The development of NFT is initiated by the formation of pre-tangles of oligomeric tau that assemble into insoluble filaments before aggregating to form NFT. Prior to, during or after this process, tau undergoes numerous, and potentially harmful, modifications. Therefore, though NFT may themselves be neurotoxic, the presence of some of these modifications may be indicative of tau-mediated damage that arose before their deposition. Indeed, tau-mediated neuronal death, in the absence of tau filaments, is observed in Drosophila and some transgenic mouse models overexpressing htau [
65‐
67]. Mice overexpressing htau with the P301L mutation (rTg4510 mice) do develop age-related NFT, neuronal loss and memory impairments. Yet, the subsequent suppression of the mutant tau stabilizes neuronal loss and improves memory function even though NFT continue to accumulate [
47]. In rTg4510, a regional dissociation between neuronal loss and the accumulation of NFT is observed; there is a loss of neurons in the dentate gyrus before NFT lesions appear and, conversely, NFT appear without major cell loss in the striatum [
68]. Likewise, many of the neurons that accumulate NFT in aged transgenic mice overexpressing normal htau seem "healthy" in terms of nuclear morphology, while a number of dying neurons do not appear to have a significant load of tau filaments [
69]. Furthermore, using models based on quantitative data on neuron loss and NFT formation as a function of disease duration, it is estimated that CA1 hippocampal neurons in AD can survive with NFT for approximately 20 years [
70]. Together, these studies suggest that tau-mediated neuronal death does not require the formation of NFT. Rather, non-filamentous tau, as well as abnormally modified tau intermediates, may be neurotoxic. Indeed, tau can undergo numerous post-translational modifications and some of these modifications, like phosphorylation and glycosylation, are believed to occur early in the development of tau pathology [
71,
72]. However, it is not yet known which tau intermediates are critical for the development of the different stages of neurodegeneration and by which mechanisms these intermediates cause cellular injury.
Tau hyperphosphorylation
The phosphorylation of tau plays a physiological role in regulating the affinity of tau for microtubules. Though less well studied, phosphorylation also regulates the binding of tau to signaling molecules and could thus influence tau-mediated signaling [
21]. Most of the phosphorylation sites on tau are present in the proline-rich and the C-terminal regions flanking the microtubule binding domains (Fig
1); (for review, see [
73]). The kinases that phosphorylate tau can be divided into two major groups, according to motif specificity: proline-directed protein kinases (PDPK) and non-proline-directed protein kinases (non-PDPK). The PDPK include cyclin-dependent kinase 5 (cdk5), mitogen-activated protein kinase, and several stress-activated protein kinases. GSK3-β is often described as a PDPK but the proline is not always required for phosphorylation by GSK3-β. Both cdk5 and GSK3-β co-purify with microtubules [
74,
75] and phosphorylate tau within a cellular environment [
76,
77]. The phosphorylation of tau by these kinases inhibits the ability of tau to promote microtubule assembly and facilitates the polymerization of tau into PHF [
78‐
81]. Among the non-PDPK are cyclic AMP-dependent protein kinase (PKA), calcium- and calmodulin-dependent protein kinase II (CaMKII), and microtubule affinity regulating kinase (MARK), the mammalian homologue of PAR-1. MARK targets KXGS motifs within the microtubule binding repeat domains (serine residues at 262, 293, 324 and 356) of tau [
82]. Tau phosphorylation at KXGS motifs induces its dissociation from microtubules and prevents its degradation [
83]. Unbound tau may then be hyperphosphorylated by other kinases. In fact, the phosphorylation of tau by MARK/PAR-1 may be a prerequisite for the action of downstream kinases, including GSK-3β and Cdk5 [
84]. There is also evidence that tau can be phosphorylated on tyrosine residues (Tyr18, Tyr29, Tyr197 and Tyr394) [
85‐
89].
Tau hyperphosphorylation is an early event in the pathogenesis of tauopathies, appearing before the development of NFT [
71]. Several missense mutations (G272V, P301L, V337M and R406W) in FTDP-17 result in tau proteins that are more favorable substrates to kinases
in vitro [
90]. In AD brains, the levels of total tau are approximately eight-fold higher than in age-matched controls, and this increase is due to higher levels of abnormally hyperphosphorylated tau, either polymerized into NFT of PHF or straight filaments, or present as a non-fibrillized form in the cytosol [
50,
91]. Elevated levels of hyperphosphorylated tau are also detected in cerebral spinal fluid of AD patients and may be predictive of neurodegeneration [
92,
93]. The increase in tau protein is not likely to result from increased transcription since several studies failed to observe increased tau mRNA levels in AD brains compared to controls [
94‐
98], though one study did report a relative downregulation of 3R-tau mRNA and an upregulation of 4R-tau mRNA in areas heavily affected by NFT [
99]. Since these studies did not examine tau mRNA expression at the cellular level, it remains possible that differences in tau mRNA levels between AD and normal cases occur in selective cell subpopulations. Interestingly, while one study found no change in tau mRNA isoform expression in AD, it did find that levels of mRNA for 4R-tau isoforms were increased in the brainstem, but not the fontal cortex or cerebellum, of patients with progressive supranuclear palsy [
98].
There is ample experimental evidence to support the view that hyperphosphorylated tau plays a pathological role in tauopathies. For example, the expression of pseudophosphorylated tau, which mimics disease-like tau hyperphosphorylation, causes apoptosis in neuronal cells, an effect not observed when cells express wild-type tau [
100]. The co-transfection of tau with GSK-3β in a cell culture model results in more cell death compared to the expression of tau and mutant (inactive) GSK-3β, suggesting that tau phosphorylation by GSK3-β is toxic [
101]. In a similar fashion, the activation of cdk5 by overexpressing p25 accelerates tau phosphorylation and aggregation in mice overexpressing mutant (P301L) tau [
102]. In fact, p25 overexpression and the ensuing cdk5 activation even contribute to tau pathology in mice expression only endogenous tau. Some studies have shown that p25 transgenic mice show increased tau phosphorylation compared to wild-type controls and, although NFT are not present, cytoskeletal components are disorganized, axonal swelling is observed, and the affected axoplasm is filled with abnormally clustered mitochondria and lysosomes, features consistent with loss of a functional microtubule network [
103,
104]. Cruz et al., (2003) also examined cdk5 activation on tau pathology and this group used bitransgenic mice that inducibly overexpress human p25 in the forebrains of mice. In these mice, a time-dependent increase in neuronal loss and astrogliosis is observed in the cerebral cortex between 5 and 12 weeks of cdk5 induction. Tau phosphorylation is increased in p25 transgenic mice compared to controls but there is no marked change in total tau protein levels. By 27 weeks of cdk5 induction, NFT pathology is visible in the cerebral cortex and hippocampus [
105]. Together, these results provide compelling evidence that aberrant tau hyperphosphorylation can lead to neurodegeneration, even in the absence of tau mutations or forced tau overexpression. Of interest, cdk5 activity is elevated in the prefrontal cortex of AD brains, where NFT are found, but not in the cerebellar cortex suggesting a relationship between deregulated cdk5 activity and tau pathology in humans [
106,
107].
Not only may increased kinase activity participate in tau hyperphosphorylation, but so may decreased tau dephosphorylation. Tau is dephosphorylated by protein phosphatase 2A (PP2A) and, to a lesser extend, by PP1, PP2B and PP5 [
19,
108‐
110]. In the human brain, PP2A, PP1, PP5 and PP2B account for approximately 71, 11, 10 and 7%, respectively, of the total tau phosphatase activity [
110]. The mRNA and protein expression of some phosphatases, as well as their activities, are decreased in affected areas of AD brain [
96,
110‐
114]. For example, in the AD hippocampus, PP2A and PP1 mRNA levels are decreased [
111] and the protein expression level of PP2A subunits is significantly and selectively decreased in AD-affected brain regions and in tangle-bearing neurons [
114]. Indeed, the progressive loss of PP2A subunit expression closely parallels the formation of tau lesions in discrete neurons [
114]. Compared to controls, phosphatase activity towards hyperphosphorylated tau is lower in gray matter extracts from AD brains [
112] and PP2A activity is decreased in homogenates from the frontal and temporal cortices [
114]. Of interest, one study found that the activities of PP2A and PP5 are decreased in the AD brain but PP2B activity is increased [
110]. Nonetheless, the total phosphatase activity in this study was significantly lower [
110] and another study has shown PP2B activity to be decreased in the AD brain [
113]. Together, these findings suggest that the downregulation of phosphatase activity, especially that of PP2A, can contribute to increasing levels of hyperphosphorylated tau. Consistent with this notion, PP2A inhibition by okadaic acid induces tau hyperphosphorylation and accumulation in rat brain slices [
109] and the inhibition of PP2A and PP1 activity by calyculin A injections into the rat hippocampus leads to tau hyperphosphorylation and defects in spatial memory retention [
115]. Moreover, transgenic mice with reduced neuronal PP2A activity exhibit increased tau hyperphosphorylation and the accumulation of tau aggregates in the soma and dendrites of cortical pyramidal cells and cerebellar Purkinje cells [
116].
Tau phosphorylation is also regulated by Pin1 (protein interacting with NIMA 1), a member of the peptidyl-prolyl
cis-trans isomerase group of proteins involved in the assembly, folding and transport of cellular proteins. The interaction between tau and Pin1 depends on the phosphorylation state of tau; Pin1 binds tau when phosphorylated at Thr231 [
117] and facilitates its dephosphorylation by PP2A [
118‐
120]. In AD neurons, Pin1 binds hyperphosphorylated tau in PHF, potentially depleting soluble Pin1 levels [
117,
121]. Pin1 is significantly down-regulated and oxidized in the AD hippocampus [
122]. Additionally, pyramidal neurons from AD brains that have lower Pin1 levels are more prone to contain tangles, whereas neurons with higher levels of Pin1 are generally tangle-free [
123]. Deregulation of Pin1 expression and activity could induce an imbalance in the phosphorylation-dephosphorylation of tau and negatively impact tau regulation and function. Indeed, Pin1 restores the ability of phosphorylated tau to bind microtubules and promote microtubule assembly
in vitro [
117]. It has been proposed that Pin1 functions as a co-chaperone and, together with HSP90 and other members of the HSP90 complex, is involved in the refolding and dephosphorylation of aberrantly phosphorylated tau [
83]. If Pin1 levels are knocked-down in Hela cells by siRNA prior to transfecting cells with wild-type tau, tau levels are decreased compared to Pin1-expressing cells [
83]. This suggests that when Pin1 levels are decreased, attempts to refold/dephosphorylate tau are subverted and tau degradation is favored. However, Pin1 knock-down
increases the stability of wild-type tau, as well as that of V337M and R406W mutant tau in SH-SY5Y cells [
124]. Differences in the results among these two studies may reflect differences in the culture models used and experimental design. It is also possible that, in the absence of Pin1 and its associated dephosphorylation and refolding activities, the degradation machinery may become overburdened, leading to tau accumulation. It should also be noted that, while knocking-down Pin1 increases the stability of wild-type tau and various mutant forms of tau in SH-SY5Y cells, it decreases the stability of P301L- and P301S-tau [
124] indicating that the effect of Pin1 on tau is mutation-dependent. Of interest, Pin1-/- mice develop age-dependent neuropathy, characterized pathologically by tau hyperphosphorylation, tau filament formation and neuronal degeneration in the brain and spinal cord [
123], thus providing another model in which the hyperphosphorylation of endogenous tau correlates with neuronal death. Conversely, Pin1 overexpression reduces tau levels and suppresses the tauopathy phenotype in transgenic mice expressing wild-type tau [
124]. However, in keeping with the opposing effects of Pin1 on wild-type tau and P301L-tau in SH-SY5Y cells, Pin1 overexpression exacerbates the tauopathy phenotype in P301L tau transgenic mice. Moreover, when Pin1-/- mice are crossed with transgenic mice overexpressing mutant (P301L) tau, P301L mutant tau levels are decreased and the robust tauopathy phenotype is abolished [
124].
Though many questions remain regarding the cause of aberrant tau phosphorylation in tauopathies, tau hyperphosphorylation is believed to play an important role in tau-mediated toxicity. Soluble hyperphosphorylated tau isolated from AD brains has lower microtubule-promoting activity
in vitro [
125] and sequesters normal tau, MAP1 (A/B) and MAP2, causing the inhibition of microtubule assembly and even the disassembly of microtubules [
126,
127]. These findings suggest that hyperphosphorylated tau can cause the breakdown of microtubules by interacting with microtubule associated proteins. Consequently, one could thus speculate that hyperphosphorylated tau is involved in the depletion and abnormal orientation of microtubules that is observed in the frontal cortex layers II and III in AD brains [
58]. An expected consequence of disarrayed or depleted microtubules is the impairment of microtubule-based transport, also an early event observed in AD [
128,
129]. As previously mentioned, loss of tau function alone may be insufficient to disrupt microtubule networks [
61]. However, the combined loss of tau and other microtubule-associated proteins could have more detrimental consequences on microtubule regulation. Consistent with this is the observation that mating tau-/- and MAP1B-/- mice leads to a lethal postnatal phenotype [
62].
Unlike the soluble form of hyperphosphorylated tau, the filamentous form of tau does not bind MAPs and does not disrupt microtubules
in vitro [
56]. Not only does this imply that tau filaments would have less of an impact on the microtubule network, the formation of filaments may, in fact, be a mechanisms adopted by neurons to sequester the toxic forms of hyperphosphorylated tau. However, if NFT
are detrimental to cells, and if tau hyperphosphorylation facilitates aggregation and filament formation, this could be one more mechanism by which tau hyperphosphorylation contributes to neuronal death. When hyperphosphorylated tau isolated from the AD brain is dephosphorylated by PP2A, the ability of tau to polymerize into PHF is inhibited. Conversely, the sequential rephosphorylation of tau by PKA, CaMKII, and GSK3-β or cdk5, as well as by GSK3-β and cdk5, promotes the assembly of tau into tangles of PHF similar to those observed in the AD brain [
130]. Yet, the
in vitro phosphorylation of recombinant tau promotes the formation of tau filaments in some studies [
130,
131] but not all [
132], putting into question the role of tau phosphorylation in enhanced filament formation.
Another mechanism by which tau hyperphosphorylation may contribute to neuronal toxicity is through its interaction with actin. In
Drosophila and mice, tau leads to the accumulation of filamentous actin into structures resembling the Hirano bodies observed in the brains of patients with AD or other tauopathies, like Pick's disease [
11]. Hirano bodies are intraneuronal inclusions that contain, among other proteins, actin and tau [
133,
134], and may play a causative role in AD [
135,
136]. The formation of Hirano body-like structures in neurons disrupts microtubules in neurites and could thus impair axonal transport and lead to synapse loss [
135]. Fulga et al., (2007) have shown that phosphorylated tau can induce changes in the actin cytoskeleton and lead to toxicity. The retinal expression of pseudophosphorylated tau in
Drosophila induces a striking accumulation of actin in the lamina and produces substantial toxicity. Conversely, the expression of phosphorylation-incompetent tau does not lead to actin accumulation and only causes mild toxicity [
11]. These results suggest that phosphorylated tau can cause neuronal death by inducing changes in the actin cytoskeleton.
Overall, though tau hyperphosphorylation is implicated in tau pathology, it is still not fully understood which of the tau phosphorylation sites are critical for the development of tauopathies, nor is it decidedly known how hyperphosphorylated tau causes neuronal death. A better understanding of the physiological roles of tau phosphorylation, as it regulates the binding of tau to microtubules and affects other less well characterized functions of tau, will likely shed light on the mechanisms by which tau hyperphosphorylation contributes to cell death.
Other tau modifications
Intimately linked to tau phosphorylation is tau glycosylation. Glycosylation is characterized by the covalent attachment of oligosaccharides to protein side chains. Glycosidic bonds are classified as either N-linked or O-linked. In N-linked glycosylation, the sugar is linked to the amide group of asparagine residues of proteins, while in O-linked glycosylation, sugars are attached to a hydroxyl group of serine or threonine residues. Hyperphosphorylated tau and PHF-tau purified from AD brains are glycosylated, mainly through N-linkage [
137,
138]. Additionally, non-hyperphosphorylated tau isolated from AD brains is also glycosylated, whereas no glycan is detected in tau purified from normal control brains [
137], suggesting that aberrant glycosylation precedes abnormal tau hyperphosphorylation. Indeed, glycosylation facilitates the site-specific phosphorylation of tau catalyzed by PKA, cdk5 and GSK-3β [
137,
139]. Conversely, glycosylation appears to inhibit the dephosphorylation of tau by PP2A and PP5 [
140]. Tau glycosylation may also coordinate with hyperphosphorylation to stabilize the filamentous structure of PHF given that deglycosylation of PHF untwists PHF into straight filaments [
137]. Together, these findings suggest that aberrant N-linked glycosylation is an early tau modification that enhances tau hyperphosphorylation, which may drive NFT formation, and also help maintain and stabilize NFT structures.
In addition to N-linked glycosylation, human brain tau can be modified by O-linked monosaccharide β-N-acetylglucosamine (O-GlcNAc) [
141]. O-GlcNAcylation regulates tau phosphorylation in a site-specific manner in both cultured cells overexpressing htau and in rodent brains; at most of the phosphorylation sites examined, O-GlcNAcylation reduces tau phosphorylation [
141]. Consistent with this finding, in neuroblastoma cells transfected with htau, O-GlcNAc mainly modifies the less-phosphorylated tau species, while highly phosphorylated tau is devoid of O-GlcNAc residues [
142]. In starved mice, a model used to mimic the reduction in glucose uptake and metabolism observed in the AD brain, O-GlcNAcylation is decreased and tau hyperphosphorylation is increased in the brains of the mice [
141]. In the AD brain, the level of O-GlcNAcylation is lower than that in control brains, indicating that O-GlcNAcylation is compromised [
141]. Based on these findings, it was proposed that impaired glucose metabolism in AD may contribute to disease pathogenesis by reducing tau O-GlcNAcylation and, consequently, increasing tau phosphorylation [
143]. Yuzwa et al., (2008) have shown that Thiamet-G, an inhibitor of O-GlcNAcase that enhances O-GlcNAcylation, markedly reduces tau phosphorylation in PC12 cells at pathologically relevant sites, like Thr231 and Ser396. Moreover, Thiamet-G also efficiently reduces phosphorylation of tau at Thr231, Ser396 and Ser422 in both the rat cortex and hippocampus [
144]. Together, these findings underscore the dynamic relationship between the O-GlcNAcylation and phosphorylation of tau.
Besides phosphorylation and glycosylation, tau undergoes other changes that could enhance tau self-assembly and filament formation and may confer toxic gains or loss of function. For instance, the proteolytic cleavage of tau coincides with the pathogenesis of AD. Granular aggregations containing tau truncated at Glu391 are detected within the somatodendritic compartment of AD brains but not in age-matched non-demented controls [
145], Glu391-truncated tau is present in PHF isolated from AD tissue [
146‐
148] and tau-truncated at Asp421 associates with neurofibrillary pathology in AD brains [
149‐
151]. Tau cleaved at Glu391 and/or Asp421 is also observed in Pick's disease, progressive supranuclear palsy and corticobasal degeneration [
152‐
154].
The truncation of tau accelerates its assembly into fibrils
in vitro [
149,
155,
156], promotes microtubule assembly
in vitro more than full-length tau [
157], and increases its association with microtubules [
158]. The effect of tau phosphorylation at Ser396/Ser404 on microtubule binding differs between full-length tau and tau truncated at Asp421, indicating that specific tau forms (e.g. intact versus cleaved tau) respond differently to site-specific phosphorylation [
158]. Notably, transgenic rats that overexpress truncated tau species (aa 151–391) in the brain and spinal cord develop neurofibrillary pathology [
157], and cultured cortical neurons derived from these rats have fewer mitochondria in neuronal processes, display higher levels of reactive oxygen species and are more susceptible to oxidative stress compared to cultures from non-transgenic rats [
159]. Consistent with these findings, the expression of tau fragments cause cell death or render cells more sensitive to insults in various culture models [
160‐
163].
Taken together, the above findings suggest that tau cleavage is neurotoxic. However, there is some debate as to whether tau cleavage occurs before or after the aggregation of tau into NFT. On the one hand, Guillozet-Bongaarts et al., (2004) have shown by immunohistochemical studies that tau truncation at Asp421 occurs only after the Alz50 conformation change in tau, the presence of which is indicative of the appearance of filamentous tau [
164]. On the other hand, the deletion of CHIP, a tau ubiquitin ligase, leads to the accumulation of non-aggregated, hyperphosphorylated and caspase-cleaved tau in mice, suggesting that tau hyperphosphorylation and caspase-3 cleavage both occur prior to aggregate formation [
165]. Indeed, Rissman et al. (2004), show that in both transgenic mice and in AD brain, caspase-cleaved tau at Asp421 associates with early and late markers of NFT and correlates with cognitive decline [
150].
In addition to the incorporation of truncated tau into NFT, the PHF and NFT in AD brains are glycated [
166] as well as ubiquitinated [
167,
168], but these modifications are believed to be later events in disease progression. Nitrated tau is also detected in cytoplasmic inclusions in AD, corticobasal degeneration, Pick's disease, progressive supranuclear palsy and FTPD-17 [
169]. Tau-nY29, an antibody specific for tau when nitrated at Tyr29, detects soluble tau and PHF-tau from severely affected AD brains but fails to recognize tau from normal aged brains, suggesting that tau nitration is disease-specific [
170]. The exact mechanisms by which nitrated tau contributes to pathology, however, remain poorly understood. Nitration can greatly affect protein folding and function [
171,
172]. Peroxynitrite (ONOO-), which is capable of both protein nitration and oxidation [
173], leads to tau oligomerization
in vitro and in neuroblastoma cells [
174,
175]. Yet, it is believed that this effect results from the oxidative role of peroxynitrite and the formation of dityrosine bonds in tau [
175]. The overall effect of tau nitration by peroxynitrite
in vitro is to delay the polymerization of tau into filaments [
175,
176]. The toxicity of tau nitration may instead result from the inhibitory effect of nitration on the ability of tau to promote tubulin assembly which could compromise microtubule function [
177].
Tau mutations
Even though no mutations in tau have been identified in AD or sporadic cases of frontotemporal dementia, understanding how mutations in tau confer toxicity in FTDP-17 should provide insight on the role of tau in the development of neurodegeneration. At least 34 mutations in the human
MAPT gene, falling into two functional classes, have been reported (Fig.
1) [
178]. The first class of mutations, which includes missense and deletion changes in the coding region of
MAPT, generates tau proteins with altered function. These mutations can reduce the binding affinity of tau for microtubules [
38,
39]. LeBoeuf et al., (2008) have shown that FTDP-17 tau mutations that map to the repeat/inter-repeat region of tau compromise its ability to regulated microtubule dynamics
in vitro [
179]. However, cells transiently expressing mutant (P301L or R406W) or wild-type tau are indistinguishable in terms of the co-localization of tau with microtubules and the generation of microtubule bundles [
180], implying that these tau mutations do not have an immediate impact on the integrity of the microtubule system. In addition to impaired microtubule binding, first class mutations enhance the ability of tau to aggregate and form filaments
in vitro [
41,
42,
44]. Insoluble aggregates in patients with the P301L mutation consist largely of mutant 4R-tau, with only small amounts of normal 4R- and 3R-tau [
181]. The selective trapping of P301L tau in the insoluble deposits is presumably caused by the increased aggregation potential conferred by the mutation. It is tempting to speculate that the combined effects of altered microtubule regulation and accelerated NFT formation caused by mutations in tau contribute to tau-mediated toxicity or, at the very least, render cells more vulnerable to age-related stressors.
The second class of mutations affects the alternative splicing of
MAPT transcripts, mainly influencing exon 10 splicing and leading to a change in the ratio of tau isoforms with three of four microtubule binding repeats. In the normal adult brain, the ratio of 4R- to 3R-tau is approximately 1. Many of the second class mutations increase this ratio [
29], suggesting that 4R-tau is the more toxic isoform. However, while only 4R-tau aggregates into twisted and straight filaments in corticobasal degeneration and progressive supranuclear palsy, NFT in AD brains contain both 3R- and 4R-tau, and 3R-tau inclusions are primarily observed in Pick's disease [
182‐
184]. Therefore, neurodegeneration may not result from one isoform being more toxic than another, but rather from an imbalance in the proper ratio of 3R- to 4R-tau. One hypothesis proposes that since splicing mutations cause an excess of a specific tau isoform and, since 3R- and 4R-tau bind microtubules at different sites [
185], a shortage of available binding sites would occur for the overexpressed tau isoform [
186]. This could lead to an excess of free tau available for filament assembly. It is also highly probable that abnormal changes in isoform expression would adversely affect tau function. Given that various tau isoforms are differentially expressed during development, differentially distributed in neuronal subpopulations and even present in distinct localizations within neurons [
187], it is likely that they have specific functions. For example, different tau isoforms have dramatically different effects on the rate and number of motors driving the cargo along microtubules [
188]. As our understanding of the functions carried out by distinct tau isoforms grows, so will our understanding of how alterations in their expression levels contribute to neuronal dysfunction.