Introduction

The microtubule (MT)-associated protein tau (MAPT) is an intrinsically disordered protein expressed at its highest levels in neurons throughout the central nervous system. Higher molecular mass isoforms generated through alternative splicing, often termed “big tau,” are expressed primarily in the peripheral nervous system, but are sometimes also observed in the spinal cord and skeletal muscle, when exon 4a and exon 6 are translated, respectively [1,2,3]. One of tau’s primary functions is to bind to and promote the assembly and stability of MTs; this binding activity can be negatively regulated by phosphorylation at select sites [1, 4].

Tauopathies refer to a wide range of phenotypically diverse diseases characterized by the aberrant aggregation of tau in neurons and/or glia, including Alzheimer’s disease (AD), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), Pick’s disease (PiD), chronic traumatic encephalopathy (CTE), and frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17) [5]. First discovered as a MT-associated protein in 1975 [6], tau was later found to be the principal component of neurofibrillary tangles (NFTs), which are hyperphosphorylated proteinaceous inclusions found in AD and other tauopathies [7].

In 1998, autosomal dominant mutations in the MAPT gene that encodes tau were found to cause some forms of FTDP-17 [8,9,10], now referred to as FTDP-17t, proving that tau dysfunction is sufficient for widespread central nervous system neurodegeneration. Disease pathology for individuals with FTDP-17t is characterized by the presence of filamentous tau inclusions throughout the frontal and temporal lobes in neurons and sometimes in glia, accompanied by atrophy in these regions, as well as ventricular dilation [11]. These MAPT mutations can cause variable cognitive, behavioral, and motor deficits, with an average age of onset of 49 years and a duration of disease of 8.5 years [12].

As of 2018, over 50 different pathogenic MAPT missense, silent, and intronic mutations have been reported (Figs. 1 and 2, Table 1) [11]. Because many of these mutations present neuropathologically in a manner consistent with different sporadic tauopathies such as PSP, CBD, and PiD [13,14,15], there have recently been calls to characterize tauopathy caused by certain mutations as familial versions of these specific diseases [16]. Additionally, some of these mutations have been found to be risk modifiers in certain tauopathies, for instance, A152T in AD [17]. In addition to the phenotypic and neuropathological variability between these mutations, there are also a number of mechanisms by which these mutations are thought to cause disease. Loss of function, including MT binding and assembly, changes in alternative splicing, shifts in protein-aggregation kinetics, and, more recently, prion-like “seeding,” have all been implicated. Thus, this review focuses on the potential biochemical and cellular mechanisms in which different tau mutants might cause disease—with an emphasis on their ability to aggregate with seeding—and the implications this might have on the study of sporadic tauopathy.

Fig. 1
figure 1

The longest tau isoform found in human brain, with its corresponding mRNA and known pathogenic missense and deletion mutations. The MAPT mRNA resulting in 2N4R tau is shown with an embedded number corresponding to the originating exon. Exon 1 contains both untranslated 5’ region and the start of the protein. Exons 2 and 3 are present in this isoform as N1 and N2 inserts; however, in the 1N and 0N isoforms of tau, exon 2 or neither exon 2 or 3 is translated, respectively. MT-binding repeat 2, or R2, is present in this isoform; however, in 3R tau, exon 10 is alternatively spliced, and this region is not present in the protein. The different colors serve to highlight the regions of the protein that are alternatively spliced as well as the MT-binding domain. The N- terminal, proline-rich, MT-binding, and C-terminal regions are indicated above. Below the protein, known pathogenic missense mutations are indicated. Many of these mutations reside in the MT-binding region, and as such the specific amino-acid sequence of this area is shown. The PHF6* and PHF6 motifs, which are important for tau aggregation, are also indicated. MT, microtubule

Full size image
Fig. 2
figure 2

Schematic of the RNA stem loop present at the end of exon 10 and the beginning of intron 10. Known pathogenic exonic (both missense and silent) and intronic mutations with their corresponding amino-acid changes or deletions, when applicable, are shown. The boundary between exon 10 and intron 10 is indicated by the partition at the top left and also by the use of lower case letters for intron 10. Mutations in this region that have been shown to increase exon 10 inclusion are indicated in red, while those that have been shown to decrease exon 10 inclusion are indicated in green. Mutations that have not been shown to affect splicing are represented in black

Full size image
Table 1 Summary of the reported effects of mutations within the MAPT-coding regions on in vitro tau amyloid aggregation, tau’s ability to promote MT assembly, tau MT binding, altered exon 10 splicing, and the major tau isoforms present as aggregates in human brains
Full size table

Tau expression and splicing

The MAPT gene, located on chromosome 17, comprises 16 exons, numbered 0–14 [5]. Exon 1 contains both 5’ untranslated region as well as the start codon of the protein, while exon 14 contains untranslated 3’ region. Splicing variants that include exon 4a are primarily present in the peripheral nervous system, while variants that include exon 6 can be found specifically in the spinal cord and skeletal muscle, resulting in a higher molecular mass protein referred to as “big tau” [1,2,3, 5]. In the human brain, six distinct isoforms of tau exist based on the alternative splicing of exons 2, 3, and 10 [18]. Alternative splicing of exons 2 and 3 yields isoforms with 0, 1, or 2 N-terminal repeats (0N, 1N, 2N), while alternative splicing of exon 10 results in tau with three or four MT-binding repeats in the MT-binding domain (3R or 4R) (Fig. 1). While 0N3R tau is the predominant isoform in fetal brain [19], the overall ratio of 3R and 4R tau isoforms is roughly equal in the adult brain [1, 18, 20], although this ratio can differ in given regions [21]. 0N and 1N tau isoforms comprise 37 and 54% of the total human brain tau, respectively, while 2N tau makes up only 9% of the total tau isoforms [22]. 4R tau isoforms show both increased affinity for MTs as well as greater levels of MT assembly in vitro compared to 3R tau isoforms [23, 24]. Although the role of N-terminal inserts is less clear, they have been implicated in regulating MT stabilization and plasma membrane interactions [25, 26]. Additionally, co-immunoprecipitation studies in mice have shown that 0, 1, and 2N isoforms interact with different proteins preferentially [27] and that there are regional differences in the expression of these isoforms in the brain [28].

Alternative RNA splicing of tau is regulated by several cis elements and trans-acting factors [29]. Exon 10, in particular, is flanked by an abnormally large intron 9 and has a weak 5’ and 3’ splice site, which can be acted upon by these cis elements and trans-acting factors [30]. Additionally, the 3’-end of exon 10 and the 5’-end of intron 10 form a highly self-complementary stem loop (Fig. 2), which inhibits the binding of the U1 small nuclear RNA (snRNA) molecule, part of the small nuclear ribonucleoprotein particle (snRNP) that functions to bind to the pre-mRNA and catalyzes the removal of introns [29]. Disruption of this loop leads to increased binding by the snRNP and higher levels of exon 10 inclusion [30]. This destabilization can also be seen in rodents, in which the presence of a guanine instead of an adenine at position IVS10 + 13 leads to the predominance of 4R tau [29]. Thus, mutations within specific cis elements can promote or suppress the inclusion of this exon, while mutations specifically within the stem loop tend to promote exon 10 inclusion (Fig. 2).

Two major haplotypes of tau, H1 and H2, are formed primarily by the 900 kb inversion in the q21 region of chromosome 17 and a 238 bp deletion in intron 9 in H2 [3]. Furthermore, a number of single-nucleotide polymorphisms in the H1 haplotype produce several sub-variants, some of which are associated with an increased risk of certain tauopathies, such as CBD and PSP [31,32,33]. Mechanistically, it is thought that distinct haplotypes, particularly H1c, can promote tauopathy through the increased expression of 4R tau, as is seen in some FTDP-17t mutant tauopathies [34].

Tau structure and aggregation

Four general domains of tau include the N-terminal acidic projection domain, the proline-rich domain, the MT-binding domain, and the C-terminal tail (Fig. 1). Although tau is intrinsically disordered and natively unfolded [35, 36], it can adopt a “paperclip” like structure, in which the MT-binding domain and the N terminus approach and interact with the C terminus [37]. Preclusion of this structure, through the interaction with other molecules, post-translational modifications, truncations, or mutations, could potentiate abnormal aggregation. The MT-binding repeats also comprise the “paired-helical filament core,” or PHF core, which serves as the primary structure of aggregated tau filaments [38]. Within this structure are two hexapeptide motifs, termed PHF6* and PHF6, which are important for aggregation, and the latter is necessary and sufficient for β-sheet aggregation in tau [39,40,41]. Recent cryo-electron microscopy studies of the PHF core from both AD and PiD tau filaments further confirm the key areas of β-sheet forming residues in this core [42, 43]. Overall, the missense MAPT mutations that disrupt the proposed “paperclip” structure of tau, or promote and stabilize the PHF core of tau, are likely to promote aggregation and inclusion formation.

The process by which tau polymerizes to form amyloid in vivo is not completely understood. Nevertheless, “nucleation-elongation” is a potential mechanism that can contribute to this process in which tau initially forms an oligomeric nucleus or “seed” before elongating into tau filaments [44, 45]. As such, the formation of this oligomeric nucleus is the rate-limiting step, after which tau can elongate rapidly by attaching to the growing ends of the fibril [45,46,47]. It has recently been proposed that tau can undergo liquid–liquid separation to form condensed liquid droplets within cellular physiological conditions, which could serve as a precursor for this initial tau β-sheet formation and aggregation [48]. Experimentally, this rate-limiting step can be potentially overcome by introducing the pre-formed “nuclei” of tau that can act as a template for soluble tau to quickly bind to and polymerize onto, in a process known as seeding [49]. This mechanism is akin to the misfolding of prion protein, in which exogenous or intrinsic pathogenic prion conformers act as templates that induce conformational changes in the native protein, inducing misfolding, further aggregation, and neurodegeneration [50]. In a similar manner, it is thought that certain neurodegenerative proteinopathies can spread to anatomically connected regions through template-assisted conformational changes, in which soluble protein is induced to become pathological. This concept is further supported based on AD autopsy series studies, as Braak stages I–VI [51], in which tau pathology in AD can be defined in a regionally specific and somewhat predictable manner. Evidence that tau aggregation can be induced by exogenous tau aggregates and subsequently spread in this manner—through some combinatorial processes of synaptic or exosomal secretion followed by endocytotic or exosomal uptake, for example [49, 52]—has been shown in vivo in various types of cell culture systems and in animal studies (Tables 2–4). Of note, many of these studies heavily utilize specific mutants that may serve to enhance or act as a primer for this process (Tables 2–4).

Table 2 Summary of tau seeding studies in cultured cells
Full size table
Table 3 Summary of in vivo tau seeding experiments
Full size table
Table 4 Summary of in vivo tau seeding experiments utilizing PS19 mice
Full size table

Tau function and post-translational modifications

Tau resides mostly in the axons of developed neurons, where it has a higher affinity for MTs than in the dendrites [53, 54]. Additionally, different isoforms of tau can have distinct localization patterns, and missorting of tau into dendrites is an early sign of neurodegeneration in AD [53, 55, 56]. Thus, altered splicing patterns may contribute to tau mislocalization and altered MT stability. Normal tau has roles in regulating axonal transport [57] and promoting neurite outgrowth [58]. Knockout mice have further demonstrated important roles for tau in neurogenesis and neuroplasticity, with significant impairments to both in the absence of tau [59, 60]. Tau also binds to and interacts with a number of other molecules. In particular, the N-terminal domain, which has a negative charge and projects away from the MTs when tau is bound [61], can act as a link to membrane components, particularly annexin 2 [62]. This region also binds to the p150 subunit of the dynactin complex, which regulates the MT motor protein dynein [63].

Given that tau functions to enhance MT assembly and regulate its stability, which play important roles in neurite outgrowth, cell stabilization, and intracellular transport, normal tau activity contributes to maintaining axonal health. Tau binds to the interface between α- and β-tubulin heterodimers specifically with residues interspersed throughout and around the MT-binding repeats [64]. Thus, tau mutations within these repeats have the ability to disrupt this interaction, resulting in the destabilization of MTs as well as a potential increase in unbound, free-floating tau, which may also promote tau aggregation (Table 1).

Normally, the process by which tau interacts with MTs is negatively regulated by phosphorylation [4]. There are over 80 potential phosphorylation sites (i.e., serine, threonine, and tyrosine residues) on the longest isoform of tau, a number of which are abnormally hyperphosphorylated in AD and other tauopathies [53]. This hyperphosphorylation may induce the missorting of tau [65], as well as potentially promote aggregation, as shown in vitro [66]. Individual missense mutations in tau can not only alter potential phosphorylation sites, but also have been shown to promote phosphorylation compared to WT tau in vitro [67]. Lysine acetylation has also been shown to be of importance with regard to tau pathology. Depending on the residue, acetylation can inhibit tau’s degradation and correlate with tauopathy, or promote its degradation and suppress aggregation [68, 69]. Other post-translational modifications include glycosylation, isomerization, methylation, ubiquitination, and truncation [70]. N-glycosylation, isomerization, and truncation are implicated in the promotion and stabilization of PHFs [71,72,73], while methylation has been shown in vitro to suppress aggregation [74].

Altered Tau mRNA splicing in disease

Normally, the ratio of 3R to 4R tau in adult human brain is approximately equal [1, 18, 20]. In AD, this ratio remains normal [75]; however, many tauopathies exhibit altered ratios of tau isoforms, especially within the pathologic inclusions. For instance, PSP and CBD are considered 4R tauopathies, while PiD is considered a 3R tauopathy [1]. Interestingly, specific MAPT mutations can cause either 4R or 3R predominant tauopathies as well as tauopathies with equal isoform ratios [76]. Intronic pathogenic mutations largely reside in intron 10 and serve to disrupt the mRNA stem loop, usually causing an increase in 4R tau (Fig. 2). For instance, the most common intronic mutation, IVS10+16 (C to U), serves to increase 4R tau expression by disrupting this stem loop [11], as do a number of other intronic mutations in this area (Fig. 2). Additionally, IVS10+11 (U to C), IVS10+12 (C to U), IVS10+13 (A to G), IVS10+14 (C to U), and IVS10+16 (C to U) reside on an intronic splicing silencer and all increase exon 10 inclusion, while IVS10+19 (C to G) resides on an adjacent intronic splicing modulator and increases exon 10 splicing [77]. Interestingly, the IVS10+13 (A to G) mutation is naturally occurring in rodents, resulting in a preponderance of 4R tau in these animals [29]. Deletion studies have shown that these two regions have opposing effects on splicing [29, 30]. Two missense mutations that affect splicing in opposing ways are N279K and Δ280K, which are both found around a polypurine enhancer, which interacts with a number of different regulatory splicing sequences [77]. While the Δ280K mutation weakens this enhancer, the N279K mutation strengthens it, producing dramatically opposing effects on exon 10 inclusion. The majority of missense mutations that are found in exon 10 are shown to increase the levels of 4R tau expression as revealed by mRNA analysis and/or the protein biochemical profile of patients’ tau isoforms (Table 1). These mutations likely cause disease either through the direct disruption of the mRNA stem loop (Fig. 2) or through the disruption of splicing enhancers and silencers [18, 77]. Furthermore, of the silent mutations found in tau, only those residing in exon 10 (L284L, N296N, or S305S) or exon 11 (L315L) have shown to be pathogenic and could also affect splicing by the disruption of the stem loop (S305S) or by disrupting or enhancing the splicing cis elements (Fig. 2) [77].

Alterations in the normal ratio of tau isoforms can lead to a number of adverse results, including impaired axonal transport [78]. Additionally, for all of the mutations that affect splicing, insoluble tau levels from affected patients’ brains are predominantly but not exclusively comprised the isoform that is over represented (Table 1) [79, 80]. Furthermore, because it is proposed that different tau isoforms harbor distinct MT-binding affinities and potentially bind unique sites, an overproduction of one isoform over another might lead to an overabundance of free-floating, unbound tau, which is primed for aggregation [23, 24, 81].

Altered MT assembly and binding due to tau mutations

Tau missense mutations can affect MT assembly and binding, thus reducing MT stability, as well as potentially leading to increased levels of unbound, free-floating tau in the cytosol of neurons. In fact, one of most common features among missense mutations is their diminished ability, at least in vitro, to promote MT assembly from tubulin compared to WT tau (Table 1). Conversely, a few mutations, such as Q336H and Q336R, have a greater ability to promote MT assembly (Table 1) [82, 83]. Additionally, some missense mutations, such as S305N, that predominantly affect splicing demonstrate little effect on MT polymerization (Table 1) [84]. The impact of tau missense mutations on its interactions with MTs has also been studied in vitro by comparing the binding affinity for taxol-stabilized MTs (Table 1). Although this interaction has been not been as extensively studied as MT assembly, typically, these data demonstrate a reduced MT-binding affinity that is consistent with a reduced ability to induce MT polymerization (Table 1).

Impacts of tau mutations on aggregation and seeding

Mutations that alter splicing dynamics and that induce loss of function with regard to MT interaction both have the potential to increase the amount of free, unbound forms of tau in the cytosol that can ultimately potentiate aggregation and inclusion formation. In vitro, recombinant tau needs a polyanionic inducer, such as heparin or arachidonic acid, in order to form tau filaments [85]. In the cytosol, similar cofactors could promote this aggregation, including RNA [86, 87]. In addition to specific cofactors, mutations in tau may help to induce aggregation, making tau more susceptible to template-assisted growth. In vitro, a number of tau mutants have been shown to increase the rate at which tau fibrillizes (Table 1).

Some tau mutations can also have marked effects on seed-induced tau aggregation in vivo. Growing evidence in cell culture supports prion-like seeding as a possible mechanism that contributes to pathogenesis (Table 2). Original studies conducted have shown that the addition of exogenous pre-formed tau fibrils can induce tau aggregation in cells expressing WT human tau protein [88, 89]. Other studies have focused on the seeding and aggregation of repeat domain (RD) tau, or tau only containing the MT-binding repeats, as these constructs contain the region responsible for tau fibrillization and are more prone to aggregation when seeded or even simply expressed in culture [44, 90,91,92]. The majority of cell culture studies, however, have utilized tau mutants, namely the P301L or P301S mutants, which robustly aggregate with seeding compared to WT tau [91,92,93,94,95,96]. Even when comparing WT RD tau with RD tau containing a P301L or P301S mutation, these differences in aggregation remain [90, 92].

Additionally, the types of tau seeds that were used to treat these tau-expressing cells were largely divided into brain lysate from transgenic mice expressing human tau with a P301L or P301S mutation, human brain lysate from patients with tauopathy, or recombinant tau protein fibrillized in vitro using a polyanionic inducer (Table 2). Comparative studies have shown that tau from brain lysates are more potent at inducing aggregation than recombinant fibrils; however, recombinant tau that was “seeded” in cell culture by detergent-insoluble tau from brain lysate can acquire this potency [94].

Evidence for seeding in vivo has been shown in a number of mouse models (Tables 3 and 4), beginning with Clavaguera et al., where pathology was found at the site of injection with insoluble P301S brain extract in human WT-overexpressing mice 6 months post-injection, with limited spread to nearby regions thereafter [97]. Subsequent studies have mostly utilized mice expressing human mutant P301L or P301S tau, which have shown significant seeding and synaptic spread of pathology with a variety of “seeds” utilized (Tables 3 and 4). These mice were treated, largely, through stereotactic hippocampal injection at 2–3 months, with brain homogenates from different tauopathies [98], lysate from aged P301L or P301S tau-expressing mice [99, 100], lysate from cells with stably expressing tau aggregates [92, 101], or recombinant fibrils, usually the K18 tau fragment (residues 244–372) [102,103,104,105,106].

Brain lysates from human tauopathy cases or from aged mutant tau transgenic mice containing tau aggregates appear to generate the most potent seeds [94]. Further, using sarkosyl-insoluble lysate or immunoprecipitating tau from these brains rather than total lysate appears to enhance their potency [94], while immunodepleting tau greatly diminishes this potency [100]. This, along with evidence that “strains” of tau aggregates can survive multiple passages through both cell culture and mice [101], points to the importance of conformational templating in this process. Isoform differences could also contribute to differential templating; because the P301 residue only exists in 4R tau and because murine tau is almost exclusively 4R [19, 53], most studies have only compared seeding of 4R-expressed tau with 4R aggregates. This differs from the physiological conditions of AD and PiD, as well as from many instances of FTDP-17t. In fact, when PiD brain lysates, which predominately contain 3R tau, were used to seed 4R tau in vivo, there was significantly less pathology [107]. This apparent seeding barrier has also been seen in vitro and in cell culture [89, 108,109,110,111], and corroborates the different proposed structures of tau filaments from AD and PiD elucidated by cryo-electron microscopy [42, 43].

Differences in the entity of seeds comprising either WT or mutant tau appear to matter less than the methods used to generate the seeds (Tables 2–4). For the monomer to be seeded, however, the data in both cell culture and in vivo show the importance of whether WT tau or mutant tau is utilized. Specifically, studies have shown consistent differences in seeding and aggregation propensities between P301L, P301S, and P301T mutant and WT tau in vitro, in cell culture, and in mouse models [90, 92, 94,95,96, 109, 112, 113].

A recent study from our laboratory investigated 19 different pathogenic tau missense mutations in the context of an established cellular seeding assay [109]. It was found that the majority of mutants, including WT tau, failed to robustly aggregate with seeding, unlike tau with mutations at the P301 site. This pattern of aggregation with seeding was similar among the known FTDP-17t mutants P301L, P301S, and P301T, and even a deletion at the P301 site; additionally, there was no difference in seeding between the 0N4R or 2N4R isoforms [109]. Because prolines can serve as inhibitors of β-sheet formation [114], the impact of other proline residues throughout the MT-binding region was investigated; it was found again that the inhibitory effect of P301 on aggregation and seeding was unique, but that re-introducing a proline at residue 302 was sufficient to inhibit seeded aggregation [109], demonstrating the importance of the local molecular environment.

Another unique mutant that affected tau aggregation was S320F, which was also able to aggregate with seeding in this assay and even showed some ability to aggregate without seeding [109]. Other studies have shown that this mutant is prone to aggregate in vitro and in vivo compared to WT and other tau mutants [112, 115]. More specifically, this mutation has conferred greater rates of tau nucleation, leading to the production of short fibrils [112], which could explain this mutant's ability to aggregate without the addition of pre-formed fibrils, since the lag time to create nucleated “seeds” is shorter than that for other tau mutants. Mechanistically, this mutation could promote aggregation in a number of different ways. First, while cryo-electron microscopy findings of tau from AD PHFs have shown that, within the amyloidogenic fold, the S320 residue could reside within a hydrophobic pocket [42], the same group found that in PiD tau filaments, cryo-electron microscopy showed S320 to most likely reside within a hydrophobic pocket [43]; thus, the S to F mutation could act by stabilizing this structure. Agreeing with the proposed structure of PiD tau filaments, S320F has been neuropathologically compared to PiD, with an abundance of Pick bodies found in a carrier’s brain [116]. Additionally, despite being intrinsically disordered, tau can adopt a protective paperclip-like global conformation, in which the MT-binding repeat and C-terminal and N-terminal domains approach each other [37]. The substitution of a large, aromatic side chain in the region where the C-terminal and MT-binding domains interact could potentially disrupt this fold, facilitating polymerization. This disruption of the paperclip-like structure of tau has been shown in vitro with pseudo-phosphorylation of tau at the AT8 and PHF1 epitopes, which are markers of tau pathology [117]. Given these two mutants’ unique properties, it was also found that a double-mutant combination of either P301L or P301S with S320F resulted in a rapid and robust aggregation even without seeding [109].

The presence of aggregated tau inclusions remains a constant between MAPT mutant cases, despite the potentially different mechanisms that led to those inclusions. The exact source of neurodegeneration, however, is unclear and could be due to a number of factors. In AD, the rate of cognitive decline correlates positively with the number of NFTs in the brain [118], which could cause neurodegeneration through the disruption of cytoplasmic organelles or by blocking axonal transport [119, 120]. However, evidence in human AD cases has shown that, while correlative, the number of NFTs can far exceed neuronal loss, and in rodent models, many neurons have been shown to die without ever forming NFTs [121,122,123]. Soluble oligomeric species of tau have also been implicated in cellular toxicity when added in culture as well as accelerated pathology when injected into the hippocampus of transgenic mice [124, 125]; however, the mechanisms of potential toxicity remain unclear. Finally, unbound, aggregate-prone, or aggregated tau are unable to perform their normal functions, namely maintaining or assembling MTs; this, in itself, could lead to neurodegeneration through MT disassembly and impaired axonal transport.

Conclusions

Mutations in the MAPT gene can exert several different effects on the functions and properties of tau. These effects can overlap or be completely distinct between mutations, but all result in the formation of aggregated tau inclusions with neuronal loss and atrophy. Mutations that affect tau mRNA splicing can alter the ratio of tau isoforms and lead to the potential dysregulation of MT dynamics as well as an isoform-specific overabundance of soluble, free-floating tau. Mutations that functionally affect tau’s ability to bind to, promote the assembly of or stabilize MTs may also lead to neurodegeneration in a similar manner: through the dysregulation of MT dynamics and/or tau mislocalization, leading to an increase in MT-unattached tau. In both of these instances, this increased amount of unbound tau could increase the usually minute chance of nucleation events occurring, or these proteins could interact with polyanionic molecules in the cytosol, leading to an eventual cascade of elongation and seeding. For other mutations that increase aggregation propensity and/or seeding, perhaps they function under a similar mechanism, but with a higher susceptibility to β-sheet formation and aggregation, speeding up the pathological process. Importantly, impaired MT function and aberrant tau aggregation are not mutually exclusive pathogenic mechanisms. However, it is noteworthy that the observed differences in seeding propensities between different tau mutants do not correlate with an earlier age of onset or shorter duration of disease [12, 116], suggesting that tau loss of function, such as impacts on MT activities or perhaps some still undiscovered functions of tau, could be more important for pathogenesis than a relative increase in aberrant aggregation propensity.

Given the wide-ranging and unique differences between tau mutants, the choice of models used is important. The question remains, however, as to whether utilizing specific tau mutants—namely those at the P301 residue—in the context of aggregation and seeding studies constitutes a model mechanistically similar and applicable to other mutations in FTD and/or sporadic tauopathy: an important fact to consider when utilizing these models to demonstrate therapeutic efficacy.