- Email: [email protected]
Tau in Alzheimer's disease
trends in CELL BIOLOGY (Vol. 8) November 1998
comment
Tau in Alzheimer’s disease
FORUM
Alzheimer’s disease (AD) is an age-dependent dementia characterized by the loss of cognitive functions and by corresponding changes in the brain. Major diagnostic problems arise because AD develops slowly, the afflicted brain tissue is not available for routine analysis and there are no biochemical markers for monitoring the progression. Thus, the ultimate diagnosis relies on neuropathological examination post mortem. The main hallmarks of AD are the aggregates formed by the extracellular amyloid peptide Ab and by intracellular tau protein. Tau deposits occur in dystrophic neurites, as fine neuropil threads, or as massive neurofibrillary tangles in neuronal cell bodies, which become extracellular ghost tangles after the death of the neuron. Amyloid occurs in diffuse or neuritic plaques (the latter contain dystrophic neurites with aggregated tau). Formerly, the neuropathological criteria for diagnosing AD were based on amyloid plaques in general without distinguishing neuritic or diffuse plaques1. The subsequent CERAD (consortium to establish a registry for Alzheimer’s disease) criteria2 referred more specifically to neuritic plaques, with the recent added emphasis on tau deposits such as neurofibrillary tangles (Consensus criteria, NIA and Reagan Institute3). A key issue is whether the neuropathological hallmarks can be correlated with the progression of the clinical symptoms, and whether they allow a backextrapolation to early pre-clinical stages. This seems possible in the case of tau deposits. Their spreading from the transentorhinal region to the hippocampus and the neocortex has been subdivided into six stages, the first two of which are pre-clinical4. The first sign in affected neurons is an unusual phosphorylation of tau, followed by aggregation. There is a good correlation between tau pathology and loss of synapses, leading to the proposal that, in AD, neuronal transport is impaired5. A strong driving force in AD research is the amyloid cascade hypothesis (see accompanying article by Selkoe6 and references therein). It states that improper processing of the membrane protein APP (amyloid precursor protein) lies at the root of AD because this generates the neurotoxic and aggregating Ab peptide. The major evidence comes from mutations in the gene encoding APP (on chromosome 21) that are linked to familial forms of AD and enhance the generation of Ab. Further support comes from mutations in the genes encoding presenilin 1 and 2 on chromosomes 14 and 1, which also enhance the level of Ab. Finally, ApoE4, a risk factor for AD, is also associated with amyloid plaques. By contrast, no AD-causing mutations have been detected thus far in the gene encoding tau (on chromosome 17). However, several mutations have emerged as the cause of related dementias (such as FTDP-17 – fronto–temporal dementia with parkinsonism linked to chromosome 17), which show pronounced tau deposits7–9. These ‘tauopathies’ added fresh fuel to the discussion on the causes of AD. In addition, despite the important evidence derived from mutations in APP or presenilin, it should be noted that altogether they account for only a minority of AD cases. The majority must be explained
Eva-Maria Mandelkow and Eckhard Mandelkow Neurofibrillar protein aggregates containing tau are one of the major hallmarks of Alzheimer’s disease (AD). In normal cells, tau stabilizes axonal microtubules, which are the tracks for intracellular traffic. In AD, tau becomes abnormally phosphorylated, aggregates into paired helical filaments and loses its ability to maintain the microtubule tracks. There is renewed interest in tau as a causative factor in neurodegenerative disease based on recently discovered mutations in the gene encoding tau. This article discusses how changes in tau protein could lead to retraction of neuronal processes and thus cell death and argues that tau pathology, rather than b-amyloid, might be the most reliable indicative factor for AD.
by non-genetic factors, which points again to the pathology of tau as being the best correlate for the clinical progression and as the most reliable neuropathological indicator. Physiological functions of tau Tau is one of the microtubule-associated proteins (MAPs) that stabilize neuronal microtubules for their role in the development of cell processes, establishment of cell polarity and intracellular transport10. Tau occurs mainly in axons (in contrast to the somatodendritic MAP2). A single gene encodes tau, which generates six isoforms, of 352 to 441 amino acid residues, in the human central nervous system by alternative splicing (Fig. 1), plus a larger variant in peripheral nerves11. The synthesis of tau is unusual in that tau mRNA is transported to the proximal axon where translation occurs12. This is one mechanism by which neuronal polarity is established. Tau contains an acidic N-terminal domain, a basic and proline-rich middle domain, a basic domain containing three or four internal repeats, and a C-terminal domain. Tau is very hydrophilic, soluble and displays a ‘natively unfolded’ structure. It can be phosphorylated at multiple sites, some of which regulate its microtubule-binding properties. The sites can be broadly subdivided into two classes: several Ser-Pro or Thr-Pro motifs occur in both regions flanking the internal repeats and are targets of prolinedirected kinases such as glycogen synthase kinase 3,
0962-8924/98/$ – see front matter © 1998 Elsevier Science. All rights reserved. PII: S0962-8924(98)01368-3
The authors are in the Max-PlanckUnit for Structural Molecular Biology, Notkestrasse 85, D-22603 Hamburg, Germany. E-mail: mand@ mpasmb.desy.de
425
FORUM
comment MAPK GSK-3b Cdk5
PKA
MARK PP2A
*
PP2B
N
C 1
2
3
4
G272V N279K
P189A
K280–
V337M
R406W
P301L
* TP
SP
KXGS
Cys
S214
Mutations
Alternative splicing
FIGURE 1 Diagram of tau protein, its domains, sites of phosphorylation and mutations. The near-N-terminal inserts (exons 2 and 3, grey) and the second of the four repeats (exon 10) can be absent owing to alternative splicing. The ‘assembly domain’ comprises the C-terminal half; it binds to microtubules and stabilizes them, whereas the N-terminal ‘projection domain’ protrudes from the microtubule surface. The regions flanking the repeats are rich in Ser-Pro or Thr-Pro motifs, which can be phosphorylated by proline-directed kinases and thus form epitopes of antibodies diagnostic of Alzheimer tau. Ser214 and Ser262 (in the KXGS motif of the first repeat) are non-proline motifs phosphorylated in Alzheimer’s disease and in vitro by PKA or MARK; this causes the detachment of tau from microtubules. Some of the mutations linked to dementias with tauopathy are indicated; others are silent at the protein level. Abbreviations: Cdk, cyclin-dependent kinase; Cys, cysteine; GSK-3, glycogen synthase kinase 3; MAPK, mitogen-activated protein kinase; MARK, microtubuleaffinity-regulating kinase; PKA, protein kinase A; PP, phosphatase; SP, serine-proline motif; TP, threonine-proline motif.
cyclin-dependent kinase Cdk5 or MAP kinase13,14. These sites have only a moderate influence on tau–microtubule interactions but are useful as diagnostic tools for the AD-like phosphorylation of tau. Other sites include targets of protein kinase A (e.g. Ser214), microtubule-affinity-regulating kinase (MARK; at KXGS motifs including Ser262, Ser356) or Ca21/calmodulin-dependent protein kinase (Ser416). Tau detaches from microtubules when phosphorylated at Ser262 or at Ser214, two sites that are phosphorylated in AD. In mitotic cells, tau shows enhanced phosphorylation at several Ser-Pro motifs and at Ser214, concomitant with its detachment from microtubules15. Phosphorylation of tau or related MAPs by MARK appears to be important for the establishment of cell polarity, but overactivity of MARK leads to cell death10. Tau is necessary for the outgrowth of neurites16. Axons show a gradient of tau concentration, with a maximum near the tip17, and a gradient of tau phosphorylation, highest in the proximal axon18. However, transgenic mice lacking tau have no major defects, possibly because other MAPs can substitute for tau19. Several observations point to other, less-well-characterized functions of tau. The size of the N-terminal domains of MAPs regulates the spacing between microtubules in neuronal processes; in tau, this domain is relatively small and would allow a close approach of microtubules (unlike MAP2)20. A fraction
426
of tau associates with the plasma membrane rather than with microtubules21. Neuronal cells possess mechanisms to sort MAPs into specific compartments. Although tau is sorted mainly into the axon, this preference is not absolute, and overloading the cell with tau can lead to its appearance in dendrites22. Some forms of tau have even been detected in the nucleus23. MAPs can also serve as anchoring devices for cellular proteins. A recent example is the role of tau as a targeting subunit of protein phosphatase 138. Finally, the ubiquitous MAP4 can regulate vesicle transport along microtubules24; a similar effect appears to operate for tau as well25. Pathological alterations of tau in AD In AD, the properties of tau change in several ways. • AD-tau shows an abnormal ‘hyperphosphorylation’ at many sites13,14,26. Some increase occurs also in foetal tissue and in mitotic cells, leading to the hypothesis that ‘mitotic’ signals received by differentiated neurons might cause hyperphosphorylation of tau and, later, death by apoptosis. Many of the abnormal phosphorylation sites are at Ser-Pro or Thr-Pro motifs; this explains why various antibodies raised against AD-tau react with such phosphorylated epitopes and are now used in the diagnosis of Alzheimer brain tissue or for the development of cell models. This analysis has revealed that abnormal phosphorylation of tau occurs before aggregation. • AD-tau also shows a loss of microtubule binding, which is probably a consequence of the hyperphosphorylation at sites (e.g. S262 or S214) that detach tau from microtubules. This could account for the disappearance of microtubules, causing the breakdown of intracellular traffic, which would result in the ‘dying back’ of axons. • The redistribution of tau from an axonal to a somatodendritic pattern suggests that, in AD, there is a defect in directing tau into the right compartment. This could arise from several mechanisms – for example, increased mRNA levels and tau synthesis – which could lead to overloading of the cell and perturbation of the sorting mechanism. • The aggregation of tau is particularly enigmatic, considering its unusual solubility. The aggregates are termed ‘paired helical filaments’ (PHFs) because of their two-stranded appearance, with widths between 10 and 20 nm and crossover repeats of 80 nm27. The PHFs in turn bundle into ‘neurofibrillary tangles’ or ‘neuropil threads’. In vitro, PHF assembly is inefficient but can be enhanced by oxidation, which leads to dimerization of tau28, and by interaction with polyanions such as heparin29 or RNA30. • Ubiquitination and proteolysis are posttranslational modifications of AD-tau that probably represent cellular attempts to degrade the aberrant protein (via the proteasome or calpain pathway31); however, since some proteolytic fragments are detected at early stages, this could also contribute to the nucleation of PHFs32. • Glycation is a consequence of oxidative damage and crosslinking, which accumulates once tangles are formed33. trends in CELL BIOLOGY (Vol. 8) November 1998
FORUM
• There is an increase of tau in the cerebrospinal fluid (from ~200 pg ml–1 to ~600 pg ml–1), which probably arises from dying neurons34. This feature could become valuable as an early diagnostic assay.
comment Cell body
1 2 3 4
Kinases Phosphatases
1 2 3 4
P 1 2 3 4
P
PHF A role for tau in AD? 1 1 2 2 The above observations can be placed 3 1 P 1 P 3 Axon 4 2 2 4 in a hypothetical scheme (Fig. 2): inap3 3 4 4 propriate signals lead to an imbalance of kinases/phosphatases, hyperphosSynapse phorylation of tau, its detachment from Axonal Tau stabilizes Tau phosphorylation PHF microtubules, microtubule breakdown, transport microtubules MT depolymerization assembly aggregation of tau into PHFs, breakdown of intracellular transport and degeneration of neurons. However, many FIGURE 2 intermediate steps are still unknown. Model showing a possible link between axonal transport, microtubules and tau in Alzheimer’s disease. For example, the effect of different ki- Transport relies on microtubules as tracks, stabilized by the ties of tau protein. Phosphorylation (P) at nases on tau has been examined in vitro crucial sites detaches tau from microtubules, leading to the breakdown of microtubules and the and in cell models10, but which of accumulation of tau aggregated into paired helical filaments (PHFs). them is responsible for tau phosphorylation in AD remains unknown. PHFs from recom12 Litman, P., Barg, J. and Ginzburg, I. (1994) Neuron 13, 1463–1474 binant tau can now be generated efficiently in vitro35, 13 Mandelkow, E-M. et al. (1995) Neurobiol. Aging 16, 355–362 but attempts to generate them in cell models have 14 Trojanowski, J. Q. and Lee, V. M. Y. (1995) FASEB J. 9, 1570–1576 failed so far. Conditions for generating ‘abnormal’ 15 Illenberger, S. et al. (1998) Mol. Biol. Cell 9, 1495–1512 phosphorylation in cells have been described (e.g. mi16 Kosik, K. S. and McConlogue, L. (1994) Cell Motil. Cytoskeleton totic signals15 or activation of signalling cascades36), 28, 195–198 but in general this does not lead to neuronal degen17 Black, M. et al. (1996) J. Neurosci. 16, 3601–3619 eration per se. Moreover, pathological markers of AD 18 Mandell, J. W. and Banker, G. A. (1996) J. Neuroscience 16, appear initially in single neurons, begging the ques5727–5740 tion of why there is selective vulnerability. Finally, 19 Harada, A. et al. (1994) Nature 369, 488–491 overexpressing human tau isoforms in transgenic 20 Matus, A. (1994) Trends Neurosci. 17, 19–22 mice has not resulted in an AD-like pathology37. 21 Brandt, R., Leger, J. and Lee, G. (1995) J. Cell Biol. 131, 1327–1340 How might the newly discovered tau mutations 22 Hirokawa, N. et al. (1996) J. Cell Biol. 132, 667–679 be involved? Remarkably, the mutations can occur in 23 Wang, Y. et al. (1993) J. Cell Biol. 121, 257–267 both coding and noncoding regions of the gene 24 Bulinski, J. C. et al. (1997) J. Cell Sci. 110, 3055–3064 encoding tau, with similar effects. One possible expla25 Ebneth, A. et al. J. Cell Biol. (in press) nation is that the mutations affect the splicing of tau 26 Delacourte, A. and Buee, L. (1997) Int. Rev. Cytol. 171, 167–224 mRNA, especially at the exon 10 boundary, so that 27 Crowther, R. A. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, a greater proportion of four-repeat tau is generated, 2288–2292 which might somehow affect one of the tau-depen28 Schweers, O. et al. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, dent reactions in the cell and lead to tau hyper8463–8467 phosphorylation, aggregation and cell death8. The 29 Goedert, M. et al. (1996) Nature 383, 550–553 dementias associated with the tau mutations are dis30 Kampers, T. et al. (1996) FEBS Lett. 399, 344–349 tinct from AD and affect different regions in the 31 Litersky, J. M. and Johnson, G. V. W. (1995) J. Neurochem. 65, brain. However, there may be a common link between 903–911 all ‘tauopathies’, which we hope will lead to a better 32 Novak, M., Kabat, J. and Wischik, C. M. (1993) EMBO J. 12, understanding of the pattern of neuronal loss in AD. 365–370
References 1 2 3 4 5 6 7 8
Khachaturian, Z. S. (1985) Arch. Neurol. 42, 1097–1105 Mirra, S. et al. (1994) J. Neuropathol. Exp. Neurol. 53, 303–315 Ball, M. et al. (1997) Neurobiol. Aging 18, 1–2 Braak, H. and Braak, E. (1991) Acta Neuropathol. 82, 239–259 Terry, R. D. (1996) J. Neuropathol. Exp. Neurol. 55, 1023–1025 Selkoe, E. (1998) Trends Cell Biol. 8, 447–453 Poorkaj, P. et al. (1998) Ann. Neurol. 43, 815–825 Spillantini, M. et al. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 7737–7741 9 Hutton, M. et al. (1998) Nature 393, 702–705 10 Drewes, G., Ebneth, A. and Mandelkow, E-M. (1998) Trends Biochem. Sci. 23, 307–311 11 Goedert, M. (1997) Neuroscientist 3, 131–141 trends in CELL BIOLOGY (Vol. 8) November 1998
33 34 35 36 37
Yan, S. D. et al. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 7787–7791 Vigo-Pelfrey, C. et al. (1995) Neurology 45, 788–793 Friedhoff, P. et al. (1998) Biochemistry 37, 10223–10230 Anderton, B. et al. (1995) Neurobiol. Aging 16, 389–397 Götz, J. et al. (1995) EMBO J. 14, 1304–1313 Acknowledgements
Reference added in proof: 38 Liao, H. et al. (1998) J. Biol. Chem. 273, 21901–21908
Authors’ note: Debates on mechanisms of neuronal degeneration in Alzheimer’s disease held during the recent International Alzheimer congress in Amsterdam can be found on the Internet at: http://www.alzforum.org/members/index.html.
We thank J. Biernat and A. Ebneth for help in designing the figures. Our work is supported by Deutsche Forschungsgemeinschaft (Alzheimer Research Group, Hamburg).
427
comment
Tau in Alzheimer’s disease
FORUM
Alzheimer’s disease (AD) is an age-dependent dementia characterized by the loss of cognitive functions and by corresponding changes in the brain. Major diagnostic problems arise because AD develops slowly, the afflicted brain tissue is not available for routine analysis and there are no biochemical markers for monitoring the progression. Thus, the ultimate diagnosis relies on neuropathological examination post mortem. The main hallmarks of AD are the aggregates formed by the extracellular amyloid peptide Ab and by intracellular tau protein. Tau deposits occur in dystrophic neurites, as fine neuropil threads, or as massive neurofibrillary tangles in neuronal cell bodies, which become extracellular ghost tangles after the death of the neuron. Amyloid occurs in diffuse or neuritic plaques (the latter contain dystrophic neurites with aggregated tau). Formerly, the neuropathological criteria for diagnosing AD were based on amyloid plaques in general without distinguishing neuritic or diffuse plaques1. The subsequent CERAD (consortium to establish a registry for Alzheimer’s disease) criteria2 referred more specifically to neuritic plaques, with the recent added emphasis on tau deposits such as neurofibrillary tangles (Consensus criteria, NIA and Reagan Institute3). A key issue is whether the neuropathological hallmarks can be correlated with the progression of the clinical symptoms, and whether they allow a backextrapolation to early pre-clinical stages. This seems possible in the case of tau deposits. Their spreading from the transentorhinal region to the hippocampus and the neocortex has been subdivided into six stages, the first two of which are pre-clinical4. The first sign in affected neurons is an unusual phosphorylation of tau, followed by aggregation. There is a good correlation between tau pathology and loss of synapses, leading to the proposal that, in AD, neuronal transport is impaired5. A strong driving force in AD research is the amyloid cascade hypothesis (see accompanying article by Selkoe6 and references therein). It states that improper processing of the membrane protein APP (amyloid precursor protein) lies at the root of AD because this generates the neurotoxic and aggregating Ab peptide. The major evidence comes from mutations in the gene encoding APP (on chromosome 21) that are linked to familial forms of AD and enhance the generation of Ab. Further support comes from mutations in the genes encoding presenilin 1 and 2 on chromosomes 14 and 1, which also enhance the level of Ab. Finally, ApoE4, a risk factor for AD, is also associated with amyloid plaques. By contrast, no AD-causing mutations have been detected thus far in the gene encoding tau (on chromosome 17). However, several mutations have emerged as the cause of related dementias (such as FTDP-17 – fronto–temporal dementia with parkinsonism linked to chromosome 17), which show pronounced tau deposits7–9. These ‘tauopathies’ added fresh fuel to the discussion on the causes of AD. In addition, despite the important evidence derived from mutations in APP or presenilin, it should be noted that altogether they account for only a minority of AD cases. The majority must be explained
Eva-Maria Mandelkow and Eckhard Mandelkow Neurofibrillar protein aggregates containing tau are one of the major hallmarks of Alzheimer’s disease (AD). In normal cells, tau stabilizes axonal microtubules, which are the tracks for intracellular traffic. In AD, tau becomes abnormally phosphorylated, aggregates into paired helical filaments and loses its ability to maintain the microtubule tracks. There is renewed interest in tau as a causative factor in neurodegenerative disease based on recently discovered mutations in the gene encoding tau. This article discusses how changes in tau protein could lead to retraction of neuronal processes and thus cell death and argues that tau pathology, rather than b-amyloid, might be the most reliable indicative factor for AD.
by non-genetic factors, which points again to the pathology of tau as being the best correlate for the clinical progression and as the most reliable neuropathological indicator. Physiological functions of tau Tau is one of the microtubule-associated proteins (MAPs) that stabilize neuronal microtubules for their role in the development of cell processes, establishment of cell polarity and intracellular transport10. Tau occurs mainly in axons (in contrast to the somatodendritic MAP2). A single gene encodes tau, which generates six isoforms, of 352 to 441 amino acid residues, in the human central nervous system by alternative splicing (Fig. 1), plus a larger variant in peripheral nerves11. The synthesis of tau is unusual in that tau mRNA is transported to the proximal axon where translation occurs12. This is one mechanism by which neuronal polarity is established. Tau contains an acidic N-terminal domain, a basic and proline-rich middle domain, a basic domain containing three or four internal repeats, and a C-terminal domain. Tau is very hydrophilic, soluble and displays a ‘natively unfolded’ structure. It can be phosphorylated at multiple sites, some of which regulate its microtubule-binding properties. The sites can be broadly subdivided into two classes: several Ser-Pro or Thr-Pro motifs occur in both regions flanking the internal repeats and are targets of prolinedirected kinases such as glycogen synthase kinase 3,
0962-8924/98/$ – see front matter © 1998 Elsevier Science. All rights reserved. PII: S0962-8924(98)01368-3
The authors are in the Max-PlanckUnit for Structural Molecular Biology, Notkestrasse 85, D-22603 Hamburg, Germany. E-mail: mand@ mpasmb.desy.de
425
FORUM
comment MAPK GSK-3b Cdk5
PKA
MARK PP2A
*
PP2B
N
C 1
2
3
4
G272V N279K
P189A
K280–
V337M
R406W
P301L
* TP
SP
KXGS
Cys
S214
Mutations
Alternative splicing
FIGURE 1 Diagram of tau protein, its domains, sites of phosphorylation and mutations. The near-N-terminal inserts (exons 2 and 3, grey) and the second of the four repeats (exon 10) can be absent owing to alternative splicing. The ‘assembly domain’ comprises the C-terminal half; it binds to microtubules and stabilizes them, whereas the N-terminal ‘projection domain’ protrudes from the microtubule surface. The regions flanking the repeats are rich in Ser-Pro or Thr-Pro motifs, which can be phosphorylated by proline-directed kinases and thus form epitopes of antibodies diagnostic of Alzheimer tau. Ser214 and Ser262 (in the KXGS motif of the first repeat) are non-proline motifs phosphorylated in Alzheimer’s disease and in vitro by PKA or MARK; this causes the detachment of tau from microtubules. Some of the mutations linked to dementias with tauopathy are indicated; others are silent at the protein level. Abbreviations: Cdk, cyclin-dependent kinase; Cys, cysteine; GSK-3, glycogen synthase kinase 3; MAPK, mitogen-activated protein kinase; MARK, microtubuleaffinity-regulating kinase; PKA, protein kinase A; PP, phosphatase; SP, serine-proline motif; TP, threonine-proline motif.
cyclin-dependent kinase Cdk5 or MAP kinase13,14. These sites have only a moderate influence on tau–microtubule interactions but are useful as diagnostic tools for the AD-like phosphorylation of tau. Other sites include targets of protein kinase A (e.g. Ser214), microtubule-affinity-regulating kinase (MARK; at KXGS motifs including Ser262, Ser356) or Ca21/calmodulin-dependent protein kinase (Ser416). Tau detaches from microtubules when phosphorylated at Ser262 or at Ser214, two sites that are phosphorylated in AD. In mitotic cells, tau shows enhanced phosphorylation at several Ser-Pro motifs and at Ser214, concomitant with its detachment from microtubules15. Phosphorylation of tau or related MAPs by MARK appears to be important for the establishment of cell polarity, but overactivity of MARK leads to cell death10. Tau is necessary for the outgrowth of neurites16. Axons show a gradient of tau concentration, with a maximum near the tip17, and a gradient of tau phosphorylation, highest in the proximal axon18. However, transgenic mice lacking tau have no major defects, possibly because other MAPs can substitute for tau19. Several observations point to other, less-well-characterized functions of tau. The size of the N-terminal domains of MAPs regulates the spacing between microtubules in neuronal processes; in tau, this domain is relatively small and would allow a close approach of microtubules (unlike MAP2)20. A fraction
426
of tau associates with the plasma membrane rather than with microtubules21. Neuronal cells possess mechanisms to sort MAPs into specific compartments. Although tau is sorted mainly into the axon, this preference is not absolute, and overloading the cell with tau can lead to its appearance in dendrites22. Some forms of tau have even been detected in the nucleus23. MAPs can also serve as anchoring devices for cellular proteins. A recent example is the role of tau as a targeting subunit of protein phosphatase 138. Finally, the ubiquitous MAP4 can regulate vesicle transport along microtubules24; a similar effect appears to operate for tau as well25. Pathological alterations of tau in AD In AD, the properties of tau change in several ways. • AD-tau shows an abnormal ‘hyperphosphorylation’ at many sites13,14,26. Some increase occurs also in foetal tissue and in mitotic cells, leading to the hypothesis that ‘mitotic’ signals received by differentiated neurons might cause hyperphosphorylation of tau and, later, death by apoptosis. Many of the abnormal phosphorylation sites are at Ser-Pro or Thr-Pro motifs; this explains why various antibodies raised against AD-tau react with such phosphorylated epitopes and are now used in the diagnosis of Alzheimer brain tissue or for the development of cell models. This analysis has revealed that abnormal phosphorylation of tau occurs before aggregation. • AD-tau also shows a loss of microtubule binding, which is probably a consequence of the hyperphosphorylation at sites (e.g. S262 or S214) that detach tau from microtubules. This could account for the disappearance of microtubules, causing the breakdown of intracellular traffic, which would result in the ‘dying back’ of axons. • The redistribution of tau from an axonal to a somatodendritic pattern suggests that, in AD, there is a defect in directing tau into the right compartment. This could arise from several mechanisms – for example, increased mRNA levels and tau synthesis – which could lead to overloading of the cell and perturbation of the sorting mechanism. • The aggregation of tau is particularly enigmatic, considering its unusual solubility. The aggregates are termed ‘paired helical filaments’ (PHFs) because of their two-stranded appearance, with widths between 10 and 20 nm and crossover repeats of 80 nm27. The PHFs in turn bundle into ‘neurofibrillary tangles’ or ‘neuropil threads’. In vitro, PHF assembly is inefficient but can be enhanced by oxidation, which leads to dimerization of tau28, and by interaction with polyanions such as heparin29 or RNA30. • Ubiquitination and proteolysis are posttranslational modifications of AD-tau that probably represent cellular attempts to degrade the aberrant protein (via the proteasome or calpain pathway31); however, since some proteolytic fragments are detected at early stages, this could also contribute to the nucleation of PHFs32. • Glycation is a consequence of oxidative damage and crosslinking, which accumulates once tangles are formed33. trends in CELL BIOLOGY (Vol. 8) November 1998
FORUM
• There is an increase of tau in the cerebrospinal fluid (from ~200 pg ml–1 to ~600 pg ml–1), which probably arises from dying neurons34. This feature could become valuable as an early diagnostic assay.
comment Cell body
1 2 3 4
Kinases Phosphatases
1 2 3 4
P 1 2 3 4
P
PHF A role for tau in AD? 1 1 2 2 The above observations can be placed 3 1 P 1 P 3 Axon 4 2 2 4 in a hypothetical scheme (Fig. 2): inap3 3 4 4 propriate signals lead to an imbalance of kinases/phosphatases, hyperphosSynapse phorylation of tau, its detachment from Axonal Tau stabilizes Tau phosphorylation PHF microtubules, microtubule breakdown, transport microtubules MT depolymerization assembly aggregation of tau into PHFs, breakdown of intracellular transport and degeneration of neurons. However, many FIGURE 2 intermediate steps are still unknown. Model showing a possible link between axonal transport, microtubules and tau in Alzheimer’s disease. For example, the effect of different ki- Transport relies on microtubules as tracks, stabilized by the ties of tau protein. Phosphorylation (P) at nases on tau has been examined in vitro crucial sites detaches tau from microtubules, leading to the breakdown of microtubules and the and in cell models10, but which of accumulation of tau aggregated into paired helical filaments (PHFs). them is responsible for tau phosphorylation in AD remains unknown. PHFs from recom12 Litman, P., Barg, J. and Ginzburg, I. (1994) Neuron 13, 1463–1474 binant tau can now be generated efficiently in vitro35, 13 Mandelkow, E-M. et al. (1995) Neurobiol. Aging 16, 355–362 but attempts to generate them in cell models have 14 Trojanowski, J. Q. and Lee, V. M. Y. (1995) FASEB J. 9, 1570–1576 failed so far. Conditions for generating ‘abnormal’ 15 Illenberger, S. et al. (1998) Mol. Biol. Cell 9, 1495–1512 phosphorylation in cells have been described (e.g. mi16 Kosik, K. S. and McConlogue, L. (1994) Cell Motil. Cytoskeleton totic signals15 or activation of signalling cascades36), 28, 195–198 but in general this does not lead to neuronal degen17 Black, M. et al. (1996) J. Neurosci. 16, 3601–3619 eration per se. Moreover, pathological markers of AD 18 Mandell, J. W. and Banker, G. A. (1996) J. Neuroscience 16, appear initially in single neurons, begging the ques5727–5740 tion of why there is selective vulnerability. Finally, 19 Harada, A. et al. (1994) Nature 369, 488–491 overexpressing human tau isoforms in transgenic 20 Matus, A. (1994) Trends Neurosci. 17, 19–22 mice has not resulted in an AD-like pathology37. 21 Brandt, R., Leger, J. and Lee, G. (1995) J. Cell Biol. 131, 1327–1340 How might the newly discovered tau mutations 22 Hirokawa, N. et al. (1996) J. Cell Biol. 132, 667–679 be involved? Remarkably, the mutations can occur in 23 Wang, Y. et al. (1993) J. Cell Biol. 121, 257–267 both coding and noncoding regions of the gene 24 Bulinski, J. C. et al. (1997) J. Cell Sci. 110, 3055–3064 encoding tau, with similar effects. One possible expla25 Ebneth, A. et al. J. Cell Biol. (in press) nation is that the mutations affect the splicing of tau 26 Delacourte, A. and Buee, L. (1997) Int. Rev. Cytol. 171, 167–224 mRNA, especially at the exon 10 boundary, so that 27 Crowther, R. A. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, a greater proportion of four-repeat tau is generated, 2288–2292 which might somehow affect one of the tau-depen28 Schweers, O. et al. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, dent reactions in the cell and lead to tau hyper8463–8467 phosphorylation, aggregation and cell death8. The 29 Goedert, M. et al. (1996) Nature 383, 550–553 dementias associated with the tau mutations are dis30 Kampers, T. et al. (1996) FEBS Lett. 399, 344–349 tinct from AD and affect different regions in the 31 Litersky, J. M. and Johnson, G. V. W. (1995) J. Neurochem. 65, brain. However, there may be a common link between 903–911 all ‘tauopathies’, which we hope will lead to a better 32 Novak, M., Kabat, J. and Wischik, C. M. (1993) EMBO J. 12, understanding of the pattern of neuronal loss in AD. 365–370
References 1 2 3 4 5 6 7 8
Khachaturian, Z. S. (1985) Arch. Neurol. 42, 1097–1105 Mirra, S. et al. (1994) J. Neuropathol. Exp. Neurol. 53, 303–315 Ball, M. et al. (1997) Neurobiol. Aging 18, 1–2 Braak, H. and Braak, E. (1991) Acta Neuropathol. 82, 239–259 Terry, R. D. (1996) J. Neuropathol. Exp. Neurol. 55, 1023–1025 Selkoe, E. (1998) Trends Cell Biol. 8, 447–453 Poorkaj, P. et al. (1998) Ann. Neurol. 43, 815–825 Spillantini, M. et al. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 7737–7741 9 Hutton, M. et al. (1998) Nature 393, 702–705 10 Drewes, G., Ebneth, A. and Mandelkow, E-M. (1998) Trends Biochem. Sci. 23, 307–311 11 Goedert, M. (1997) Neuroscientist 3, 131–141 trends in CELL BIOLOGY (Vol. 8) November 1998
33 34 35 36 37
Yan, S. D. et al. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 7787–7791 Vigo-Pelfrey, C. et al. (1995) Neurology 45, 788–793 Friedhoff, P. et al. (1998) Biochemistry 37, 10223–10230 Anderton, B. et al. (1995) Neurobiol. Aging 16, 389–397 Götz, J. et al. (1995) EMBO J. 14, 1304–1313 Acknowledgements
Reference added in proof: 38 Liao, H. et al. (1998) J. Biol. Chem. 273, 21901–21908
Authors’ note: Debates on mechanisms of neuronal degeneration in Alzheimer’s disease held during the recent International Alzheimer congress in Amsterdam can be found on the Internet at: http://www.alzforum.org/members/index.html.
We thank J. Biernat and A. Ebneth for help in designing the figures. Our work is supported by Deutsche Forschungsgemeinschaft (Alzheimer Research Group, Hamburg).
427