01 · Biology
What tau does — and what goes wrong
Tau is a microtubule-associated protein expressed predominantly in neurons. Its normal job is to stabilise the microtubule tracks along which motor proteins ferry cargo up and down axons. To do this, tau needs to bind the microtubule — and that binding is regulated by phosphorylation: adding phosphate groups reduces binding affinity and promotes tau release when the microtubule network needs to remodel.
In Alzheimer's disease, kinases (especially CDK5 and GSK3β) hyperphosphorylate tau at dozens of sites. Hyperphosphorylated tau can no longer bind microtubules — the tracks destabilise and axonal transport fails. Worse, the freed tau forms oligomers and then insoluble filaments (neurofibrillary tangles, NFTs) that accumulate in the cell body and are toxic to the neuron.
Tau tangles spread through the brain in a stereotyped pattern (Braak stages I–VI) that begins in the entorhinal cortex and hippocampus — explaining why memory is the first casualty — before spreading to temporal, parietal, and eventually frontal cortices. This spread appears to happen via synaptic transmission: tau seeds released from one neuron are taken up by its neighbours and template further aggregation, analogous to prion propagation.
For the underlying biochemistry of phosphorylation and its role in cell signalling, see the Protein synthesis guide and the Cells and organelles guide. For how neuroplasticity is affected when this transport fails, see the Neuroplasticity article.
02 · The amyloid hypothesis
Amyloid, tau, and the two-hit model
The amyloid cascade hypothesis (1992) proposes that amyloid-beta (Aβ) accumulation is the initiating event in Alzheimer's, upstream of tau pathology, neuroinflammation, and cell death. The hypothesis is supported by the genetics: all three familial early-onset genes (APP, PSEN1, PSEN2) increase Aβ42 production or aggregation.
But clinical trials targeting Aβ repeatedly showed plaque clearance without cognitive benefit — until lecanemab and donanemab demonstrated that clearing Aβ early enough (before substantial tau pathology) does slow decline. The emerging two-hit model positions Aβ as necessary but insufficient: Aβ accumulation triggers tau hyperphosphorylation, and tau spread drives neuronal death. Targeting both simultaneously is now the goal.
03 · Drug pipeline
Anti-amyloid approvals and the tau pipeline
Lecanemab
Anti-Aβ protofibrils monoclonal antibody. FDA approved Jan 2023. Slows decline 27 % over 18 months. ARIA risk ~21 %.
Donanemab
Anti-Aβ plaque monoclonal antibody. FDA approved Jul 2024. Slows decline 35 % in low/medium tau group. ARIA risk ~24 %.
JNJ-63733657
Anti-pTau217 antibody (J&J). Phase II. Reduces CSF p-tau217 up to 60 %. Cognitive endpoints read out 2026.
BIIB080
ASO targeting tau mRNA (Biogen). Phase II. Reduces CSF total tau ~50 %. Safety and tolerability data encouraging.
The antisense oligonucleotide (ASO) approach targets tau at the mRNA level — reducing production rather than clearing existing tangles. For the molecular biology of how mRNA is produced and how ASOs interfere with it, see the Protein synthesis guide. CRISPR-based approaches to silencing tau expression are in preclinical development — see the CRISPR base editing article.
04 · Early detection
Blood biomarkers: detecting Alzheimer's before symptoms
The biggest shift in Alzheimer's research in 2025–2026 is the validation of blood-based biomarkers. A single blood draw can now measure plasma p-tau217 and the Aβ42/40 ratio with accuracy comparable to PET scans (AUC ~0.96), at a fraction of the cost.
This matters because anti-amyloid drugs only work in early-stage disease — before widespread tau pathology. Identifying patients a decade before symptoms appear opens a treatment window that currently does not exist clinically. Population-scale screening using blood biomarkers, combined with polygenic risk scores from genome-wide association data, could identify the 10–15 % of middle-aged adults who carry the highest Alzheimer's risk.
The genetic risk architecture also links to genetics and inheritance: the APOE ε4 allele (see the Allele glossary entry) modulates both amyloid accumulation rate and tau propagation speed, making it the most important single genetic determinant of sporadic Alzheimer's risk.
Alzheimer's disease — common questions
- Amyloid-beta plaques form outside neurons; tau tangles form inside. Tau normally stabilises microtubules in the axon. When hyperphosphorylated, it detaches and aggregates. Tau spread follows a predictable staging pattern (Braak stages I–VI) that correlates with symptom severity more closely than amyloid burden. See the Protein synthesis guide for how protein misfolding arises.
- Lecanemab and donanemab slow cognitive decline by ~27–35 % in early Alzheimer's but do not reverse damage. Both carry risk of ARIA (brain oedema and micro-bleeds), particularly in APOE ε4 carriers. Clinical benefit vs. risk must be weighed individually. Anti-tau therapies in Phase II/III trials aim to complement anti-amyloid approaches.
- Plasma biomarkers — p-tau217, Aβ42/40 ratio, GFAP — can detect pathology 10–15 years before symptoms with accuracy matching PET/CSF tests. This window opens the possibility of intervention before significant neuronal loss. Biomarker-based screening is being evaluated in the AHEAD 3-45 and DIAN-TU trials.
- Familial early-onset Alzheimer's is caused by dominant mutations in APP, PSEN1, or PSEN2. Sporadic late-onset disease is influenced by many genetic variants, with APOE ε4 being the strongest single-gene risk factor. See Genetics and inheritance for how dominant mutations and risk alleles differ.