The Microtubule-associated Protein Tau Forms a Triple-stranded Left ...

Vol. 266, No. 32, Issue of November 15, PP. 22019-22027, 1991 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY [CI

1991 by The American Society for Biochemistry and Molecular Biology, Inc.

The Microtubule-associated Protein Tau Forms a Triple-stranded Left-hand Helical Polymer* (Received for publication, June 18, 1991)

George C. Ruben*#, Khalid IqbalQV,Inge Grundke-Iqball, Henryk M. WisniewskiBl, Thomas L. CiardelliII, and John E. Johnson, Jr.** From the $Department of Biological Sciences, Dartmouth College, Hanouer, New Hampshire 03755, the 7New York State Institute for Basic Research in Developmental Disabilities, Staten Island, New York 10314, the [[Departmentof Pharmacology, Dartmouth Medical School, Hanouer, New Hampshire, 03756, and the **Department of Integrative Biology, University of California, Berkeley, California 94720 and Department of Neuroscience, SRZ International, Menlo Park, California 94025

High resolution transmission electron microscopy (TEM) has shown that bovine tau are 2.1 f 0.2-nm diameter filaments which are triple-stranded left-hand helical structures composedof three 1.0 f 0.2-nm strands. The reported amino acid sequence of human and bovine tau have been computer processed to predict secondary structure. Within the constraints imposed by the images, the secondary structure models and other structural information have been used to calculate tau’s maximumandminimum length. The length calculations and secondary structure form the basis for image interpretation. This work indicates that each -1.0-nm strand is a tau polypeptide chain and that the -2.1-nm filament is composed of three separate tau chains (tau,). Bovine tau length measurements indicate that tau trimer filaments are generally longer than a fully extendedtau monomer. These measurements indicate that each trimer, tau3, is joined with other trimers to form long tau polymers, (tau3),. An inverse temperature transition has been found in the circular dichroism spectrum of tau indicating that its structure is less ordered below 20 “C and moreordered at 37 “C. The implications of this phenomenon with respect to tau’s temperature-dependent ability to reconstitute microtubules is discussed and a mechanism for the possible abnormal aggregation of tau into neurofibrillary tangles in Alzheimers disease is proposed.

have been reported. All of these reports indicate that tau is a family of proteins derived from a single gene and that the heterogeneity in the amino acid chain length is due to alternative RNA splicing (Himmler, 1989). Bovine tau has a sequence of448 amino acids (46,332 daltons) with variable deletions that can reduce its length by as much as 146 amino acids (Himmler et al., 1989; Himmler, 1989). Human tau has a sequence of 441 amino acids (45,850 daltons) with deletions of 29, 31, 58, and as many as 89 amino acids (Goedert et al., 1989). Sodium dodecyl sulfate (SDS)-polyacrylamidegel electrophoresis (PAGE) of tau produces four bands ranging from 55,000 to 62,000 daltons (Cleveland et al., 1977a, 1977b). When tau is partially phosphorylated there can be six to eight bands by SDS-PAGE (Lindwall and Cole, 1984b), with the phosphorylated bands shifted to what appear to be higher molecular weights. Fully denatured tau has a higher apparent molecular weight than a fully denatured equivalent standard globular protein marker by SDS-PAGE. Tau is much less hydrophobic than globular proteins (see “Results andDiscussion”), binds less negatively charged SDS, andruns more slowly in an electric field applied to a polyacrylamide gel. The structure of tau was first studied by ultracentrifugation (Cleveland et al., 1977b). This work suggested that it was a rod-shaped molecule with an axial ratio of 20:l. Morerecently (Hagestedt et al., 1989), paracrystals of phosphorylated and nonphosphorylated tau have been reported. Phosphorylated tau was 90-95 nm in length and 3-6 nm in diameter whereas nonphosphorylated tau was69-75 nm in length. An even In neurons, the microtubule-associated proteins, MAP’-2 shorter length of30 nm was reported for undamaged tau and tau, are found in the dendrites andthe axons, respectively indicating that it is an extremely flexible molecule. Tau was (Binder et al., 1985; Peng et al., 1986; Kosik and Finch, 1987). also studied in relation to microtubules, and its length was Stability to heat and solubility in perchloric acid form the found to be 56.1 f 14.1 nm (Hirokawa et al., 1988). No basis for tau’s isolation (Grundke-Iqbal et al., 1986a; Lindwall reference wasmade in this work to tau’s phosphorylation and Cole, 198413). The cDNA-predicted amino acid sequence state. The study of freeze-dried vertically platinum-carbon (Ptof mouse tau (Lee et al., 1988), bovine tau (Himmler et al., C) replicated isolated bovine tau was undertaken to charac1989; Himmler, 1989), and human tau (Goedert et al., 1989) terize tau’s structure with high resolution transmission elec* This work was supported in part by New York State Office of tron microscopy (TEM) (Ruben, 1989). Since tau has been Mental Retardation and Developmental Disabilities, National Insti- found in both Alzheimer neurofibrillary tangles and in paired tutes of Health Grants AG05892, AG08076, AG04220, and NS18105, helical filaments (Grundke-Iqbalet al., 1986a, 1986b;Ihara et and a grant from the Alzheimer’s Disease Research Program of the American Health Assistance Foundation, Rockville, MD. The costs al., 1986; Nukina and Ihara, 1986; Yen et al. 1987; Goedert et of publication of this article were defrayed in part by the payment of al., 1988, 1989;Wischik et al., 1988; Leeet al., 1991),the study page charges. This article must therefore be hereby marked “adver- of tau’s normal structure had to preceed TEM studies of tisement” in accordance with 18U.S.C. Section 1734 solelyto indicate neurofibrillary tangles and paired helical filaments so that this fact. their structuralrelationship to taucould be properly accessed § To whom correspondence should be addressed. subsequent work. ’ The abbreviations used are: MAP, microtubule-associated pro- in In the present study isolated bovine tau preparations are tein; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel eletrophoresis; Pt-C, platinum-carbon; TEM, transmission electron mi- shown to contain 2.1 f 0.2-nm filaments. It is shown that tau croscopy. is triple-stranded and left-handhelical. Using the amino acid

22019

The MAP Tau Forms a Three-stranded Left-hand Helical Polymer

22020

measured with a map measurer a t a print magnification ofX69,OOO. sequence of bovine tau and/or human tau, computer programs have been used to predict protein secondary structure under A JEM lOOCX transmission electron microscope with a lanthanum hexaboride filament,a 400-pm condenser aperture,and a 40-pm "Results and Discussion." Limits imposed by the tau images objective aperture was used. The fact that this arrangement limited and circular dichroism estimates of a helix and p structure resolution a t 80 kV to 0.66 nm was of no consequence since replica were used to temper these secondary structure predictions. resolution was not better than0.6-0.7 nm (Ruben, 1989). The microThe secondary structure, even with its limitedaccuracy, made graphs were reversed and printed asdescribed before (Ruben, 1989). it possible to identify the three separate strands as different Circular Dichroism-The bovine tau prepared for circular dichroism was 160 pg of protein/ml in 25 mM sodium phosphate, pH 6.5, tau sequences.

and was also in a mixed state of phosphorylation averaging 78 -I- 9 ( n = 3) nmol of phosphate/mg of protein or 3.1 0.4 ( n = 3) phosphates/ 384 amino acid tau (average number of amino acids in tau due to Bovine Tau Preparationfor TEM-The bovine tau was isolated by alternative RNA splicing, see table IV) as measured by Iqbal and heat treatment of three cycled microtubules at pH 2.7, followed by Grundke-Iqbal (1990). The circular dichroism spectra were recorded extraction in 2.5% perchloric acid according to Grundke-Iqbal et al. as a function of temperature a t 5" intervals from 10 to 80 "C. The (1986a). These preparations generally revealed in SDS-PAGE three sample was equilibrated at each temperature for 15-20 min before to four bands in the 48,000-62,000-dalton range shown in Fig. 1 (lane two separate CD spectra were recorded and averaged. The circular I ) and no bands corresponding to microtubule-associated protein 2 dichroism equipment and methods have been described before (Ciar(MAP-2). Theline at thetop of lane 1 is the protein entrance to the delli et al., 1988). gel. Western blots of bovine tau (Fig. 1, lane 2) were developed with Secondary Structure Analysis-The analysis of the published bomonoclonal antibody Tau-1 (Binder et al., 1985) using avidin biotin vine tau sequence (Himmler et al.,1989; Himmler, 1989) and the reagents of Vector (Burlingame, CA) according to Grundke-Iqbal et published human tau sequence (Goedert et al., 1989) was accomal. (1986a). Three samples were measured and averaged to yield 78 -t plished using the computer software PC Gene (6.1). @-Turnswere 9 ( n= 3) nmolof phosphate/mg of protein (Iqbal and Grundke-Iqbal, analyzed with the Chou and Fasman (1979) @-turn program. The 1990). Garnier program for the prediction of protein secondary structure The tau used for freeze-drying and vertical Pt-C replication in a (Garnier et al., 1978) was used to predict CY helix, @-sheet,and coil mixed state of phosphorylation was at a concentration of 100 pg of structure. The evaluation of hydropathic index as afunction of protein/ml in 0.15 M NaCl, pH 7, and was deposited on the surface sequence was averaged over a five to nine amino acid interval and of a 13-mm filter disc with a 0.1-pm porosity (mixture of cellulose was done with the Kyte and Doolittle (1982) program SOAP. This acetate and cellulose nitrate) from Millipore Corporation (type: VC, programalsocalculatedagrand average of hydrophobicity score catalog no. WP01300). This sample was washed with distilled water (GRAVY) which is either positive or negative and is a measure of a at 18-20 "C, blotted with ashless filterpaper to remove excess water, protein's overall hydrophobic (+) or hydrophilic (-) character. This and frozen in liquid propane. It was then freeze-dried in the modified work was used to interpret theimages of tau. Baker's 300 for 2.5 h a t -80 "C, vertically (80O angle) replicated with Longest and Shortest TauMonomer Length Calculations Using the 1.04 nm of Pt-C, and backed with rotary deposited (100" angle) 13.8 Idealized Tau Model in Fig. 6-The triple helix model has a diameter nm of evaporated carbon (Ruben, 1989). The samples were digested of 2.1 nm (Fig. 6a) where each strand is roughly 1.05 nm in diameter on 80% sulfuric acid, washed, and mounted as previously described and where the center of each strand is 0.525 nm from the centralaxis (Ruben, 1989). of the 2.1-nm filament. Each tau strand forms a super helix around A second sample of tau in 5 mM Tris-HC1 buffer, pH 7, was the filamentaxiswitha 5.4-nm pitch. Using the extendedchain prepared by placing three to four large drops on a mica disc at the spacing of 0.3-0.35 nm (0.325 nm) for each amino acid and 0.95 nm center of a spinning table top centrifuge. The drops spread rapidly for each @-turn,the longest tau axiallengths in the triple helix and radially from the center until they disappeared from the disc's filaments were calcuated for bovine tau (448 aminoacids, 42 &turns) surface. The sample disc was then frozen in liquid propane. This and human tau(441 amino acids, 42 @-turns). The shortest lengths tau sample was freeze-dried and replicated with 0.93 nm of Pt-C and were calculated by assuming that the amino acid sequence has the backed with 12.2 nm of rotary deposited carbon. The very long lengths axial spacings of an CY helix or 0.15 nm between amino acids. This of tau reported in the results were taken from this sample and were calculation is shown inTable IV along with theshortest length calculation for the spiral conformation first described in elastin by Urry et al. (1969). This helical coil ( p spiral) has a -1.66-nm diameter 1 Stds (could be 0.5 nm less, see Table I11 footnotes) with a pitch of 0.945 nm (Changand Urry, 1989). The first calculationassumes an CY helical configuration with a tau monomer diameter of 1.05 nm and a pitch in the polymer of 5.4 nm. In the CY helix of each monomer the amino acid chain would follow a path -1.16-nm long/0.54-nm pitch. Each @-turnwould occupy a distance of 0.95 nm along this helical path. The length calculations for these two models is shown in Table IV. Since tau contains -20% CY helix and -38% @-turn structure, and -42% @ spiral residues (or coil) (excludes &turns)(TableII), a composite model (containing the afore mentioned secondary structure) shortest length was also calculated and reported in Table IV. In this calculation, the CY helical regions do not contain @-turns and the @ spiral regions contain all the &turns. MATERIALS AND METHODS

2

RESULTS AND DISCUSSION

j

18 14

...;.,

1:.

1" ' r.. ~

,

~

; h ' :a

.. ,

FIG. 1. SDS-PAGE of bovine tau. Sodium dodecyl sulfate-polyacrylamide gel patterns of isolated bovine tau protein: lane 1 (0.75 p g of protein) stained with silver; lane 2 (0.055 pg of protein). Western blot developed with monoclonal antiboby Tau-1 (0.1 pg of IgG/ml). The standards used for molecular mass gel markers were as follows: phosphorylase b (97.4 kDa), bovine serum albumin (68 kDa), ovalbumin (43 kDa),0-chymotrypsinogen (25.7 kDa), P-lactoglobin (18.4 kDa), and lysozyme (14.3 kDa).

TEM of Isolated Bovine Tau Protein Examination of the replicas of tau spread on 0.1-pm filter discs reveal the presence of long narrow filaments that are 2.1 f 0.2 nm in diameter(2.7 nm with0.6 nm of Pt-C coating; Ruben, 1989) (Fig. 2a). The length of the filaments are 120, 300, and 800 nm as far as they be canmeasured on thesurface of the 0.1-pm Millipore filter. The filter surface without any tau present does not contain any of these 2.1-nm filaments. In Fig. 2b, sections of tau filaments extending across the holes in the filter were examined a t high magnification. Although the filament inpanel A was measured as 2.2 nm (2.8

T h e MAP T a u Forms a Three-stranded Left-handHelical Polymer

22021

crossing the axis ina left-hand helical direction (see arrows). These strands average 1.0 f 0.2 nm in diameter (1.6 nm with 0.6 nm of Pt-C coating). In panel B three 1.0 f 0.2-nm strands are left-hand wrapped around a 2.1 f 0.2-nm filament axis. In panel C, a 2.1 f 0.4-nm filament shows a left-hand helical substructure. Finally, inpanel D, three strands, 1.0 f 0.3 nm, cross the filament axis in a left-handed direction. Fig.2b indicates that tau proteina is triple-stranded left-handhelical filament with a 2.1 f 0.2-nm diameter which is composed of three 1.0 f 0.2-nm strands. The replicas of tau spread on mica discs revealed a wide range of lengths of tau filaments(Fig. 3) Thelongest filament lengths were 2030 nm ( n = 2) with other lengths of 1740 ( n = 2), 1600,1380,1330,1240,914,682 nm( n= 2), andsmaller. A’

The Structure of T a u by Circular Dichroism The CD spectrum in Fig. 4a shows a large trough in the spectrum at 197 nm which is identified as a coil polypeptide conformation (Greenfield and Fasman, 1969; Johnson, 1987). Because bovine tau is mainlycoil conformation (60-66%) the empirical formulas for calculating LY helix and p-sheet are not considered accurate in quantitating these secondary structures. These formulas are onlyreliable when these structures are in abundance (Taylor and Kaiser, 1987). Nonetheless, these formulas were used on the 10, 40, or 80 “CCD spectra with nearly identical estimates of CY helix (10-12%) and p structure (24-28%) which were also similar to theroom temperature values reportedby Cleveland et al. (1977b). The CD spectum of bovine tau in Fig. 4a was taken as a function of temperature. Fig. 4b shows that the mean residue ellipticity (6’)a t 197-nm increases by 52% from 10 to 80 “C indicating that with increasing temperature random coil is disappearing and isbeing replacedwith a moreordered secondary structure.

8’

c’

=’

,d 5,

/

FIG. 2. Panel a, bovine tau suspended ona Millipore filter. Isolated bovine tau on a 0.1-m Millipore filter was washed with distilled water (18-20 “C) before freezing. This sample was freeze-dried and vertically replicated with 1.04 nmof Pt/C and backed with 13.8 nm of evaporatedcarbon. The arrows point to long thin 2.1 f 0.2-nm filaments (2.7 nm with 0.6 nm of Pt-C coating; Ruben, 1989)which are identified as tau. The larger filaments are 10 f 1 and 13 f 1 nm in width. Filters without tau did not contain long thin -2.1-nm or larger filaments extending across the filter holes. X 92,500. b, the structure of bovine tau. Bovine tau suspended over 0.1-pm Millipore filter pores was replicated with 1.04 nm of Pt-C. Panels A-D contain tau filamentsextending from lefttoright. Panels A‘-D’ contain tracings of tau adjacent to the panels with the same letter. In panels A and A’ a 2.2f.3-nm filament (2.8nm with 0.6 nm of Pt-C coating) shows subfilaments that cross the filamentaxisin a left-handed direction. The strands have a diameter averaging 1.0 f 0.2 nm (1.6 nm with 0.6 nm of Pt-C coating). Just to the left of the arrows in panels B and B’ the filament measures2.1 f 0.2 nm and shows a lefthanded helical structure. The filament separates into three strands where the arrows are labeled 1-3. These left-hand wrapped strands average 1.0 f 0.2 nm in diameter. The panel C and C‘ tau filament averages 2.1 f 0.5 nm andcontains strands twistedaround the filament axis in a left-handed direction. The edges of these strands were too poorly defined to measure. In panels D and D’, a left-hand wrapped filament separates into three strands marked 1-3. These strands also average 1.0 f 0.3 nm in diameter. This tau strand is a t the bottom of a and is an enlargement of the junction with the 13nm filament. This image was taken from a different micrograph in the tilt series of a. X 500.000.

- 0.6 nm for the Pt-C coating) in the region where it is marked, a good portion of this filament averages 2.1 f 0.3 nm. To the right of center of this figure, there are strands

Interpretation of Tau Images Using the Secondary Structure of Bovine and Human cDNA-derived Amino Acid Sequences The long filamentous propertiesof tau suggest that itshould have an unusual distributionof hydrophobic and hydrophilic amino acids in comparison to globular proteins. In Fig. 5a, the hydropathic index is plotted against the amino acid seet al., 1989). A similar plot for quence in human tau (Goedert bovine tau (not shown)looked almost identical to human tau. The tubulin-binding domains of tau, the four 31-amino acid repeats from about aminoacid 244 to 368 are less hydrophilic than the sequence from 1 to 244 and not as hydrophobic as

FIG.3. The length of bovine tau on mica. Tau in 5 mM Tris, pH 7.0, was freeze-dried and replicated as described under “Materials and Methods.” The irregular thickness of the tau filaments does not contain any of the tau fine structure of Fig. 2b, and the filaments appear to be coated with condensed buffer. Panel a, tau length of 2,030 nm;, panel b, tau length of 1,235 nm; panel c, tau length of 2,088 nm; panel d, tau length of 1,378 nm; and panel e, tau length of 1,740 nm. X 29,000.

22022

The MAP Tau Forms a Three-stranded Left-hand Helical Polymer b

-

a

;;

-

-

E"-

0.0

L

E

D 111

u

E

5

u

-3-

-

-

-2.0

5*

u

-

a -41

D

*I

2

-2 -

-

-=*r.

-4.0

I

.

3

X

31c

-5-6.0

190

200

22 10

Waveleneth

230

lnm)

240

/

10

20

I

I

I

x) 40 50 Tempero1ure:C

,

60

I

X)

l

80

FIG.4. Panel a, CD spectrum of bovine tau. Circular dichroism spectrum of 160 pg/ml of bovine tau in 25 mM sodium phosphate, pH 6.5, as a function of temperature from 10 to 70 "Cat 10 "C intervals. The curve for 80 "C was veryclose to that at 70 "C,and for figureclarity it was not included. The mean residue ellipticity (e) was based on the bovine tau sequence and an average amino acid molecular weight of 103.4. This bovine tau has a mixed phosphorylation state averaging 3.1 f 0.4 (n = 3) phosphates/384 amino acid tau (average tau sequence with deletions, Table IV) (Iqbal and Grundke-Iqbal, 1990). Panel b, inverse temperature transition in bovine tau. The ellipticity at 197 nm as a function of temperature from 10 to 80 "Cat 5 "C intervals for the bovine tau in punel a. The progressive increase at 197 nm with increasing temperature indicates that random coil is being replaced with a more orderedstructure. This figure indicates that bovine tau undergoes an inverse temperature transition. Most proteins heated from 10 to 80 "C show an increasing negative ellipticity at 197 nm or increasing random coil secondary structure. the carboxy-terminal (400-441). The grand average of hy- analyses, the evidence suggests that the three strands are dropathy, GRAVY, of human tau and bovine tau (Table 11) separate tau monomers in an extended filamentous conforis -8.68 and -8.52, repectively, whereas water-soluble globu- mation. Assuming that each 2.1 f 0.2-nm filament is comlar proteins are between +0.3 and -1.0 and average -0.4 on posed of three adjacent tau monomers (tau,), the longest and this index (Kyte andDoolittle, 1982).Tau's hydropathic index the shortest tau monomer strand lengths can be calculated and the grand average of hydropathy are consistent with an (see "Materials and Methods"). The longest tau axial lengths extended conformation exposed along its surface to water and in the triple helix filaments are 106-117 nm (112 nm for containing no buried globular hydrophobic regions. 0.325-nm amino acid spacing) or 104-116 nm (110 nm for Secondary structure predicting programs are useful for 0.325-nm amino acid spacing) for bovine tau (448 amino acids, showing trends in protein structure, even though they are no 42 @-turns) and human tau (441 amino acids, 42 @-turns), better than 63% (Garnier et al., 1978; Busetta and Hospital, respectively. Assuming that the full-length tau and tau with 1982; Kabsch and Sander, 1983) to 70% accurate (Chou and deletions are equally represented then the longest average Fasman, 1979). Estimates from our circular dichroism meas- length of human tau is96.2 nm and thelongest average length urements and Cleveland et al. (1977b) suggest that tau is 10- of bovine tau is 95.8 nm. In Table 111, the longest measured 12% a helical and 20-27% @ conformation whereas the Gar- porcine tau monomer paracrystalline length was reported as nier program (Garnier et al., 1978) suggests that taucontains 90-95 nm. Clearly the tau monomer length within a triple 31% a helix and 37-39% @ conformation (Tables I and11).It helix model can accomodate the longest measured tau monis unlikely that triple stranded tau (-2.1-nm diameter) con- omer. The shortest tau lengths were calculated by assuming that tains @-sheetsince each sheet strand could contain only two amino acid chains roughly 1.0-1.3 nm X 0.3-0.6 nm in cross- the amino acid sequence has the axial spacings of an a helix. section to approximate the 1.0 f 0.2-nm strands. Second, it This first calculation is shown in Table IV along with the is difficult to understand how these strands could remain as length calculation for the @ spiral conformation first described separatestrands with unfulfilled hydrogen bonds on both in elastin by Urry et al. (1969). This helical coil (@spiral) has sides of a strand. Third, &sheet is a fully extended confor- a -1.66-nm diameter (its diameter could be 0.5 nm less, see mation (0.3-0.35-nm spacing between amino acids) which Table I11 footnotes) with a pitch of 0.945 nm. In the a helix would not produce an elastic tau asdescribed by Hagestedt et the amino acid chain would follow a path -1.16-nm long/ al. (1989). Finally it has been shown that @-sheetcan have a 0.54-nm pitch. The length calculations for these two models right-handed helical twist with a pitch as short as9.2-9.6 nm is shown in Table IV. Neither of these models by themselves (Fraser andMacrae, 1974; Fraser et al., 1971; Stewart, 1977), yields lengths close to 30 nm. Since tau contains -20% a but itis hard to understand how these two amino acid strands helix and -38% @-turn structure, and-42% @ spiral residues could cross the 2.1 f 0.2-nm filament axisat 5.4-nm intervals (excludes @-turns) (Table11), a composite model (containing and maintain their@-sheetstructure and their separate strandthe aforementioned secondary structure) shortest length was identity. The @ structure in tau ismore likely to take theform calculated in Table IV. The composite model gives estimates of tau's monomer length slightly shorter than 30 nm. The of @-turnsalso called @-bends orreverse turns. shortest paracrystalline length reported for porcine tau in Tau Lengths Calculated Using Its Secondary Structure and Table I11 is 30 nm which can be accomodated by the tau an Idealized Periodic Conformation monomer length in the triple-helical tau polymer model. A The analysis of the amino acid sequence is consistent with nonphosphorylated porcine tau length of 69-75 nm (Hegestau's filamentous conformation. The images in Figs. 2 and 3 tedt et al.,1989) or the otherporcine tau length of 56.1 f 14.1 reinforce this prediction, and inconjunction with the previous nm (Hirokawa et al., 1988) have also been reported (Table

The MAP Tau FormsThree-stranded a Left-hand

Helical Polymer

22023

T a u F o r m a Polymeric Structure: a New Class of Triple-helical Fibrous Proteins

could also be about this size. A frequently observed 1.8 f 0.2 nm distancebetween strands thatcross the -2.1-nm filament axis at 30-55" angle is generated by the -1.0-nm strands (see Fig. 6, a and d ) . The images, sequence analyses, and length calculations suggest that thequaternary structureof the -2.1nm filament is composed of three left-hand helical extended tau monomers (taus) between 29.2-95.8 nm in length. If it is assumed that tau29 normal length is intermediate or -63 nm (Hirokawa et al., 1988; Hagestedt et al., 19891, then it can be concluded that tau3 forms polymers of many different sizes. Using -63 nm as an estimate of taus's length,then the polymer, (tau,), can have positive integer n values from 2 to 33 (see range of tau lengths in Table 111). Its not likely that n is just limited to these values. Since (tau,), appears to be a continuous -2.1-nm filament (Fig. 2), the ends of one taua smoothly intermesh with each consecutive tau, unit. The four to six tau sequence lengths (Goedert et al., 1989; Himmler et al., 1989) make this possible if each tau strand is assembled in register on the region with the three or four 31 amino acids repeat sequences, with the NHz-terminals at one end and the COOH-terminals at the other. This should align the 2 cysteines in each tau sequence to form a disulfide bond with an adjacent tau strand. Joining the tau trimers, NH2-terminal to COOH-terminal would orient the taus within the (tau3), with the same chain polarity and equidistant spacing between repeat regions. In addition the COOH-terminal domains are predominantly positively charged (17+ basic versus 12- acidic amino acids) and the NH2-terminal domains are negatively charged (27-acidic versus 4+ basic amino acids) (Himmler et al., 1989) so that opposite charge attraction could drive the polymerization process (see Table V). However compelling these arguments are, the alternative method of coupling NHz-terminal to NH2terminal and COOH-terminal to COOH-terminal, although less probable, cannot be completely ruled out. Nevertheless, each tau strand has a complex secondary structure, has little tertiary structure, and assembles with a triple-stranded lefthand helical quaternary structure. This structure helps explain why tau needs to be isolated in a disulfide bond-reducing medium and is stable at temperatures of -100 "C. Tau is also soluble and stable when heated in apH 2.7 buffer (GrundkeIqbal et al., 1986a) because it contains roughly equal numbers of basic and acidic amino acids leading to a soluble positively charged molecule at high temperature whereas MAP-2 precipitates with only 213 the number of basic amino acids versus acidic residues. Globular proteins with extensive hydrophobic domains also precipitate under these conditions. It should also be pointed out that this kind of protein structure is new and is quite different from the 1.5-nm triple-stranded righthanded helix of collagen (see Fig. 6b) and a 2-2.5-nm triplestranded a helix-coiled coil suggested for a-keratin and T3 tail fibers (see Fig. 6c) (Crick, 1953; Fraser et al., 1962; Takahashiand Ooi, 1988). Neither of thesestructures is consistent with the secondary structure predicted from tau's primary sequence, with the tau images or with circular dichroism results. Strong evidence for a triple-stranded lefthand helical fibrous protein like tau has not been reported before. Although the model of tau is presented as a regular structure to facilitate the calculation of its maximum and minimum length, allof the tau samples frozen from an initial temperature of 18-20 "C were only partially regular (Fig. 6d).

The images in Fig. 2 indicate that tau forms -2.1-nm filaments that are triple-stranded and left-hand helical. The strand diameter averages -1.0 nm, which is about the diameter of a right-handed a helix (-0.46 nm) enlarged an additional 0.5 nm due to amino acid side chains. An extended coil

Implications of a n Elastin-like Inverse Temperature Transition in T a u Elastin's Similarities to Tau Suggest That Tau Shortens from 20 to 37 "Cand Becomes Elastic-Elastin is filamentous,

2e

le

e -10

-29

M a - t u g proWility pmfile of TAU441G.froa anino acid 1 to amino acid 44l., Ihc y ax15 values npresrnt the pmbablllty p(tum) * 1W FIG. 5. Panel a, hydropathic index of human tau. The hydropathic index on the left-hand scale for an averaged interval of five amino acids is plotted against the amino acid sequence in human tau (Goedert et al., 1989). Positive values are hydrophobic and negative values are hydrophilic. The grand average of hydropathy (GRAVY) for human tau is -8.68 and for bovine tau is -8.52 (Table 11). Panel b, 0-turn probability in the human tau sequence. The probability of a tetrapeptide forming @-turns in human tau as a function of its amino acid sequence. The dashed horizontal l i n e corresponds to a cut off probability of 0.75 X 10" below which a &turn is not predicted. The grouped peaks are often overlapping amino acid sequences from which the most likely &turn has been selected in Table I. Although not shown, the plot for bovine tau looked almost identical to this human tau plot.

111).These lengths roughly correspond to the average of the longest and shortest tau monomer length calculations which are 61.8 nm for bovine tau and 62.4 nm for human tau. The triple-stranded left-handhelical model for tau can accomodate the variety of paracrystalline tau monomer lengths cited in the literature (see Table 111).The shortest length calculation uses the estimated secondary structure of human tau in Table 11. The tau polymer model is not only compatible with tau protein secondary structure estimates in Table 11, but it is also consistant with the reported elastic behavior of porcine tau (Hegestedt et al., 1989). The /3 spiral protein (in contrast to a helix) secondary structure (Table IV) according to Urry (1988) has elastic properties.

The MAP TauForms a Three-stranded Left-hand Helical Polymer

22024

TABLE I Predicted &turn and a helical regionsof human tau Four consecutive amino acids are underlined showing the location in the human tau sequence of the predicted @-turnsfrom the Chou and Fasman (1979) computer program. The 8-turn probability map is shown in Fig. 5b. The predicted sequences of a helix are in boldface letters in the sequence. Although these sequences ( 2 4 residues) frequently overlap @-turns(1-4 residues), we have chosen to ignore the incompatibility of these two structures and report the results of the Garnier et al. (1978) program. In compiling Table 11, we have retained all the &turn predictions at theexpense of a! helix except for the one at residue 3-6 which was omitted. If a helix sequences are 4 residues or fewer it has been assumed arbitrarily that they are not likely to retain this configuration. 8 sheet secondary structure is assumed to be incompatible with the 1.0-nm strand diameter in the tau images (see "Discussion"). Secondary structure which is neither 8-turn or a helix we assume to be coil or 8 spiral. This same approach was used in judging secondary structure for all the proteins listed in Table 11. 1 nUEPRQFEVmED~~LGDR~Y~~EG~GLKESPLQTPTE~SEEPGSETSD~STPTAEDVTAPLV 81 D E G ~ Q A M Q P H T E I P E G T T A E E A G I G D ~ E D E M G ~ ~ A ~ S K S K D G T G S 159 PPGQKGQANATRIPAKTPPAPKTPPSSGEPPKSGDRSGYSSPGSPGTPGSRSRTPSLPTPP~EP~A~TPPK~~ 241 S R L Q T A P V P M P D L K N V K S K I G ~ L ~ ~ ~ Q I I N K K L D L S N V Q S K C G S K D N I ~ ~ S V Q I W K P V D L S K V 320

sKCcSLGNI~~~VEVKSEKLDFKDRVQSKIGSLDNI~~NKKIET~TFRENA~K~AEI~~W 400

SGDTSPRHLSNVSEIDMVDEATLADEVSASLAKQGL

TABLE I1 Protein properties and secondary structure commsitwn No. of amino acids, full sequence

Protein and source

Sequence

M,

@-Turns" % of sequence

(No.)

Tau Human' Bovine'

441 448

MAP-2 Mouse8

1828-7.99

a Helixb

% of sequence

45,850 46,332

38 (42) 38 (42)

-20 -23

198,978

-3835 (161)

-27

Coil or @spiral

Inverse temperature transition at 197 nm in CD suectrum'

Grand average of hydropathy,

GRAVYd

Remining % of sequence

-42 -39

Yes, Fig. 4

-8.68 -8.52

Elastin a Bovineh64,229

747 35 (67) -21b.i -44 yesL.J +6.85 Chou and Fasman, 1979. Garnier et al., 1979. e The coil trough at 197 nm diminishes as the temperature is raised from 10 to 80 "C, the inverse of normal protein denaturation which increases the trough at 197 nm in the CD spectrum. Kyte and Doolittle, 1982; (-) hydrophilic, (+) hydrophobic. e Goedert et al., 1989. Himmler et al., 1989. Lewis et al., 1988. Raju and Anwar, 1987. Starcher et al. (1973), estimated from CD spectrum. Urry et al., 1969, 1985.

'

J

has a secondary and quaternary structure similar to tau's (Tables I1 and 111), and its length like porcine tau's (Table 111) is variable. Because elastin, tropoelastin, and bovine a elastin are very hydrophobic (GRAVY = +6.85, Table 11) and first assemble into small filaments, at temperatures of 2440 "C the small filaments condense into larger 0.5-9-pm wide fibers. Nonetheless, bovine elastin, tropoelastin, and a synthetic peptide containing -200 copies of elastin's pentapeptide repeat arewater soluble at temperatures less than 24 "C. At temperatures between 24-40 "C, these proteins condense into aggregates of parallel fibrils which are a coascervate of -37% protein and -63% water (Volpin, 1977; Urry, 1988). During the temperature increase from 24 to 40 "C, circular dichroism has shown that the elastin structure or pentapeptide polymer change from a less ordered internal state to a more ordered internal state (Urry et al., 1969, 1985; Starcher et al., 1973). This entropic process has been called an inverse temperature transition where ordered clathrate water leaves the protein at higher temperature, the protein becomes more ordered and theremoved water becomes disordered increasing the water and protein's overall entropy (Urry, 1990). In this process elastin and a polypentapeptide elastin-like polymer

are shortened by 22 and by 55%, respectively (Table 111) and develop elastomeric force (Urry et al., 1986; Urry, 1988). This process is reversible. By lowering the temperature to 20 "C, elastin andthe elastin-like pentapeptide polymer extend (with a (3 spiral rise of 2.16 nm/turn of 2.9-3.0 pentapeptide repeats; Chang and Urry, 1989) and looses its elastomeric properties. By rewarming to 40 "C, elastin and the elastin-like pentapeptide polymer resume their shortened length (with a (3 spiral rise of 0.945 nm/turn of 2.9-3.0 pentapeptide repeats; Chang and Urry, 1989) with the return of elastomeric properties. The inverse temperature transition is a finely tuned entropic transition which begins at 24 "C and is complete at 37 "C or body temperature. By chemically synthesizing an additional hydrophobic group in the pentapeptide repeat, the inverse temperature transition occurs at a temperature 15 "C lower (Urry, 1988). If a hydrophilic group is chemically synthesized into thepentapeptide repeat, the inverse temperature transition occurs at a temperature 45 "C higher (Urry et al., 1988). The circular dichroism spectra showing with increasing temperature a decrease in the random coil, in the present study, directly establishes that the tau protein also undergoes an inverse temperature transition which should also be affected

22025

The MAP Tau Forms a Three-stranded Left-hand Helical Polymer TABLEI11 Filament length measurements and structural properties Partiallybor undefined' Fully state of phosphorylation phosphorylated" Normal Shortest Or Average longest Range

Filament quaternary structure and diameter

Unphosphorylated" Protein temperature and source

nm

nm

nm

130-2088b

-30"

69-75"

MAP-2 Porcine Crayfish

90-95" 185d 170'

20 to 40 "C

nm

nm

Tau Bovine

Porcine

Shortening of length, with inverse transition, temperature increased from

Triple-stranded left-helical diameter 2.1 f 0.2 nm at 20 "C, 1.0 f 0.2 nmb Diameter nm"3-6

56.1 f 14.1' 90 f 30' 100-18tjd 40-170' 104 f 22'

Elastin Bovine

Not measured, inverse temperature transition present, see Fig. 4

Diameter 1.6 nmd

Triple-stranded helixg -1.8 nm with strand diameter -0.9 nm8+ & larger fibers 0.5-9-pm diameter Strand diameter -1.48 nm, 20 "C Strand diameter -1.66 nm, 40 "C

Shortened by -22%'

Shortened by Pentapeptide repeat, 200 -55%' copies in sequence, 100% j3 spiral secondary structure a Hagestedt et al., 1989. Mixed phosphorylation state, 78 -C 9 nmoles phosphate/mg protein or 3.1 k 0.4 phosphates/384 amino acid tau, thispaper. e Hirokawa et al., 1988. Voter and Erickson, 1982. e Gottlieb and Murphy, 1985. Hirokawa, 1986. Ranachandran and Santhanam, 1957. Cleary and Cliff, 1978. ' Gotte et al., 1968. j Starcher et al., 1973. Serafini-Fracassini et al., 1977. ' Urry et al., 1986. Using numbers in Chang and Urry, 1989, the 8 spiral strand diameters were calculated assuming amino acid side chains added 0.5 nm to the diameter like they do with an a helix. The diameter of the 8 spiral would be reduced if the side chains were parallel to or pointed inwards toward the filament axis. The diameters for these configurations would be 0.98 and 1.16 nm and would be closer to the size of 1.0 nm reported for the triple-helix strand diameter.

'

'

TABLE IV Calculated tau lengths assuming tau model in Fig. 6a Shortest length models a Helix

Extended length

(62%a helix and 38% &turns) 0.325 nmlaminn acid

Bovine tau Full-length (448 amino acids) Average length (384 amino acids) 27.8

112 95.8 24.1

nm

52 44.4

p Spiral" ( p spiral with 38% &turns)

nm

28

Composite model' nm

32.5

Human tau Full-length (441 amino acids) 51.1 110 32 27.6 Average length acids) (396 28.7amino 24.8 45.8 96.2 @ spiral incorporates the &turns within the spiral. According to the original definition this model would be 100% fl spiral. The length of the composite model is calculated assuming that the shortest tau would contain 38% 8-turns, -42% j3 spiral residues (coil) not including the &turns, and-20% a helix. According to the definition of B spiral, this model would be-80% B spiral, and -20% a helix.

by similar changes in amino acid composition. Microtubule Assembly and Stability-Since the circular dichroism of tau demonstrates a progressive increase in mean residue ellipticity (e) a t 197 nm with increasing temperature from 10 to 80 "C,tau, like elastin and the elastin-like penta-

peptide polymer, undergoes an inverse temperature transition (Fig, 4,a and b ) which starts in a less ordered extended state at 10 "Cand is transformed to a shortenedmore ordered state at 37 "C(Table 111) and athigher temperatures with an upper limit of -80 "C.This process has important implications for

The MAP Tau F o r m s a Three-stranded Left-hand Helical Polymer

22026

TABLE V Charged residues a t the amino or carboxyl-terminals within the first 100 amino acids Protein No. of amino NH, terminal charged COOH terminal acids and Source (full sequence)

Tau Human" acidic Bovine' acidic MAP-2 Mouse' acidic

441 448

1828

residues/100 amino acids

charged residues/100 amino acids

24- acidic/9+ basic 21- acidic/4+ 17+ basic

basic/l316+

19- acidic/l3+ basic

basic/lO17+

basic/l2-

Goedert et al., 1989.

'Himmler et al., 1989. Lewis et al., 1988.

how tau functions in reconstituting microtubules. Tubulin, at 37 "C, in the presence of buffers and tau, assembles into microtubules. If the temperature is lowered below 20 "C, the microtubulesdisassemble. This process also occurswith MAP-2 and other microtubule-associated proteins. MAP-2s secondary structure suggests that it should, like tauand elastin, undergo aninverse temperature transition from 20 to 37 "C (see Tables 11, 111, and V). We plan to investigate this CD spectrum-temperature relationshipin the near future. Additional phosphorylation shifts the inverse temperature transition to higher mean temperature (Urry, 1990). It is not surprising, then, that heavily phosphorylated tau (Lindwall and Cole, 1984a) and other heavily phosphorylated microtubule-associatedproteins (Jamesonet al., 1980) do not promote microtubule assembly a t 37 "C. This suggests that the shorter, more ordered tau or MAPS a t 37 "C binds more strongly than the longer less ordered state a t lower temperature and that heavy phosphorylation can make tau or MAPS more hydrophilic and prevent their transition toa shorter more ordered state at 37 "C. An elastic tauis needed to stabilize very long axonal microtubules that are subject to constant bending, as at knee and elbow joints. It is unlikely that individual tau monomerswith only a single binding domaincould restore microtubule structure elastically after a bending insult. Each tau has a single region of three orfour repeating sequences, totaling 93 or124 amino acids, that bind to tubulinin microtubules (Aizawa et al., 1988). A tau polymer oriented axially along the microtubule would connect tau monomerswhich each extend over a t with least seven or eight (-63 nm)8-nmtubulindimers binding domains approximately 13.1 or 17.5-nm long. Such 1 a n elastic tau polymer repeated around the microtubule periphery would stabilize the microtubule and maintain microFIG. 6. Panel a, idealized periodic triple-stranded left-handhelical tubule structure through a bending process. We are investitau model. This taumodel idealizes the structural featuresof tau that

b 3

have beenobserved in Figs. 2 and 6d. The diameter of2.1 nm is composed of three -1.0-nm strands which cross the filament axisa t -1.8-nm intervals with a pitch of -5.4 nm. Panel b, triple-stranded collagen model. In b the model for collagen is shown with a 1.5-nm diameter. Three 0.7-nm strands cross the axis a t -2.9-nm intervals along the axis with a pitch of8.6 nm. The collagen strands also contain a tripletrepeat motive, (Gly-X-Y), where X and Y are frequently proline and hydroxy proline. These residues form a lefthanded helix with a pitch of0.87-0.90 nm with each amino acid extending 0.295-0.30 nm axially (Ramachandran, 1963; Schulz and Schirmer, 1979). This is very near the fully extended amino acid distance of 0.35 nm (-17% longer than 0.3 nm) found in 8-sheet and probably explains why rat tail collagen can only extend about 1.517% before it ruptures (Kastelic and Baer, 1980). Panel c, triplestranded a-keratinmodel. In panelc, the model fora-keratin isshown witha1.9-nm diameter. Three 0.95-nm strands cross the axis at -6.3-nm intervals along the axis with a pitch of 19 nm. The coiled

coil of three a helices has a diameter of 1.9-2.5 nm and is composed of three helices each 0.95-1.0 nm in diameter which left-hand twist around each other (Crick, 1953) with a pitch of 18.6-20 nm (Crick, 1953; Fraser et al., 1962). Panel d, triple-stranded left-hand helical tau is not perfectly regular. Isolated freeze-dried vertically replicated (1.04 nm of Pt-C) bovine tau prepared as in Fig. 2a. The 2.1-nm diameter has been enlarged by 0.6 nm of Pt-C (Ruben, 1989) to 2.7 nm and the -1.0-nm strands, enlarged to 1.6 nm by 0.6 nm of Pt-C, cross the axis a t -1.8-nm intervals in a partially regular region. An optical diffraction pattern of this filament was attempted and was unsuccessful. The traced tau filament is recorded below the halftone image. The edges of the strands are black and the strand widths are not perfectly regular. The arrowslabeled 1-3 point to separate strands in the triple-stranded left-handed tau helix. The coil structure or 8 spiral structure in the strands of tau monomer canin theory be stretched as thin as the backbone aminoacid chain. X 1,800,000.

The MAP

Tau Forms Three-stranded a Left-hand

gating the prediction that tau polymer is longitudinally oriented on microtubules and that ithas a functionalrole in the axons.

Alzheirner Neurofibrillary Tangle Formation The inverse temperature transition has important theoretical implications for neurofibrillary tangle formationfrom tau in Alzheimer's disease. Elastin and theelastin-like pentapeptide polymer form coacervates at 24-37 "C. It is possible by making the amino acid composition of tau more hydrophobic that tau can aggregate like elastin at 24-37 "C instead of promoting microtubule assembly. Shifting the coascervation phenomena to lower temperatures by 10-15 "C may require the replacement or elimination of only 1 aspartate or glutamate/100 amino acid residues (Urry, 1990). It seems probable then that some tau produced in the nerve cell body of Alzheimer's disease victims contains a slighty more hydrophobic sequence which self-aggregates and condenses into neurofibrillary tangles at 37 "C and neutral pH. This mechanism could be triggered if tau was phosphorylated in the wrong position leaving a long sequence of about 100-150 amino acids more hydrophobic. Tau from Alzheimer neurofibrillary tangles is abnormally phosphorylated (Grundke-Iqbal et al., 1986b; Iqbal et al., 1986, 1989) and is unable to reconstitute microtubules a t 37 "C (Iqbal et al., 1986; Nieto et al., 1990). We know that tau that is functionally active in microtubule assembly contains a number of phosphate groups (Lindwall and Cole, 1984a;Iqbal and Grundke-Iqbal,1990).The number and location of the phosphorylation sites on both normal and abnormally phosphorylated tau areyet to be determined. Acknowledgments-We thank GeoM Co. and IBR for its support and the Dartmouth Rippel Electron Microscope Facility for the use of the JEM lOOCX and the modified Balzers 300. Isolation of bovine tau was carried out by T. Zaidi and M. S. Zaidi at IBR. REFERENCES Aizawa, H., Kawasaki, H., Murofushi, H., Kotani, S., Suzuki, K., and Sakai, H. (1988)J. Bwl. Chem. 263.7703-7707 Binder, L. I., Frankfurter, A., and Rebhun, L. I. (1985)J. Cell Biol. 101, 1371197n A" I "

Busetta, B., and Hospital, M. (1982)Biochm. Biophys. Acta 701,111-118 Chang, D. K., and Urry, D. W. (1989)J. Computational Chem. 10,850-855 Chou, P. Y., and Fasman, G. D.(1979)Bio hys J 26,367-384 Ciardelli, T. L.,Landgraf, B., Gadski, R., &rand,'J., Cohen, F. E., and Smith, K. A. (1988)J. Mol. Recognrtron. 1, 42-47 Cleary, E. G., and Cliff, W. J. (1978)Exp. Mol. Pathol. 28,227-246 Cleveland, D. W., Hwo, S. Y., and Kirschner, M. W. (1977a)J. Mol. Biol. 116, 207-225 Cleveland, D. W., Hwo, S. Y., and Kirschner, M. W. (1977h)J. Mol. Biol. 116,

". ".

337-947

Crick, F. H. C. (1953)Acta Crystallogr. 6, 689-697 Fraser, R. D. B., and Macrae, T. P. (1974)in Proceedings of the 16th International Ornithological Congress (Frith, H. J.. and Calabv. J. H.. eds) DD. 443451,AustralianAcademfofScience,.Canberra,Austraiia Fraser, R. D. B., Macrae, T. P., and Rogers, G. E. (1962)Nature 193, 10521055 Fraser, R. D. B., MacRae, T. P., Parry, D. A. D., and Suzuki, E.(1971)Polymer '

"."_

'

~

~

12.35-5A

"

Garnier, J., OsEthofRe, D. J., and Robson,B. (1978)J. Mol. Biol. 120,97-120 Goedert, M., lschl , C. M.,, Crowther, R. A., Walker, J. E., and Klug, A. (1988)Proc. Natl. Acud. Scr. U. S. A. 86,4051-4055

Helical Polymer

22027

Goedert, M., Spillantini, M. G., Jakes, R., Rutherford, D., and Crowther, R. A. (1989)Neuron 3,519-526 Gotte, L.,Mammi, M., and Pezzin, G . (1968)in Symposium on Fibrous Proteins, 3rd, Canberra, Australia, August 14-18, 1967 (Crewther, W. G., ed) pp. 236245,Plenum Publishing Co., New York Gottlieb, R. A,, and Murphy, D. B. (1985)J. Cell Biol. 101,1782-1789 Greenfield, N., and Fasman, G . D. (1969)Biochemistry 8,4108-4116 Grundke-Iqhal, I., Iqbal, K., Quinlan, M., Tung, Y.-C., Zaidi, M. S., and Wisniewski, H. M. (1986a)J. Biol. Chem. 261,6084-6089 Grundke-Iqbal, I., Iqbal, K., Tung, Y.-C., Quinlan, M., Wisniewski, H. M., and Binder, L.I. (1986b)Proc. Natl. Acud. Set. U. S. A. 83,4913-4917 Hagestedt, T., Lichtenberg,B., Wille, H., Mandelkow, E."., and Mandelkow, E. (1989)J. Cell Biol. 109, 1643-1651 Himmler, A. (1989)Mol. Cell Bhl. 9,1389-1396 Himmler. A,. Drechsel. D.. Kirschner. M. W.. and Martin. D. W... Jr. (1989) , . Mol. Cell Biol. 9,1381-1368 Hirokawa, N. (1986)J. Cell Biol. 103, 33-39 Hirokawa, N., Shiomura, Y., and Okabe, S. (1988)J. Cell. Biol. 107, 14491A59

Ih&,-Y.,

Nukina, N., Miura, R., and Ogawara, M. (1986)J. Biochem. (Tokyo)

- .

--

Q)I)~ ~ n 7 -" 1 RI n , 1" 1I

Iqbal, K., Grundke-Iqbal, I., Zaidi, T., Merz, P. A., Wen, G. Y., Shaikh, S. S., Wisniewski, H. M., Alafuzoff, I., and Winblad, B. (1986)Lancet 2,421-426 Iqbal, K., Grundke-Iqbal, I., Smith, A. J., George, L., Tung, Y-C., and Zaidi, T. (1989)Proc. Natl. Acud. Sci. U. S. A. 86, 5646-5650 Iqbal, K., and Grundke-Iqbal, I. (1990)J. Neuropathol. Exp. Neurol. 49,270 (ahstr.) Jameson, L.,Frey, T., Zeeberg, B., Dalldorf, F., and Caplow, M. (1980)Biochemistry 19,2472-2479 Johnson, W. C., Jr. (1987)Methods Biochem. Anal. 31.61-163 Kabsch, W., and Sander, C. (1983)FEBS Lett. 156, 179-182 Kastelic, J., and Baer, E. (1980)in The Mechanical Properties of Biological Materials, SEB Symposia (Vincent, J. F. V., and Currey, J. D., eds) No. 34, pp. 397-433,Cambridge University Press, London Kosik, K. S., and Finch, E. A. (1987)J. Neurosci. 7,3142-3153 Kyte, J., and Doolittle, R. F. (1982)J. Mol. Biol. 157, 105-132 Lee, G., Cowan, N., and Kirschner, M. (1988)Science 239,285-288 Lewis, S. A., Wan ,D , and Cowan, N. J. (1988)Science 242,936-939 Lindwall, G., and tole, R. D.(1984a)J. Biol. Chem. 269,5301-5305 Lindwall, G., and Cole, R. D. (1984b)J. Biol. Chem. 259,12241-12245 Nieto, A., Montejo De Garcini,E., Correas, I., and Avila, J. (1990)Neuroscience 37. 1m-17n _.,"_

Nukina, N., and Ihara, Y. (1986)J. Biochem. (Tokyo) 99,1541-1544 Peng, I., Binder, L. I. and Black, M. M. (1986)J. Cell Biol. 102,252-262 Raju, K., and Anwar,'R. A. (1987)J . Biol. Chem. 262,5755-5762 Ramachandran, G. N. (1963)in Aspects of Protein Structure, the Proceedings of the International Symposium on Protein Structure and Crystallography m Madras Janua 14 18,1963 (Ramachandran,G. N., ed)pp. 39-54,Academic Press, New Y O X Ramachandran, G. N., and Santhanam, M. S. (1957)Proc. Indian Acud. Sci. A 46,124-134 Ruben, G. C. (1989)J. Elect. Micros. Tech. 13,335-354 Schulz, G. E., and Schirmer, R. H. (1979)in Principles of Protein Structure: $mmgeerAduanced Texts in Chemistry (Canter, C. R., ed) pp. 66-84, Springererlag, New York Serafini-Fracassini, A., Field, J. M., and Delf, B. (1977)Adu. Exp. Med. Biol. A", 2 R7Q-GRl "" I."

Starcher, B. C., Saccomani, G., and Urry, D. W. (1973)Bioehim. Biophys. Acta 310,481-486 Stewart, M. (1977)J. Ultrastruct. Res. 60,27-33 Takahashi, S., and Ooi, T. (1988)Biochem. Biophys. Res. Commun. 150,12441250

Tay&, J. W., and Kaiser, E. T. (1987)Methods Enzymol. 154,473-511 Urry, D. W., Starcher, B., and Partridge, S. M. (1969)Nature 222!795-796 Urry, D.W., Shaw, R. G., and Prasad, K. U. (1985)Biochem.Bmphys.Res. Commun. 130,50-57 Urry, D. W., Haynes, B., andHarris, R.D. (1986)Biochem.Biophys.Res. Commun. 141,749-755 Urry, D. W. (1988)J. Protein Chem. 7,1-34 Urry, D. W. (1990)inProtein Folding: Decipheri the Second Half of the Genetic Code (Gierasch, L., and King, J. A., pp. 63-74, American Association for Advanced Science, Washington, D. C. Urry, D. W., Haynes, B.. Zhang, H., Harris, R. D., and Prasad, K. U. (1988) Proc. Natl. Acud. Scr. U. S. A. 85.3407-3411 .~.~ . " Volpin, D. (1977)Adu. E& Med. B h . 79,119-126 Voter, W. A., and Erickson, H. P. (1982)J. Ultrastruct. Res. 80,374-382 Wischik, C. M., Novak, M., Thogersen, H. C., Edwards, P. C., Runswick, M. J., Jakes, R., Walker, J. E., Milstein, C., Roth, M., and Klug, A. (1988)Proc. Natl. Acud. Sa. U. S. A. 85.4506-4510 Yen. S. H.,Dixon, D. W., Crowe, A., Butler, M., and Sbelanski, M. L. (1987) Am. J. PQthol. 126.81-91

3s)

~

~~

~

~