J. Mol. Biol. (1992) 223, 6’73-682

Metal Ion-induced Conformational Changes of Phosphorylated Fragments of Human Neurofilament (NF-M) Protein M. Hollbsi’,

L. arge I, A. Percze12, J. Kajthrl, I. Teplhn3, L. bvijs, and G. D. Fasman2T

Jr4

‘Institute of Organic ChenGtry, L. EG%iis University H-1518 Budapest 112, P.O.B. 32, Hungary 2Department of Biochemistry, Brandeis University Waltham, MA 02254-9110, U.S.A. 31st Institute of Biochemistry Semmelweis University Medical School H-1444 Budapest 8, P.O.B. 260, Hungary 4The Wistar Institute, 36th Street at Spruce, Philadephia, PA 19104-4268, U.S.A. (Received 29 April

1991; accepted 10 October 1991)

The NF-M subunit of human neurofilaments has a C-terminal repeating 13-mer sequence. The 13-mer (Lys-Ser-Pro-Val-Pro-Lys-Ser-Pro-Val-Glu-Glu-Lys-Gly) (NF-M13) and 17-mer (Glu-Glu-Lys-Gly)-(NF-M13) sequences were synthesized, as were both the monoand diphosphorylated Ser species. Circular dichroism (c.d.) studies and c.d. titrations with A13+ and Ca2+ were performed. The conformation of the phosphorylated and unphosphorylated material was random in water. Deconvolution of the c.d. spectra, in trifluoroethanol, of the untitrated samples yielded a high content of unordered structure, similar to the poly-L-proline II structure. Titration of the phosphorylated species with A13+ or Ca2+ caused a surprising conformational change to occur, yielding a high content of b-pleated sheet structure. A mechanism of metal binding to the phosphofragments is proposed which may be relevant to the formation of neurofibrillary tangles in Alzheimer’s disease.

Keywords: Alzheimer’s

neurofilaments; /l-pleated

circular dichroism; phosphorylated sheets; metal ion titrations

1. Introduction Neurofilaments (NFs$) are abundant in cytoskeletal structures. NF proteins form filaments, via coiled-coil interactions, between structurally conservative “rod” sequences of the high, medium and low J& (NF-H, NF-M and NF-L) subunits. In NF-H and NF-M proteins, long polypeptide sections (400 to 600 residues) follow the homologous rod regions on the C-terminal side (Tokutake, 1990; Geisler et al., 1983; Lewis & Cowan, 1985). It is these regions protruding from the NF backbone that most likely determine the surface properties of NFs, and thus, t Author to whom reprint requests should be sent. 1 Abbreviations used: NF, neurofilament; c.d. circular dichroism; h.p.l.c., high pressure liquid chromatography; n.m.r., nuclear magnetic resonance; TFE, trifluoroethanol; PPII, poly-L-proline II; NFT, neurofibrillary tangles.

NF-M

fragments;

at the molecular level, their tendency to form aggregates. The amino acid sequence of human NF-M has been determined (Myers et al., 1987). The C-terminal domain was found to contain a 13-mer sequence, Lys-Ser-Pro-Val-Pro-Lys-Ser-Pro-Val-Glu-Glu-LysGly (NF-M13), which was repeated contiguously six times. In this ‘i&residue block the Lys-Ser-Pro-Val tetramer occurs 12 times. A related Lys-Ser-Pro-Ala tetramer is tandemly repeated many times in the analogous C-terminal domain of rat NF-H (Robinson et al., 1986). On the basis of circular dichroism (c.d.) studies (Geisler et al., 1983), the rod region of NF proteins adopts an E-helical conformation, while the C-terminal (tail) domains show cd. spectra characteristic of the unordered conformation. According to comparative in vivo and in vitro studies, the N-terminal (head) and C-terminal (tail) domains of NF-M appear to be selectively phosphor-

673 0022-2836/92/030673-10

$03.00/O

0

1992

Academic

Press

Limited

674

Circular

M. Holldsi

dichroism

spectra

of NP-M

fragments

et al.

Table 1 and phosphorylated

XF-Al

(PI

c.d. parameters Sequence 2 KS PVPKS -._ s -

P&H,

iv (nm)

Abbreviation 7

f ra g ments in trijuoroethanol

in TFEt

(A(nm). [S],) WC* bands

1 @ml

WI,

1 @ml

(-6400) ( - 6500)

2045 205

( - 8500) ( - 8800)

-187 -196

(5100) (6400)

(-4000)

105

i-4400)

-191

(3750)

(-7294) ( - 6050)

205 208

(~ 10,510) ( - 6550)

1x9 -190

(6d48) (8350)

(-5200)

207.5

(-5500)

- 189

(5700)

(- 3550)

201

(-5100)

5 equivalents of metal ion. Typical b-pleated sheet spectra could not be measured even at a higher (> 15 times) excess of metal ions. This finding suggests that the Lys-Ser-Pro-Val tetramers preserve their natural conformation and do not undergo an unordered-+/&sheet conformational transition if they are situated between two Ca2+ (A13+)-binding Glu-Glu-Lys-Gly tetramers, as found in the 78 residue block of NF-M. Phosphorylation at Ser7 or both Ser2 and Ser7 positions, however, destroys the conformational stability of the unphosphorylated repeating segments and causes a ppleated sheet-inducing process in the presence of metal ions.

4. Discussion Selectively phosphorylated fragments of the NF-M protein yielded class C c.d. spectra in TFE (Table 1). Class C spectra, as defined by Woody (1974, 1985) show a resemblance to that of an a-helix, except that the band intensities of C spectra are much lower. Class C spectra have also been measured for p-turns whose type I (4i+l=--BO, ~i+l=-30, bi+2=-90, $i+z=O) or type III (~i+l,i+z= -60, $i+r,i+z= -30) character (Venkatachalam, 1968; Smith & Pease, 1980) was shown by X-ray diffraction and/or spectroscopic studies (Woody, 1974, 1985; Smith & Pease, 1980). 31,, helices (4 = - 60, $ = - 30; repeats of type III /?-turns) also give low intensity helix-like cd. spectra (Woody, 1974; Smith & Pease, 1980). On the basis of cd. curve deconvolution, computer-assisted secondary structure prediction and molecular mechanical calculations, the cd. spectrum in NF-Ml3 in TFE (Table l), which is similar to that of NF-M39, can be assigned to an assembly of type I and type III b-turns or a 310 helix located in the C-terminal part (-Val-Pro-LysSer-Pro-Val-Glu-Glu-Lys-) of the molecule (Otvos et For NF-Ml7 (Glu-Glu-Lys-Glyal., 1988). (NF-M/113)), phosphorylated NF-Ml3 and phosphorylated NF-Ml7 fragments, similar c.d. spectra with almost symmetrical negative nn* couplets and a negative m* band near 220 nm were found (Table 1). However, the absolute band intensities in the spectra of Ser7 phosphorylated fragments (NF-Ml3 Ser7P and NF-Ml7 Ser7P) were definitely lower than in the case of the other fragments (Table 1). In aqueous solutions of NF-M13, NF-Ml7 and their monophosphorylated derivatives, the cd. spectra have a strong negative band below 200 nm. This spectrum is similar to that found for the unordered conformation (Woody, 1974, 1985). Ionic forces between H,O-solvated NH3+ and PO3H2 groups are presumably not strong enough to alter the “unordered” conformation of the parent peptides. Ca2+ and A13+ titrations of phosphorylated NF-Ml7 fragments, measured in TFE, showed remarkable changes in the c.d. spectra (Fig. l(a)

Changes

qf (NF-M)

Fragments

677

and (b)). The deconvolution (using the method described by Perczel et al., 1991) of the c.d. titration curves of NF-M Ser2P indicates an increase of the unordered structure component at low Ca2+ excess (75 o/0 at ratio 2), and at high Ca2+ excesses a decrease of the unordered structure component and a simultaneous gradual increase of the B-pleated sheet component ( N 55 o/o unordered structure and N 45 o/e P-pleated sheet at ratio 30). The c.d. titration curves of NF-Ml7 Ser7P with Ca2+ yield similar results to those found with A13+; however, they show significant differences (Fig. 2(a) and (b)). Upon Ca ‘+ titration, the unordered struccomponent increased tural slightly before decreasing while the Al 3+ titration did not exhibit this initial decrease. With the diphospho derivative, NF-Ml7 SerSPSer7, Ca’+ titration first caused the appearance of more of the unordered structure component of the c.d. curves (not shown), as found with the monophosphates. However, at high concentrations of Ca2+ ([Ca”]/[pep]> 3), a gradual increase in the amount of the b-pleated sheet spectrum was observed, peaking at a ratio of N 10. Further increases in Ca2+ concentration causes precipitation. Al 3+ titration of the diphospho derivative, NF-Ml7 SerBPSer’lP shows a continuous increase in the B-pleated sheet component spectrum (not shown). The maximum is reached at [A13’]/[pep]7.5. c.d. spectra of aqueous solutions of SerBP, Ser7P and SerBPSer7P derivatives of NF-M17, all yielded a negative band below 200 nm, indicating the unordered conformation (Woody, 1974, 1985). These spectra do not change upon the addition of metal ions up to ratios of [metal ions]/[pep] - 30. Metal ion binding cannot be excluded even in aqueous solution; however, it is not effective enough to change the backbone conformation of the highly solvated phosphopeptides. TFE is well known to have a helix-promoting effect (Urry et aZ., 1971). On the basis of highresolution n.m.r. experiments (Dyson et al., 1988), TFE appears to stabilize incipient structures previously detected in aqueous solution. c.d. titration curves measured in TFE can be deconvoluted into two spectral components (Fig. 4). Allowing for a contribution of a third subspectrum (3-component deconvolution) did not result in a significant decrease of the error of the deconvolution. This suggests that the secondary structure of phosphorylated NF-M fragments is composed of two major components having differing chiroptical contributions. c.d. spectra marked by a strong negative band near 200 nm are generally expected to reflect the presence of an unordered (or aperiodic) conformation of peptides (Woody, 1974, 1985). However, considering the unexpected red shift (Parrish & Blout, 1971) of the negative bands of the unordered structure spectra in TFE (Figs l(a) and 2(a)), it is reasonable to suppose that a portion of the peptide sequence adopts a poly-n-proline II (PPII) (Ronish & Krimm, 1974; Sasisekharan, 1959; Mandelkern, 1967) conformation instead of an aper-

ill. HoZZdsi et al.

678

Log ([Co2+]/Nf-Ml7

S2IP)l)

(Cl!

- 251 180

I 190

I 200

I I 210 220 Wavelength (b)

/ 230 (nm)

I 240

I 250

I 260

I

Figure 4. Two-component convex constraint deconvolution (Perczel et al., 1991) of c.d. titration curves of NF-Ml7 SerZP with Ca2+. (For selected experimental spectra see Fig. l(a)). (a) Contribution (%) of subspectrum (unordered structure, A) and /?-pleated sheet (0) as a function of log ([Ca2’]/[pepJ). (b) Subspectraof unordered structure (A) and B-pleated sheet (0) results obtained by deconvolution.

one. The intermediate (Ser-Pro-Val-Pro-LysSer-Pro-Val) segment of ?JF-,M17 contains three proline residues. Serine also has a high probability of adopting a PPII conformation (Glover et al., 1983; Tonan et aZ., 1990). An X-ray analysis (Glover et at., 1983) of the avian pancreatic polypeptide, at 098 A resolution (1 A =O.l nm), showed that residues 2 to 8 (Pro-Ser-Gln-Pro-Thr-Tyr-Pro) adopt a PPIT conformation with mean torsion angles d= -72”, $= 140”. The cd. spectrum of PPII shows a strong negative extremum near

iodic

206 nm and a weak positive one above 226 nm (Ronish Ss Krimm, j 974; Sasisekharan: 1959; ,Mandelkern, 1967). In the spectra reported here the positive band is obhterat,ed by the stronger negative c.d. band above 220 nm, of the other P-pleated sheet spectral component. On the basis of its spectral parameters, the deconvoluted cd. curve (Fig, S(b)), can be assigned to segment(s) adopting extended or B-pleated sheetlike conformation. The predominant c.d. feature of b-pleated sheet st,ructures is the appearance of a negative band near 216 nm and a posit,ive band between 1.95and 200 nm (Woody, 1974, 1985). To discriminate the conformational effects Gf binding metal ions to the t’wo glutamic acid residues at positions 10 and 11 in SF-M13 from binding to the serine-phosphate (SerSP or Ser7P) sites, c.d. titrations with Ca2+ were carried out on shorter phosphorylated fragments: Pro-Lys-SerP-Pro-Va’lN-a,cetyi-Pro-Lys-SerP-Pro-6/‘al-PUTH2, and KH,, non-phosphoryla,ted NF-MIS. The first two listed pept,ides contain a serine-phosphate, while KF-M 13 contains a C-terminal Glu-Qlu binding site. The pentapeptide fragments Pro-Lys-SerP-Pro-ValNH, and N-acetyl-Pro-Lys-SerP-Pro-Val-Pu’N, yielded unordered structure type c.d. spect,ra (Table 1) in TFE, which did not change on addition of ten equivalents of Ga’+ . Ca*+ titrations of KF-Ml3 in TFE (Fig. 3) demonstrated the presence of a high-affinity binding site localized in the C-terminal Glu-Glu residues. Presumably, metal ion binding at this binding site is responsible for the class C-+/&pleated sheet spectral change in NF-Ml3 and a,lsofor the unordered structure--+/?-pleated sheet transition in mono- and diphosphorylated fragments of NF-M17. The serinephosphate sites may also have a high-affinity for Ca’+ but binding of Ca2+ to these sites does not result in the appearance of the B-pleated sheet spectra and the predominance of the corresponding $-conformation. 5. Conclusisns Complexing of Ca2+ resulted in a ma.rked change i.n the c.d. spectra of all three phosphorylated fragment,s of NF-Ml7 in TFE. The conformation shifts from an unordered (or poly-L-proline II) structure to one that has a high B-pleated sheet content’. There is, however, an important difference in the Ca2+ dependence of the spectra of these fra,gments. Titration curves (Figs l(a) and 2(a)) and deconvolution data (e.g. Fig. 4) indicate that phosphorylation at Ser2 initially delays the conformational change giving rise to the P-pleated sheet spectra. Moreover, at high Ca2+ excess the a,mount of P-pleated sheet’ component is much lower for &IF-M17 Ser2P than for its Ser7P isomer. The spectra of NF-Ml7 SerBPSer7P show an intermediate Ca2+ sensitivit,y (not shown). On the basis of the Ca2+ titration curves of XF-Ml3 (containing no phosphate binding site) in TFE, the conformationa! change responsible for the

Metal Ion-induced

Conformational

30% PPII.~

P

3 x A13+ 13

11 1

65% PP II, 33

@

I

f$er2

phosphorylation

Repeating turns (or PPII)

1

13

Ser7 phosphorylation hyperphosphorylation

B O

or

a 3

1

6

P

P

35% PPII ,m >I0 x ca*+ >5 x bP+

a 1

3 56

Aggregation a

T

poly-L-Pro

M

P-sheet

-

random @

II (PPII)

Ca*’ or Al ‘+

Figure 5. Proposed mechanism of metal ion-induced b-pleated sheet formation of the C-terminal repeating domain of phosphorylated NF-M protein. Route A, retarding effect of Ser2 (normal) phosphorylation; route B, b-pleated sheet-inducing effect of Ser7 (abnormal). B-Pleated sheet formation of hyperphosphorylation [SerBPSer7P] and the PPII-+B transition induced by metal ions.

Changes

qf (NF-M)

Fragments

679

appearance of a b-pleated sheet spectrum can be accounted for by Ca ‘+ binding to the C-terminal glutamate binding site. Ca2+ complexing gives rise to a rather intensive p-pleated sheet spectrum with a weak negative shoulder at a low [Ca’+]/(pep]2 ratio. It is the competition between the glutamate and phosphate binding sites that explains why, in the case of phosphorylated NF-Ml7 fragments, the second conformational change, resulting in the predominance of the /?-pleated sheet spectrum, occurs only at higher (>5 times) Ca2+ excess. Ca2+ and A13+ titration data suggest that the N-terminal glutamate binding site (Glu-Glu-Lys-Gly) in NF-Ml7 also has a delaying, if not a protecting effect, against P-pleated sheet formation and this effect most likely also manifests itself in the fulllength 78-residue repeating block. It was found herein that Ser2 phosphorylation decreases the sensitivity towards Ca2+ relative to the unphosphorylated sample. The prevailing PPII-like conformation ( > 60%) of NF-Ml7 Ser2P is preserved until a [Ca”]/[pep] -20 ratio is reached. By contrast, phosphorylation at Ser7 results in a steep increase of the P-pleated conformation above a ratio of 2. (For NF-Ml7 SerZPSer7P, the steep increase of the contribution of P-pleated sheet component starts at ratio 3 (not shown)). In a poorly solvating environment (e.g. membranevicinity, simulated by TFE (Urry et al., 1971) in the cd. experiments presented here), Ca2+ binding destabilizes the loosened PPII structure of the phosphorylated NF-Ml7 fragment and, most likely, also in the repeating 13-mer domain of the NF-M protein, phosphorylated in the mid-chain (Ser7) or both Ser2 and Ser7 positions. On the basis of the cd. titration data reported here, Ca2+ and Al3 ’ binding by p hosphorylated NF-Ml7 fragments is accompanied by the adoption of an extended or p-pleated sheet conformation above a critical level of the metal ion concentration, which also depends on the position of the phosphoserine residue (Ser2 or Ser7). A13+ binding brings about a direct increase of the p-pleated sheet conformation and the predominance of this conformation is reached at lower concentrations than in the case of Ca2+ binding. The /?-p leated sheets so formed precipitate on standing. An hypothesis for the mechanism of P-pleated sheet formation is the following: the metal ions bind to phosphate or carboxylate molecules on one polypeptide fragment, neutralizing the negative charge, and then cross-link in the same fashion to a second fragment. This causes chain association which produces the b-pleated sheet conformation, a highly insoluble structure. The main difference between the Ca2+ and A13+-induced backbone extension process of phosphorylated NF-Ml7 fragments is that low Ca” concentrations first cause a decrease in the amount of the extended (p-pleated sheet) conformation. The gradual increase of the /?-content occurs only above a critical [Ca2+] level, which depends on the site and number of phosphoryl group(s) in the molecule

680

M. HolMsi

(Figs l(a) and 2(a)). In the presence of A13+, a rather steep increase of the /? content can be observed for all three phosphorylated fragments (Figs l(b) and 2(b)). The difference in the conformational effect of Ca2+ and A13+ complexing may be accounted for by the enhanced A13+ binding affinity of the glutamate (and presumably phosphate) site(s) and by the differing geometric and thermodynamic conditions of trivalent metal ion complexing (Siegel & Haug, 1983). On the basis of the circular dichroism evidence discussed here, the following model is proposed for the backbone extension to form p-pleated sheets of the repeating domain of NF-M protein (Fig. 5). Phosphorylation in the natural position (Ser2) conformation (Fig. 5, stabilizes a PPII-like route A). Moreover, the phosphoserine residue may act as an internal Ca2+ storage site. By binding Ca2+, it buffers flucbuations of the Ca2+ concentration and delays Ca2+ binding to the glutamate site, which most likely is responsible for the PPII+/3-pleated sheet conformational change. Abnormal (Ser7) and hyperphosphorylation (Fig. 5, route B) of the peptide, by decreasing the stability of the central PPII structure, results in higher P-pleated sheet content (65 to 70%) and an expressed sensitivity towards Ca2+. Thus, a small increase of the Ca2+ concentration can give rise to a predominantly P-pleated sheet conformation of the repeating domain. As pointed out earlier, aluminum is a more potent P-pleated sheet inducer. While a ten times molar excess of Ca2+ induces only a 30 y0 p-pleated sheet conformation in the Ser2P fragment, three equivalences of A13+ increases the b-content to 70%. In abnormally (Ser7) phosphorylated and hyperphosphorylated fragments a relatively small Al”+ excess leads to a prevailing P-pleated sheet structure. “H n.m.r. experiments on the binding of Ca2+ to a 34-residue peptide representing one of the Ca2+ binding sites of the muscle contractile protein troponin C, resulted in B-sheet formation and subsequent association of the Ca2+-bound monomers (Shaw et al., 1990). P-sheet formation induced by multivalent cation binding appears to be a necessary step in the aggregation process of negatively charged (glutamate and aspartate-rich) polypeptide chains. It would be highly desirable to perform n.m.r. experiments to follow this B-pleated sheet formation. Unfortunately the concentrations necessary for n.m.r. studies are much higher than required for cd. studies, and the polypeptides precipitate out on the initial addition of Ca2+ or A13+. Segments of the rod-like region of NF proteins, composed of three subunits (Tokutake, 1990), may be similar to the triple helix of collagen and both may be composed of PPII chains. The PPII conformation appears to be a general structural principle for the C-terminal domains of NF proteins, and on the basis of sequence similarities, also for the long and repeating sequencesof tau and MAP2. On the other hand, NF and MAP proteins are rich in glutamate and aspartate and contain a great number of

et al. Glu-Glu, Glu-Glu-Glu, Glu-Asp (Asp-Glu) or Asp-Gly-Asp units that have a high affinity for multivalent metal ions.“ Their expressed anionic character could place them within the family of Ca2+ binding (buffer and/or storage) proteins Phosphorylation, playing an essential role in the dynamic interactions of neurofilaments during axoplasmic transport (Nixon et al., 1989), enhances their Ca2+ binding potential. Most notably, suggested phosphorylation sites of NF and MAP proteins are embedded in anionic (glutamate and aspartate-rich) domains. Experiments have indicated that no high-affinity Ca2+ binding sites were found in the soluble C-terminal domains of NF-H and NF-M (Lefebvre & Mushynski, 1988). However, phosphoserine residues located in the C-terminal domains of NF subunits appear to have a regulat,ory effect on the accessibility of the high-affinity Ca2+ binding sites in the N-terminal core region. It was also shown that A13+ binds not only to the high but also to t,he lowaffinity binding sites on NFs. In A13+-induced neuropathies, the NFs of neuronal perikarya of animals injected with A13+ are in a hyperphosphorylated state (Bizzi & Gambetti, 1986). Aluminum salts cause argyrophilie accumulations in perikarya of many NB2a( - ) and NB2a( 4 ) neuroblastoma cells (Shea et al., 1989). The Als+-induced direct PPII-+/? pleated sheet conformational change of Ser7P a,nd Ser2PSer7P hyperphosphorylated NF-M fragment,s, reported here, strongly suggests t,hat it is the C-terminal part of NF proteins that is involved in early stages of the aggregation. Perhaps it would not be premature t.o attempt to correlate the results reported here with observations relative to Alzheimer’s and related diseases. The common feature of these neurodegenerative diseases is the accumulations of fibrous proteinaceous structures such as neurofibrillary tangles (NFT) and senile plaques (Tomlinson & Corsellis, 1984). These NFTs contain NF prot,eins as well as other componenm (Tokutake, 1990). These tangles are organized in a cross-b-pleated sheet structure, although none of the components of NFTs is predicted to assumethe P-sheet conformation (&vijs et aZ., 1988; Lee et al., 1988a,6; Lewis et al., 1988). It has been suggested that an imbalance within specific kinases responsible for the phosphorylation of different sites in neurofilaments may be involved in the formation of Alzheimer’s lesions (Sternberger et al., 1985). The selected sequencesof the NF-M, which were synthesized and studied (Table I), allowed the investigation of the role of phosphorylation at t,wo different sites (Ser2 and Ser7). Aberrant phosphorylation of the NF-M protein may lead to an increasing number of SerP residues in the $er7 position, thus causing their aggregation. An electron microscopic study (Troncoso et al,, 1990) shows that the accumulation of neurofilaments may be caused by abnormal levels of multivalent cations. However, &here is a wide variation in the level of the effective eation concentration (0-l mM for Gd 3t, 0.75 rnM for A13+, 10 rnM for Ca,2C

Metal Ion-induced

Conformational

and 30 mM for Mg ‘+. . Troncoso et al., 1990). It has also been observed that some neurofilament proteins become less soluble in the presence of Ca*+ (Day, 1977). These findings agree with the results herein and can be explained satisfactorily by the increased affinity of trivalent cations towards binding sites, whose saturation triggers backbone p-pleated sheet formation and chain association. Multivalent metal ion binding may play a role not only in the backbone extension (or backbone P-pleated sheet formation) but also in promoting aggregation of neighboring b-chains by binding to the carboxylate and phosphate side-chain groups and shielding their repulsion. Cytoskeletal proteins of nerve cells probably consist of large a-helical (Geisler et al., 1983) and poly-L-proline II-like conformational domains. Associations of /?-pleated sheet chains due to conformation-sensitive mono-phosphorylation or hyperphosphorylation and subsequent Ca2+ or A13+ binding, may serve as cores for progressive aggregation of NF proteins or the cross-aggregation of NF and microtubule-associated proteins characteristic of neurofibrillary tangles. The A13+ concentration used herein (@l mM to 1.0 mM) to produce the observed conformational changes may not accurately reflect concentrations of free A13+ in solution. As pointed out by Nixon et al. (1990) the concentration of free A13+ in specific cells is uncertain because of ion selective permeability and specific binding to proteins, and may be higher than that reported for the whole tissue. The concentration of A13+ in a degenerating cell (Per1 & Pendebury, 1986), where the selectivity has been altered, may be higher than in the normal cell. By activating kinases or phosphatases, multivalent metal ions may also play a crucial role in determining the phosphorylation state of NF-M and microtubule-associated proteins. Thus, multivalent cations may affect the process of aggregation in a twofold manner. The model suggested in Figure 5 gives a possible explanation for both the complexity of the tangle-inducing aggregation process and the differing roles that Ca2+ and A13+ are likely to play in it. Although the model suggests that abnormal or hyper-phosphorylation is necessary to cause aggregation, it is also feasible that normal phosphorylated NF proteins may aggregate if the Ca2+ or A13+ concentration is increased in the cell. The observations discussed here may have relevance to the mechanism of tangle formation associated with Alzheimer’s disease. This work was supported by grants from the Il’ational Science Foundation (to G.D.F., NSF DMB-8512570), a grant from the National Institutes of Health (to L.O., GM4501 1) and a National Scientific Research Foundation grant (to M.H., OTKA 1-600-2-88-l-591, Hungary).

References Baudier, J. & Cole, R. D. (1987). Separation different microtubule-associated tau protein

of the species

Changes qf (NF-M)

Fragments

681

from bovine brain and their mode II phosphorylation by Ca2+/phospholipid-dependent protein kinase C. J. Biol. Chem. 262, 17577-17583. Bertolf, R. L. (1987). CRC Crit. Rev. Clin. Lab. Sci. 25, 195-210. Bizzi, A. & Gambetti, P. (1986). Acta NeuropathoE. 71, 154-158. Crapper, McLachlan, D. R. (1986). Neurobiol. Aging, 7, 525. Day, W. A. (1977). Solubilization of neurofilaments from central nervous system myelinated nerve. J.

Ultrastruct.

Res. 60,

362-372.

Dyson, H. J., Rance, M., Houghten, R. A., Wright, P. E. & Lerner, R. A. (1988). Folding of immunogenic peptide fragments of proteins in water solution. J. Mol. Biol. 201, 201-217. Elgavish, G. A., Hay, D. I. & Schlesinger, D. H. (1984). ‘H and 31P nuclear magnetic resonance studies of human salivary statherin. Int. J. Pept. Protein Res. 23, 23&234. Geisler, N., Kaufmann, E., Fischer, S., Plessmann, U. & Weber, K. (1983). Neurofilament architecture combines structural principles of intermediate filaments with carboxyl-terminal extensions increasing in size between triplet proteins. EMBO J. 2, 1295-1302. Glover, I., Haneef, I., Pitts, J., Wood, S., Moss, D., Tickle, I. & Blundell, T. (1983). Conformational flexibility in a small glolular hormone: X-ray analysis of avian pancreatic polypeptide at, 0.98 A resolution. Biopolymers,

22, 293-304.

Greenfield, N. & Fasman, G. D. (1969). Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry, 8, 4108-4116. Hoshi, M., Nishida, E., Miyata, Y., Sakai, H., Miyoshi, T., Ogawara, H. & Akiyama, Y. (1987). Protein kinase C phosphorylates tau and induces its functional alterations. FEBS Letters, 217, 237-241, Khachaturian, Z. S. (1989). The role of calcium regulation in brain aging: re-examination of a hypothesis. Aging, 1, 17-34. Kosik, K. S., Orecchio, L. D., Bakalis, S., Duffy, L. & Neve, R. L. (1988). Partial sequence of MAP2 in the region of a shared epitope with Alzheimer neurofibrillary tangles. J. Neurochem. 51, 587-598. Ksiezak-Reding, H., Dickson, D. W., Davies, P. & Yen, S.-H. (1987). Recognition of tau epitopes by antineurofilament antibodies that bind to Alzheimer neurofilament tangles. Proc. Nat. Acad. Sci., U.S.A. 84, 3410-3414. Lee, G., Cowan, N. & Kirschner, M. (1988). The primary structure and heterogeneity of tau protein from mouse brain. Science, 239, 285-288. Lee, V. M.-Y., &vb;s, L., Jr, Carden, M. J., Ho116si, M., Dietzschold, B. & Lazzarini, R. A. (1988a). Identification of the major multiphosphorylation site in mammalian neurofilaments. Proc. Nat. Acad. Sci., U.S.A. 85, 1998-2002. Lee, V. M.-Y., &vSs, L., Jr, Schmidt, M. L. & Trojanowski, J. Q. (1988b). Alzheimer disease tangles share immunological similarities with multiphosphorylation repeats in the two large neurofilament proteins. Proc. Nat. Acad. Sci., U.S.A. 85, 7384-7388. Lefebvre, S. & Mushynski, W. E. (1987). Calcium binding to untreated and dephosphorylated porcine neurofilaments. Biochem. Biophys. Res. Commun. 145, 1006-1011. Lefebvre, S. & Mushynski, W. (1988). Characterization of

682

M.

Holldsi

the cation-binding properties of porcine neurofilaments. Biochemistry, 27, 8503-8508. Lewis, S. A. & Cowan, N. J. (1985). Genetics, evolution, and expression of the 68,000-mol-wt neurofilament protein: isolation of a cloned cDNA probe. J. Cell. Biol.

100,

843-850.

Lewis, S. A., Wang, D. & Cowan, N. J. (1988). protein MAP2 shares Microtubule-associated microtubule binding motif with tau protein. Science; 242, 936-939. Mandelkern, L. (1967). Poly-r-proline. In Poly-&Amino Acids (Fasman; G., ed.), pp. 675724, Marcel Dekker Inc., New York. Mohan, M. S. & Abbott, E. H. (1978). Metal complexes of amino acid phosphate esters. Inorg. Chem. 17, 220332207. Myers, M. W., Lazzarini, R. A., Lee, V. M.-Y., Schlaepfer, W. W. & Nelson, D. L., (1987). The human mid-size neurofilament subunit: a repeated protein sequence and the relationship of its gene to the intermediate filament gene family. EMBO d. 6; 1617-1626. Nixon, R. A., Lewis, S. E., Dahl, D., Marotta, C. A. & Drager, K. C. (1989). Early postranslational modifications of the three-neurofilament subunits in mouse retinal ganglion cells: neuronal sites and time course in relation to subunit polymerization and axial transport. Mol. Brain Res. 5, 93-108. Nixon, R. A., Clarke, J. F., Logvinenko, K. B., Tan, M. K. H., Hoult, M. 8: Grynspan, F. (1990). Alumina inhibits calpain-mediated proteolysis and induces human neurofilament proteins to form proteaseresistant high molecular weight complexes. J. Neurochem. 55, 1950-1959. Nukina, N., Kosik, K. S. & Selkoe, D. J. (1987). Recognition of Alzheimer paired helical filaments by monoclonal neurofilament antibodies is due to crossreaction with tau protein. Proc. Nat. Acad. Sci., U.S.A. 84, 3415-3419. &viis, IL., Jr, Hollosi, M., Perczel, A., Dietzschold, B. & Fasman, G. D. (1988). Phosphorylation loops in synthetic peptides of the human neurofilament protein middle-sized proteins. J. Protein Chem. 7, 365376. otviis, L., Jr, Tangoren, I. A., Wroblewski, K., Hollosi, M. & Lee; V. M.-Y. (1990). Reversed-phase high performance liquid chroma,tograph separation of synthetic phosphopeptide isomers. J. Chromatogr. 512,

265-272.

Parrish, J. R. Jr & Blout, E. R. (1971). Spectroscopic studies of random chains and a-helical polypeptides in hexafluorisopropanol. Biopolymers, 10, 1491-1512. Perczel, A., Tusnady, G., Hollosi, M. & Fasman, G. D. (1991). Convex constraint, analysis: a natural deconvolution of circular dichroism curves of proteins. Protein Eng. 4, 669-679. Perl, D. P. & Pendebury, W. W. (1986). Aluminum neurotoxicity-potential role in the pathogens of neurofibrillary tangle formation. Canad. J. Neurol. sci. 13, 441445. Prevelige, P. Jr & Fasman, G. D. (1987). Structural st,udies of acetylated and control inner core histones. Biochemistry, 26, 2944-2955. Robinson, P. A., Wion, D. & Anderton, B. H. (1986). Isolation of a cDNA for the rat heavy neurofilament polypeptide (NF-H). FEBS Letters, 209. 203-205. Ronish, E. W. & Krimm, S. (1974). The calculated circular dichroism of polyproline II in the Edited

by P.

et al.

poiarizability approximation. 16351651. Sasisekharan, V. (1959). Structure Acta

Crystallogr.

Biopolymers,

13;

of poly-n-proline

IF.

12; 897-903.

Shaw, 61. S., Hodges, R. S. & Sykes, B. D. (1990). Calcium-induced peptide association to form intact protein domain: ‘H NMR structural evidence, Science,

249,

280-283.

Shea, T. B.; Clarke, J. F., Wheelock, T. R., Paskervich, P. A. & Nixon, R. A. (1989). Aluminum salts induce the accumulation of neurofilaments in perikarya of NBBa/dl neuroblastoma. Brain Res. 492, 53-64. Siegel, N. & Haug. A. (1983). Aluminum interaction with calmodulin: evidence for altered structure and function from optical and enzymatic studies. Biochim.

Biophys.

Acta,

744,

3tX5.

Sihag, R. K. & Nixon, R. 9. (1990). Phosphorylation of the amino-terminal head domain of the middle molecular mass 145-kDa subunit of neurofilaments. J. Biol. Chem. 265, 41664171. Sloboda, R. D., Rudolph, S. A., Rosenbaum, J. L. & Greengard, P. (1975). Cyclic AMP-dependent endogenous phosphorylation of a microtubuleassociated proteins. Proc. Nat. Acad. Xci., U.S.A. 72: 1777181. Smith, F. A. & Pease, L. G. (1980). Circular dichroism of biological membranes. I. Mitochondria and red blood cell ghosts. CRC Crit. Rev. Biochem. 8; 315-399. Sternberger, N. H., Sternberger, L. A. & Ulrich, J. (1985). gberrant neurofilament phosphorylation in Alzheimer’s disease. Proc. Nat. Acad. Sci., U.S.A. 82, 4274-4276. Tokutake, S. (1990). On the assembly mechanism of neurofilaments. Int. J. Biochem. 22, l-6. Tomiinson, B. E. & Corsel!is; J. A. N. (1984). Aging and Dementia. In @reen$eld’s Neuropathology (Adams, J. H.; Corsellis, J. A. N. & Duchen, L. W.; eds), 4th edit,., pp. 951, Arnold; London. Tonan, K., Dawata, Y. & Hamaguchi, K. (1999). Conformations of isolated fragment,s of pancreatic polypeptide. Biochemistry, 29, 4424-4429. Troneoso, J. C., March, J. L., HLner, M. & Aebi, U. (1990). Effect of aluminum and other multivalent cations on neurofilaments in vitro: an electron microscopy study. J. Struct. Biol. 103, 2-12. Urry, D. W., Masotti, L. & Krivacis, J. R. (1971). Circular dichroism of biological membranes. I. Mitochondria and red blood cells. Biochim. Biophys. Acta, 2 600-612. Venkatachalam, C. M. (1968). Stereochemical criteria for polypeptides and proteins. V. Conformation of a system of three linked peptide units. Biopolymers, 6, 1425-1436. Woody, R. W. (1974). Studies of theoretical circular dichroism of polypeptides: contributions of b-burns. In Peptides, PoEypeptides and Proteins, (Blout? E. R., Bovey, F. A., Goodman, M. & Lotan, M.; eds), pp. 338-350, Wiley, New York. Woody, R. W. (1985). Circular dichroism of peptides. In The Peptides (Hruby, V. J., ed.), vol. 7, pp. 15-l 14, Academic Press, Orlando. Pamauchi, T. & Fujisawa, H. (1982). Phosphorylation of microtubule-associated protein 2 by calmodulindependent protein kinase (kinase II) which occurs only in the brain tissue. Biochem. Biophys. Res. Commun. 109, 975-981.

E.

Wright