahara et al. 2007; Mondragón-Rodríguez et al. 2008; Lasagna-Reeves et al. 2010; Patterson et al. 2011). Scientific evidence shows that NFT itself may not be involved in inducing neurotoxicity; however, the intermediate tau oligomer has been demonstrated to be the most toxic form of tau that results in synaptic impairment in AD (Fá et al. 2016). Besides, in vivo and in vitro studies have also shown that tau pathology may spread to different brain regions (Clavaguera et al. 2009; Holmes and Diamond 2014; Fá et al. 2016). As tauopathies are indicated in dementia and AD, hereby we review the physiological role of tau proteins in brains, the key factors contributing to tauopathies, and the mechanisms involved, as well as potential tau-targeted therapies for AD.

Tau and Tauopathies

Tau Physiology and Function

In the past decade, research in AD has been focused on Aβ. Recently, the research target began to shift from Aβ to tau when studies revealed that tau proteins were the main component that made up NFTs. Tau is a microtubule-associated protein which plays a crucial role in stabilizing microtubules by promoting tubulin assembly to regulate the normal functions of neurons (Giacobini and Gold 2013). The location of tau gene, MAPT, was found on the long arm of human chromosome 17. In human brain, tau proteins encode six isoforms which vary in size, ranging from 352 to 441 amino acids. These isoforms are different from each other by the presence of three repeats (3R) or four repeats (4R) in the C terminal and the presence or absence of one (29 amino acids) or two inserts (58 amino acids) in the N-terminal region. Studies have suggested that the repeat regions (244–268 amino acids) in the C terminal are the domain that is responsible for the binding of tau to microtubule (Goedert et al. 1989a, b; Himmler et al. 1989). The six isoforms have their specific physiological role as they are differently expressed during brain development. The shortest isoforms of tau with the presence of 3R and without the presence of N-terminal inserts have been found to express during the foetal stages. However, the rest of the isoforms with either 3R or 4R and one or two inserts in C terminal and N terminal, respectively, have been revealed to be expressed during adulthood (Kosik et al. 1989; Stanford et al. 2003). Studies have discovered that tau isoforms that are expressed during adulthood are about 40-fold more efficient than the foetal isoform in promoting microtubule assembly (Lindwall and Cole 1984).

Tau protein has been demonstrated to play a pivotal role in binding to the microtubule for its stabilization (Daun et al. 2017). Microtubule is a cellular important structure that transports organelle for the plus-end-directed motor kinesin and the minus-end-directed motor dynein. The plus-end-directed motor kinesin is responsible for the delivery of their cargoes such as mitochondria, endocytic or endocytic vesicles, and lysosomes towards the peripheral cells, whereas the minus-end-directed motor dynein is responsible for delivering the cargoes backwards to the microtubule organizing centre. The role of these motors helps in regulating the axonal transport in the brain. Study has shown that tau proteins which bind to the microtubule tightly regulate the transport of cargoes which in turn affect the axonal transport (Callahan et al. 2002). Due to the strong tau’s affinity to the microtubule, it could detach the cargoes from the kinesin and hence inhibit the axonal transport and result in the symptoms of AD. This indicates that tau phosphorylation plays a crucial role because it regulates the tau’s affinity to microtubule (Brady and Sperry 1995; Lippincott-Schwartz et al. 1995).

Role of Tau Protein in Tauopathies

Conformational Changes of Tau

In AD patients, the misfoldings of tau would affect the microtubule binding, and hence the tau is neither bound to tubulin nor promotes the microtubule assembly, thus disrupting the organization of microtubule. Tau in its abnormal phosphorylated form would form tau dimer, followed by the formation of tau oligomer. A number of studies have provided evidence that the dimerization of tau monomers is an essential step for the formation of tau oligomers (Chirita et al. 2005; Sahara et al. 2007; Mondragón-Rodríguez et al. 2008; Lasagna-Reeves et al. 2010; Patterson et al. 2011). The tau oligomer then aggregates and forms the paired helical filaments (PHF), which eventually leads to the formation of NFTs that are abundant in the brain of AD patient (Grundke-Iqbal et al. 1986).

Spreading of Tau Pathology Extracellularly

Tau protein is mainly localized in the axonal and somatodendritic compartment (Binder et al. 1985). However, previous studies have shown a wider distribution of tau in other compartments including the nucleus (Sultan et al. 2011) and dendrites (Hoover et al. 2010). The aggregates were first seen in the entorhinal cortex, then the pathology would propagate to the hippocampus and spread to other brain regions including the limbic cortex and association cortices (DeKosky and Scheff 1990; Terry et al. 1991). In an in vivo study, the trans-synaptic propagation of tau pathology through anatomically connected neuronal networks was demonstrated in the transgenic mouse model that expressed human tau in the entorhinal cortex (Liu et al. 2012). Several studies have proposed that tau oligomers induce neurotoxicity more than the aggregated form of tau and also play a more prominent role in spreading the pathology. In vitro studies have shown that in the frontal cortex of the AD brain, T22 and AT8 antibodies have been detected in different brain regions apart from the regions that were exclusively stained for tau oligomers. These data suggested that tau oligomers can transmit from one region to another in the AD brain. The presence of oligomeric tau in the extracellular space has been found to be associated with memory deficits in AD. In this case, several in vitro and in vivo studies have revealed that tau oligomerization can induce the aggregation of tau prior to the formation of NFTs (Clavaguera et al. 2009; Fá et al. 2016).

Synaptic Impairment

The symptoms of AD include progressive memory loss, cognitive decline, and learning difficulty. These symptoms are strongly associated with the dysfunction of synapses which eventually lead to irreparable synaptic loss in AD. Previous research mainly focused on the toxicity of Aβ plaques which were thought to be involved in the synaptic damage in AD; however, a later study has found an emerging role of tau in synapses (Delacourte et al. 2002). In the healthy brain, the presence of tau plays a prominent role in regulating normal synaptic function. Previous study has discovered that the presence of tau oligomers in synapses are toxic to the neurons and can cause synaptic impairment prior to the formation of NFTs, an event that would trigger the neurodegeneration such as those in AD (Tai et al. 2014). Studies on drosophila model have demonstrated that overexpressing tau results in neuronal loss with the absence of NFTs. In agreement with that, other studies have also shown that when reducing the levels of overexpressing tau in mutant tau transgenic mice neuronal loss is also decreased, despite the continuous formation of NFTs (Williams et al. 2000, 2001; Mershin et al. 2004). Although various studies have claimed that tau oligomer is responsible for the neurotoxicity in tauopathies, its underlying mechanism still awaits further elaboration. Past studies also show that NFT formation rather than the deposition of Aβ plaques correlates well with a cognitive decline in AD. Besides that, cognitive decline has been shown to correlate well with the synaptic loss in AD in parallel with the severity of AD. This is supported by the study that shows the extensive synaptic loss in the AD brain (Masliah et al. 1989). Tau affects the synaptic function in several ways. Axonal transport of mitochondria can be disrupted by the detachment of tau from the microtubule, synaptophysin mRNA was found to be reduced by 35–37% in the AD brain with the presence of NFTs (Callahan et al. 2002), and, most importantly, deregulation of synapses was observed in the AD brain which is independent from the Aβ pathology (Masliah et al. 1993). In transgenic mouse models, results such as loss of synaptophysin, dendritic targeting of Fyn kinase, and impairment of synaptic transmission were reported (Alldred et al. 2012; Harris et al. 2012). Previous evidence has evidenced the role of Aβ, tau, and Fyn in dysregulating synaptic function, which causes neuronal synchrony. In an in vivo study carried out in hAPP mice, the results have shown that by reducing tau, synaptic impairment, cognitive deficits, and behavioural abnormalities induced by the synergistic effects of Aβ and Fyn can be prevented. The involvement of Aβ, tau, and Fyn in synaptic impairment is strongly supported by the previous studies; however, the mechanism by which these components cooperate in AD-related pathogenesis still remains to be elucidated (Roberson et al. 2011).

Key Factors Contribute to Tauopathies

It is evident that, in many neurodegenerative disorders including AD, microtubule-associated tau has undergone several post-translational modifications. These modifications are highly associated with the aggregation of tau in AD. The post-translational modifications that are identified in tauopathies include tau hyperphosphorylation, truncation (Wischik et al. 1988; Novak et al. 1993; Gamblin et al. 2003), glycosylation (Wang et al. 1996; Liu et al. 2004), glycation (Yan et al. 1994, 1995), nitration (Zhang et al. 2005), and ubiquitination (Iqbal and Grundke-Iqbal 1991). We will also discuss other post-translational modifications associated with tau aggregation in this review.

Hyperphosphorylation of Tau

Tau phosphorylation is essential for the neurons to maintain its normal physiological function, including the binding of normal tau to the microtubule and the regulation of microtubule assembly and its stabilization. In the studies performed by Ksiezak-Reding et al., tau was found to contain 2–3 mol of phosphates per molecule of tau in the normal brain; however, in the autopsied AD brain, the level of tau phosphorylation is at least 3–4 times higher than that in the normal brain, which is around 6–8 mol of phosphates per molecule of tau (Ksiezak-Reding et al. 1992). Hence, hyperphosphorylation of tau plays a critical role in tau aggregation and contributes to the pathogenesis of AD (Ihara et al. 1986). It should be noted that the actual causes that lead to hyperphosphorylation are still in debate. Studies have reported that the largest brain tau isoform, tau 441, contains 80 serine or threonine and five tyrosine residue phosphorylation sites which are potentially phosphorylated in AD (Hasegawa et al. 1992). With the use of phosphorylation-dependent monoclonal antibodies against tau and mass spectrometric analysis, the results have shown that at least 39 serine or threonine (Hanger et al. 2007) and two tyrosine residues are phosphorylated in PHF (Lee et al. 2004).

The normal tau phosphorylation is regulated by the activities of both cellular enzymes, protein kinases, and protein phosphatases. Both enzymes are crucial in maintaining microtubule assembly and normal neuron function in the human brain. Past studies have shown that the imbalance between kinases and phosphatases is the potential cause of the formation of hyperphosphorylated tau in the AD brain (Wang et al. 2007). The major protein kinases involved in tau phosphorylation are part of the proline-directed protein kinases, including cyclin-dependent kinase 5 (Cdk5) (Baumann et al. 1993) and mitogen-activated protein kinase (MAP). However, kinases that phosphorylate at non-Ser/Thr-Pro sites include cyclic AMP-dependent protein kinase (PKA) (Jicha et al. 1999), microtubule-associated regulatory kinase (MARK) (Drewes et al. 1997), casein kinase I and II, and calmodulin-dependent protein kinase II (CaMKII) (Baudier and Cole 1987). Other than that, glycogen synthase kinase 3β (GSK3β) is able to phosphorylate at both proline-directed sites and non-proline-directed sites (Hanger et al. 1992). Tau phosphorylation occurred in different sites can be regulated by different kinases. Having different tau phosphorylation sites can have various impacts on neuronal function. In the in vitro studies demonstrated by Liu et al., Thr231, Ser262, and Ser356 phosphorylation sites have been shown to inhibit the binding of normal tau to microtubule, leading to the disruption of microtubule assembly. The results of the study show that tau phosphorylation that occurred at proline-rich region inhibits microtubule assembly and promotes self-aggregation into tangles of filaments. At the C-terminal tail region, tau phosphorylation was shown to increase its microtubule-stimulating activity and promote self-aggregation. The results, taken together, revealed that tau phosphorylation occurred at all the proline-rich region, C-terminal tail region, and microtubule-binding region has diminished the normal function of microtubules (Liu et al. 2007). Luna-Monoz et al. have reported that normal tau appeared to undergo conformational changes at the phosphorylation site Thr231. The structurally changed tau was first detected by monoclonal antibody TG-3 and Alz-50 subsequently (Luna-Muñoz et al. 2005, 2007).

Besides, tau has also shown its involvement in neurotransmission via the Fyn pathway (Lee et al. 1998). A previous study has also suggested that the src family kinase Fyn is upregulated in the AD brain (Shirazi and Wood 1993). In the study, Fyn is involved in the tau phosphorylation at tyr18 with the use of antibody probes specific for phosphor-tyr18. Results of immunocytochemical studies also indicated the presence of tyrosine-phosphorylated tau in NFTs in the AD brain. Taken together, all these findings supported the role of Fyn in the neurodegenerative diseases (Lee et al. 2004). Wood and Zinsmeister have reported the involvement of the tyrosine phosphorylation system in AD pathology. Elevated levels of phosphotyrosine were observed in the somatodendritic compartment of neuritic plaque and dystrophic neurites, which reflects the role of the tyrosine kinase system in neurodegeneration process (Wood and Zinsmeister 1991).

Alonso et al. suggested that the products of mutant tau gene with alterations in the tau conformation are more prone to hyperphosphorylation. Hence, hyperphosphorylation of mutant tau would result in further conformational changes, enables them to sequester normal microtubule-associated tau proteins, and aggregates to form PHFs. However, the actual mechanism by which the mutations lead to tau hyperphosphorylation is not fully understood (Alonso et al. 2004).

Besides kinases, a decrease in the levels of phosphatases has also been shown in the autopsied AD brain. A previous study has revealed that the activity of phosphatases in the isolated AD brain is reduced by approximately 20% in grey matter and 40% in white matter (Gong et al. 1993). Phosphatases play an opposite role of kinases and hence promote tau dephosphorylation instead. In the AD brain, the phosphate levels were found to be 3- to 4-fold higher than those in the normal brain (Ksiezak-Reding et al. 1992). The underlying cause of the elevated levels of phosphates is the downregulation of phosphatases, and thus the efficiency of the neurons to remove the phosphate group is dampened, which eventually leads to the accumulation of phosphates in the AD brain. The major phosphatase involved in tau phosphorylation is PP2A. I1PP2A and I2PP2A are the endogenous PP2A inhibitors that play a crucial role in the regulation of PP2A activity. However, both inhibitors have been shown to be downregulated in the AD brain, leading to increased levels of inhibitors and decreased PP2A activity (Sontag et al. 2004; Chen et al. 2008). A previous study has reported that brain acidosis activates and induces the translocation of asparaginyl endopeptidase from the neuronal lysosomes to the cytoplasm, inhibiting PP2A through the cleavage of I2PP2A, causing hyperphosphorylation of tau, and likely contributes to the pathogenesis of AD (Basurto-Islas et al. 2013).

Truncation

Other than tau hyperphosphorylation, truncation of tau is also one of the important modifications that have been associated with tau aggregation. It is defined as the proteolytic cleavage of tau which leads to aberrant aggregation. In the AD brain, caspase 3, a member of the caspase family of cysteine proteases, has been proved to be involved in the proteolytic cleavage of tau at aspartic acid 421 residue (D421) in vitro (Gamblin et al. 2003; Rissman et al. 2004). The truncated tau at D421 aggregates more readily than the full-length tau (Derisbourg et al. 2015). A previous study also reported that the occurrence of tau truncation at D421 has increased cell death via apoptosis significantly (Stadelmann et al. 1999). Interestingly, the caspase 3 cleavage site can be recognized by the monoclonal antibody Tau-C3. There is also in vitro evidence suggesting that phosphorylated tau at serine 422 can inhibit caspase cleavage at D421 (Guillozet-Bongaarts et al. 2006). Besides, a previous study has shown that the truncation of tau also occurred at glutamic acid 391 residue (E391) and the cleavage site can be recognized by antibody MN243 due to the formation of MN243 epitope (Novak et al. 1993). Other than that, the N terminus of tau is also cleaved at D13 and D402 by caspase 6 in the AD brain (Horowitz et al. 2004). Studies also suggested that caspase 6 is implicated in the apoptosis and neuronal degeneration in AD (Guo et al. 2004; Albrecht et al. 2007). Garcia-Sierra et al. reported that the truncation of the C terminus of tau is correlated with the cognitive impairment in AD (García-Sierra et al. 2001). When all these were taken together, cumulative evidence has suggested that the truncation of tau plays a critical role in the tau aggregation, as it can result in tau polymerization to form insoluble PHF.

Glycosylation

In the AD brain, tau is abnormally modified by glycosylation which is associated with tau aggregation. Study has shown that hyperphosphorylated tau was glycosylated mainly through N-glycosylation. The role of N-glycosylation in the AD brain is to maintain and stabilize the structure of PHF, which also helps in facilitating kinases to promote tau hyperphosphorylation and suppressing the tau dephosphorylation by phosphatases (Liu et al. 2004). Non-hyperphosphorylated tau isolated from the AD brain was found to be glycosylated, as opposed to the absence of glycan in the normal brain. Thus, this suggests that tau glycosylation precedes tau hyperphosphorylation (Wang et al. 1996). In vitro studies have shown that glycosylated tau acts as a better substrate for cAMP-dependent protein kinase than the non-glycosylated tau. This suggests that glycosylated tau facilitates the subsequent hyperphosphorylation in the AD brain, a process probably catalysed by cyclin-dependent kinase 5 (cdk5) and glycogen synthase kinase-3 beta (GSK3β) at site-specific phosphorylation sites (Liu et al. 2002a, b). In contrast to N-glycosylation, O-GlcNAcylation, a type of O-glycosylation helps in regulating tau phosphorylation in a site-specific manner. In the AD brain, the level of O-GlcNAcylation was found to be reduced, which probably explains the presence of tau hyperphosphorylation and tau aggregation. Using starved mice as the model, impaired glucose uptake or metabolism is shown to decrease the level of O-GlcNAcylation and promote tau hyperphosphorylation (Liu et al. 2004). This suggests that patients with diabetes mellitus, which characterized a similar impairment of glucose metabolism, are likely to have a higher risk of developing AD (Ott et al. 1999). In vivo studies have shown that the reduction of O-GlcNAcylation suppressed PKA–CREB signalling, which in turn impairs the ability of learning and memory in AD (Xie et al. 2016). This is because PKA catalytic subunits play a pivotal role in learning and memory and were post-translationally modified by O-GlcNAc.

Glycation

Glycation is a process of non-enzymatic post-translational modification which may facilitate the aggregation of tau with the formation of advanced glycation end products (AGEs). Previous studies have shown that the accumulation of AGEs in PHF induces oxidative stress which promotes neuronal dysfunction in AD patients (Yan et al. 1994, 1995). Li et al. have reported that the level of methylglyoxal (MG), a reactive carbonyl compound, is increased in the AD brain. The data have suggested that MG is involved in inducing tau hyperphosphorylation through AGE formation, upregulation of receptors of AGEs (RAGE), and activation of GSK-3β and p38 MAPK (Li et al. 2012). Other studies have also shown that MG could enhance tangle formation in vivo and this process could be slightly accelerated by tau hyperphosphorylation (Kuhla et al. 2007). Recent studies have also revealed that the level of 4R2N tau glycation is increased with ageing and/or development of diabetes mellitus. This would result in the promotion of tau phosphorylation and aggregation, which in turn contribute to the pathogenesis of AD (Liu et al. 2016).

Nitration

Tau nitration is also a possible modification which may be involved in tau aggregation. Abnormal nitration of tau was found in NFTs in the AD brain. Tyrosine residues that are involved in tau nitration are Tyr18, Tyr29, Tyr197, and Tyr394. Studies showed that nitrated tau Tyr197 can be found in the normal healthy brain and may play a crucial physiological role in human body. However, the rest of the Tyr residues are only detected in the AD brain. However, the involvement of tau nitration in AD pathology remains unclear. Zhang et al. have reported that peroxynitrite treatment can result in tau nitration and oligomerization in a dose-dependent manner. The study revealed that the nitrated tau undergoes conformational change, which inhibits its binding to microtubule and promotes tau aggregation (Zhang et al. 2005). With the use of quantitative electron microscopy, Reynolds et al. have demonstrated that nitration of residues Tyr29 and Tyr197 increases the average length of tau filaments without alteration in the steady-state polymer mass. However, nitration of residues Tyr18 and Tyr394 has been shown to decrease the average length of tau filaments, causing filamentous mass reduction. Taken together, in vitro evidence suggested that nitration affects microtubule assembly differently, based on the site of nitration (Reynolds et al. 2005). In another study, the monoclonal antibody Tau-nY29 was used to detect soluble tau and PHF isolated from the severely affected AD brain that has nitration at Tyr29. It was shown that nitration at Tyr29 is a disease-related event which may cause alteration on tau, inducing self-polymerization in neurodegenerative tauopathies (Reynolds et al. 2006).

Ubiquitination

Tau is also known to be ubiquitylated and involved in tau aggregation into tangles of filaments, leading to neuronal death (Iqbal and Grundke-Iqbal 1991). The impairment of ubiquitin ligases, carboxyl terminus of the Hsc70-interacting protein (CHIP), has been implicated in neurodegenerative tauopathies. The studies have reported that the interaction of CHIP with Hsp70/90 can induce ubiquitination of tau, resulting in tau aggregation. Conversely, the mouse model that overexpressed inducible Hsp70 has been shown to have a significant reduction in tau levels. Taken together, these data suggest that the Hsp70/CHIP chaperone system plays a critical role in regulating tau ubiquitination and degradation as well as preventing tau aggregation (Petrucelli et al. 2004). Besides, the CHIP–Hsc70 complex could regulate the ubiquitination of phosphorylated tau species, enhancing neuronal cell survival (Petrucelli et al. 2004; Shimura et al. 2004).

Other Post-translational Modifications Associated with Tau Aggregation

Other post-translational modifications including lysine methylation (Funk et al. 2014) and sumoylation (Dorval and Fraser 2006; Luo et al. 2014) are also indicated in the mechanism of tau aggregation. According to that study, lysine methylation retains tau’s ability to promote microtubule assembly which in turn attenuates tau aggregation propensity. This methylation could help in suppressing tau aggregation in vitro which might confer neuroprotection in AD (Funk et al. 2014). Besides, sumoylation is also another important post-translational modification that has been indicated in AD pathology. It is known that tau contains two small ubiquitin-like modifier (SUMO) consensus sequences and Lys(340) being the major sumoylation site (Dorval and Fraser 2006). A previous study has reported that tau sumoylation induces tau hyperphosphorylation at multiple AD-associated sites. This would in turn promote further sumoylation and inhibit tau ubiquitination and degradation. These data suggested that sumoylation may play a pathogenic role in the neurodegenerative tauopathies (Luo et al. 2014).

Therapeutic Approaches for Tauopathies in AD

In the past 40 years, studies have been performed to investigate the feasibility of targeting Aβ pathology as a potential treatment for AD. Positive results such as cognition improvement and reduction of tau hyperphosphorylation with low adverse effects have been observed in the AD transgenic mice treated with anti-amyloid therapies. However, the results reported in animal studies were not parallel with the results obtained in the clinical trials (Giannakopoulos et al. 2003; Desikan et al. 2012). On the hand, more and more emerging evidence suggests that tauopathies might play a key role in the pathogenesis of AD, and hence tau protein has now received substantial attention to be the next treatment target for AD. Many studies have been conducted to find ways to reduce tau hyperphosphorylation by inhibiting the activity of kinases and restoring the activity of phosphatases. On top of that, neuroscientists are also exploring other potential approaches such as inhibiting tau aggregation, restoring tau GlcNAcylation, and deploying tau immunotherapy to target tauopathies in AD.

Tau Aggregation Inhibitors

Previous studies have shown that methylene blue (MB) can block tau–tau binding interaction through the repeat domain. This has been shown to effectively reduce the stability of the protease-resistant PHF and inhibit tau aggregation in vitro (Wischik et al. 1996). Previous studies have demonstrated that MB treatment preserves cognition in transgenic mice expressing full-length pro-aggregant human tau before the onset of cognition impairment. Therefore, MB that serves as a tau aggregation inhibitor can be used as a potential drug for AD treatment before the onset of cognitive impairments (Hochgräfe et al. 2015). TRx0237 (LMTX™), methylthioninium chloride, is a derivative of MB which is also capable of inhibiting tau aggregation (Wischik et al. 2015).

Microtubule Stabilizers

In AD, microtubule is destabilized due to the inhibition of tubulin assembly caused by the conformational changes of tau, which would contribute to the neuronal dysfunction. Therefore, treatment with microtubule stabilizer is a potential therapeutic approach for AD. Microtubule stabilizers such as paclitaxel and other taxanes are not suitable for the treatment of tauopathies due to their poor blood–brain barrier permeability. In contrast to paclitaxel, BMS-241027, which is also known as Epothilone D (EpoD), is a microtubule-stabilizing agent with better blood–brain barrier permeability. Brunden et al. have reported that treatment with BMS-241027 improved microtubule density and axonal integrity, and reduced cognitive deficits in the transgenic mouse model of tauopathies without inducing any side-effects (Brunden et al. 2010; Zhang et al. 2012). Due to its positive effects on animal studies, the therapeutic effect of BMS-241027 is recommended for clinical testing; however, the clinical trial was discontinued in 2011 and the results have been inconclusive. NAP or AL-108, which is also known as davunetide, is another microtubule-stabilizing peptide derived from activity-dependent neuroprotective protein. It helps in promoting microtubule stability and has shown a reduction effect on tau phosphorylation in preclinical studies. These findings were supported by Shiryaev et al.’s study in which memory improvement and tau hyperphosphorylation reduction have been observed in NAP-treated transgenic mice (Shiryaev et al. 2009). In phase II clinical study of NAP, the results suggest that intranasal administration of NAP was safe, well tolerated and showed improvement in cognitive function in subjects with mild cognitive impairment (Morimoto et al. 2013).

Kinase Inhibitors and Phosphatase Activators

Kinases and phosphatases have been targeted for the novel therapeutic approaches in AD due to their involvement in tau hyperphosphorylation (Wang et al. 2007). Therefore, these two enzymes have been targeted for the novel therapeutic approaches in AD. Lee et al. suggested that the Src family tyrosine kinase, Fyn, has shown its involvement in AD by inducing synaptic loss and memory impairments in the transgenic mouse model (Lee et al. 2004; Chin et al. 2005). Elevated Fyn in AD patients has been found to be responsible for tau phosphorylation and the mediated neurodegeneration process in AD. Hence, treatment targeting Fyn kinase can be a novel therapeutic approach for AD. Saracatinib (AZD0530), a Fyn kinase inhibitor, which is a potential candidate drug for treating AD, has now entered clinical trial. It has passed phase I clinical trial and now currently enters into phase II (Nygaard et al. 2015). Le Corre et al. have reported that the kinase inhibitor named K252a is effective at clearing hyperphosphorylated tau effects against ERK2 in the tau transgenic mouse model. The in vivo findings have also shown that a reduction of tau hyperphosphorylation directly helped in preventing motor impairment in the transgenic mice. However, NFTs are not reduced in the treated mice (Le Corre et al. 2006). A kinase inhibitor, diaminothiazole, which targets Cdk5/p25 and GSK3β, has been shown to improve memories and reduce the amount of PHFs in the AD mouse models (Zhang et al. 2013). A previous study has reported that microglial activation promotes tau hyperphosphorylation via activation of the p38α MAPK signalling pathway in the mouse model of tauopathies (Roy et al. 2015). Microglial activation also promotes MAPT hyperphosphorylation in the neurons of non-transgenic mice lacking microglia-specific fractalkine receptor (CX3CR1). In addition, in the humanized MAPT transgenic mice lacking CX3CR1 also showed increased MAPT hyperphosphorylation and behavioural impairments which correlated with increased levels of active p38 MAPK. The same results were obtained in other in vitro studies (Bhaskar et al. 2010; Maphis et al. 2015). Besides, studies have also suggested that synaptic dysfunction in AD is associated with the activation of p38α MAPK-mediated signalling pathway (Watterson et al. 2013). Because of that, MW181, a p38α MAPK inhibitor, has been developed. A recent study reported that the treatment with MW181 has multiple advantages: (1) it induces selectively the suppression of α-isoform of p38 MAPK, (2) it reduces tau hyperphosphorylation, (3) it reduces proinflammatory cytokine production, and (4) it prevents cognitive deficits in aged hTau mice. Thus, MW181 is a potential drug for further evaluation for treatment of AD (Maphis et al. 2016). In another study, Branca et al. reported that the Dyrk1A inhibitor reduces amyloid precursor protein (APP) and tau phosphorylation, thus making it another ideal candidate drug for the treatment of AD (Coutadeur et al. 2015; Branca et al. 2017). Another therapeutic approach is to induce the upregulation of protein phosphatases. Previous study has shown that the eicosanoyl-5-hydroxytryptamide (EHT), a minor component of coffee, has therapeutic benefits in the rat model of AD. Dietary supplementation with EHT inhibits tau hyperphosphorylation, reduces Aβ pathology, and also enhances cognition in the AD rat model (Basurto-Islas et al. 2014). The antidiabetic drug biguanide such as metformin has also been shown to promote the activity of PP2A and decrease tau hyperphosphorylation at PP2A-dependent epitope in vitro and in vivo, suggesting its potential therapeutic role for AD (Kickstein et al. 2010). VEL015 (sodium selenate) is another drug candidate that reduces the level of hyperphosphorylated tau and NFTs. In the VEL015-treated transgenic mice, it has been shown that VEL015 not only reduces the levels of hyperphosphorylated tau and NFTs, but also improves memory and motor functions. On top of that, VEL015 also stabilizes the PP2A activity (van Eersel et al. 2010). VEL015 is currently in phase III clinical trial for its evaluation as a potential disease-modifying therapy for AD (Malpas et al. 2016).

Insulin Sensitizer

In AD, brain glucose metabolism is impaired, implying that diabetes mellitus increases the risk of AD. However, the underlying mechanism by which the impairment leads to AD is not well understood. Previous studies have reported that hyperphosphorylated tau contains fourfold less levels of O-GlcNAc in comparison to non-hyperphosphorylated tau, suggesting that the impairment of brain glucose metabolism causes downregulation of O-GlcNacylation, resulting in abnormal hyperphosphorylation of tau, eventually leading to neurofibrillary degeneration in AD. Hence, restoration of brain tau O-GlcNacylation activity could help reverse tau hyperphosphorylation, offering a promising therapeutic target for AD treatment (Liu et al. 2009a, b). Previous studies have reported that thiazolidinediones including troglitazone and rosiglitazone have been shown to reduce tau phosphorylation in in vitro model (d’Abramo et al. 2006). Pedersen et al. have reported that rosiglitazone, which is an insulin sensitizer, attenuates learning and memory deficits in Tg2576 mice with AD. Based on the findings, the rosiglitazone-treated mice exhibited better spatial learning and memory abilities than the untreated mice (Pedersen et al. 2006). Due to the positive effects of rosiglitazone in attenuating learning and memory abilities in mice, the drug serves as a potential therapy for AD and has now been further tested in clinical trial. Another insulin sensitizer, pioglitazone, was also tested in clinical trial due to its anti-inflammatory effects. Both rosiglitazone and pioglitazone were well tolerated; however, oedema was observed in patients who were receiving these two drugs. Pioglitazone has produced conflicting results regarding its efficacy in AD due to numerous trial limitations. Thus, the current findings from the clinical trials suggested that rosiglitazone and pioglitazone are not recommended to be used for AD treatment (Miller et al. 2011).

O-GlcNAcase Inhibitors

Tau O-GlcNAcylation activity is regulated by both O-GlcNAc transferase (an enzyme which catalyse the addition of GlcNAc to proteins) and β-N-acetylglucosaminidase (OGA, an enzyme which catalyse the removal of GlcNAc from proteins). Studies have reported that the administration of the selective OGA inhibitor Thiamet G to JNPL3 transgenic mice exhibited an increase in tau O-GlcNAcylation and reduced NFT formation, eventually leading to reduced neuronal loss in mice. Hence, OGA inhibitors could serve as a potential therapeutic drug for AD treatment (Graham et al. 2014; Yuzwa et al. 2012). Recent studies have investigated the effects of Thiamet G in inhibiting OGA in rTg4510 mice. Evidence has shown that OGA inhibition leads to the activation of O-ClcNAc tau, which in turn reduces pathological tau in the cerebrospinal fluid of mice. Due to the positive effects of OGA inhibitors in mice, it has provided support for further clinical development of OGA inhibitors for treatment of tauopathies (Hastings et al. 2017).

Tau Immunotherapy

Immunotherapy has emerged to be a promising therapeutic approach for AD treatment. Many studies have been conducted using active and passive immunotherapy targeting Aβ peptide. Studies have revealed that Aβ immunotherapy has shown efficacy not only in the Aβ clearance but also in the clearance of early tau pathology. However, the Aβ immunization cannot clear the hyperphosphorylated tau (Oddo et al. 2004; Serrano-Pozo et al. 2010). In phase III clinical trial, the first Aβ vaccination trial has failed to show efficacy on clearing tau pathology as well as halting and slowing down disease progression in subjects with mild to moderate AD (Boche et al. 2010). Hence, recent studies have shifted the focus to tau pathology. Both active and passive immunization have been shown to reduce tau pathology with brain function improvement in the tauopathy mouse model. Studies have shown that neuronal uptake of tau antibodies occurs via clathrin-dependent Fcγ II/III receptor endocytosis. The intracellular interaction of antibodies and tau aggregates results in the clearance of tau pathology (Congdon et al. 2013). Other than that, an in vivo study has revealed that peripheral administration of anti-tau/pS422 antibody formed a complex with tau/pS422 which is present in membrane microdomains on neuronal cell surface, inducing the clearance of tau pathology intracellularly (Collin et al. 2014). Yanamandra et al. have demonstrated that anti-tau antibodies have reduced hyperphosphorylated and aggregated tau in P301S mice effectively. The antibodies are also capable of blocking the development of tau seeding activity and transcellular propagation of tau pathology in mice (Yanamandra et al. 2013). A previous study showed that neuronal uptake of cis tau-specific antibody via Fcγ receptors helps prevent tauopathy development and propagation of tau pathology following traumatic brain injury (Kondo et al. 2015).

Asuni et al. have reported that active immunization against phospho-tau epitope is capable of reducing tau aggregates and improving motor performance in P301L tangle model mice (Asuni et al. 2007). In addition, a study has demonstrated the effects of peripheral administration of antibody against pathological tau in both JNPL3 and P301S mouse models. The findings have shown that passive immunization reduces tau pathology and delays disease progression in both mouse models (Chai et al. 2011). Another study has shown that immunotherapy targeting tau pathology is capable of reducing aggregated tau which can help in preventing cognitive impairments in htau/PS1, a new mouse model with accelerated tangle development. Taken together, the findings of the studies have paved the ways for immunotherapy to be assessed in clinical trials (Boutajangout et al. 2010). Besides, Rasool et al. also reported that systemic vaccination using anti-oligomeric monoclonal antibodies showed great potential in reducing amyloid load and hyperphosphorylated tau, which in turn improved the cognitive functions of 3xTg-AD mice (Rasool et al. 2013). Recent studies have also demonstrated that the 3xTg-AD mice treated with tau antibodies 43D and 77E9 showed successfully reduced total tau levels and tau hyperphosphorylation at different phosphorylation sites. This finding suggested that the passive immunotherapy could possibly reduce Aβ pathology in the AD brain. The passive immunization targeting the N-terminal projection domain of tau has shown memory and cognition improvements in 3xTg-AD mice (Dai et al. 2015, 2017).

The monoclonal antibody DC8E8 is able to recognize pretangles, intracellular and extracellular tangles in AD. DC8E8 can effectively reduce the amount of tau oligomers and NFTs in the transgenic mice. The binding of DC8E8 to the determinants inhibits tau–tau interaction in vitro and in vivo. Besides, DC8E8 also selectively targets the diseased tau but not the healthy tau (Kontsekova et al. 2014b). A tau peptide which encompasses the epitope targeted by DC8E8 was selected to develop an active vaccine, AADvac1. It is a vaccine candidate that targets the structural determinant that is essential for pathological tau–tau interaction on tau. Based on the results from preclinical study, the vaccine effectively reduced tau oligomers and decreased neurofibrillary pathology in transgenic rat model. The findings also showed that active immunotherapy reduced the level of hyperphosphorylated tau by approximately 95%. AADvac1 has entered phase I clinical trial in which its safety and tolerability have been assessed in patients with mild to moderate AD (Kontsekova et al. 2014a). Patients involved in the study have received 3–6 immunization doses of AADvac1. AADvac1 showed an excellent safety and tolerability profile in the first-in-man study (Novak et al. 2017). An 18-month safety follow-up study of AADvac1 in patients who have completed the phase 1 clinical trial of AADvac1 was completed in March 2017. AADvac1 is currently entering phase II clinical trial (Novak et al. 2017). A 24-month safety and efficacy study of AADvac1 in patients with mild AD is ongoing and the primary completion date is expected to be June 2019. In this study, 60% of the patients will receive AADvac1, while the rest of the patients will receive placebo (ClinicalTrials.gov. 2016).

Tauopathies in Other Diseases

Tau deposition in NFTs is the hallmark of AD, and therefore numerous therapeutic approaches are aimed at reducing tau levels in the AD brain. It is known that cardiovascular diseases (CVD) including stroke, high blood pressure, and high cholesterol level are the risk factors of AD (Eriksson et al. 2010; de Bruijn and Ikram 2014; Hersi et al. 2017). In a recent study making use of western blot and immunohistochemistry staining, tau proteins have been shown to be present in heart tissue and play a crucial functional role in the cardiovascular system (Betrie et al. 2017). Previously, tau proteins have been proved to play functional roles in the formation of endothelial cell lumen and induce post-translational modifications, microtubule assembly, and asymmetric cytoskeletal polarization (Kim et al. 2013). Hence, the reduction of tau proteins will likely lead to the suppression of these functions. Regulation of cardiac function, cytoskeleton organization, and myocardium survival by endothelial cell–cardiac myocyte interactions could also be affected by loss of tau proteins (Narmoneva et al. 2004). Betrie et al. conducted the study on wild-type (WT) and tau-deficient (KO) mice of two different age groups (aged 13 and 23 months) for cardiovascular phenotype assessment. Both the right and left atria and small mesenteric arteries were isolated from both mice. According to the results reported, 13-month-old tau KO mice showed evidence of increased systolic blood pressure and cardiac hypertrophy, in addition to decreased right atrial rate and decreased left atrial contractility. In ageing 23-month-old tau KO mice, they exhibited similar results to 13-month-old tau KO mice; however, the condition was more critical compared to young tau KO mice. In concordance with this, a significant reduction of left atrial contractility could be seen in the ageing tau KO mice. In addition, a significant reduction of contractile response to calcium, isoprenaline, and electrical nerve stimulation leading to left atrial decompensation had been observed in the ageing tau KO mice. However, cardiac hypertrophy in 13-month-old tau KO mice did not result in decompensation stage. In ageing WT mice, cardiac hypertrophy has been shown in different chambers of the heart, in which the heart-to-body ratio has increased. Based on the results reported from the contractile response assessment, ageing WT mice also showed a decrease in inotropic responses to calcium and isoprenaline. The sensitivity of mesenteric arteries to contraction by methoxamine has also decreased (Betrie et al. 2017). To sum up, this study showed the functional role of tau in cardiac tissue and the loss of tau induced by AD treatment would lead to damage of cardiovascular performance which worsens with age.

Previous research has suggested that increased tau aggregates were often accompanied by a decrease in the levels of soluble tau in the affected brain regions of patients with AD and PD (Lei et al. 2012). Soluble tau plays an important role in the brain as it helps in promoting the trafficking of APP to the neuronal surface in order to lower the iron levels present in the brain by promoting iron efflux (Brady and Sperry 1995; Lippincott-Schwartz et al. 1995). This prevents iron from accumulating in the brain over time. However, tau-directed therapeutic approaches for AD reduce tau aggregates and soluble tau in the brain. The loss of soluble tau would attenuate the efflux of iron, causing iron to accumulate in the brain, which in turn results in iron neurotoxicity (Lei et al. 2012, 2017). Other than that, neuronal iron retention was also seen in AD patients due to the inhibition of APP ferroxidase activity that helps in iron efflux by the accumulating zinc within the amyloid plaques found in AD (Duce et al. 2010; Lei et al. 2012).

Supporting evidence for the above statement was shown in the investigations conducted by Lei et al. The team had examined the presence of tau in the post-mortem brain samples collected from patients with Parkinson’s disease (PD) and healthy controls. With the use of blot densitometry, the results have shown lower soluble tau levels in the substantia nigra (SN) of PD patients when compared with healthy controls. In this study, results also demonstrated that iron levels in SN were 39% higher than those in the healthy controls, indicating the abnormal accumulation of iron in the SN region. The data also demonstrated that the deficiency of soluble tau could result in the accumulation of iron due to the impairment of APP trafficking to the neuronal surface. When comparing tau KO mice with age-matched WT mice of different ages (6, 12, 15, 18, 24 months old), brain atrophy was seen in tau KO mice with the conditions worsening with age. Age-dependent neuronal loss of SN, cognitive decline, and parkinsonism were seen in the mice. Iron levels in different regions of the brain have been assessed. It showed that the iron levels were significantly higher in the SN, hippocampus, and cortex (around 20% higher in each case) in 12-month-old tau KO mice compared to the WT mice of the same age. Both Fe(II) and Fe(III) were found to be abundant in the regions of the hippocampus and SN of tau KO mice, suggesting that the accumulation of iron causes age-dependent neurodegeneration in tau KO mice. Supporting this, there were no significant changes observed in 6-month-old young mice. This study has emphasized the role of tau in AD, PD, and other tauopathies whereby tau is not only a neurotoxic aggregate in the brain, but it also plays an important role in preventing age-related damage by regulating APP trafficking (Lei et al. 2012).

In a recent study published by Lei et al., lithium can suppress tau levels but can induce iron accumulation in the brain, resulting in neurodegeneration (Lei et al. 2017). By knocking down tau levels using lithium, iron-induced parkinsonism with cognitive impairment can be observed among mice. Considering both studies conducted by Lei et al., it is likely that decreased soluble tau could lead to neurotoxic iron accumulation in AD, PD, and other tauopathies, and hence more caution is needed when providing tau-directed therapies for AD due to tau’s unexpected physiological role in the central nervous system (Lei et al. 2012).

In September 2017, two papers published by Bi et al. and Tuo et al. have confirmed that young tau KO mice (3 months old) were protected against ischaemia–reperfusion injury after stroke, compared to WT mice where extensive cell death had been observed after stroke. The authors have demonstrated that it is likely that tau KO mice were protected against excitotoxicity due to the accumulation of a protective protein called synaptic RAS GTP-activating protein (SynGAP1). This protein exhibits silencing activity towards the excitotoxic signalling pathway in the post synapses of mice. The silencing of the excitotoxic signalling pathway enhanced the recovery of 3-month-old tau KO mice after stroke. However, the neuroprotection effect was not observed in older tau KO mice probably due to the accumulation of iron in the brain. When looking at the effect of excitotoxicity-induced brain damage in mouse stroke model created by blockage of middle cerebral arteries, data have shown that both groups of mice experienced local cell death at the blockage site and exhibited similar minor movement deficit after 1 h of injury. No further cell loss was observed over time in tau KO mice. In comparison to tau KO mice, the changes occurred in the brain of WT mice were more critical as the condition continued to worsen over time. Supporting this, massive cell death was observed. There is also evidence of a pattern of cell loss spreading throughout the affected hemisphere, causing severe motor deficits. The differences in the conditions observed between the tau KO mice and WT mice may be explained by the differential gene expressions in both mice after stroke. After administration of pentylenetetrazole (PTZ), the findings showed that ERK signalling was activated, which in turn results in the activation of Ras GTPase, causing excitotoxic genes in WT mice to be expressed. This mechanism explained why the condition worsens in WT mice over time. In contrast to tau KO mice, the presence of SynGAP1 leads to the inactivation of Ras GTPase, causing Ras deactivation through binding to PSD-95 in the post synapses. As the in vitro study showed that tau would compete with SynGAP1 for the binding site of PSD-95 in the post synapses, the increased levels of SynGAP1 would result in the silencing of the ERK signalling pathway. Therefore, the excitotoxic pathway was inactivated, which helps prevent further brain damage in tau KO mice. However, in the case of AD, further investigations are required to identify whether tau would compete with SynGAP1 binding to PSD-95 in the post synapses which resulting unable silencing excitotoxic signalling pathway and prevent further brain damage as what happening in stroke. Other than that, further investigation is also needed to determine whether SynGAP1 could be a good therapeutic target to prevent excitotoxicity (Bi et al. 2017).

Tuo et al.’s study has reported that there was no evidence of increased iron levels in the affected hemisphere in tau KO mice after blockage of middle cerebral arteries. Enhanced recovery with only little neurodegeneration was observed after ischaemic stroke. Hence, the authors hypothesized that the suppression of tau may prevent ferroptosis damage, which is a form of non-apoptotic cell death programme dependent upon intracellular iron level. However, differences have been observed in WT mice whereby the iron levels in the affected hemisphere had been shown to increase 50% several hours after the blockage, meanwhile tau levels in the particular area had decreased. Unlike tau KO mice, extensive neuronal cell death and motor deficits were observed in WT mice. In this case, the authors proposed a hypothesized statement whereby the presence of compensatory mechanisms in young tau KO mice may compensate for the loss of tau and help in iron export, reducing the accumulation of iron in the brain. However, neurotoxic iron accumulation increased with age. Supporting this, the findings reported that ageing tau KO mice accumulated 50% iron more than the WT mice. This has negated the neuroprotective effect of tau suppression against hemispheric reperfusion injury following middle cerebral artery occlusion (Tuo et al. 2017).

Findings also showed that the inhibition of ferroptosis exerts a neuroprotective effect on organisms (Dixon et al. 2012). Tuo et al. reported that 3-month-old WT mice treated with ferroptosis inhibitors, liproxstatin-1 and ferrostatin-1, after blockage showed better performance compared to untreated mice within 6 h. Taken together, further investigations are needed to assess the effects of the combination therapy to lower both iron and tau in AD patients (Tuo et al. 2017).

Conclusion Remarks

This article has discussed the pivotal roles of tau pathology in causing AD. Tau has played an important role in regulating the normal function of neurons, but it can also be involved in the pathogenesis of AD under certain conditions. The balance of the activities of both kinase and phosphatase is crucial to regulate normal tau phosphorylation level in the human brain. In AD, the imbalance of their activities leads to tau hyperphosphorylation. Tau hyperphosphorylation induced tau conformational changes, leading to synaptic impairment in AD. The propagation of tau pathology extracellularly was also discussed in this review. Hence, tau-directed therapies might be useful to slow down the progression of AD and other tauopathies. Accumulating evidence has shown that many drugs have been developed as inhibitors for therapeutic targets of AD; however, most of the drugs have only shown some symptomatic benefits. Some of the drugs such as TRx0237 and VEL015 have entered clinical trials for further evaluation as a potential disease-modifying therapy for AD. As a future perspective, drugs targeting tauopathies and drugs targeting amyloid deposition can be administered together to serve as a combination therapy that would be a potential therapeutic approach for AD. It is anticipated that the combination therapy of these drugs could be more efficacious than the anti-amyloid or anti-tau monotherapy. Most importantly, further understanding of the exact mechanism of AD is still of great importance in order to develop an effective treatment for AD. Only then, a more effective treatment regimen can be designed to cure the incurable disease.