Functional Studies of Alzheimer's Disease Tau Protein

Home , Neurofibrillary tangle, Tau protein

The Journal of Neuroscience, February 1993, 13(2): 508415

Functional Studies of Alzheimer’s Disease Tau Protein

Qun Lu and John G. Wood Department of Anatomy and Cell Biology, Emory University School of Medicine, Atlanta, Georgia 30322

In vitroassays were used to monitor and compare the kinetic concentration at which pure tubulin assembles(Weingarten et behavior of bovine tubulin polymerization enhanced by tau al., 1975; Cleveland et al., 1977). In cultured fibroblasts, which proteins isolated from Alrheimer’s disease (AD) and nonde- do not contain endogenoustau, microinjected tau can incor- mented (ND) age-matched control brains. Tau from AD cases porate into microtubules and stabilize them against depoly- induced slower polymerization and a steady state turbidity merization conditions (Drubin and Kirschner, 1986; Lu and value approximately 50% of that stimulated by tau from con- Wood, 1991b). trol cases. Tau from the most severe AD case was least In brain, tau is largely localized in axons (Binder et al., 1985; effective at promoting polymerization. Dark-field light mi- Brion et al., 1988). However, in AD brain tau becomes an croscopy of the control samples revealed abundant micro- integral part of paired helical filaments (PHFs) in neurofibrillary tubule formation and many microtubule bundles. Microtubule tangles of neuronal cell bodies as well as dystrophic neurites assembly was observed in AD samples as well, but bundling associatedwith neuritic plaques (Brion et al., 1985; Grundke- was not obvious. These results were confirmed by negative- Iqbal et al., 1986, 1988; Kosik et al., 1986; Wood et al., 1986). stain electron microscopy. Morphological analysis showed This dislocation is accompanied by abnormal phosphorylation that AD tau-induced microtubules were longer than control (Grundke-Iqbal et al., 1986; Wood et al., 1986; Iqbal et al., microtubules. Furthermore, our initial results suggest that 1989; Brion et al., 199la; Lee et al., 199 1). Recently, it was the reduction of AD tau activity is correlated with neurofi- found that A68, a putative AD-specific protein (Wolozin et al., brillary pathology in AD brains. Earlier reports indicated that 1986), is an abnormally phosphorylated tau and is the only AD tau is modified by phosphorylation (Grundke-lqbal et al., component of a classof PHFs (Lee et al., 1991). The tubulins, 1988; Wood et al., 1988; lqbal et al., 1989; Brion et al., 1991 a,b; the main building block of microtubules from AD brain, are Lee et al., 1991). Our results support the hypothesis that tau functionally competent to reassembleand form microtubules modification compromises its function by altering its ability in vitro, but microtubule assemblyfrom AD brain homogenates to nucleate and bundle microtubules. is not observed (Iqbal et al., 1986). Thus, modification of tau [Key words: Alzheimer’s disease, tau protein, microtubule may be responsiblefor the defective microtubule assembly,as assembly, functional alteration, kinetic assay, light and elec- phosphorylated bovine tau was found to be less efficient in tron microscopy] promoting tubulin polymerization than dephosphorylated tau (Lindwall and Cole, 1984). Although it is reasonableto hy- pothesize that the alteration of tau distribution and chemistry Microtubules are the fundamental organelle for fast axonal contributes to the disruption of neuronal microtubule integrity transport, which is essentialfor the renewal of axons and mem- and the formation of AD pathology, it has not been established branes in the nerve terminal. Defective microtubule assembly whether AD tau, before it transforms into PHFs, is still func- and stabilization in neurons, therefore, could lead to impaired tional. In this study, we used in vitro assaysincluding kinetic axonal transport and abnormal synaptic transmission.The sta- analysisand subsequentdark-field light microscopy and electron bility of microtubules in neuronscan be achieved in a number microscopy to addressthe questionswhether AD tau in soluble of ways, including tubulin posttranslational modification and form is functionally competent to promote microtubule for- the regulation of microtubule-associatedproteins (MAPS) of mation and whether the microtubules thus formed expressdis- either the high-molecular-weight MAPS or the low-molecular- tinct behavior and morphology when compared with that stim- weight tau proteins (for review, seeMatus, 1988; Mitchison and ulated by ND tau. The results show that AD tau can stimulate Kirschner, 1988). microtubule assembly but with slower kinetics and different Tau proteins are a heat-stablefamily of developmentally reg- microtubule morphology, which supports the hypothesis that ulated phosphoproteinsthat are generatedby alternative splic- abnormal tau is involved in modification of the neuronal mi- ing of a singlegene (Drubin et al., 1984; Lee et al., 1988; Himm- crotubule system and AD pathogenesis. let-, 1989;Kosik et al., 1989). Zn vitro, tau stimulatesthe assembly of microtubules at tubulin concentrations well below the critical Materials and Methods Protein puriJicution. Bovinetubulin was isolated through temperature- dependent microtubule polymerization-depolymerization cycles and Received May 4, 1992; revised July 2 1, 199 1; accepted July 23, 1992. further purified by DEAE-Sephacelion exchangecolumn chromatog- We thank Dr. L. Binder for providing mAb Tau- 1, and Ms. J. Soria and Mr. raphy (Detrich. 1986). Tau was isolated using a modified method (Pol- R. Gopal for technical assistance. This work was supported by NIH Grants AG lock and Wood, 1988) of perchloric acid extraction of heat-stable frac- 06383, AG 11123, and NS 27847. tions (HSF) by Baudier et al. (1987). Fresh half AD or ND brains with Correspondence should be. addressed to J. G. Wood at the above address. averagepostmortem times of 7 hr were homogenizedin cold 20 mM Copyright 0 1993 Society for Neuroscience 0270-6474/93/130508-08$05.00/O 2-[N-morpholinolethanesulfonic acid (MES) buffer, pH 6.85 with 2 mM The Journal of Neuroscience, February 1993, 73(2) 509 ethylene glycol bis(@aminoethyl) ether-N,N,N’,N’-tetra-acetic acid (EGTA), 1 mM MgSO,, 0.75 M NaCl, and 1 mM fi-mercaptoethanol. A B Protease inhibitors leupeptin (10 pg/ml), pepstatin A (10 pg/ml), apro- tinin (125 KIU/mk KIU = Kallikrein inhibitorv unit), and 2 mM phen- vlmethylsulfonyl fluoride (PMSF) were added fresh to the buffer before 4 homogenization. The 75,000 x g supematant of brain homogenates was boiled at 95°C for 5 min in the same buffer with 0.75 M NaCl and 2 mM m-dithiothreitol (DTT) to generate HSF. Proteins other than tau 4 were precipitated by perchloric acid treatment and removed by cen- trifugation. Tau was also separated from high-molecular-weight microtubule-associated protein MAP-2 by molecular sieve chromatography 4 of the HSF on a Bio-Gel A 1.5 M column. Tau protein used for kinetic experiments was concentrated using an ultrafiltration unit (Amicon Corp., Danvers, MA) to 3-5 mg/ml and centrifuged at 45,000 x g for 20 min to remove aggregates before experiments. Proteinconcentrations were 4 determined using a Bio-Rad protein assay based on the Bradford method (1976). Western blot analysis was performed as described previously (Wood et al., 1986). Quantitative analysis of the bands and the ratio of 4 tau to tau fragments was performed using a Bio-Rad 620 video densi- tometer. Putholonv. Hippocampal slices from AD and ND brains were fixed in fresh PEP (p&formaldehyde, lysine, sodium periodate, and NaCl) mixture of McLean and Nakane ( 1974). Vibratome sections, 50 Brn, of hippocampus were stained using‘a modified Sevier-Munger silver im- pregnation method (Sevier and Munger, 1965). The occurrences of tangles and plaques were counted primarily in CA1 and CA4 subdivision of hippocampi at 200 x under the Zeiss Axiovert 35 light microscope. Figure 1. Western Blots of the proteins isolated for in vitro assays. A, The densities of tangles and plaques were determined by averaging the Bovine tubulin immunostained with monoclonal anti-a tubulin DM 1A, data from total area of 2.1-10.8 mm2 with statistical analysis performed B, isolated human tau immunoblotted with mAb Tau- 1. Gel lanes were using CRICKET GRAPH software on the Apple Macintosh IIci computer. loaded with 4.2 pg total protein. Molecular weight standards are indi- Kineticanalysis. DEAE-Sephacel tubulin was mixed with AD or ND cated on the right (arrowheads);from top to bottom: 205 kDa, 116 kDa, tau at 4°C and polymerization was initiated in 0.1 M piperazine-N,N’- 66 kDa, 45 kDa, and 36 kDa. bis[2-ethanesulfonic acid] (PIPES) with 2 mM EGTA and 1 mM MgSO,, pH 6.72, by adding 1 mM guanosine 5’-triphosphate (GTP) and raising the temperature to 37°C. The kinetic behavior was monitored by re- Western blots of typical samplesof bovine tubulin and human cordine turbiditv changes at 350 nm o.d. using a Beckman DU-64 UV tau usedfor kinetic analysis.The bovine tubulin wasdevoid of spectrgphotomeier with a Kinetics Soft-Pa& module and temperature controller. microtubule-associatedproteins asjudged by Coomassieblue- Light andelectron microscopy. After the kinetic measurements were stained gels (data not shown). AD and ND tau were isolated completed, aliquots were taken from cuvettes for dark-field light mi- basedon the heat-stableand perchloric acid-soluble properties croscopic examination. In addition, 5 ~1 aliquots were dropped onto of tau proteins. Human tau proteins isolated by our method Formvar-coated grids and negatively stained with 2% uranyl acetate for electron microscopy. Some samples were fixed in 0.2% glutaraldehyde contained the major isoforms at 55-70 kDa MW with some prior to electron microscopic (EM) examination. For thin-section elec- degradation products. These bands were reactive with Tau-l tron microscopy, aliquots were centrifuged at 45,000 x g for 30 min monoclonalanti-tau (a gift generouslyprovided by Dr. L. Binder at 32°C to nellet microtubules. These samples were then fixed in 0.2% at the University of Alabama, Birmingham) and polyclonal anti- glutaraldehyde, postfixed in 1% OsO,, and embedded for thin sectioning. tau (Sigma, St. Louis, MO). MAP-2 was not detected on Coo- Sections were examined and photographed using a Jeol 100’ CX II transmission electron microscope. massieblue-stained gels or on Western blots (data not shown). Microtubulelength determination. Microtubule numbers and lengths Densitometry was employed to compare the amount of tau were determined first using thin-section electron microscopy. The ratio isoforms and their proteolytic fragments among casesand we of long microtubules versus short microtubules (LMT/SMT) was de- found no significant differencesby statistical analysis (t test, p termined on photographic prints. The numbers shown are averaged > 0.01, n = 4). values f SEM. Long microtubules were arbitrarily defined as microtubules longer than 0.5 pm. In more recent studies, to obtain more Above the critical concentration, tubulin can self-polymerize global views of the microtubule population, microtubule lengths were in vitro in an appropriate buffer with addition of GTP and raising measured on the photographs taken under dark-field light microscopy. the temperature to 37°C (Weingarten et al., 1975). When mi- In this case, single microtubules were counted and thick bundles of crotubules are formed, they scatter light and the solution be- microtubules were omitted because it was not possible to determine the number and length of individual microtubules in the bundles. A mi- comesviscous. The light scattering can be measuredquantita- crotubule length distribution was generated and statistical analysis per- tively by recording the turbidity changesat 350 nm o.d. When formed using CRICKET GRAPH and cANvAs0 software on the Apple ND tau and AD tau are mixed with DEAE-Sephacel tubulin Macintosh II ci computer. and polymerization is initiated as described above, a microtubule assemblykinetics recording is obtained (Fig. 2A). This Results figure comparesone casepair of AD and ND tau for their ability D@erential kinetic activity of AD and ND tau to induce to stimulate microtubule assembly,and it showsthat AD tau microtubule assembly can induce tubulin polymerization but at a slowerrate and less To study whether AD tau is capableof stimulating microtubule effectively than ND tau. In fact, the level of polymerization at assembly,it is necessaryto use purified tubulin and tau in the steady state for the AD sample was only 68% of that for the assembly assays.Fresh bovine brain microtubules were ob- ND control. Figure 2B summarizeskinetic experiments testing tained through temperature-dependent polymerization-depo- a total of eight cases(five AD samplesand three ND controls) lymerization cycles,and tubulin wasfurther purified by DEAE- for their tau activities. Again, AD tau appeared to be able to Sephacelion exchangecolumn chromatography. Figure 1 shows induce tubulin to polymerize when compared with tubulin self- 510 Lu and Wood - Functional Studies of Alzheimer’s Tau Protein

0.8 - ND tau 1.0 - ADtau A Figure 2. Turbidimetric assayillus- trating the kinetic behavior of bovine tubulin polymerizationinduced by AD 0.8 tau and age-matchedND control tau. -E 0.6 A, One pair of AD and ND tau was c testedfor their activity to promotemi- 0.6 crotubuleassembly monitored by tur- 5: bidity changesat o.d. 350nm. Squares, pJ 0.4 z ND tau activity; triangles, AD tau ac- - tivity. [tubulinj, 2.99 &g/ml; [tau], 0.2 . - 0.4 mg/ml.B, Summaryof the tau activity 9 ti for eight AD and ND cases.Open 0 0.2 d squares, ND tau activity andeach point is an averagevalue from three cases; solid squares, AD tau activity and each pointis an averagevalue from five cases; triangles, tubulinself assembly and each 0.0 L:f10 20 30 40 0 2 4 6 8 1012141618202224262830 point representstwo independentex- periments. Time (Minutes) Time (Minutes) polymerization. However, their activities were greatly reduced and the mean microtubule lengthswere 2.04 & 0.17 pm for ND to an average value for all casesof 50.57% of that for ND microtubulesversus 3.58 f 0.41 wrn for AD microtubules.Thus, controls. These results suggestedthat the AD tau ability to in- microtubule length analysisfrom both EM and dark-field light duce microtubule assembly in vitro was altered quantitatively microscopic data demonstrated a correlation of higher ampli- but not qualitatively. tude of assemblyto shorter microtubule lengths (Fig. 4). A rea- sonable interpretation of these results is that when there are Microtubule morphology and behavior finite tubulin moleculesavailable for polymerization, a greater Although recording of turbidity changesenables us to measure number of microtubule seedsare formed and stabilized in the quantitatively and therefore differentiate AD tau and ND tau presenceof more assembly-competentND tau moleculesso that for their abilities to induce tubulin polymerization, alternative at steady state there are more microtubules of shorter size. assaysare required to demonstrate microtubule formation and Dark-field light microscopy revealed another interesting dif- structure. In earlier studies,microtubules were pelleted by cen- ference between AD and ND tau-induced microtubule forma- trifugation after the kinetic measurementwas completed and tion. Extensive bundling or grouping of microtubules was ob- then prepared for thin-section electron microscopy. Figure 3 served in ND controls (Fig. 5A,C), but this was rarely seenin showsthat both ND and AD tau were able to induce microtubule AD samples(Fig. 5&D). Negative-stain electron microscopy formation. The ultrastructure of the individual microtubules confirmed the bundling in ND controls, and also confirmed from ND control and AD samplesappeared similar. However, minimal microtubule bundling in AD samples(Fig. 7A,B). large numbers of microtubules in ND controls were of relatively short size (Fig. 3B), while self-assembledmicrotubules were Tau activity in vitro is correlated with Alzheimer pathology significantly longer (Fig. 3A). When the ratio of long microtu- We then examined the relationship between the reduction of bules to short microtubules (LMT:SMT) was correlated with AD tau activity and the severity of the disease.The presence the amplitude of assembly(represented as the percentagechange of neurofibrillary tanglesand neuritic plaquesin the hippocam- of turbidity at the steady state of tubulin polymerization; for pus, the amygdala, and frontal and temporal cortices is the ND tau, the activity was assignedas 100% turbidity changein hallmark pathology in AD and is correlated with the degreeof each paired experiment), the ND tau control had a very low dementia (Rewcastle, 199 1). We selectedhippocampus to mea- LMT:SMT ratio (Fig. 4A). Interestingly, AD tau-induced mi- sure tangle and plaque densities in 7-10 areas each of -0.3 crotubule populations showed an LMT:SMT between ND con- mmZ. The average value was compared against the amplitude trols and self-assembly(Figs. 3C, 4A), which agreeswith the of microtubule assemblyobtained as describedearlier. Table 1 kinetic data. This observation is consistent with other work shows the correlation of tangle and plaque density with the showing that under steady state conditions of assembly, MAP- amplitude of microtubule assembly. The ability of tau protein depletedmicrotubules were longer than MAP-rich microtubules to stimulatemicrotubule assemblywas correlated with the tangle (Farrell et al., 1987). density such that a greater number of tangleswas reflected in To obtain a more global view of microtubule populations, lower tau activity. dark-field light microscopy was employed to comparethe length distribution of microtubules assembledin the presenceof AD Discussion or ND tau. When aliquots taken from the cuvettes where tubulin In this article we have demonstratedthat AD and ND tau exhibit polymerization had reached steady state were examined, nu- differential abilities to stimulate microtubule assembly,and ND merous microtubules were observed in both ND controls and tau-induced microtubules are both shorter and more bundled AD samples(Fig. 5). However, calculation of microtubule length than thoseinduced by AD tau. Furthermore, initial comparisons distribution indicated that ND controls had more, but shorter, suggestthat the reduction of AD tau activity is correlated with microtubules than AD samples(Fig. 6). Five casesin two paired an increase in neurofibrillary tangle density in AD brain hip- experiments were studied for microtubule length distribution, pocampus. The results predict consequencesfor microtubule The Journal of Neuroscience, February 1993, 73(2) 511

0.8

iO.6 2 SO.4

Amplitude of MT Assembly

10% AD 52% AD 100% ND Amplitude of MT Assembly Figure 4. Histogramsshowing the relationshipbetween microtubule (MT) lengthand amplitude of microtubuleassembly. A, Microtubule lengthratio correlatedwith amplitudeof microtubuleassembly. LMTI SMT, the ratio of longmicrotubule number to shortmicrotubule number. Longmicrotubules are arbitrarily definedas longer than -0.5 pm. The microtubulenumbers and lengthswere determinedon electron micrographprints. B, Mean of microtubulelength correlated with amplitude of microtubuleassembly. The microtubulelengths were mea- suredon dark-fieldlight micrographprints. All the datashown here are averagedvalue f SEM. The amplitudeof assemblyis representedas the percentagechange of turbidity at steadystate of tubulin polymerization, and the amplitudeof assemblyfor ND tau activity is assigned to 100%for eachpaired experiment. The LMT:SMT for self-assembled Figure 3. Thin-sectionelectron micrographs of the assembledmicro- microtubules(SA) is assignedas 1. tubules.A, Self-polymerizedmicrotubules. B, ND tau-inducedmicro- tubuleformation. C, AD tau-inducedmicrotubule formation. Self-polymerizedmicrotubules are long. Note that the ultrastructureof ND and of tau-related fragments isolated from PHFs was reduced. The AD microtubulesappeared similar; however, the majority of ND mi- prediction is further supported by the observations of Ellisman crotubulesare of relatively shortsize while the AD microtubulefraction containsboth longand short microtubules. Scale bar, 0.3 pm. et al. (1987) who used three-dimensionalreconstruction from serialsections of biopsy tissueto showa paucity of microtubules in the cytoplasm of tangle-bearingneurons. In the presentstudy, stability, axonal transport functions, and potential mechanisms we provide evidence that directly points to AD tau as the cause for PHF formation in AD brain. of the defective microtubule system in AD brain. This conclu- The assemblydata show that the AD tau function of pro- sion is strengthenedby our useof a defined in vitro systemthat moting microtubule assemblyis compromisedcompared to ND usedonly purified bovine tubulin and AD or ND tau proteins. tau. A likely consequenceis that the microtubule system in The reduced ability of AD tau to stimulate microtubule as- disease-affectedneurons would be defective. This prediction is sembly and the subsequentpaucity of microtubules in affected supported by studiesof Iqbal et al. (1986). They showed that neurons would have direct consequencesfor the cell. Axons microtubule reassemblyfrom AD brain homogenateswas not require microtubule-basedfast axonal transport for renewal of observed, whereasneurofilaments could be reassembled.They membrane components. The efficient operation of this system further showed that tubulins from AD brains were competent requires a relatively stablepopulation of microtubules. The un- to form microtubules with the addition of DEAE-Dextran. Nie- stablepopulation predicted from our results and those of other to et al. (1990) showed that HSFs, prepared from AD brains groupswould directly and adversely affect axonal transport pro- and incubated with purified porcine tubulin, generatedprotein cesses.Over time and with significant diminution in microtu- aggregatesand that microtubules were rarely observed. Nieto et bule number, the neuron would reach a state of inability to al. (1991) further demonstratedthat the binding to microtubules maintain distal axonal processes.The cell would degenerate Figure 5. Assembled microtubule populations visualized by dark-field light microscopy. A and C show the morphology of assembled microtubule population induced by ND tau. Note that in A there are both single microtubules and microtubule bundles (arrowheads). In C, there are many short microtubules grouped together (arrowheads) with some long microtubules. B and D show the morphology of assembled microtubules induced by AD tau. Note that in B and D there are many microtubules (arrowheads) but microtubule bundles similar to A and C are not seen. Scale bar. 10 pm.

101 :A Mean= 3.91 f 0.23 urn B Mean= 1.64 + 0.1 I urn :8- z ‘5 . -3 P . 3 20- 36. 4 0 2 ’ O4- b E * E lo- e G2- 9 If 0 I 0 I 0123456789’ 0123456789 Length Cum ) Length (urn 1

10 Mean= 2.29 f 0,35 urn a.tn D Mean= 3.272 0.43 urn

0 u 0123456789’ 1 0123456789 3 Length (urn > Length (urn 1 Figure 6. Histograms showing examples of length distribution of microtubules measured from dark-field light microscopy at steady states. A, Self- assembled microtubules. B, ND tau-induced microtubules. C and D, AD mu-induced microtubules. The Journal of Neuroscience, February 1993, 13(2) 513

Table 1. Relationship between amplitude of microtubule assembly and Alzheimer cytoskeletal pathology

Ampli- tude* (MT as- Clinical Case Sex/age Tangles= Plaquesa sembly) diagnosis 1 F/55 - - 100 ND 2 M/75 - - 100 ND 3 M/92 - - 100 ND 4 F/77 6 z!c0.80 9 f 0.17 68 AD 5 F/85 4 k 2.02 31 -t 8.13 52 AD 6 M/64 59 -t 0.85 22 rfr 0.90 8 AD nAD cytoskeletal pathology: shown asnumber of tanglesorplaques in hippo-pi persquare millimeter + SEM,-, veryfew/not detected. b Amplitude of microtubule (MT) assembly is represented as percentage change ofturbidity at the steady state ofmicrotubule assembly; amplitude ofMT assembly for ND tau activity is assigned as 100%.

of their normal interaction with microtubules, may accumulate Figure 7. Ultrastructureof assembledmicrotubules visualized by electron microscopywith negativestaining. A, ND tau-inducedmicrotu- as free proteins where they would be available to self-associate bulesshowing distinct bundling; B, AD tau-inducedmicrotubules. Note as polymers. Over time and with additional phosphorylation, that the microtubulebundling is not obviousalthough there are possible these polymers could progressively alter into the final SDS- associationsbetween adjacent microtubules (arrowhead). Scalebar, 0.5 insolubleneurofibrillary tangle. It should be noted that the func- rm. tional deficit in AD tau may be more pronounced than we have demonstrated. Although we make every effort to ensure that those processes,thus severing the functional connection of the our starting brain material is selected to allow enrichment in disease-affectedneuron to its ensembleof neurons forming a tau from regions of abundant pathology, our samplesundoubt- circuit involved in mediating cognitive function and behavior. edly contain some normal tau from nonaffected neurons.This The observation that soluble tau not yet incorporated into would shift the kinetic data toward that observed for the control solubleor insolublePHFs has compromisedability to assemble samples,so that the significant decreasein functional activity microtubules has implications for AD pathogenesis.Neurofi- we observe may actually be greater for the modified tau. brillary tangles from AD brains are highly insoluble even in We useddark-field light microscopy and EM to demonstrate strong detergents such as SDS. Greenberg and Davies (1990) that microtubules assembledin the presenceof AD tau were isolated a distinct group of PHFs soluble in SDS buffer and longer than self-assembledmicrotubules but shorter than mi- sharing similarities with tau proteins. Lee et al. (199 1) directly crotubules assembledby ND tau. This supports the hypothesis demonstratedthat A68, a putative AD-specific protein (Wolozin that AD tau function is altered since, at steady stateconditions, et al., 1986), is a modified form of normal tau and is the major MAP-deficient microtubules are generally longer than MAP- (or only) component of a group of SDS-soluble PHFs. At pres- rich microtubules (Farrell et al., 1987). ent, we do not know how normal tau, which is solublein aqueous We also presentedevidence that AD tau may be lessefficient solution, transforms into SDS-insoluble PHFs, but the presence at bundling microtubules than ND tau. In addition to its role of several intermediates, including the tau reported in this ar- in microtubule assemblyin vitro (Weingarten et al., 1975; Cleve- ticle, suggestsa sequenceof tau modifications. However, Bram- land et al., 1977), tau can bind to and stabilize microtubules in blett et al. (1992) suggestedthat soluble tau in AD brain may cultured RAT- 1 cells (Drubin and Kirschner, 1986) and CHO be transformed to A68 rapidly without involvement of an in- cells (Lu and Wood, 1991b). Transfected cells overexpressing termediate stage. We have previously demonstrated that the tau exhibit distinct bundles of microtubules that are highly re- solubletau isolated by our methods is modified by phosphor- sistant to microtubule depolymerizing drug treatments (Kanai ylation at the site recognized by the Tau-1 antibody (Pollock et al., 1989; Lewis et al., 1989; Basset al., 1991; Knops et al., and Wood, 1988). Lee et al. (1991) have shown that their SDS- 1991). The functional consequencesof microtubule bundling solublePHF tau is modified by phosphorylation on at least two induced by tau is presently unclear, but it could play a role in acceptor sites.Thus, a plausiblemodel for PHF formation would microtubule stabilization in axons. This is particularly possible include a seriesof phosphorylation events at several aberrant at the initial segment where microtubules are tightly packed acceptor sites (Grundke-Iqbal et al., 1986; Wood et al., 1986; (Peters et al., 1991). We are currently usingmicroinjection pro- Iqbal et al., 1989; Brion et al., 1991a; Kopke-Secundo et al., tocols to investigate the properties of AD and ND tau with 1991; Lee et al., 1991). It may be of interest that the soluble respect to microtubule bundling in living cells. Regardlessof tau we are studying is apparently an early modified form, yet is the ultimate significanceof microtubule bundling to neuronal already deficient in its functional ability to assemblemicrotu- function, the present results offer one more property whereby bules. This property likely affects microtubule function as elab- AD tau is altered in its functional capacity. orated above, but other offshoots are possibleas well. One par- Our initial results suggestthat the ability of tau to stimulate ticularly meaningfulpossibility isthat AD tau proteins, deprived microtubule assembly is negatively correlated with the degree 514 Lu and Wood - Functional Studies of Alzheimer’s Tau Protein of cytoskeletal pathology, particularly the neurofibrillary tangle. Farrell KW, Jordan MA, Miller HP, Wilson L (1987) Phase dynamics Although this is clearly a satisfying trend, additional cases will at microtubule ends: the coexistence of microtubule length changes and treadmilling. J Cell Biol 104: 1035-1046. be required to illustrate the detailed specifics of this correlation. Greenberg S, Davies P (1990) A preparation of Alzheimer paired In summary, we have shown that a soluble form of AD tau helical filaments that displays distinct tau proteins by polyacrylamide is less efficient than ND tau at promoting microtubule assembly. gel electrophoresis. Proc Nat1 Acad Sci USA 87:5827-583 1. This AD tau is in an early stage of modification that is most Grundke-Iqbal I, Iqbal K, Tung YCH, Quinlan M, Wisniewski HM, likely an aberrant phosphorylation event at one or more sites. Binder LI (1986) Abnormal phosphorylation of the microtubule- associated protein tau in Alzheimer cytoskeletal pathology. Proc Nat1 A predicted consequence supported by available evidence is that Acad Sci USA 83~49 13-49 17. axonal microtubules would be unstable. This in turn would Grundke-Iqbal I, Vorbrodt AW, Iqbal K, Tung YCH, Wang GP, Wis- adversely affect fast axonal transport processes critical to sur- niewski HM (1988) Microtubule-associated polypeptides tau are vival of the neuron. Consistent with the kinetic data, micro- altered in Alzheimer naired helical filaments. Mol Brain Res 4:43- 52. tubule morphology and behavior are different in the presence Himmler A (1989) Structure of the bovine tau gene: alternative spliced of AD compared to ND tau. An additional functional conse- transcripts generate a nrotein familv. Mol Cell Biol 9: 1389-l 396. quence possible from these observations is that tau not com- Iqbal K, Grundke-Iqbal I, Zaidi T, Merz PA, Wen GY, Shaikh SS, petent to promote assembly may exist as a free monomer that Wisniewski HM (1986) Defective brain microtubule assembly in could self-associate to form polymer. In a stepwise fashion ac- Alzheimer’s disease. Lancet Aug 23:421-426. Iqbal K, Grundke-Iqbal I, Smith AJ, George L, Tung YCH, Zaidi T companied by additional phosphorylation events, the polymer -( 1989) Identification and localization of ar peptideto paired helical could be converted into the SDS-insoluble neurofibrillary tangle. filaments of Alzheimer disease. Proc Nat1 Acad Sci USA 86:5646- Experiments to address the latter possibility are underway. 5650. Kanai Y, Takemura R, Oshima T, Mori H, Ihara Y, Yanagissawa M, Masaki T, Hirokawa N (1989) Expression of multiple tau isoforms References and microtubule bundle formation in fibroblasts transfected with a Bass PW, Pienkowski TP, Kosik KS (199 1) Processes induced by tau single tau cDNA. J Cell Biol 109: 1173-l 184. expression in Sf 9 cells have an axon-like microtubule organization. Knops J, Kosik KS, Lee G, Pardee JD, Cohen-Gould L, McConlogue J Cell Biol 115:342a. L (199 1) Overexpression of tau in a nonneuronal cell induces long Baudier J, Lee S-H, Cole RD (1987) Separation of the different mi- cellular processes. J Cell Biol 114:725-733. crotubule-associated tau protein species from bovine brain and their Kopke-Secundo E, Grundke-Iqbal I, Iqbal K (1991) Abnormal phos- mode II phosphorylation by Ca++/phospholipid-dependent protein phorylation of tau is one of the earliest events in Alzheimer neuro- kinase C. J Biol Chem 262:17584-17590. fibrillary pathology. Sot Neurosci Abstr 17: 1070. Binder LI, Frankfurter A, Rebhun LI (1985) The distribution of tau Kosik KS, Joachim CL, Selkoe DJ (1986) Microtubule-associated pro- in the mammalian central nervous svstem. J Cell Biol 10 1: 137 l- tein tau is a major antigenic component of paired helical filaments 1378. in Alzheimer’s disease. Proc Nat1 Acad Sci USA 83:4044-4048. Bradford MM (1976) A rapid and sensitive method for the quanti- Kosik KS, Orecchio LD, Bakalis S, Neve RL (1989) Developmentally tation of microgram quantities of protein utilizing the principle of regulated expression of specific tau sequences. Neuron 2: 1389-1397. protein-dye binding. Anal Biochem 72~248-254. Lee G, Cowan N, Kirschner M (1988) The primarv structure and Bramblett GT, Trojanowski JQ, Lee VM-Y (1992) Regions with abun- heterogeneity oftau protein from‘mouse brain: Science 239:285-288. dant neurofibrillary pathology in human brain exhibit a selective Lee VM-Y. Balin BJ. Otvos Jr L. Troianowski JO (1991) A 68: a reduction in levels of binding-competent 7 and accumulation of ab- major subunit of paired hehcal’filaments and h&vatized forms of normal T-isoforms (A68 proteins). Lab Invest 66:212-222. normal tau. Science 25 1:675-678. Brion JP. Passareiro JP. Nunez J. Flament-Durand J (1985) Mise en Lewis SA, Ivanov IE, Lee GH, Cowan NJ (1989) Organization of evidence immunolbgique de la’proteine tau au niveau des’lesions de microtubules in dendrites and axons is determined by a short hydro- degenerescence neurofibrillaire de la maladie d’Alzheimer. Arch Biol phobic zipper in microtubule-associated proteins MAP2 and tau. Na- (Bruxells) 95229-235. ture 342:498-505. Brion JP, Guilleminot J, Couchi D, Flament-Durand J, Nunez J (1988) Lindwall G, Cole RD (1984) Phosphorylation affects the ability of tau Both adult and juvenile tau microtubule-associated proteins are axon protein to promote microtubule assembly. J Biol Chem 259:5301- specific in the developing and adult rat cerebellum. Neuroscience 25: 5305. 139-146. Lu Q, Wood JG (199 1) Properties of fluorescently derivatized bovine Brion JP, Hanger DP, Bruce MT, Couck A, Flament-Durand J, An- tau protein. J Cell Biol 115:384a. derton BH (199 1a) Tau in Alzheimer neurofibrillary tangles: N- and Matus-A (1988) Microtubule-associated proteins: their potential role C-terminal regions are differentially associated with paired helical in determining neuronal morphology. Annu Rev Neurosci 11:2944. filaments and the location of a putative abnormal phosphorylation McLean IW, Nakane PK (1974) Periodate-lysine-paraformaldehyde site. Biochem J 273:127-133. fixative. A new fixative for immunoelectron microscopy. J Histochem Brion JP, Hanger DP, Couck A, Anderton BH (199 1b) A68 proteins Cytochem 22: 1077-1083. in Alzheimer’s disease are composed of several tau isoforms in a Mitchison T, Kirschner MW (1988) Cytoskeletal dynamics and nerve phosphorylated state which affects their electrophoretic mobilities. growth. Neuron 1:761-772. Biochem J 279:831-836. Nieto A, Montejo de Garcini E, Correas I, Avila J (1990) Character- Cleveland DW, Hwo S-Y, Kirschner MW (1977) Purification of tau, ization of tau protein present in microtubules and paired helical fil- a microtubule-associated protein that induces assembly of microtu- aments of Alzheimer’s disease patient’s brain. Neuroscience 37: 163- bules from purified tubulm. J Mol Biol 116:207-225. 170. Detrich III HW (1986) Isolation of sea urchin ege,-- tubulin. Methods Nieto A, Correas I, Lopez-Otin C, Avila J (199 1) Tau-related protein Enzymol 134:128-138. present in paired helical filaments has a decreased tubulin binding Drubin D, Kirschner MW (1986) Tau function in living cells. J Cell capacity as compared with microtubule-associated protein tau. Bio- Biol 103:2739-2746. chim Biophys Acta 1096: 197-204. Drubin D, Kirschner MW, Feinstein S (1984) Microtubule-associated Peters A. Palav S. Webster HD ( 199 1) The fine structure of the nervous tau protein induction by nerve growth factor during neurite outgrowth system, 3d kd.‘New York: Oxford ‘UP. in PC 12 cells. In: Molecular biology of the cytoskeleton (Borisy G, Pollock NJ, Wood JG (1988) Differential sensitivity of the microtu- Cleveland D, Murphy D, eds), pp 343-355. Cold Spring Harbor, NY: bule-associated protein, tau, in Alzheimer’s disease tissue to formalin Cold Spring Harbor Laboratories. fixation. J Histochem Cytochem 36: 1117-l 12 1. Ellisman M, Raganathan R, Deerinck T, et al. (1987) Neurofibrillary Rewcastle NB (199 1) Degenerative disease of the central nervous sys- cytoskeleton and endomembrane system organization in Alzheimer’s tem. In: Textbook of neuropathology (Davis RL, Robertson DM, eds), disease. Adv Behav Biol 34~6 l-73. pp 904-9 18. Baltimore: Williams and Wilkins. The Journal of Neuroscience, February 1993, 13(2) 515

Sevier AC, Munger BL (1965) A silver method for paraffin sections Wood JG, Mirra SS, Pollock NJ, Binder LI (1986) Neurofibrillary of neural tissue. J Neuropathol Exp Neurol 24: 130-l 35. tangles of Alzheimer disease share antigenic determinants with the Weingarten MD, Lockwood AH, Hwo SY, Kirschner MW (1975) A axonal microtubule associated protein, tau. Proc Nat1 Acad Sci USA protein factor essential for microtubule assembly. Proc Nat1 Acad Sci 83:40404043. USA 72:1858-1862. Wolozin BL, Pruchnicki A, Dickson DW, Davies P (1986) A neuronal antigen in the brains of Alzheimer patients. Science 232:648-650.