- Email: [email protected]
Molecular characterization of microtubule-associated proteins tau and map2
(1988) Brain Res. 439, 211-221 39 Bouton, M. S. and Bittner, G. D. (1981) Cell Tissue Res. 219, 379-392 40 Hay, R. R., Bittner, G. D. and Kennedy, D. (1967) Science 156, 251-252 41 Nordlander, R. H. and Singer, M. (1972) Z. Zellforsch. Mikrosk. Anat. 126, 157-181 42 Clark, R. D. (1976) J. Camp. Neural. 170, 253-266 43 Clark, R. D. (1976) J. Camp. Neural. 170, 267-278 44 Boulton, P. S. (1969) Z. Zellforsch. Mikrosk. Anat. 101, 98-118 45 Tung, A. S. C. and Pipa, R. L. (1971) J. Ultrastruct. Res. 36, 694-707 46 Dereimer, S. A., Elliot, E. J., Macagno, E. R. and Muller, K. J. (1983) Brain Res. 272, 157-161 47 Birse, S. C. and Bittner, G. D. (1980) J. Neurophysiol. 45, 724-742 48 Bittner, G. D. and Brown, M. A. (1981) Brain Res. 218, 357-364 49 Muller, K. H. and Carbonetto, S. (1979) J. Camp. Neural. 185, 485-516 50 Elliot, E. J. and Muller, K. J. (1981) Brain Res. 218, 99-114 51 Elliot, E. J. and Muller, K. J. (1983) J. Neurosci. 3, 1994-2006 52 Murphy, A. D. and Kater, S. B. (1978) Brain Res. 156,
322-328 53 Wilson, D. (1960) J. Exp. BioL 37, 57-72 54 Young, J. Z. (1942) PhysioL Rev. 22, 318-373 55 Koopowicz, H. and Holman, M. (1988) Am. Zool. 28, 1065--1076 56 Perry, V. H., Brown, M. C., Lunn, E. R., Tree, P. and Gordon, S. (1990) Eur. J. Neurosci. 2, 802-808 57 Perry, V. H., Lunn, E. R., Brown, M. C., Cahusac, S. and Gordon, S. (1990) Eur. J. Neurosci. 2, 408-413 58 Strand, F. L., Rose, K. J., King, J. A., Segarra, A. C. and Zuccarelli, L. A. (1989) Prog. Neurobiol. 33, 45-85 59 Velez, S., Bittner, G. D., Govind, C. K. and Atwood, H. L. (1981) Exp. Neural. 71,307-325 60 Krasne, F. B. and Lee, S. H. (1977) Science 198, 517-519 61 Muller, K. J. (1988) Am. ZooL 28, 1091-1097 62 Mason, A. and Muller, K. J. (1982) Nature 296, 655-657 63 Fernandez, J. and Fernandez, M. S. (1974) Nature 251, 428-430 64 Nordlander, R. H, and Singer, M. (1976) Cell Tissue Res. 166, 445-460 65 Krause, L. and Bittner, G. D. (1990) Proc. NatlAcad. Sci. USA 87, 1471-1475 66 Eidetberg, E., Straehley, D. and Erspamer, R. (1967) Exp. Neural. 56, 312-322
Acknowledgements I wouldlike to thank my colleagues M. Ballinger,J. Blundon, T. Krause, J. Moehlenbruckand R. Shellerfor their commentsand contribub'ons to this manuscript. Much of this recent work was funded by the Texas Advanced TechnologyProgram and the National ScienceFoundation.
Molecularcharaderizationofmicrotubule-assodatedproteins tau andMAP2 M. Goedert,
R. A . C r o w t h e r
Tau and MAP2 are two of the major microtubuleassociated proteins in the vertebrate nervous system. They promote microtubule assembly and stability, and might be involved in the establishment and maintenance of neuronal polarity. In nerve cells immunohistochemistry shows complementary distributions, with tau being concentrated in axons and high molecular mass MAP2 being confined to dendrites. Each protein consists of multiple isoforms that contain three or four homologous tandem repeats near the carboxy-terminus, which constitute microtubule-binding domains. In humans, tau consists of at least six isoforms of related amino acid sequences that are produced from a single gene by alternative mRNA splicing and that are expressed in a stage- and cell type-specific manner. Tau is also a component of the paired helical filaments associated with Alzheimer's disease and other disorders of the CNS. Rat MAP2 consists of at least three isoforms produced from a single gene: high molecular mass MAP2a and MAP2b, and low molecular mass MAP2c. MAP2c is expressed only during early development and has so far been seen only in axons; MAP2a appears to replace MAP2c, whereas MAP2b is expressed throughout life. Messenger RNAs for MAP2 of high molecular mass are expressed both in cell bodies and in dendrites, consistent with the dendritic localization of the corresponding protein isoforms. Brain microtubules contain a variety of microtubuleassociated proteins that co-purify with tubulin during repeated cycles of assembly 1. Knowledge of the properties of these proteins is essential for an understanding of the function of microtubules in specialized cells. Microtubule-associated proteins are characterized functionally by their ability to promote the assembly and stability of microtubules in axons TINS, Vol. 14, No. 5, 1991
a n d C. C. G a r n e r
and dendrites, suggesting a major role in the determination of neuronal morphology and plasticity. Microtubules are known to undergo extensive growth and rearrangement in nerve cells, especially during periods of neurite outgrowth. Tau and MAP2 are two of the major and moststudied microtubule-associated proteins of the vertebrate nervous system 2. Immunohistochemistry shows complementary distributions with tau being found mostly in axons and high molecular mass MAP2 being confined to dendrites. Electron micrographs show these proteins as lateral projections that appear to cross-bridge adjacent microtubules (Fig. 1)3'4. The spacing between adjacent microtubules in the presence of tan is about 20 nm (Fig. 1A), while that for high molecular mass MAP2 is about 100 nm (Fig. 1B), which is similar to the cross-bridging structures found in a dendrite (Fig. 1C). This suggests that tau and MAP2 each contain a microtubule-binding domain and a projection domain. Since microtubules are inherently unstable structures, the stabilizing effects of tan and MAP2 are likely to be important for the integrity of processes that depend on stable microtubules, such as rapid axonal transport. The finding that tau is a constituent of paired helical filaments, the major component of the neurofibrillary tangle of Alzheimer's disease 5, indicates that some microtubule-associated proteins are involved in pathophysiological processes. Although tau and MAP2 were the first microtubuleassociated proteins to be discovered some 15 years ago6'7, only recently have their primary amino acid sequences been determined s-19, thus opening the way to detailed molecular study. Tan and MAP2 are each encoded by a single gene, from which multiple isoforms are produced in a developmentally regulated manner by alternative mRNA splicing. The primary
© 1991, ElsevierScience PublishersLtd, (UK) O166- 2236/91/$02.00
M. 6oedert and R. A. Crowther are at the MRC Laboratoryof MolecularBiology, Hills Road, Cambridge CB22QH, UK, and C C Garner is atthe Centre for Molecular Neurobiology, University of Hamburg, Martinistrasse.52, 2000Hamburg20, FRO.
193
molar concentration of peptide required is two orders of magnitude higher than that of full length tau protein22'23'~5. This difference might be explained by less than a full repeat being used, by the reiteration of the repeat sequence in the complete protein and by conformational changes imposed by sequences lying outside the repeat region. Recombinant tau isoforms with four repeats induce faster microtubule assembly than those with three repeats ~1, as might be expected if one tan molecule links a number of tubulin subunits. Tan is thought to be an elongated molecule without much secondary structure. However, images of tau paracrystals have shown it to be unusually elastic, ' i . . . . . !,~,~, "~ .:~., ',., suggesting that interactions between different parts of the molecule could occur26. Besides being distinFig. 1. Tau and MAP2 form projections that appear to cross-bridge adjacent guished by the presence of three or four tandem microtubules. Quick-frozen, deep-etched pellets of microtubules polymerized repeats, some tau isoforms isolated from human brain with (A) tau or (B) MAP,?. (C) Similarly prepared dendrite of a motor neurone tissue contain inserts of 29 or 58 amino acids located from bovine spinal cord, in which neurofilaments are seen as thin filaments near the amino-terminus (Fig. 2) 16. These sequences running parallel to microtubules. Note the frequent cross-bridges between are rich in proline and charged amino acid residues and microtubules in (,4) and (B) and the cross-bridges between microtubules and share some similarity with the carboxy-terminus of neurofilaments in (C). Scale bar is 100 nm. (Courtesy of N. Hirokawa.) the middle neurofilament subunit. Their presence might well modulate the presently unknown function amino acid sequences are highly similar at the of the projection domain, the sequence of which carboxy-terminus, which contains three or four tan- shows a striking charge separation, with acidic dem repeats and represents the microtubule-binding residues grouped towards the amino-terminus and domain. The amino-terminal parts of tau and MAP2, basic residues towards the carboxy-terminus. representing the projection domains, appear unWhen individual human tan isoforms that are related but both contain large numbers of prolines and expressed in E. coil ~1 are mixed together, they give a charged amino acids, which suggests an extended characteristic set of six bands (Fig. 2), ranging from structure. MAP-U, a ubiquitous microtubule- 48 to 67 kDa in apparent molecular masses. True associated protein, shows a similar structure with molecular masses of the different isoforms range from tandem repeats at the carboxy-terminus that are 37 to 46 kDa21; therefore, tau is anomalously retarded similar to those of tau and MAP22°; this indicates the on denaturing gels, as has been clear since the existence of a family of microtubule-associated pro- sequencing of the first tau isoforms s'9. This behaviour teins with similar microtubule-binding domains. has been attributed to the extended configuration of taus'26. The purification of tan from brain tissue makes A family of tau proteins use of its association with microtubules in repeated Ever since tau was first purified from adult brain cycles of assembly, its heat stability and its solubility tissue it has been known to exist in multiple forms, in perchloric acid; tan protein extracted from adult which were later shown to be subject to develop- mammalian brain tissue in this way runs as 4-6 bands mental regulation 1. Molecular cloning has now defined on SDS-PAGE. Comparison of the pattern of recomat least six tau isoforms that are generated from a binant tau isoforms with that of native tan shows that single gene by alternative mRNA splicing. To date, the four major tau isoforms in adult human brain cDNA cloning has yielded tau sequences from mouse, correspond to isoforms with three and four repeats rat, cow and humanss-16. The six human tan isoforms without amino-terminal insertions, and to isoforms that have so far been isolated as full length cDNA with three and four repeats containing the aminoclones have sequences that range from 352 to 441 terminal insertion of 29 amino acids (Fig. 2, bands A, amino acids in length and differ from each other by the B, D, E)~1. presence or absence of three inserts (Fig. 2) 9'1°'16,21. Phosphorylation also contributes to the complex Variable phosphorylation of the different isoforms pattern of tau bands, since some bands exhibit an adds to the complexity. increased mobility on denaturing gels after alkaline The most striking feature of the tau sequences is phosphatase treatment. Recombinant tau proteins can the presence of three or four tandem repeats of 31 or be phosphorylated in vitro by cPuMP-dependent 32 amino acids located at the carboxy-terminus, each protein kinase, multifunctional Ca2+/calmodulincontaining a characteristic Pro-Gly-Gly-Gly motif. dependent protein kinase II (CaM kinase II), protein The extra repeat in the isoforms with four repeats is kinase C and casein kinase I127. Interestingly, only inserted within the first repeat of the isoforms with phosphorylation by CaM kinase II results in a reduced three repeats in a way that preserves the periodic mobility on denaturing gels; the same holds true for pattern. Twelve residues are completely conserved native tan extracted from brain tissue 28. Although between the four repeats and a further four residues there are many potential phosphorylation sites for the show conservative changes (Fig. 3). Experiments above protein kinases throughout the tau sequence, it with synthetic peptides and with tan fragments is only the carboxy-terminal half that becomes phosproduced by expression in E. coli show that the phorylated. Whereas cAMP-dependent protein repeats constitute microtubule-binding domains22-25. kinase, protein kinase C and casein kinase II phosMicrotubule polymerization is promoted by synthetic phorylate tau at multiple sites, CaM kinase II phos18-mer peptides from the repeat region, although the phorylates a single serine residue that is located 26 194
TINS, VoL 14, No. 5, 1991
amino acids from the carboxy-terminus of human tau 27. Phosphorylation of tau proteins is likely to be of functional importance, since it leads to a reduced affinity for microtubules 29. It is interesting to speculate that differential phosphorylation of tau in axons and dendrites might be responsible for its axonal location. The human cDNA clones isolated to date account for most, if not all, tau isoforms present in adult human brain. The screening of mouse and cow cDNA and genomic libraries has suggested the existence of other forms of taus'u'12. These include bovine isoforms with an alternative amino-terminus or an alternative extended carboxy-terminus, and a mouse isoform with an extension of 25 amino acids at the carboxy-terminus. The status of these reported clones is unclear, since the corresponding proteins (if they exist) might not necessarily be isolated by
Tandem
repeat
r e g i o n of t a u DL
K N V K S
K I G S T E N L
K H Q P G G G K
[V Q i I N K K L D L S N V Q S K C G S K D N I K H V P G G G S] VQ
i V Y K P V D L
S KVTS
K CG
S LGN
I HHK
P G G G Q
V E V K S E K L D F KDR V Q S K I G S L D N I T H V P G G G N K K I E T H K L T F REN A K A K #
Tandem
#*
repeat
#
region
*
i VTK
K I DL
**
*#
*
* * * *
of M A P 2
D L KNVK VQ
**
S K I GS
S HVT
S K CG
TDN
I K YQP
S L KN
KGGQ
I R H R P G G G R
V K I E S V K L D F KEK A Q A K V G S L D N A H H V P G G G N V K I D S Q K L N F REH A K A K V ,
*
#
*
#*
#
#
*
**
*
#
*
**
Consensus
352
VK
A 1
381
B 1
410
C 383
1
D 412
1
E 441
mll
F 1
234
94--
56
~,ii¸)
67-43--
30-
Fig. 2. (Upper panel) Schematic representation of six human tau isoforms. The region common to all isoforms is tinted and the three inserts that distinguish them are shown in white. The three or four tandem repeats are marked by black bars. (Lower panel) A comparison of recombinant tau isoforms with human brain tau. Lanes 3 and 6, mixture of recombinant tau isoforms with each band identified by letter; lanes I and 2, tau extracted from fetal brain before and after alkaline phosphatase treatment; and lanes 4 and 5, tau extracted from adult cerebral cortex before and after alkaline phosphatase treatment. After electrophoresis on a 10% SDS-PAGE gel the tau isoforms were visualized by western blotting. Note the alignment of major brain tau bands with the recombinant isoforms (labelled A-F). (Modified, with permission, from Ref. 21 .)
TINS, Vol. 14, No. 5, 1991
I E S K KLDL
K N V Q S
K I GS
5DN
I K H V P G G G N
Fig. 3. Four tandem repeats in the human tau sequence and three homologous tandem repeats in the rat MAP2 sequence, with a derived consensus sequence. These tandem repeats are located near the carboxy-terminus and constitute the microtubule-binding domains of tau and MAP2. Tau can exist with three or four tandem repeats; the sequence giving rise to the extra tau repeat is bracketed. Similarity is indicated as follows: *, identical residues in all repeats; #, identical residues in all repeats except one, which shows a conservative replacement. Note that the MAP2 repeats show more conservation towards the amino-terminal end of each repeat, while the tau repeats show more conservation towards the carboxy-terminal end of each repeat. The sequences have been aligned optimally; thus, in two of the tandem repeats of both tau and MAP2, there appear extra inserted residues. The sequence of bovine tau is identical to that of human tau in the repeat region, whereas in mouse and rat tau an arginine replaces a lysine at position 6 of the first repeat. The sequences of mouse and rat MAP2 are identical in the repeat region.
standard tau extraction procedures. Moreover, the nucleotide sequence with a high degree of similarity to that claimed to be coding the longer carboxy-terminus in the extended mouse protein forms part of an intron in the cow gene. There is no intron in the human gene at the point corresponding to where the two mouse sequences diverge (Goedert, M., Crowther, R. A. and Garner, C. C., unpublished observation). In the nervous system, mRNA transcripts encoding the various tau isoforms are expressed in a manner that is stage- and cell type-specific 1°'16. Transcripts encoding isoforms with three and four repeats are found throughout the adult human brain and their levels do not vary much between different regions. By contrast, in fetal brain, although transcripts containing three repeats are abundant, no transcripts containing four repeats are detectable. The developmental shift of tau bands from a simple fetal pattern to a more complex adult pattern thus involves the transition from the expression of the isoform with three repeats and no insertions to the expression of all six isoforms. In situ hybridization has shown that tau protein mRNAs occur only in neurones, and that different neuronal cell types express different isoforms 1°'14. Transcripts encoding isoforms with three or four repeats are found in pyramidal cell bodies throughout all layers of the adult human cerebral cortex. In the hippocampal formation, transcripts for isoforms with three repeats are present in granule cells of the dentate gyrns, pyramidal cells throughout the cornu 195
It is not known why this otherwise soluble axonal protein ends up in an insoluble form in affected nerve cells. A simple overproduction of tau does not appear to be responsible, since transcript levels are not significantly changed 16. Tan is probably posttranslationally modified in Alzheimer's disease; indirect evidence suggests that it is abnormally phosphorylated35-39, since certain tan antibodies stain Q neurofibrillary tangles only after treatment of tissue sections with alkaline phosphatase 35. Moreover, extraction of tan protein with SDS from affected brain regions and from paired helical filaments gives tan bands that are not observed in tau extracted from unaffected brain regions or from brain tissue of control patients 36-39. These molecular species show an increased gel mobility after alkaline phosphatase treatment. At present, it is not known whether abnormal / phosphorylation represents a causal event in the formation of paired helical filaments, or whether it is a mere consequence of a proportion of tau being Fig. 4. Cellular localization of tau protein mRNAs in adult immobilized in cell bodies and dendrites rather than human hippocampal formation. A probe spedfic for moving into the axon. transcripts encoding isoforms with three repeats labels Pathology involving tan protein is not limited to numerous nerve cell bodies. Abbreviations: CA, cornu Alzheimer's disease, but is also observed in a number ammonis; DG, dentate gyms; 5U, subiculum. Scale bar is of disorders of the CNS, such as Pick's disease and 350t~m. (Taken, with permission, from Ref. 10.) progressive supranuclear palsy4°. This suggests that tan might be affected in a relatively nonspecific manner in response to various kinds of nerve cell ammonis (CA) layers and pyramidal cells in the damage. Alternatively, different modifications of tau, subiculum (Fig. 4). Transcripts encoding isoforms each specific to a particular disease, might lead to the with four repeats are also found in pyramidal cells in formation of similar neurofibrillary tangles. the subiculum and hippocampus, but not in granule cells of the dentate gyms. Isoforms of MAP2 The availability of sequenced cDNA clones for tan The term MAP2 generically refers to three proprotein has enabled preliminary functional studies to teins of related amino acid sequence that are produced be undertaken. Experiments using antisense oligo- from a single gene. MAP2a and MAP2b have apparent nucleotides to block tau expression in cultured nerve molecular masses of 288 and 280 kDa on SDS-PAGE cells suggest that axonal morphology might be depen- gels, whereas MAP2c runs at 70 kDa 1'2. The exdent on tau expression 3°, though it is unclear whether pression in E. coli of a rat cDNA encoding the high tan plays a crucial role in the establishment of molecular mass MAP2 of 1830 amino acids results in a neuronal polarity. Recently, tau protein has been protein that aligns with rat MAP2b on gels (Fig. 5) 19. expressed from cDNA clones in non-neuronal cells is. It is not known whether MAP2a is produced by As expected, transfection of fibroblasts results in the alternative splicing or whether it is the product of binding of the expressed tau protein to microtubules. post-translational modification of MAP2b. Similar In addition, microtubules are aggregated into thick experiments with cDNA encoding MAP2c have longitudinal bundles, suggesting an additional bundling demonstrated that the expressed protein aligns on effect of tan when expressed at high levels. gels with dephosphorylated MAP2c that has been extracted from rat brain at postnatal day 5 (P5) (Fig. Tau proteins and Alzheimer's disease 5) 19. As is the case for the tau isoforms, MAP2b and Alzheimer's disease is characterized neuropatho- MAP2c are anomalously retarded on SDS-PAGE logically by the presence of abundant amyloid plaques gels. MAP2c, which is made up of 467 amino acids, and neurofibrillary tangles in cerebral cortex and arises via alternative mRNA splicing and lacks 1363 hippocampus, as well as in some subcortical nuclei. amino acids from the middle of MAP2b41. MAP2b is Neurofibrillary tangles consist of insoluble filamentous expressed in rat brain throughout life, while MAP2c is material that is deposited mostly in cell bodies and expressed in most brain regions only before P1042. dendrites of affected nerve cells. Paired helical This is in contrast with MAP2a, which appears after filaments constitute the major component of the P10 and appears to replace MAP2c in mature neurofibrillary tangle and both immunohistochemical neurones. The forms of MAP2 that have high and biochemical studies have demonstrated that these molecular masses are expressed almost exclusively in filaments contain various tau protein isoforms 5'16'31-33. the dendrites of nerve cells 1'2, while MAP2c is found At present, it is unclear whether tan is the only in axons during development along with tau 43, constituent of paired helical filaments or whether although it is not known whether it is also present in more components remain to be discovered. A recent dendrites. All three isoforms appear to be expressed study has provided the strongest evidence yet that it in macroglia 43'44. The complete amino acid sequences is in fact modified tan that forms these filaments ~. of mouse and rat high molecular mass MAP2 and Final proof must await the assembly of paired helical rat MAP2c have been determined by molecular cloning ]7-19. filaments from defined components. 196
TINS, Vol. 14, No. 5, 1991
Limited proteolysis of high molecular mass MAP2 bound to microtubules demonstrated that it is composed of a small, microtubule-binding domain and a much larger projection domain that is thought to cross-link various cytoskeletal components. The 210 amino acids at the carboxy-terminus of MAP2 show a high degree of similarity with the carboxy-terminus of tau, and contain in particular three tandem repeats that bind to microtubules (Fig. 3) 17-19. As with tau, an octadecapeptide from the second MAP2 repeat promotes microtubule assembly in vitro45. Subsequent work has demonstrated that the repeat region of MAP2 is also able to bind to microtubules when expressed in cultured cell lines, and has suggested that sequences flanking the repeats are important for the binding of MAP2 to microtubules 46'47. Although it is unclear what role these neighbouring sequences might play, the presence of an upstream proline-rich stretch suggests that this coil region might function as a hinge between the projection domain and the microtubule-binding domain19. What role does the projection domain of MAP2 play in developing and mature dendrites? Possible functions include the cross-linking or spacing of various components of the neuronal cytoskeleton and the provision of an anchoring surface for enzymes. The latter has been demonstrated for the regulatory subunit of cAMP-dependent protein kinase II, which binds within the 100 amino acids at the amino terminus of all three MAP2 isoforms 48'49. A second protein, calmodulin, is thought to bind near the tubulin-binding domain at the carboxy-terminus 19'5°. The effects of both proteins influence the affinity of MAP2 for microtubules. Phosphorylation of MAP2 is likely to play an important role in the modulation of its function, as indicated by the finding that the activation of glutamate receptors of the N-methyl-o-aspartate subtype in hippocampal slices leads to a rapid and specific dephosphorylation of MAP251. The precise role of MAP2 in cross-linking the cytoskeleton is unclear. Ultrastructurally, MAP2 extends from the microtubule surface by some 100 nm and appears to form cross-links with the surrounding cytoskeleton (Fig. 1C)4. Since neuronal microtubules are arranged in longitudinal bundles, it was proposed that one function of MAP2 in dendrites might be to bundle microtubules. Experiments in which parts of MAP2 are expressed in cultured non-neuronal cells have identified a short region on the carboxy-terminal side of the tubulin-binding repeats that participates in bundling microtubules with a separation of about 25nm 46'47. This distance is shorter than that observed in vivo in nerve cells (Fig. 1C), and if the bundling activity does indeed reside in this sequence, then it leaves open the question as to what role the long arm of MAP2 might play in the dendritic cytoskeleton. Although microtubule bundling is clearly observed following overexpression of tau or MAP2, it might be a consequence of microtubule stabilization and be mediated by an endogenous bundling protein rather than by the overexpressed tau or MAP252. An added twist to the story comes from in situ hybridization, which shows that transcripts encoding both MAP2b and MAP2c are expressed in cell bodies of nerve cells, such as cerebellar granule cells, pyramidal ceils in cerebral cortex and hippocampus and dentate gyms granule cells (Fig. 6) 53,54. Strikingly, TINS, Vol. 14, No. 5, 1991
1
151
1514
o
1830
....---
MAP2c 1 151
467
1
..., 11111 MAP2b t MAP2a
MAP2c
2
~
3
4
5
~
.._
--
IR
Fig. 5. (Upper panel) Schematic representation of MAP2b and MAP2c. The region common to MAP2b and MAP2c is tinted and the three tandem repeats are marked by black bars. MAP2c is generated by an alternative splicing event that deletes 1363 amino acids from the middle of MAP2b. (Lower panel) A comparison of recombinant MAP2 isoforms with rat brain MAP2. Lanes 2 and 5, recombinant high molecular mass MAP2b and MAP2c, respectively; lane 1, MAP2 extracted from adult rat brain; and lanes 3 and 4, MAP2 extracted from rat brain at P5 before and after alkaline phosphatase treatment. After electrophoresis on a 3-15% SDS-PAGE gel, the MAP,? isoforms were visualized by western blotting. Note the alignment of brain MAP2b and MAP2c with the recombinant MAP2 bands. (Taken, with permission, from Ref. 19.)
the 9 kb mRNA encoding the MAP2 isoforms of high molecular mass is also seen in the dendritic compartment (Fig. 6), whereas the 6 kb mRNA encoding MAP2c is not is. The compartmentalization of specific mRNAs to dendrites allows local regulation of where particular proteins are synthesized, thereby possibly altering the stability of the dendritic cytoskeleton. The occurrence of high molecular mass MAP2 mRNA in the dendritic compartment represented the first example of the spatial segregation of an mRNA in nerve cells 53. Recently, mRNA encoding the o~ subunit of the multifunctional protein CaM kinase II has also been localized in cell bodies and dendrites of nerve cells 54'55. Spatial segregation of an mRNA could arise co-translationally, with protein-targeting signal., conferring specificity. Alternatively, specificity could lie at the mRNA level, with cis-acting signals in the RNA interacting with a receptor at the localization site. Evidence from localized maternal mRNAs in Xenopus laevis and Drosophila melanogaster eggs is in favour of the RNA targeting model 56'57. The dendritic localization both of MAP2 mRNAs and MAP2 proteins 197
Concluding remarks Tau and MAP2 provide prototypes for other microtubule-associated proteins found in neuronal and non-neuronal cells. They define a family of structurally related proteins that presently also includes MAP-U; there are probably more members of this family to be discovered. A second class is represented by MAP1B, a microtubule-associated protein of high molecular mass with no sequence similarity to tau or MAP2 59. The molecular characterization of tau and MAP2 has given us a glimpse of the complex and diverse organization and modification of the cytoskeleton. Both proteins have similar microtubule-binding domains, but possess unique projections at the aminoterminus that might modulate the local environment. Such modulation might involve active bridging between neighbouring filaments, or passive repulsion due to electrostatic effects, as has been proposed for the side arms of neurofilaments6°. Diversity of tau and MAP2 is generated by alternative splicing, and the capacity to regulate when and where they bind microtubules might depend on post-translational modifications and, for MAP2, on targeting of mRNAs.
Selected references
Fig. 6. A comparison of the cellular localization o f (A) MAP2 mRNAs, (B) nuclear staining and (C) MAP2 proteins in adult rat hippocampus. Note that MAP2 mRNAs are found in the granule cell layer and in the proximal portion of the dentate gyrus molecular layer, indicating a somatodendritic localization. By contrast, MAP2 proteins are found throughout the molecular layer, indicating a dendritic localization. Abbreviations: g, granule cell layer; m, molecular layer. Scale bar is 250 t~m. (Taken, with permission, from Ref. 54.)
of high molecular mass suggests that MAP2 might owe its restricted distribution to the sequestration of its mRNAs within the dendrite. However, there is also evidence indicating that the information for MAP2 segregation might reside within the protein itself 98. Thus, biotinylated MAP2, when injected into cultured nerve cells, is initially uniformly distributed in axons and dendrites, but then progressively disappears from the axon. There must therefore be some mechanism to clear MAP2 protein from inappropriate regions of the cell. 198
1 Matus, A. (1988)Annu. Rev. Neurosci. 11, 29-44 2 Nunez, J. (1988) Trends Neurosci. 11,477-479 3 Hirokawa, N., Shiomura, Y. and Okabe, S. (1988) J. CellBioL 107, 1449-1459 4 Hirokawa, N., Hisanaga, S. I. and Shiomura, Y. (1988) J. Neurosci. 8, 2769-2779 5 Brion, J. P. (1990) Semm. Neurosci. 2, 89-100 6 Weingarten, M.D., Lockwood, A. H., Hwo, S. Y. and Kirschner, M. W. (1975) Proc. Natl Acad. Sci. USA 72, 1858-1862 7 Sloboda, R. D., Rudolph, S. A., Rosenbaum, J. L. and Greengard, P. (1975) Proc. NatlAcad. Sci. USA 72, 177-181 8 Lee, G., Cowan, N. and Kirschner, M. (1988) Science 239, 285-288 9 Goedert, M., Wischik, C. M., Crowther, R. A., Walker, J. E. and Klug, A. (1988) Proc. NatlAcad. Sci. USA 85, 4051-4055 10 Goedert, M., Spillantini, M. G., Potier, M. C., Ulrich, J. and Crowther, R. A. (1989) EMBO J. 8, 393-399 11 Himmler, A. (1989) Mol. Cell. Biol. 9, 1389-1396 12 Himmler, A., Drechsel, D., Kirschner, M. W. and Martin, D. W. (1989) Mol. Cell. Biol. 9, 1381-1388 13 Mori, H. eta/. (1989) Biochem. Biophys. Res. Commun. 159, 1221-1226 14 Kosik, K. S., Orecchio, L. D., Bakalis, S. and Neve, R. L. (1989) Neuron 2, 1389-1397 15 Kanai, Y. et aL (1989) J. Cell Biol. 109, 1173-1184 16 Goedert, M., Spillantini, M. G., Jakes, R., Rutherford, D. and Crowther, R. A. (1989) Neuron 3, 519-526 17 Lewis, S. A., Wang, D. and Cowan, N. J. (1988) Science242, 936-939 18 Papandrikopoulou, A., Doll, T., Tucker, R. P., Garner, C. C. and Matus, A. (1989) Nature 340, 650-652 19 Kindler, S., Schulz, B., Goeclert, M. and Garner, C. C. (1990) J. Biol. Chem, 265, 19679-19684 20 Aizawa, H. etal. (1990)J. BioL Chem. 265, 13849-13855. 21 Goedert, M. and Jakes, R. (1990) EMBOJ. 9, 4225-4230 22 Ennulat, D. J., Liem, R. K. H., Hashim, G. A. and Shelanski, M. L. (1989) J. Biol. Chem. 264, 5327-5330 23 Aizawa, H. etal. (1989)J. Biol. Chem. 264, 5885-5890 24 Lee, G., Neve, R. L. and Kosik, K. S. (1989) Neuron 2, 1615-1624 25 Maccioni, R. 8., Vera, J. C., Dominguez, J. and Avila, J. (1989) Arch. Biochem. Biophys. 275, 568-579 26 Lichtenberg, 8, Mandelkow, E. M., Hagestedt, T. and Mandelkow, E. (1988) Nature 334, 359-362 27 Steiner, B. etaL (1990) EMBOJ. 9, 3539-3544 28 Baudier, J. and Cole, R, D. (1987) J. Biol. Chem. 262, 17577-17583 29 Lindwall, G. and Cole, R. D. (1984) J. Biol. Chem. 259, 5301-5305 30 Caceres, A. and Kosik, K. S. (1990) Nature 343,461-463 TINS, VoL 14, No. 5, 1991
31 Wischik, C. M. et al. (1988) Proc. Natl Acad. Sci. USA 85, 4506-4510 32 Wischik, C. M. eta/. (1988) Proc. Natl Acad. Sci. USA 85, 4884-4888 33 Kondo, J. et al. (1988) Neuron 1,827-834 34 Lee, V. M. Y., Balin, B. J., Otovos, L. and Trojanowski, J. Q. (1991) Science 251,675-678 35 Wood, J. G., Mirra, S. S., Pollock, N. J. and Binder, L. I. (1986) Proc Natl Acad. Sci. USA 83, 4040-4043 36 Wolozin, B. L., Pruchnicki, A., Dickson, D. W. and Davies, P. (1986) Science 232,648-650 37 Flament, S. and Delacourte, A. (1989) FEBS Left. 247, 213-216 38 Ksiezak-Reding, H., Binder, L. I. and Yen, S. H. (1990) J. Neurosci. Res. 25,420-430 39 Greenberg, S. G. and Davies, P. (1990) Proc. NatlAcad. Sci. LISA 87, 5827-5831 40 Joachim, C. L., Morris, J. H., Kosik, K. S. and Selkoe, D. J. (1987) Ann. Neurol. 22, 514-520 41 Garner, C. C. and Matus, A. (1988) J. CellBiol. 106, 779-783 42 Riederer, B. and Matus, A. (1985) Proc. NatlAcad. Sci. USA 82, 6006-6009 43 Tucker, R. P. and Matus, A. (1988) J. Comp. Neurol. 271, 44-55. 44 Geisert, E. E., Johnson, H. G. and Binder, L. I. (1990) Proc. Natl Acad. Sci. USA 87, 3967-3971 45 Joly, J. C. and Prurich, D. L. (1990) Biochemistry 29,
8916-8920: 46 Lewis, S. A., Ivanov, I. E., Lee, G. H. and Cowan, N. J. (1989) Nature 342,498-505 47 Lewis, S. A. and Cowan, N. (1990) Nature 345, 674 48 Rubino, H. M., Dammerman, M., Shafit-Zagardo, B. and Edichman, J. (1989) Neuron 3,631-638 49 Obar, R. A., Dingus, J., Bayley, H. and Vallee, R. B. (1989) Neuron 3,639-645 50 Lee, Y. C. and Wolff, J. (1984) J. Biol. Chem. 259, 1226-1230 51 Halpain, S. and Greengard, P. (1990) Neuron 5, 237-246 52 Chapin, S. J., Bulinski, J. C. and Gundersen, G. G. (1991) Nature 349, 24 53 Garner, C. C., Tucker, R. P. and Matus, A. (1988) Nature 336, 674-677 54 Tucker, R. P., Garner, C. C. and Matus, A. (1989) Neuron 2, 1245-1256 55 Burgin, K. E. et al. (1990)J. Neurosci. 10, 1788-1798 56 Yisraeli, J. and Melton, D. A. (1988) Nature 336, 592-595 57 Macdonald, P. M. and Struhl, G. (1988) Nature 336, 595-598 58 Okabe, S. and Hirokawa, N. (1989) Proc. NatlAcad. Sci. USA 86, 4127-4131 59 Noble, M., Lewis, S. A. and Cowan, N. J. (1989) J. Cell Biol. 109, 3367-3376 60 Chin, T. K., Harding, S. E. and Eagles, P. A. M. (t989) Biochem. J. 264, 53-60
Acknowledgements We are mostgrateful to Dr N. Hirokawa for providing Fig. 1, especiallypanel B, which has not been pubhshedbefore. We thank Dr L. A. Amos for help&~comments on the manuscript.
Song-learningbehavior: the interface with neuroethology Peter Marler
Behavioral studies of song learning in birds continue to raise new problems for neuroethological investigation. Evidence is emerging for a new form of vocal plasticity, called 'action-based learning'. Motor patterns are overproduced during a particular phase of ontogeny, and are then subjected to attrition and selective reinforcement by various kinds of social stimulation as the young bird matures. This form of learning, analogous to operant conditioning, can occur at phases of development when the more traditional form of 'memory-based learning' is no longer possible. There is evidence that different physiological mechanisms are involved in the development and the maintenance of mature singing, with a transition occurring at the time of song crystallization. Neural events associated with the developmental stabilization of motor patterns are worthy of more study. Understanding the neural basis of learning is all the more challenging when we consider the span of disciplines, from molecular genetics to behavior, that must be encompassed. Neuroethologists work strategically near the middle of this spectrum. They find their behavioral roots, not so much in the evolutionary theorizing of behavioral ecology, but in the classical ethology of Lorenz 1 and Tinbergen 2. Much progress has been made in understanding the neural basis of song learning3 but we still know little about the physiological basis of basic phenomena such as innate release mechanisms and their associated sign stimuli. In this review of some recent behavioral studies on song learning in birds (concentrating especially on comparative approaches to song ontogeny), I argue that ethological conceptions are still heuristically valuable and that they sometimes serve, not as prescriptions for designing animals as automata, but as a basis for learning4. TINS, Vol. 14, No. 5, 1991
Learningp r e f e r e n c e s The normal songs of oscine birds and some of their calls are learned in the sense that they develop abnormally in isolation. If we present young birds with different song types to learn, are they accepted indiscriminately, as though the bird brain is a tabula rasa, or are there preferences? To address this question, male swamp sparrows (Melospiza georgiana) and song sparrows (Melospiza melodia) were raised from the egg by canaries (Serinus canarius), placed in individual isolation, and then exposed as fledglings (20-60 days old) to taped songs that consisted of equal numbers of song sparrow and swamp sparrow songs. These songs differ in overall 'syntax', in the 'phonology' of individual notes, and in the size of individual male repertoires, which are three times larger in song sparrows than in swamp sparrows (Fig. 1). As in nature, the repertoires that the song sparrows developed were about three times larger than those of swamp sparrows. Each bird displayed a preference for learning songs of its own species 5'6, but with some interesting details. The preference was complete (100%) in swamp sparrows but incomplete (80%) in song sparrows, both with birds raised in the laboratory from the egg, and also with birds taken as nestlings that had ample opportunity to hear songs of their own species for up to ten days of age before the onset of tape training. Evidently song experience as a nestling does not result in acquisition, and has no effect on this learning preference. How should the weaker preference of song sparrows be interpreted? Neither species is known to copy the other in nature, even though they live within earshot. Probably the opportunity for social interaction with live tutors, which is known to exert an additional and in some cases a crucial influence on
© 1991, Elsevier Science Publishers Ltd, (UK)
0166 - 2236/91/$02.00
PeterMarler is at the Dept of Zoology, University of California, Davis, CA 95616, USA.
199
322-328 53 Wilson, D. (1960) J. Exp. BioL 37, 57-72 54 Young, J. Z. (1942) PhysioL Rev. 22, 318-373 55 Koopowicz, H. and Holman, M. (1988) Am. Zool. 28, 1065--1076 56 Perry, V. H., Brown, M. C., Lunn, E. R., Tree, P. and Gordon, S. (1990) Eur. J. Neurosci. 2, 802-808 57 Perry, V. H., Lunn, E. R., Brown, M. C., Cahusac, S. and Gordon, S. (1990) Eur. J. Neurosci. 2, 408-413 58 Strand, F. L., Rose, K. J., King, J. A., Segarra, A. C. and Zuccarelli, L. A. (1989) Prog. Neurobiol. 33, 45-85 59 Velez, S., Bittner, G. D., Govind, C. K. and Atwood, H. L. (1981) Exp. Neural. 71,307-325 60 Krasne, F. B. and Lee, S. H. (1977) Science 198, 517-519 61 Muller, K. J. (1988) Am. ZooL 28, 1091-1097 62 Mason, A. and Muller, K. J. (1982) Nature 296, 655-657 63 Fernandez, J. and Fernandez, M. S. (1974) Nature 251, 428-430 64 Nordlander, R. H, and Singer, M. (1976) Cell Tissue Res. 166, 445-460 65 Krause, L. and Bittner, G. D. (1990) Proc. NatlAcad. Sci. USA 87, 1471-1475 66 Eidetberg, E., Straehley, D. and Erspamer, R. (1967) Exp. Neural. 56, 312-322
Acknowledgements I wouldlike to thank my colleagues M. Ballinger,J. Blundon, T. Krause, J. Moehlenbruckand R. Shellerfor their commentsand contribub'ons to this manuscript. Much of this recent work was funded by the Texas Advanced TechnologyProgram and the National ScienceFoundation.
Molecularcharaderizationofmicrotubule-assodatedproteins tau andMAP2 M. Goedert,
R. A . C r o w t h e r
Tau and MAP2 are two of the major microtubuleassociated proteins in the vertebrate nervous system. They promote microtubule assembly and stability, and might be involved in the establishment and maintenance of neuronal polarity. In nerve cells immunohistochemistry shows complementary distributions, with tau being concentrated in axons and high molecular mass MAP2 being confined to dendrites. Each protein consists of multiple isoforms that contain three or four homologous tandem repeats near the carboxy-terminus, which constitute microtubule-binding domains. In humans, tau consists of at least six isoforms of related amino acid sequences that are produced from a single gene by alternative mRNA splicing and that are expressed in a stage- and cell type-specific manner. Tau is also a component of the paired helical filaments associated with Alzheimer's disease and other disorders of the CNS. Rat MAP2 consists of at least three isoforms produced from a single gene: high molecular mass MAP2a and MAP2b, and low molecular mass MAP2c. MAP2c is expressed only during early development and has so far been seen only in axons; MAP2a appears to replace MAP2c, whereas MAP2b is expressed throughout life. Messenger RNAs for MAP2 of high molecular mass are expressed both in cell bodies and in dendrites, consistent with the dendritic localization of the corresponding protein isoforms. Brain microtubules contain a variety of microtubuleassociated proteins that co-purify with tubulin during repeated cycles of assembly 1. Knowledge of the properties of these proteins is essential for an understanding of the function of microtubules in specialized cells. Microtubule-associated proteins are characterized functionally by their ability to promote the assembly and stability of microtubules in axons TINS, Vol. 14, No. 5, 1991
a n d C. C. G a r n e r
and dendrites, suggesting a major role in the determination of neuronal morphology and plasticity. Microtubules are known to undergo extensive growth and rearrangement in nerve cells, especially during periods of neurite outgrowth. Tau and MAP2 are two of the major and moststudied microtubule-associated proteins of the vertebrate nervous system 2. Immunohistochemistry shows complementary distributions with tau being found mostly in axons and high molecular mass MAP2 being confined to dendrites. Electron micrographs show these proteins as lateral projections that appear to cross-bridge adjacent microtubules (Fig. 1)3'4. The spacing between adjacent microtubules in the presence of tan is about 20 nm (Fig. 1A), while that for high molecular mass MAP2 is about 100 nm (Fig. 1B), which is similar to the cross-bridging structures found in a dendrite (Fig. 1C). This suggests that tau and MAP2 each contain a microtubule-binding domain and a projection domain. Since microtubules are inherently unstable structures, the stabilizing effects of tan and MAP2 are likely to be important for the integrity of processes that depend on stable microtubules, such as rapid axonal transport. The finding that tau is a constituent of paired helical filaments, the major component of the neurofibrillary tangle of Alzheimer's disease 5, indicates that some microtubule-associated proteins are involved in pathophysiological processes. Although tau and MAP2 were the first microtubuleassociated proteins to be discovered some 15 years ago6'7, only recently have their primary amino acid sequences been determined s-19, thus opening the way to detailed molecular study. Tan and MAP2 are each encoded by a single gene, from which multiple isoforms are produced in a developmentally regulated manner by alternative mRNA splicing. The primary
© 1991, ElsevierScience PublishersLtd, (UK) O166- 2236/91/$02.00
M. 6oedert and R. A. Crowther are at the MRC Laboratoryof MolecularBiology, Hills Road, Cambridge CB22QH, UK, and C C Garner is atthe Centre for Molecular Neurobiology, University of Hamburg, Martinistrasse.52, 2000Hamburg20, FRO.
193
molar concentration of peptide required is two orders of magnitude higher than that of full length tau protein22'23'~5. This difference might be explained by less than a full repeat being used, by the reiteration of the repeat sequence in the complete protein and by conformational changes imposed by sequences lying outside the repeat region. Recombinant tau isoforms with four repeats induce faster microtubule assembly than those with three repeats ~1, as might be expected if one tan molecule links a number of tubulin subunits. Tan is thought to be an elongated molecule without much secondary structure. However, images of tau paracrystals have shown it to be unusually elastic, ' i . . . . . !,~,~, "~ .:~., ',., suggesting that interactions between different parts of the molecule could occur26. Besides being distinFig. 1. Tau and MAP2 form projections that appear to cross-bridge adjacent guished by the presence of three or four tandem microtubules. Quick-frozen, deep-etched pellets of microtubules polymerized repeats, some tau isoforms isolated from human brain with (A) tau or (B) MAP,?. (C) Similarly prepared dendrite of a motor neurone tissue contain inserts of 29 or 58 amino acids located from bovine spinal cord, in which neurofilaments are seen as thin filaments near the amino-terminus (Fig. 2) 16. These sequences running parallel to microtubules. Note the frequent cross-bridges between are rich in proline and charged amino acid residues and microtubules in (,4) and (B) and the cross-bridges between microtubules and share some similarity with the carboxy-terminus of neurofilaments in (C). Scale bar is 100 nm. (Courtesy of N. Hirokawa.) the middle neurofilament subunit. Their presence might well modulate the presently unknown function amino acid sequences are highly similar at the of the projection domain, the sequence of which carboxy-terminus, which contains three or four tan- shows a striking charge separation, with acidic dem repeats and represents the microtubule-binding residues grouped towards the amino-terminus and domain. The amino-terminal parts of tau and MAP2, basic residues towards the carboxy-terminus. representing the projection domains, appear unWhen individual human tan isoforms that are related but both contain large numbers of prolines and expressed in E. coil ~1 are mixed together, they give a charged amino acids, which suggests an extended characteristic set of six bands (Fig. 2), ranging from structure. MAP-U, a ubiquitous microtubule- 48 to 67 kDa in apparent molecular masses. True associated protein, shows a similar structure with molecular masses of the different isoforms range from tandem repeats at the carboxy-terminus that are 37 to 46 kDa21; therefore, tau is anomalously retarded similar to those of tau and MAP22°; this indicates the on denaturing gels, as has been clear since the existence of a family of microtubule-associated pro- sequencing of the first tau isoforms s'9. This behaviour teins with similar microtubule-binding domains. has been attributed to the extended configuration of taus'26. The purification of tan from brain tissue makes A family of tau proteins use of its association with microtubules in repeated Ever since tau was first purified from adult brain cycles of assembly, its heat stability and its solubility tissue it has been known to exist in multiple forms, in perchloric acid; tan protein extracted from adult which were later shown to be subject to develop- mammalian brain tissue in this way runs as 4-6 bands mental regulation 1. Molecular cloning has now defined on SDS-PAGE. Comparison of the pattern of recomat least six tau isoforms that are generated from a binant tau isoforms with that of native tan shows that single gene by alternative mRNA splicing. To date, the four major tau isoforms in adult human brain cDNA cloning has yielded tau sequences from mouse, correspond to isoforms with three and four repeats rat, cow and humanss-16. The six human tan isoforms without amino-terminal insertions, and to isoforms that have so far been isolated as full length cDNA with three and four repeats containing the aminoclones have sequences that range from 352 to 441 terminal insertion of 29 amino acids (Fig. 2, bands A, amino acids in length and differ from each other by the B, D, E)~1. presence or absence of three inserts (Fig. 2) 9'1°'16,21. Phosphorylation also contributes to the complex Variable phosphorylation of the different isoforms pattern of tau bands, since some bands exhibit an adds to the complexity. increased mobility on denaturing gels after alkaline The most striking feature of the tau sequences is phosphatase treatment. Recombinant tau proteins can the presence of three or four tandem repeats of 31 or be phosphorylated in vitro by cPuMP-dependent 32 amino acids located at the carboxy-terminus, each protein kinase, multifunctional Ca2+/calmodulincontaining a characteristic Pro-Gly-Gly-Gly motif. dependent protein kinase II (CaM kinase II), protein The extra repeat in the isoforms with four repeats is kinase C and casein kinase I127. Interestingly, only inserted within the first repeat of the isoforms with phosphorylation by CaM kinase II results in a reduced three repeats in a way that preserves the periodic mobility on denaturing gels; the same holds true for pattern. Twelve residues are completely conserved native tan extracted from brain tissue 28. Although between the four repeats and a further four residues there are many potential phosphorylation sites for the show conservative changes (Fig. 3). Experiments above protein kinases throughout the tau sequence, it with synthetic peptides and with tan fragments is only the carboxy-terminal half that becomes phosproduced by expression in E. coli show that the phorylated. Whereas cAMP-dependent protein repeats constitute microtubule-binding domains22-25. kinase, protein kinase C and casein kinase II phosMicrotubule polymerization is promoted by synthetic phorylate tau at multiple sites, CaM kinase II phos18-mer peptides from the repeat region, although the phorylates a single serine residue that is located 26 194
TINS, VoL 14, No. 5, 1991
amino acids from the carboxy-terminus of human tau 27. Phosphorylation of tau proteins is likely to be of functional importance, since it leads to a reduced affinity for microtubules 29. It is interesting to speculate that differential phosphorylation of tau in axons and dendrites might be responsible for its axonal location. The human cDNA clones isolated to date account for most, if not all, tau isoforms present in adult human brain. The screening of mouse and cow cDNA and genomic libraries has suggested the existence of other forms of taus'u'12. These include bovine isoforms with an alternative amino-terminus or an alternative extended carboxy-terminus, and a mouse isoform with an extension of 25 amino acids at the carboxy-terminus. The status of these reported clones is unclear, since the corresponding proteins (if they exist) might not necessarily be isolated by
Tandem
repeat
r e g i o n of t a u DL
K N V K S
K I G S T E N L
K H Q P G G G K
[V Q i I N K K L D L S N V Q S K C G S K D N I K H V P G G G S] VQ
i V Y K P V D L
S KVTS
K CG
S LGN
I HHK
P G G G Q
V E V K S E K L D F KDR V Q S K I G S L D N I T H V P G G G N K K I E T H K L T F REN A K A K #
Tandem
#*
repeat
#
region
*
i VTK
K I DL
**
*#
*
* * * *
of M A P 2
D L KNVK VQ
**
S K I GS
S HVT
S K CG
TDN
I K YQP
S L KN
KGGQ
I R H R P G G G R
V K I E S V K L D F KEK A Q A K V G S L D N A H H V P G G G N V K I D S Q K L N F REH A K A K V ,
*
#
*
#*
#
#
*
**
*
#
*
**
Consensus
352
VK
A 1
381
B 1
410
C 383
1
D 412
1
E 441
mll
F 1
234
94--
56
~,ii¸)
67-43--
30-
Fig. 2. (Upper panel) Schematic representation of six human tau isoforms. The region common to all isoforms is tinted and the three inserts that distinguish them are shown in white. The three or four tandem repeats are marked by black bars. (Lower panel) A comparison of recombinant tau isoforms with human brain tau. Lanes 3 and 6, mixture of recombinant tau isoforms with each band identified by letter; lanes I and 2, tau extracted from fetal brain before and after alkaline phosphatase treatment; and lanes 4 and 5, tau extracted from adult cerebral cortex before and after alkaline phosphatase treatment. After electrophoresis on a 10% SDS-PAGE gel the tau isoforms were visualized by western blotting. Note the alignment of major brain tau bands with the recombinant isoforms (labelled A-F). (Modified, with permission, from Ref. 21 .)
TINS, Vol. 14, No. 5, 1991
I E S K KLDL
K N V Q S
K I GS
5DN
I K H V P G G G N
Fig. 3. Four tandem repeats in the human tau sequence and three homologous tandem repeats in the rat MAP2 sequence, with a derived consensus sequence. These tandem repeats are located near the carboxy-terminus and constitute the microtubule-binding domains of tau and MAP2. Tau can exist with three or four tandem repeats; the sequence giving rise to the extra tau repeat is bracketed. Similarity is indicated as follows: *, identical residues in all repeats; #, identical residues in all repeats except one, which shows a conservative replacement. Note that the MAP2 repeats show more conservation towards the amino-terminal end of each repeat, while the tau repeats show more conservation towards the carboxy-terminal end of each repeat. The sequences have been aligned optimally; thus, in two of the tandem repeats of both tau and MAP2, there appear extra inserted residues. The sequence of bovine tau is identical to that of human tau in the repeat region, whereas in mouse and rat tau an arginine replaces a lysine at position 6 of the first repeat. The sequences of mouse and rat MAP2 are identical in the repeat region.
standard tau extraction procedures. Moreover, the nucleotide sequence with a high degree of similarity to that claimed to be coding the longer carboxy-terminus in the extended mouse protein forms part of an intron in the cow gene. There is no intron in the human gene at the point corresponding to where the two mouse sequences diverge (Goedert, M., Crowther, R. A. and Garner, C. C., unpublished observation). In the nervous system, mRNA transcripts encoding the various tau isoforms are expressed in a manner that is stage- and cell type-specific 1°'16. Transcripts encoding isoforms with three and four repeats are found throughout the adult human brain and their levels do not vary much between different regions. By contrast, in fetal brain, although transcripts containing three repeats are abundant, no transcripts containing four repeats are detectable. The developmental shift of tau bands from a simple fetal pattern to a more complex adult pattern thus involves the transition from the expression of the isoform with three repeats and no insertions to the expression of all six isoforms. In situ hybridization has shown that tau protein mRNAs occur only in neurones, and that different neuronal cell types express different isoforms 1°'14. Transcripts encoding isoforms with three or four repeats are found in pyramidal cell bodies throughout all layers of the adult human cerebral cortex. In the hippocampal formation, transcripts for isoforms with three repeats are present in granule cells of the dentate gyrns, pyramidal cells throughout the cornu 195
It is not known why this otherwise soluble axonal protein ends up in an insoluble form in affected nerve cells. A simple overproduction of tau does not appear to be responsible, since transcript levels are not significantly changed 16. Tan is probably posttranslationally modified in Alzheimer's disease; indirect evidence suggests that it is abnormally phosphorylated35-39, since certain tan antibodies stain Q neurofibrillary tangles only after treatment of tissue sections with alkaline phosphatase 35. Moreover, extraction of tan protein with SDS from affected brain regions and from paired helical filaments gives tan bands that are not observed in tau extracted from unaffected brain regions or from brain tissue of control patients 36-39. These molecular species show an increased gel mobility after alkaline phosphatase treatment. At present, it is not known whether abnormal / phosphorylation represents a causal event in the formation of paired helical filaments, or whether it is a mere consequence of a proportion of tau being Fig. 4. Cellular localization of tau protein mRNAs in adult immobilized in cell bodies and dendrites rather than human hippocampal formation. A probe spedfic for moving into the axon. transcripts encoding isoforms with three repeats labels Pathology involving tan protein is not limited to numerous nerve cell bodies. Abbreviations: CA, cornu Alzheimer's disease, but is also observed in a number ammonis; DG, dentate gyms; 5U, subiculum. Scale bar is of disorders of the CNS, such as Pick's disease and 350t~m. (Taken, with permission, from Ref. 10.) progressive supranuclear palsy4°. This suggests that tan might be affected in a relatively nonspecific manner in response to various kinds of nerve cell ammonis (CA) layers and pyramidal cells in the damage. Alternatively, different modifications of tau, subiculum (Fig. 4). Transcripts encoding isoforms each specific to a particular disease, might lead to the with four repeats are also found in pyramidal cells in formation of similar neurofibrillary tangles. the subiculum and hippocampus, but not in granule cells of the dentate gyms. Isoforms of MAP2 The availability of sequenced cDNA clones for tan The term MAP2 generically refers to three proprotein has enabled preliminary functional studies to teins of related amino acid sequence that are produced be undertaken. Experiments using antisense oligo- from a single gene. MAP2a and MAP2b have apparent nucleotides to block tau expression in cultured nerve molecular masses of 288 and 280 kDa on SDS-PAGE cells suggest that axonal morphology might be depen- gels, whereas MAP2c runs at 70 kDa 1'2. The exdent on tau expression 3°, though it is unclear whether pression in E. coli of a rat cDNA encoding the high tan plays a crucial role in the establishment of molecular mass MAP2 of 1830 amino acids results in a neuronal polarity. Recently, tau protein has been protein that aligns with rat MAP2b on gels (Fig. 5) 19. expressed from cDNA clones in non-neuronal cells is. It is not known whether MAP2a is produced by As expected, transfection of fibroblasts results in the alternative splicing or whether it is the product of binding of the expressed tau protein to microtubules. post-translational modification of MAP2b. Similar In addition, microtubules are aggregated into thick experiments with cDNA encoding MAP2c have longitudinal bundles, suggesting an additional bundling demonstrated that the expressed protein aligns on effect of tan when expressed at high levels. gels with dephosphorylated MAP2c that has been extracted from rat brain at postnatal day 5 (P5) (Fig. Tau proteins and Alzheimer's disease 5) 19. As is the case for the tau isoforms, MAP2b and Alzheimer's disease is characterized neuropatho- MAP2c are anomalously retarded on SDS-PAGE logically by the presence of abundant amyloid plaques gels. MAP2c, which is made up of 467 amino acids, and neurofibrillary tangles in cerebral cortex and arises via alternative mRNA splicing and lacks 1363 hippocampus, as well as in some subcortical nuclei. amino acids from the middle of MAP2b41. MAP2b is Neurofibrillary tangles consist of insoluble filamentous expressed in rat brain throughout life, while MAP2c is material that is deposited mostly in cell bodies and expressed in most brain regions only before P1042. dendrites of affected nerve cells. Paired helical This is in contrast with MAP2a, which appears after filaments constitute the major component of the P10 and appears to replace MAP2c in mature neurofibrillary tangle and both immunohistochemical neurones. The forms of MAP2 that have high and biochemical studies have demonstrated that these molecular masses are expressed almost exclusively in filaments contain various tau protein isoforms 5'16'31-33. the dendrites of nerve cells 1'2, while MAP2c is found At present, it is unclear whether tan is the only in axons during development along with tau 43, constituent of paired helical filaments or whether although it is not known whether it is also present in more components remain to be discovered. A recent dendrites. All three isoforms appear to be expressed study has provided the strongest evidence yet that it in macroglia 43'44. The complete amino acid sequences is in fact modified tan that forms these filaments ~. of mouse and rat high molecular mass MAP2 and Final proof must await the assembly of paired helical rat MAP2c have been determined by molecular cloning ]7-19. filaments from defined components. 196
TINS, Vol. 14, No. 5, 1991
Limited proteolysis of high molecular mass MAP2 bound to microtubules demonstrated that it is composed of a small, microtubule-binding domain and a much larger projection domain that is thought to cross-link various cytoskeletal components. The 210 amino acids at the carboxy-terminus of MAP2 show a high degree of similarity with the carboxy-terminus of tau, and contain in particular three tandem repeats that bind to microtubules (Fig. 3) 17-19. As with tau, an octadecapeptide from the second MAP2 repeat promotes microtubule assembly in vitro45. Subsequent work has demonstrated that the repeat region of MAP2 is also able to bind to microtubules when expressed in cultured cell lines, and has suggested that sequences flanking the repeats are important for the binding of MAP2 to microtubules 46'47. Although it is unclear what role these neighbouring sequences might play, the presence of an upstream proline-rich stretch suggests that this coil region might function as a hinge between the projection domain and the microtubule-binding domain19. What role does the projection domain of MAP2 play in developing and mature dendrites? Possible functions include the cross-linking or spacing of various components of the neuronal cytoskeleton and the provision of an anchoring surface for enzymes. The latter has been demonstrated for the regulatory subunit of cAMP-dependent protein kinase II, which binds within the 100 amino acids at the amino terminus of all three MAP2 isoforms 48'49. A second protein, calmodulin, is thought to bind near the tubulin-binding domain at the carboxy-terminus 19'5°. The effects of both proteins influence the affinity of MAP2 for microtubules. Phosphorylation of MAP2 is likely to play an important role in the modulation of its function, as indicated by the finding that the activation of glutamate receptors of the N-methyl-o-aspartate subtype in hippocampal slices leads to a rapid and specific dephosphorylation of MAP251. The precise role of MAP2 in cross-linking the cytoskeleton is unclear. Ultrastructurally, MAP2 extends from the microtubule surface by some 100 nm and appears to form cross-links with the surrounding cytoskeleton (Fig. 1C)4. Since neuronal microtubules are arranged in longitudinal bundles, it was proposed that one function of MAP2 in dendrites might be to bundle microtubules. Experiments in which parts of MAP2 are expressed in cultured non-neuronal cells have identified a short region on the carboxy-terminal side of the tubulin-binding repeats that participates in bundling microtubules with a separation of about 25nm 46'47. This distance is shorter than that observed in vivo in nerve cells (Fig. 1C), and if the bundling activity does indeed reside in this sequence, then it leaves open the question as to what role the long arm of MAP2 might play in the dendritic cytoskeleton. Although microtubule bundling is clearly observed following overexpression of tau or MAP2, it might be a consequence of microtubule stabilization and be mediated by an endogenous bundling protein rather than by the overexpressed tau or MAP252. An added twist to the story comes from in situ hybridization, which shows that transcripts encoding both MAP2b and MAP2c are expressed in cell bodies of nerve cells, such as cerebellar granule cells, pyramidal ceils in cerebral cortex and hippocampus and dentate gyms granule cells (Fig. 6) 53,54. Strikingly, TINS, Vol. 14, No. 5, 1991
1
151
1514
o
1830
....---
MAP2c 1 151
467
1
..., 11111 MAP2b t MAP2a
MAP2c
2
~
3
4
5
~
.._
--
IR
Fig. 5. (Upper panel) Schematic representation of MAP2b and MAP2c. The region common to MAP2b and MAP2c is tinted and the three tandem repeats are marked by black bars. MAP2c is generated by an alternative splicing event that deletes 1363 amino acids from the middle of MAP2b. (Lower panel) A comparison of recombinant MAP2 isoforms with rat brain MAP2. Lanes 2 and 5, recombinant high molecular mass MAP2b and MAP2c, respectively; lane 1, MAP2 extracted from adult rat brain; and lanes 3 and 4, MAP2 extracted from rat brain at P5 before and after alkaline phosphatase treatment. After electrophoresis on a 3-15% SDS-PAGE gel, the MAP,? isoforms were visualized by western blotting. Note the alignment of brain MAP2b and MAP2c with the recombinant MAP2 bands. (Taken, with permission, from Ref. 19.)
the 9 kb mRNA encoding the MAP2 isoforms of high molecular mass is also seen in the dendritic compartment (Fig. 6), whereas the 6 kb mRNA encoding MAP2c is not is. The compartmentalization of specific mRNAs to dendrites allows local regulation of where particular proteins are synthesized, thereby possibly altering the stability of the dendritic cytoskeleton. The occurrence of high molecular mass MAP2 mRNA in the dendritic compartment represented the first example of the spatial segregation of an mRNA in nerve cells 53. Recently, mRNA encoding the o~ subunit of the multifunctional protein CaM kinase II has also been localized in cell bodies and dendrites of nerve cells 54'55. Spatial segregation of an mRNA could arise co-translationally, with protein-targeting signal., conferring specificity. Alternatively, specificity could lie at the mRNA level, with cis-acting signals in the RNA interacting with a receptor at the localization site. Evidence from localized maternal mRNAs in Xenopus laevis and Drosophila melanogaster eggs is in favour of the RNA targeting model 56'57. The dendritic localization both of MAP2 mRNAs and MAP2 proteins 197
Concluding remarks Tau and MAP2 provide prototypes for other microtubule-associated proteins found in neuronal and non-neuronal cells. They define a family of structurally related proteins that presently also includes MAP-U; there are probably more members of this family to be discovered. A second class is represented by MAP1B, a microtubule-associated protein of high molecular mass with no sequence similarity to tau or MAP2 59. The molecular characterization of tau and MAP2 has given us a glimpse of the complex and diverse organization and modification of the cytoskeleton. Both proteins have similar microtubule-binding domains, but possess unique projections at the aminoterminus that might modulate the local environment. Such modulation might involve active bridging between neighbouring filaments, or passive repulsion due to electrostatic effects, as has been proposed for the side arms of neurofilaments6°. Diversity of tau and MAP2 is generated by alternative splicing, and the capacity to regulate when and where they bind microtubules might depend on post-translational modifications and, for MAP2, on targeting of mRNAs.
Selected references
Fig. 6. A comparison of the cellular localization o f (A) MAP2 mRNAs, (B) nuclear staining and (C) MAP2 proteins in adult rat hippocampus. Note that MAP2 mRNAs are found in the granule cell layer and in the proximal portion of the dentate gyrus molecular layer, indicating a somatodendritic localization. By contrast, MAP2 proteins are found throughout the molecular layer, indicating a dendritic localization. Abbreviations: g, granule cell layer; m, molecular layer. Scale bar is 250 t~m. (Taken, with permission, from Ref. 54.)
of high molecular mass suggests that MAP2 might owe its restricted distribution to the sequestration of its mRNAs within the dendrite. However, there is also evidence indicating that the information for MAP2 segregation might reside within the protein itself 98. Thus, biotinylated MAP2, when injected into cultured nerve cells, is initially uniformly distributed in axons and dendrites, but then progressively disappears from the axon. There must therefore be some mechanism to clear MAP2 protein from inappropriate regions of the cell. 198
1 Matus, A. (1988)Annu. Rev. Neurosci. 11, 29-44 2 Nunez, J. (1988) Trends Neurosci. 11,477-479 3 Hirokawa, N., Shiomura, Y. and Okabe, S. (1988) J. CellBioL 107, 1449-1459 4 Hirokawa, N., Hisanaga, S. I. and Shiomura, Y. (1988) J. Neurosci. 8, 2769-2779 5 Brion, J. P. (1990) Semm. Neurosci. 2, 89-100 6 Weingarten, M.D., Lockwood, A. H., Hwo, S. Y. and Kirschner, M. W. (1975) Proc. Natl Acad. Sci. USA 72, 1858-1862 7 Sloboda, R. D., Rudolph, S. A., Rosenbaum, J. L. and Greengard, P. (1975) Proc. NatlAcad. Sci. USA 72, 177-181 8 Lee, G., Cowan, N. and Kirschner, M. (1988) Science 239, 285-288 9 Goedert, M., Wischik, C. M., Crowther, R. A., Walker, J. E. and Klug, A. (1988) Proc. NatlAcad. Sci. USA 85, 4051-4055 10 Goedert, M., Spillantini, M. G., Potier, M. C., Ulrich, J. and Crowther, R. A. (1989) EMBO J. 8, 393-399 11 Himmler, A. (1989) Mol. Cell. Biol. 9, 1389-1396 12 Himmler, A., Drechsel, D., Kirschner, M. W. and Martin, D. W. (1989) Mol. Cell. Biol. 9, 1381-1388 13 Mori, H. eta/. (1989) Biochem. Biophys. Res. Commun. 159, 1221-1226 14 Kosik, K. S., Orecchio, L. D., Bakalis, S. and Neve, R. L. (1989) Neuron 2, 1389-1397 15 Kanai, Y. et aL (1989) J. Cell Biol. 109, 1173-1184 16 Goedert, M., Spillantini, M. G., Jakes, R., Rutherford, D. and Crowther, R. A. (1989) Neuron 3, 519-526 17 Lewis, S. A., Wang, D. and Cowan, N. J. (1988) Science242, 936-939 18 Papandrikopoulou, A., Doll, T., Tucker, R. P., Garner, C. C. and Matus, A. (1989) Nature 340, 650-652 19 Kindler, S., Schulz, B., Goeclert, M. and Garner, C. C. (1990) J. Biol. Chem, 265, 19679-19684 20 Aizawa, H. etal. (1990)J. BioL Chem. 265, 13849-13855. 21 Goedert, M. and Jakes, R. (1990) EMBOJ. 9, 4225-4230 22 Ennulat, D. J., Liem, R. K. H., Hashim, G. A. and Shelanski, M. L. (1989) J. Biol. Chem. 264, 5327-5330 23 Aizawa, H. etal. (1989)J. Biol. Chem. 264, 5885-5890 24 Lee, G., Neve, R. L. and Kosik, K. S. (1989) Neuron 2, 1615-1624 25 Maccioni, R. 8., Vera, J. C., Dominguez, J. and Avila, J. (1989) Arch. Biochem. Biophys. 275, 568-579 26 Lichtenberg, 8, Mandelkow, E. M., Hagestedt, T. and Mandelkow, E. (1988) Nature 334, 359-362 27 Steiner, B. etaL (1990) EMBOJ. 9, 3539-3544 28 Baudier, J. and Cole, R, D. (1987) J. Biol. Chem. 262, 17577-17583 29 Lindwall, G. and Cole, R. D. (1984) J. Biol. Chem. 259, 5301-5305 30 Caceres, A. and Kosik, K. S. (1990) Nature 343,461-463 TINS, VoL 14, No. 5, 1991
31 Wischik, C. M. et al. (1988) Proc. Natl Acad. Sci. USA 85, 4506-4510 32 Wischik, C. M. eta/. (1988) Proc. Natl Acad. Sci. USA 85, 4884-4888 33 Kondo, J. et al. (1988) Neuron 1,827-834 34 Lee, V. M. Y., Balin, B. J., Otovos, L. and Trojanowski, J. Q. (1991) Science 251,675-678 35 Wood, J. G., Mirra, S. S., Pollock, N. J. and Binder, L. I. (1986) Proc Natl Acad. Sci. USA 83, 4040-4043 36 Wolozin, B. L., Pruchnicki, A., Dickson, D. W. and Davies, P. (1986) Science 232,648-650 37 Flament, S. and Delacourte, A. (1989) FEBS Left. 247, 213-216 38 Ksiezak-Reding, H., Binder, L. I. and Yen, S. H. (1990) J. Neurosci. Res. 25,420-430 39 Greenberg, S. G. and Davies, P. (1990) Proc. NatlAcad. Sci. LISA 87, 5827-5831 40 Joachim, C. L., Morris, J. H., Kosik, K. S. and Selkoe, D. J. (1987) Ann. Neurol. 22, 514-520 41 Garner, C. C. and Matus, A. (1988) J. CellBiol. 106, 779-783 42 Riederer, B. and Matus, A. (1985) Proc. NatlAcad. Sci. USA 82, 6006-6009 43 Tucker, R. P. and Matus, A. (1988) J. Comp. Neurol. 271, 44-55. 44 Geisert, E. E., Johnson, H. G. and Binder, L. I. (1990) Proc. Natl Acad. Sci. USA 87, 3967-3971 45 Joly, J. C. and Prurich, D. L. (1990) Biochemistry 29,
8916-8920: 46 Lewis, S. A., Ivanov, I. E., Lee, G. H. and Cowan, N. J. (1989) Nature 342,498-505 47 Lewis, S. A. and Cowan, N. (1990) Nature 345, 674 48 Rubino, H. M., Dammerman, M., Shafit-Zagardo, B. and Edichman, J. (1989) Neuron 3,631-638 49 Obar, R. A., Dingus, J., Bayley, H. and Vallee, R. B. (1989) Neuron 3,639-645 50 Lee, Y. C. and Wolff, J. (1984) J. Biol. Chem. 259, 1226-1230 51 Halpain, S. and Greengard, P. (1990) Neuron 5, 237-246 52 Chapin, S. J., Bulinski, J. C. and Gundersen, G. G. (1991) Nature 349, 24 53 Garner, C. C., Tucker, R. P. and Matus, A. (1988) Nature 336, 674-677 54 Tucker, R. P., Garner, C. C. and Matus, A. (1989) Neuron 2, 1245-1256 55 Burgin, K. E. et al. (1990)J. Neurosci. 10, 1788-1798 56 Yisraeli, J. and Melton, D. A. (1988) Nature 336, 592-595 57 Macdonald, P. M. and Struhl, G. (1988) Nature 336, 595-598 58 Okabe, S. and Hirokawa, N. (1989) Proc. NatlAcad. Sci. USA 86, 4127-4131 59 Noble, M., Lewis, S. A. and Cowan, N. J. (1989) J. Cell Biol. 109, 3367-3376 60 Chin, T. K., Harding, S. E. and Eagles, P. A. M. (t989) Biochem. J. 264, 53-60
Acknowledgements We are mostgrateful to Dr N. Hirokawa for providing Fig. 1, especiallypanel B, which has not been pubhshedbefore. We thank Dr L. A. Amos for help&~comments on the manuscript.
Song-learningbehavior: the interface with neuroethology Peter Marler
Behavioral studies of song learning in birds continue to raise new problems for neuroethological investigation. Evidence is emerging for a new form of vocal plasticity, called 'action-based learning'. Motor patterns are overproduced during a particular phase of ontogeny, and are then subjected to attrition and selective reinforcement by various kinds of social stimulation as the young bird matures. This form of learning, analogous to operant conditioning, can occur at phases of development when the more traditional form of 'memory-based learning' is no longer possible. There is evidence that different physiological mechanisms are involved in the development and the maintenance of mature singing, with a transition occurring at the time of song crystallization. Neural events associated with the developmental stabilization of motor patterns are worthy of more study. Understanding the neural basis of learning is all the more challenging when we consider the span of disciplines, from molecular genetics to behavior, that must be encompassed. Neuroethologists work strategically near the middle of this spectrum. They find their behavioral roots, not so much in the evolutionary theorizing of behavioral ecology, but in the classical ethology of Lorenz 1 and Tinbergen 2. Much progress has been made in understanding the neural basis of song learning3 but we still know little about the physiological basis of basic phenomena such as innate release mechanisms and their associated sign stimuli. In this review of some recent behavioral studies on song learning in birds (concentrating especially on comparative approaches to song ontogeny), I argue that ethological conceptions are still heuristically valuable and that they sometimes serve, not as prescriptions for designing animals as automata, but as a basis for learning4. TINS, Vol. 14, No. 5, 1991
Learningp r e f e r e n c e s The normal songs of oscine birds and some of their calls are learned in the sense that they develop abnormally in isolation. If we present young birds with different song types to learn, are they accepted indiscriminately, as though the bird brain is a tabula rasa, or are there preferences? To address this question, male swamp sparrows (Melospiza georgiana) and song sparrows (Melospiza melodia) were raised from the egg by canaries (Serinus canarius), placed in individual isolation, and then exposed as fledglings (20-60 days old) to taped songs that consisted of equal numbers of song sparrow and swamp sparrow songs. These songs differ in overall 'syntax', in the 'phonology' of individual notes, and in the size of individual male repertoires, which are three times larger in song sparrows than in swamp sparrows (Fig. 1). As in nature, the repertoires that the song sparrows developed were about three times larger than those of swamp sparrows. Each bird displayed a preference for learning songs of its own species 5'6, but with some interesting details. The preference was complete (100%) in swamp sparrows but incomplete (80%) in song sparrows, both with birds raised in the laboratory from the egg, and also with birds taken as nestlings that had ample opportunity to hear songs of their own species for up to ten days of age before the onset of tape training. Evidently song experience as a nestling does not result in acquisition, and has no effect on this learning preference. How should the weaker preference of song sparrows be interpreted? Neither species is known to copy the other in nature, even though they live within earshot. Probably the opportunity for social interaction with live tutors, which is known to exert an additional and in some cases a crucial influence on
© 1991, Elsevier Science Publishers Ltd, (UK)
0166 - 2236/91/$02.00
PeterMarler is at the Dept of Zoology, University of California, Davis, CA 95616, USA.
199