Alzheimer's disease

The tangled tale of tau

Article metrics

Alzheimer's disease is the main form of dementia in today's ageing population. It is characterized by abnormal protein deposits in the brain, such as the extracellular amyloid plaques and the intracellular neurofibrillary tangles. The tangles are made up of a protein called tau, which is modified by phosphorylation and aggregated into so-called paired helical filaments. Because hyperphosphorylation seems to occur before aggregation, it is considered one of the earliest signs of neuronal degeneration. But which protein kinases are responsible for it?

Two papers in this issue go some way towards answering this question. Tsai and colleagues1 (page 615) report that the cyclin-dependent kinase 5 (Cdk5) is deregulated in the brains of people with Alzheimer's disease. Not only that, but Cdk5 also hyperphosphorylates tau and accumulates in tangle-bearing neurons. And Greengard and colleagues2 (page 669) show that Cdk5 can alter signalling processes in neurons by altering the balance between its kinase and phosphatase activities.

The tau protein is associated with neuronal microtubules. It stabilizes them and regulates the transport of vesicles or organelles along them, supports the outgrowth of axons and serves as an anchor for enzymes3. The human central nervous system contains six isoforms of tau, generated from a single tau gene by alternative splicing. Embryos contain only one (the smallest) isoform; the larger isoforms are induced only when neurons differentiate and cell processes (axons and dendrites) are generated. Tau is a basic protein, rich in serine and threonine residues, and it can be phosphorylated by many kinases. A common hypothesis is that phosphorylation weakens the affinity of tau for microtubules. As a result tau detaches and then aggregates, eventually leading to neuronal degeneration (Fig. 1).

Figure 1: Hyperphosphorylation and aggregation of tau.

Phosphorylation of tau is thought to weaken its affinity for the microtubules to which it normally binds. Tau then detaches and forms the intracellular aggregates characteristic of Alzheimer's disease.

Which kinase initiates this chain of events? Proteins related to the cell-cycle kinase Cdc2 came under suspicion early on for several reasons. First, tau from the brains of patients with Alzheimer's disease is phosphorylated at serine or threonine residues followed by a proline. This points to the activity of a ‘proline-directed’ kinase, which includes members of the Cdc2 family, but also other kinases such as the mitogen-activated protein kinases or GSK-3. Second, one Cdc2-family kinase, Cdk5 (also known as NCLK, for neuronal Cdc2-like kinase), is one of two main tau-protein kinases isolated from brain tissue (the other being GSK-3b4). Third, proline-directed phosphorylation of tau increases strongly during mitosis, and correlates with other markers of mitosis5. This is consistent with the idea that degeneration could be the response of a differentiated neuron to inappropriate mitotic signals, which forces that neuron to commit suicide (apoptosis). Indeed, inhibitors of Cdc2-like kinases can prevent an apoptotic response6.

Arguments against this circumstantial evidence include the fact that proline-directed phosphorylation of tau does not strongly influence its functions (binding to microtubules, for example) compared with the effects of other kinases such as the microtubule-affinity regulating kinase or protein kinase A7. Moreover, cells transfected with Cdk5 show no real change in tau phosphorylation. Nonetheless, Tsai and colleagues1 now provide compelling evidence for the involvement of Cdk5 in the phosphorylation of tau.

The authors first show that Cdk5 is not properly regulated in the brains of people with Alzheimer's disease. Normally the kinase is activated by binding a partner protein, p35 (also known as Nck5a). The function of p35 is analogous to that of the cyclins, which bind to other Cdc2-like kinases. The activity of Cdk5 is tightly regulated through spatial and temporal constraints (Fig. 2) — p35 is usually attached to the plasma membrane via a myristoylation anchor at its amino terminus. It has a short lifetime and is degraded rapidly via the ubiquitination–proteasome pathway.

Figure 2: Regulation of the cyclin-dependent kinase 5 (Cdk5).

Cdk5 is activated by binding to p35, which anchors it in the plasma membrane. If p35 is cleaved, however, Tsai and colleagues1 have found that the Cdk5–p25 complex is a highly active, long-lived form of the enzyme. It can break free from the plasma membrane and phosphorylate a number of substrates, including tau and, as Greengard and colleagues2 have shown, DARPP-32. (MAPs, microtubule-associated proteins; Rb, retinoblastoma; APC, adenomatous polyposis coli gene product.)

Surprisingly, all of these constraints are lifted when p35 is cleaved into a smaller, amino-terminal part (p10) and a larger, carboxy-terminal part (p25), although the mechanism of this aberrant cleavage is not yet known. The p10 portion disappears rapidly, but p25 remains bound to the kinase. This p25–Cdk5 complex is constitutively active, long-lived and no longer restricted to the plasma membrane. In other words, Cdk5 can now go out on a ‘phosphorylation spree’, phosphorylating everything in sight — including tau. In support of this, Tsai and colleagues have observed tau co-precipitated with p25 and Cdk5 in the degenerating neurons of Alzheimer's disease.

Other substrates of Cdk5, such as the proline-rich domains of neurofilaments8, might suffer a similar fate. Indeed, deposits of highly phosphorylated neurofilaments have been found, together with Cdk5, in patients with Parkinson's disease and amyotrophic lateral sclerosis9. The hypothesis linking Cdk5 to neurofibrillary deposits is attractive for another reason — both Cdk5 and tau are normally involved in the development of axons and dendrites. The level of Cdk5 increases transiently during brain development, then decays again. However, levels of Cdk5 remain high in regions with the highest plasticity, such as the entorhinal region or the olfactory system10. This correlates with the regions in which the hyperphosphorylation and aggregation of tau are first seen in Alzheimer's disease11.

Many questions remain. For example, what is the physiological function of Cdk5? Is it regulation of the neuronal actin cytoskeleton, as suggested by Tsai and colleagues' results1? Or does Cdk5 modulate signalling cascades via kinase/phosphatase activities, as indicated by Greengard and colleagues' study2? These authors showed that Cdk5 turns a modulator protein called DARPP-32 (dopamine and cyclic AMP-regulated phosphoprotein, relative molecular mass 32,000) into an inhibitor of protein kinase A. Another study12 implies that Cdk5 may control axonal transport. But all of them bear on the observed influence of Cdk5 on neuronal differentiation10.

Which protease is responsible for the aberrant cleavage of p35? Does the hyperphosphorylation of tau by Cdk5 really cause microtubule breakdown and tau aggregation, or are these indirect effects involving other kinases? After all, microtubule dynamics in axons is determined not only by tau, but by other microtubule-associated proteins as well, and phosphorylation of tau does not necessarily lead to aggregation13. How does the activity of Cdk5 relate to the newly discovered14 tau mutations that cause fronto-temporal dementias without changing the phosphorylatable residues? What is the crosstalk between tau phosphorylation by Cdk5 at a serine–proline motif and proline isomerization15?

Answers will come with time but, independently of them, Tsai and colleagues' results provide potential targets for diagnosis or therapy. These may come either at the level of p35 cleavage or at the level of Cdk5 inhibition, and therapy should be possible with the development of highly specific inhibitors.


  1. 1

    Patrick, G. et al. Nature 402, 615–622 (1999).

  2. 2

    Bibb, J. et al. Nature 402, 669–671 (1999).

  3. 3

    Mandelkow, E.-M. & Mandelkow, E. Trends Cell Biol. 8, 425–427 (1998).

  4. 4

    Imahori, K. & Uchida, T. J. Biochem. (Tokyo) 121, 179–188 (1997).

  5. 5

    Vincent, I., Jicha, G., Rosado, M. & Dickson, D. W. J. Neurosci. 17, 3588–3598 (1997).

  6. 6

    Padmanabhan, J., Park, D., Greene, L. & Shelanski, M. J. Neurosci. 19, 8747–8756 (1999).

  7. 7

    Drewes, G., Ebneth, A. & Mandelkow, E.-M. Trends Biochem. Sci. 23, 307–311 (1998).

  8. 8

    Sharma, P. et al. J. Biol. Chem. 274, 9600–9606 (1999).

  9. 9

    Nakamura, S. et al. Neurology 48, 267–270 (1997).

  10. 10

    Delalle, I., Bhide, P., Caviness, V. & Tsai, L.-H. J. Neurocytol. 26, 283–296 (1997).

  11. 11

    Braak, H. & Braak, E. Acta Neuropathol. 82, 239–259 (1991).

  12. 12

    Ratner, N., Bloom, G. & Brady, S. T. J. Neurosci. 18, 7717–7726 (1998).

  13. 13

    Schneider, A., Biernat, J., von Bergen, M., Mandelkow, E. & Mandelkow, E.-M. Biochemistry 38, 3549–3558 (1999).

  14. 14

    Wilhelmsen, K. Proc. Natl Acad. Sci. USA 96, 7120–7121 (1999).

  15. 15

    Lu, P.-J., Wulf, G., Zhou, X. Z., Davies, P. & Lu, K. P. Nature 399, 784–788 (1999).

Download references

Author information

Correspondence to E. Mandelkow.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Mandelkow, E. The tangled tale of tau. Nature 402, 588–589 (1999) doi:10.1038/45095

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.