Microtubule-Associated Protein Tau (MAPT)

Microtubule-associated protein tau (MAPT) is a hydrophilic, intrinsically disordered neuronal protein encoded by the MAPT gene on human chromosome 17q21.31 (16 exons spanning ~134 kb) that binds along the outer surface of axonal α/β-tubulin protofilaments to regulate microtubule assembly, dynamic instability, and kinesin/dynein axonal transport. In the healthy adult human brain, alternative splicing of three cassette exons—Exon 2, Exon 3, and Exon 10—generates six canonical central nervous system (CNS) tau isoforms ranging from 352 to 441 amino acids in length (36.8 to 45.9 kDa, nomenclature standardized against the longest 441-residue adult brain isoform, 2N4R tau or tau441). Because tau contains a high proportion of polar and charged residues (>25% Gly, Pro, Lys, Ser, and Thr) and very few hydrophobic side chains, monomeric soluble tau lacks a fixed globular tertiary fold, existing in solution as a dynamic ensemble of transient "paperclip" conformations in which the acidic N-terminal and C-terminal tails fold back over the positively charged proline-rich and microtubule-binding repeat domains.

Under pathological conditions, hyperphosphorylation of Ser/Thr residues, post-translational acetylation/truncation, or inherited MAPT coding and splice-site mutations detach tau from axonal microtubules and expose two amyloidogenic hexapeptide motifs—PHF6* (275VQIINK280 in Repeat 2) and PHF6 (306VQIVYK311 in Repeat 3)—nucleating the assembly of cross-β amyloid filaments. Over 20 distinct clinicopathological syndromes, collectively termed tauopathies, are defined by the isoform composition (3R, 4R, or mixed 3R+4R) and atomic cryo-electron microscopy (cryo-EM) protofilament fold of their intracellular tau inclusions, including Alzheimer disease (AD), Pick disease (PiD), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), chronic traumatic encephalopathy (CTE), and frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17).

Chromosome 17q21.31 H1/H2 Inversion Haplotypes and the Six CNS Isoforms

The human MAPT locus resides within a ~1.5-Mb linkage disequilibrium block on Chr17q21.31 flanked by complex low-copy segmental duplications (LCRs) that mediated an ancient ~900-kb chromosomal inversion diverging ~2.3 million years ago into two non-recombining haplotypes: H1 (direct orientation) and H2 (inverted orientation, present in ~20–25% of Europeans via Neanderthal or early hominin introgression/persistence, and nearly absent in East Asian and African populations). In genome-wide association studies, the H1 haplotype (specifically the H1c subhaplotype, which increases MAPT and Exon 10 transcriptional expression) confers significant susceptibility to 4R tauopathies (OR ≈ 5.5 in PSP and OR ≈ 3.7 in CBD), whereas the H2 inversion haplotype is protective against neurodegeneration.

Within the CNS, alternative inclusion of Exon 2 (29 aa) and Exon 3 (29 aa, which requires prior inclusion of Exon 2) near the N-terminus produces isoforms with zero (0N), one (1N), or two (2N) acidic N-terminal inserts (residues 45–102). Simultaneously, alternative splicing of Exon 10 (93 bp, encoding the 31-amino-acid second microtubule-binding repeat R2, residues 275–305) divides the six isoforms into two functional classes:

• Three-Repeat (3R) Tau (0N3R, 1N3R, 2N3R): Lacks Exon 10 (R1–R3–R4), comprising 352, 381, and 410 amino acids. Fetal human brain expresses almost exclusively the shortest isoform (0N3R) to permit high neurite plasticity during development.
• Four-Repeat (4R) Tau (0N4R, 1N4R, 2N4R): Includes Exon 10 (R1–R2–R3–R4), comprising 383, 412, and 441 amino acids. Because the extra R2 repeat and the R1–R2 inter-repeat linker (Lys274) provide an additional high-affinity tubulin-binding module, 4R tau binds and polymerizes microtubules ~3-fold more tightly than 3R tau. In the healthy adult human cerebral cortex, spliceosome regulation maintains a 1:1 stoichiometric ratio of 3R to 4R tau.

The Exon 10–Intron 10 5′ Splice-Site Stem-Loop and FTDP-17 Mutations

The 1:1 balance between 3R and 4R tau in adult human neurons is governed by an RNA secondary structure splicing switch located right at the Exon 10–Intron 10 5′ splice junction (positions −6 to +21 relative to the GU splice donor). In pre-mRNA, this 27-nucleotide sequence folds into a thermodynamically stable 6-bp RNA stem-loop hairpin (capped by a 6-nt apical loop and interrupted by a single bulging A at position -3, with folding free energy Δ*G*° ≈ −9.8 kcal/mol) that partially sequesters the 5′ splice donor sequence from base-pairing with the U1 snRNP RNA of the spliceosome, resulting in ~50% skipping and ~50% inclusion of Exon 10.

In FTDP-17, over 20 autosomal dominant pathogenic mutations—including intronic mutations (+3 G>A, +11 T>C, +12 C>T, +13 A>G, +14 C>T, +16 C>T) and silent or missense exonic mutations (N279K, ΔK280, L284L, N296H, P301L, P301S, S305N)—cluster inside or adjacent to this Exon 10 stem-loop. Intronic mutations such as +14 C>T and +16 C>T disrupt canonical Watson–Crick G•C base pairs in the RNA stem (reducing stem stability by ΔΔ*G* ≈ +3.2 to +5.1 kcal/mol and lowering the melting temperature T_m by 12–18 °C), unfolding the hairpin so U1 snRNP binds constitutively and shifts the isoform ratio from 1:1 (3R:4R) to 1:3 or 1:6 in favor of 4R tau. Even without a protein coding change, chronic 4R tau overproduction alone is sufficient to cause fatal frontotemporal neurodegeneration.

High-resolution cryo-EM structures resolved by Sjors Scheres, Michel Goedert, and colleagues demonstrated that each human tauopathy is defined by a conserved, disease-specific cross-β protofilament fold (identical across sporadic and inherited cases of the same disease, but distinct between diseases):

1. Alzheimer Disease (3R+4R C-Shaped Fold): Both paired helical filaments (PHFs, ~90% of AD filaments) and straight filaments (SFs, ~10%) share an identical C-shaped protofilament core spanning Val306–Phe378 (73 residues, comprising the entirety of R3 and R4 plus 10 C-terminal residues). Because this core begins at Val306 (PHF6 at the start of R3), both 3R and 4R tau isoforms can be incorporated indiscriminately into the growing fibril, whereas pure 3R or 4R folds include residues from R1 or R2 that sterically exclude the opposite isoform.
2. Hyperphosphorylation and Liquid-to-Solid Transition: Longest-isoform 2N4R tau contains 85 potential phosphorylation sites (45 Ser, 35 Thr, 5 Tyr). Phosphorylation within the proline-rich region (Thr181, Ser202/Thr205 recognized by the AT8 antibody, Thr217, Thr231) and microtubule-binding repeats (Ser262, Ser396/Ser404) by GSK-3β, CDK5, and MARK neutralizes the positive charge of the repeat domain, releases tau from tubulin, promotes liquid–liquid phase separation into high-concentration tau droplets, and accelerates nucleation of the PHF6 (VQIVYK) steric zipper.

Why can Alzheimer disease tau filaments incorporate both 3R and 4R isoforms whereas Pick disease and CBD filaments incorporate only one?

In Pick disease (3R), the ordered cross-β filament core spans Lys254–Phe378, joining the C-terminus of R1 (Lys274) directly to the N-terminus of R3 (Val306); presence of the 31-residue R2 domain in 4R tau sterically clashes with this junction. Conversely, in CBD and PSP (4R), the ordered core starts at Gly272/Lys274 and incorporates the entire R2 domain (Val275–Ser305) into three- or four-layered β-sheet folds. Only the AD/CTE C-shaped fold (Val306–Phe378) starts downstream of the R1/R2–R3 splice junction, permitting incorporation of both 3R and 4R monomers.

Why do adult mice fail to develop spontaneous 3R+4R Alzheimer-like neurofibrillary tangles without human transgenes?

Unlike adult human neurons, which splice MAPT Exon 10 to maintain an equimolar 1:1 ratio of 3R and 4R tau, post-natal mouse neurons splice in Exon 10 with ~100% efficiency, expressing almost exclusively 4R tau (0N4R, 1N4R, 2N4R) in adulthood, alongside three sequence differences in the N-terminal domain (residues 17–28). Modelling human 3R+4R splicing dynamics or Exon 10 stem-loop therapeutics therefore requires humanized whole-locus MAPT knock-in models or human iPSC-derived neurons.

Why is plasma p-tau217 superior to total tau for diagnosing early Alzheimer disease?

Total plasma or CSF tau rises non-specifically after acute neuronal injury (such as ischemic stroke or traumatic brain injury). By contrast, soluble N-terminal tau fragments phosphorylated specifically at Thr217 (p-tau217) and Thr181 (p-tau181) are actively hyperphosphorylated and secreted by neurons in direct response to early cortical amyloid-β plaque deposition (10–15 years before dementia onset), while MTBR-tau243 in CSF tracks insoluble neurofibrillary tangle burden (tau-PET).

1. Hutton, M., Lendon, C. L., Rizzu, P., et al. (1998). "Association of missense and 5′-splice-site mutations in tau with the inherited dementia FTDP-17." Nature, 393(6686), 702–705. https://doi.org/10.1038/31508
2. Fitzpatrick, A. W. P., Falcon, B., He, S., et al. (2017). "Cryo-EM structures of tau filaments from Alzheimer's disease." Nature, 547(7662), 185–190. https://doi.org/10.1038/nature23002
3. Shi, Y., Zhang, W., Yang, Y., et al. (2021). "Structure-based classification of tauopathies." Nature, 598(7880), 359–363. https://doi.org/10.1038/s41586-021-03911-7
4. Goedert, M., Eisenberg, D. S., & Crowther, R. A. (2017). "Propagation of tau aggregates and neurodegeneration." Annual Review of Neuroscience, 40, 189–210. https://doi.org/10.1146/annurev-neuro-072116-031153