transglutaminase cross-linking of the τ protein

Journal ofNeurochemistryLippincott-Raven Publishers, Philadelphia© 1995 International Society for Neurochemistry

Transglutaminase Cross-Linking of the T Protein

Michael L. Miller and Gail V. W. Johnson

Department of Psychiatry and Behavioral Neurobiology, University ofAlabama at Birmingham,

Abstract : Tissue transglutaminase (EC 2.3.2.13) is a cal-cium-activated enzyme that cross-links specific substrateproteins into insoluble, protease-resistant, high molecularweight complexes. Because the neurofibrillary tangles inAlzheimer disease have similar biochemical characteris-tics, and because the microtubule-associated protein Tis the predominant component of these structures, thesubstrate properties of Twith respect to transglutaminasewere investigated . Bovine T and recombinant human Tisoforms rapidly form high molecular weight, cross-linkedpolymers on incubation with transglutaminase . Poly-amine incorporation assays indicate that bovine T is anexcellent substrate of transglutaminase, with a KR, of 10.4± 2.2 pM and a Vm. of 40.9 ± 4.5 nmol/mg of enzyme/min . Individual recombinant human T isoforms are notequivalent with respect to transglutaminase, as the small-est isoform T3 (352 amino acids) is not as good a sub-strate as the larger isoforms T4 (383 amino acids) andT4L (441 amino acids) . To determine which segments ofthe T protein are susceptible to modification by transglu-taminase, T was labeled with [3H]putrescine by transglu-taminase and proteolyzed with a-chymotrypsin, and thebreakdown products were analyzed . These experimentsdemonstrate that the enzyme modifies T at only one ora few discrete sites, primarily in the carboxyl half of themolecule. Thus, the reaction is specific for only a smallnumber of the many glutamine residues in T. Furthermore,a T deletion construct (T264) containing a portion of themicrotubule-binding domains, which is a substrate oftransglutaminase, cannot be cross-linked by the enzyme.This provides evidence that the cross-linking reaction isspecific, and requires that the substrates be appropriatelyassociated for cross-linking to occur . Key Words: Alzhei-mer disease-Neurofibrillary tangle-Calcium-Poly-amines .J. Neurochem. 65,1760-1770 (1995) .

One of the primary hallmarks of Alzheimer diseaseis the accumulation of neurofibrillary tangles in thesoma of affected neurons in the neocortex and hippo-campus (Goedert et al., 1991a ; Goedert, 1993) . Neu-rofibrillary tangles are composed of paired helical fil-aments (PHFs), which are resistant to proteolysis(Kondo et al ., 1988 ; Wischik et al ., 1988 ; Jakes et al .,1991), and insoluble in harsh solvents, denaturants,and reducing agents (Selkoe et al ., 1982b) . This char-

Birmingham, Alabama, U.S.A .

1760

acteristic insolubility of the PHFs in neurofibrillarytangles has been of interest to researchers for sometime . More than 10 years ago, it was proposed thatthese insoluble PHFs arise from abnormal covalentcross-linking of constituent proteins at nondisulfide-bonded amino acid residues (Selkoe et al ., 1982a ;Miller and Anderton, 1986) . Although it was origi-nally believed that PHFs were composed of neurofil-aments (Gambetti et al ., 1981 ; Sternberger et al .,1985), it has subsequently been determined that PHFsare composed primarily, if not entirely, of an abnor-mally phosphorylated form of the T protein, known asPHF-T (Kosik et al ., 1986 ; Wood et al ., 1986 ; Goedertet al ., 1988 ; Greenberg and Davies, 1990 ; Lee et al .,1991) .T is a family of closely related microtubule-associ-

ated phosphoproteins found primarily in the axons andcell bodies of neurons (Papasozomenos and Binder,1987) . The six T isoforms found in adult human brainare generated by alternative mRNA splicing of a singlegene and contain either three or four microtubule-bind-ing domains in the carboxy-terminal half of the mole-cule (Goedert et al ., 1989) . In addition, mature iso-forms can contain inserts of 29 or 58 amino acids nearthe amino terminus (Goedert et al ., 1989) . Normally,T is a soluble protein that is involved in the polymeriza-tion and stabilization of microtubules (Drubin andKirschner, 1986 ; Knops et al ., 1991) . However, inAlzheimer disease, 7- becomes localized to the somato-dendritic compartment as well as the axon, is abnor-mally phosphorylated, loses its ability to polymerizemicrotubules efficiently (Lu and Wood, 1993), andassembles into PHFs and insoluble neurofibrillary tan-gles (for a review, see Goedert, 1993) . The mecha-nisms and interrelationships of these events are un-

Received November 9, 1994; revised manuscript received April5, 1995 ; accepted April 11, 1995 .

Address correspondence and reprint requests to Dr . G. V. W.Johnson at Department of Psychiatry, SC 1011, 1720 7th Ave. South,University of Alabama at Birmingham, Birmingham, AL 35294-0017, U.S.A .

Abbreviations used: BSA, bovine serum albumin; DMC, caseinN,N-dimethylated ; PHF, paired helical filament ; PMSF, phenylmeth-ylsulfonyl fluoride ; SDS, sodium dodecyl sulfate; TCA, trichloroace-tic acid.

TRANSGLUTAMINASE CROSS-LINKING OF THE T PROTEIN

known . Protease digestion studies have indicated thatneurofibrillary tangles consist ofan insoluble PHF corethat contains the carboxyl 40% of the T protein and a"fuzzy outer coating" around the core, formed by theamino terminus of T projecting away from the core(Wischik et al., 1988 ; Ksiezak-Reding and Yen,1991). It is also possible to isolate sodium dodecylsulfate (SDS)-soluble PHFs from Alzheimer diseasebrains, and these PHFs are also composed of PHF-T(Greenberg and Davies, 1990 ; Lee et al ., 1991) . Thesesoluble PHFs may be uncondensed precursors of theinsoluble PHFs found in neurofibrillary tangles(Crowther et al ., 1992 ; Goedert et al ., 1992) ; however,the mechanisms underlying such a conversion to theinsoluble state are unknown .One enzyme that has the potential to be involved in

the aggregation of 7 - and its conversion into insolubleneurofibrillary tangles is tissue transglutaminase (EC2.3.2.13) . The transglutaminases are a family of cal-cium-activated enzymes that cross-link substrate pro-teins into high molecular weight complexes that areresistant to proteolysis and insoluble in detergents, de-naturants, and reducing agents (Folk and Finlayson,1977) . Specific transglutaminases are involved in thecomification of skin, hair, and nails, the cross-linkingof fibrin in clot formation, and the formation of cata-racts in the eye lens (for a review, see Greenberg etal ., 1991) .

Tissue transglutaminase is primarily a cytosolicenzyme found in various organ-specific cells . It hasbeen immunocytochemically localized within neurons(Miller and Anderton, 1986), and is abundant withinnerve terminals (Gilad and Varon, 1985; Facchiano etal ., 1993) . Significant transglutaminase activity hasbeen demonstrated in the frontal and temporal neocor-tex, hippocampus, cerebellum, and white matter of hu-man brain (Selkoe et al ., 1982a) . Tissue transglutami-nase is an inducible enzyme that is up-regulated bothduring neuronal development (Gilad and Varon, 1985 ;Maccioni and Seeds, 1986) and in apoptosis, or pro-grammed cell death (Fesus et al ., 1987, 1989; ElAlaoui et al ., 1992), and it has recently been shownto modify synapsin I, another neuron-specific phospho-protein (Facchiano et al ., 1993) . As transglutaminasecatalyzes the formation of E-(y-glutamyl)lysine iso-peptide bonds between substrate proteins rather thandisulfide bonds, the resulting cross-linked protein com-plexes are insoluble (Folk and Finlayson, 1977), likethe neurofibrillary tangles . Indeed, the idea that trans-glutaminase may be involved in the formation of neu-rofibrillary tangles is not a novel one . However, theearly transglutaminase studies, performed before it wasknown that tangles were composed primarily of T, fo-cused on neurofilaments as the potential substrate andproduced equivocal results (Selkoe et al ., 1982a ;Miller and Anderton, 1986) .

Considering these previous findings, the extent towhich T is a substrate for transglutaminase was investi-gated. A preliminary study from this laboratory sug-

EXPERIMENTAL PROCEDURES

1761

gested that T is able to be cross-linked by transglutami-nase (Dudek and Johnson, 1993) ; however, substrateproperties of T were not determined . In this study itis demonstrated that T is an excellent substrate fortransglutaminase, based on both kinetic experimentsand biochemical analysis . Although all of the T iso-forms investigated are cross-linked by transglutami-nase, kinetic studies demonstrate that they are notequivalent substrates of the enzyme . Furthermore, evi-dence is presented that transglutaminase modifies T atonly a small number of specific glutamine residues, asdetermined by proteolytic experiments with recombi-nant human T isoforms . The primary site or sites modi-fied by transglutaminase occur in the carboxy half ofthe T molecule.

MaterialsGuinea pig liver tissue transglutaminase, casein N,N-di-

methylated (DMC), bovine serum albumin (BSA), putres-cine dihydrochloride, and a-chymotrypsin were purchasedfrom Sigma (St . Louis, MO, U.S.A .) ; phenylmethylsulfonylfluoride (PMSF) was purchased from Boehringer Mannheim(Indianapolis, IN, U.S.A .) ; and [1,4(N)- 3H]putrescine di-hydrochloride (1 mCi/ml) was purchased from Amersham(Arlington Heights, IL, U.S.A .) . EWHANCE autoradiogra-phy enhancer for tritium was purchased from Du Pont (Bos-ton, MA, U.S .A .), UniverSol scintillation fluid was pur-chased from ICN (Costa Mesa, CA, U.S.A .), and the En-hanced Colloidal Gold Total Protein Detection Kit waspurchased from Bio-Rad (Hercules, CA, U.S.A.) . r frombovine brains was purified as previously described (Johnsonet al., 1989) . Recombinant human T isoforms and a deletionconstruct containing amino acids 264-386 (T264 ; aminoacid numbering based on the largest human T isoform ; Goed-ert et al ., 1989) were generously provided by C . Scott ofZeneca, Inc . (Scott et al., 1991, 1992) . The isoforms usedin this study contained either three (T3), or four (T4), mi-crotubule-binding domains in the carboxy-terminal half ofthe molecule, or four microtubule-binding domains and a 58amino acid insert near the amino terminus (T4L) (Goedertet al ., 1989) . Autopsy samples of the prefrontal cortex of10 Alzheimer disease patients and nine age-matched controls(male and female, ages 57-88 years) were obtained fromthe UAB Brain Resource Program .

AntibodiesThe following T antibodies were used in this study : 5E2,

T46.1, T14, and Tau-1 (Kosik et al ., 1988), Alz-50 (Goedertet al ., 1991b), Tau-2 (Watanabe et al ., 1992), and 7C11and 8C11 (Vigo-Pelfrey et al ., 1995) . Other antibodies in-cluded the transglutaminase antibody CUB 7402 (Birck-bichler et al ., 1985), and a monoclonal antibody to ubiquitin,purchased from Zymed . The titers used are indicated in thetext .

Immunoblotting of samples from human brainSamples of autopsy tissue from the prefrontal cortex of

Alzheimer disease patients and age-matched controls werehomogenized in Laemmli sample buffer without dye, soni-cated, and incubated in a boiling water bath for 5 min. Pro-tein concentrations were determined using the method ofLowry et al . (1951) , after acid precipitation of the protein .

J. Neurochem., Vol. 65, No. 4, 1995

1762

Samples were then further diluted into Laemmli buffer withor without 8 M urea, and equivalent amounts of proteinwere run on an 8% SDS-polyacrylamide gel, transferred ontonitrocellulose (Towbin et al ., 1979), and immunoblottedwith antibodies to the T protein or to ubiquitin.

7 cross-linkingPurified bovine T, human T isoforms T3, T4, and T4L,

and the construct T264 (2 .5 /W) were cross-linked individu-ally by tissue transglutaminase (0.25 pM) in a buffer con-taining 50 mM Tris (pH 8.6) and 5 mM CaC12 (see Fig. 1,reaction A) . Samples were incubated at 37°C for the timesindicated, and the reactions were stopped by the addition ofLaemmli sample buffer. Samples were placed in a boilingwater bath for 5 min, run on an 8% SDS-polyacrylamidegel, transferred onto nitrocellulose (Towbin et al ., 1979),and immunoblotted with antibodies to T, as indicated. T264was visualized by colloidal gold staining, as specified by themanufacturer's instructions . Micromolar concentrations ofthe T isoforms were determined from their predicted molecu-lar masses, and an average molecular mass of 42,000 daltonswas used for bovine T (Himmler, 1989) .

Putrescine incorporation assayThe specificity of the transglutaminase reaction was deter-

mined by measuring the incorporation of ['H] putrescine intopotential substrates (see Fig. 1, reaction B) . Reactions wereperformed in triplicate, in a buffer containing 50 mM Tris(pH 8.6), 1 mM CaC12, 20 mM dithiothreitol, and 12.5%glycerol (modified from Achyuthan and Greenberg, 1987) .Initial experiments contained 250 pM unlabeled putrescine(Achyuthan and Greenberg, 1987) ; however, it was discov-ered that a high degree of T self-association and cross-linkingwas inhibiting putrescine incorporation, and the concentra-tion of unlabeled putrescine was subsequently increased to1 mM (see below) . This higher concentration of putrescine


M. L. MILLER AND G. V. W. JOHNSON

FIGA . Calcium-dependent reactions oftransglutaminase. Schematic diagramof two competing transglutaminase re-actions . Transglutaminases are calcium-dependent enzymes that react with freeglutamine (Q) residues on substrateproteins (box in upper left). On forminga complex with the glutamine residueand releasing ammonia, the enzyme-substrate intermediate reacts with anappropriate, nearby primary amine. Thiscan be eitherthe e-amino groupof lysineon a nearby protein, resulting in cross-linking of the two proteins (reaction A),or the primary amino group of a poly-amine, such as putrescine, resulting inpolyamine incorporation into the sub-strate (reaction B) .

was used in all assays, for both T and DMC. Transglutami-nase (0.25 MM, final concentration) and [3H]putrescine (2pCi) were added to each tube and the samples were incu-bated at 37C for 10 min. The reactions were stopped byadding an equal volume of 50% trichloroacetic acid (TCA),and proteins were precipitated on ice for 15 min. Sampleswere collected on Whatman GF/B filters and rinsed with5% cold TCA. Filters were dried and the incorporation of[ 3H]putrescine into the proteins was determined by liquidscintillation spectrometry . Putrescine incorporation, initialvelocities, and stoichiometries of incorporation were calcu-lated for each protein concentration, and data were analyzedusing the Michaelis-Menten equation. Results obtained forbovine T were compared with the results of the same assayperformed on DMC, one of the best transglutaminase sub-strates known to date (Facchiano et al ., 1993). All datawere analyzed using the Student's t test, and values wereconsidered significantly different when p < 0.05.

Proteolysis of radiolabeled -rTo determine which segments of the T molecule are sus-

ceptible to modification by transglutaminase, T was labeledwith [ 3H ]putrescine by transglutaminase, proteolyzed, andthe breakdown products were analyzed . Human T isoformsT3 and T4 (0 .1 pg/pl) were incubated at 37°Cwith transglu-taminase (0.25 pM) and [3H]putrescine (250 pCi/ml, finalconcentration), in the same buffer used for the putrescineincorporation assay, above. After 30 min, a-chymotrypsin(1 /cg/ml, final concentration), which preferentially cleavesproteins on the carboxyl side of aromatic amino acids (Tyr,Phe, and Trp), was added to the reaction and proteolysis wasallowed to proceed for an additional 30 min. As a control,transglutaminase alone (0.25 pM) was incubated with a-chymotrypsin under the same conditions . Reactions werestopped by the addition of PMSF (2 mM, final concentra-tion) and Laemmli sample buffer . Samples were incubated


FIG. 2. High molecular weight "smearing" of T in Alzheimerdisease brain homogenates and in bovine T cross-linked withtransglutaminase . Immunoblot of bovine T incubated in the pres-ence (+) or absence (-) of transglutaminase, and whole-ho-mogenate samples of prefrontal cortex from control (C) andAlzheimer disease (AD) brains, probed with antibodies to T. Bo-vine T wasincubated with or without tissue transglutaminase for30 min, and aliquots containing 0.6 wg of T were loaded onto an8% polyacrylamide gel. Whole-homogenate samples of humanprefrontal cortex in Laemmli sample buffer, from control (male,age 64 years) and Alzheimer disease (male, age 66 years) brainswere loaded onto the same gel (30 wg of total protein) . Theresultant immunoblot was probed with an antibody cocktail con-taining Tau-2 (1 :4,000), 7C11 (1 :3,000), and 8C11 (1 :10,000),as Tau-2 recognizes bovine T but not human T (Watanabe etal ., 1992), and 7C11 and 8C11 recognize human T selectively(Vigo-Pelfrey et al ., 1995). The high molecular weight smearing,seen at -100-150 kDa in the Alzheimer disease homogenate,is similar in appearance to bovine 7, which has been cross-linkedby transglutaminase .

in a boiling water bath for 5 min and run on a 12.5% SDS-polyacrylamide gel. Gels were stained with Coomassie Bril-liant Blue, impregnated with EWHANCE autoradiographyenhancer, dried, and exposed to x-ray film. Aliquots of thesame samples were also transferred onto nitrocellulose asabove and probed with antibodies to T or transglutaminase.The same experiments were also performed with unlabeledputrescine, and the resulting blots were immunolabeled ordouble labeled with one or more antibodies to r, as indicated.

RESULTS

High molecular weight aggregates of T inAlzheimer disease frontal cortexImmunoblots of human prefrontal cortex homoge-

nates, from Alzheimer disease patients and age-matched controls, were probed with antibodies to the7 protein . T immunoreactivity indicated that the pre-frontal cortex of Alzheimer disease patients containsa form of 7 that migrates as a high molecular weight"smear" of immunoreactivity above the bands of mo-nomeric T (Fig . 2) . Although the smearing is generallyseen only in Alzheimer disease brain (Fig . 2), it isoccasionally found, to a much lesser extent, in certain

1763

samples from age-matched controls (data not shown) .Such a smearing of T immunoreactivity in Alzheimerdisease was reported previously (Morishima-Kawa-shima et al ., 1993), and a subset of this T was report-edly conjugated with ubiquitin . When the immunoblotsfrom control and Alzheimer disease brain were probedwith a monoclonal antibody to ubiquitin, high molecu-lar weight ubiquitin-reactive bands and "smears"were also observed, but the extent to which the ubiqui-tin immunoreactivity overlapped the 7 immunoreactiv-ity was minimal (data not shown) . However, the highmolecular weight smearing of T in Alzheimer diseasehad an appearance similar to bovine T, which had beenincubated in vitro with transglutaminase (Fig . 2, andalso see Dudek and Johnson, 1993) . Although thisresult does not necessarily implicate transglutaminasein Alzheimer disease, the intriguing similarity led usto investigate the extent to which 7 is a substrate fortransglutaminase .

Putrescine incorporation assayAs shown in Fig . 1, transglutaminase catalyzes two

competing nucleophilic displacement reactions, pro-tein-protein cross-linking and the incorporation ofpolyamines into protein substrates . In the presence ofexcess polyamines, transglutaminase will catalyze theincorporation of a polyamine (reaction B), rather thana protein-bound lysine (reaction A) into a substrateprotein .To determine the extent to which T is a substrate of

transglutaminase, the incorporation of ['H ] putrescineinto purified bovine T was measured. Initial measure-ments resulted in kinetic values for 7 that were lowerthan expected . Further analysis revealed that a highdegree of T self-association resulted in transglutami-nase-mediated cross-linking, a reaction that interferedwith the incorporation of radiolabeled putrescine (seeFig . 1) . To overcome this competing cross-linking re-action, the concentration of unlabeled putrescine wasincreased from 250 jLM to 1 mM in the putrescineincorporation assay . Corresponding immunoblots indi-cate that most of the cross-linking is inhibited at thisconcentration of putrescine (data not shown) . Com-plete inhibition of T cross-linking by polyamines maynot be possible, however, as samples incubated withconcentrations of putrescine as high as 5 mM stillcontain detectable amounts of dimeric and polymeric7 (data not shown) .

Increasing the concentration of putrescine from 250yM to 1 mM greatly improved the kinetic data for T.The amount of putrescine incorporated into T (10 MM)by transglutaminase more than doubled (from 0.249- 0.029 nmol at 250 MM, to 0.568 ± 0.048 nmol at1 mM, p < 0.001) with this experimental modification .A plot of initial velocity versus substrate concentrationfor four potential transglutaminase substrates (Fig . 3)indicates that 7 and DMC are both good substrates fortransglutaminase and that myoglobin and BSA are notsubstrates . Thus, the transglutaminase-catalyzed incor-

J. Neurochem ., Vol. 65, No. 4, 1995

1764

poration of putrescine is limited to specific substrateproteins .A Hanes-Woolf plot (Fig . 4) of data from the pu-

trescine incorporation assays indicates that T is indeedan excellent substrate of transglutaminase, having a Kmof 10.4 ± 2.2 pM, and a Vmax of 40.9 ± 4.5 nmol/mg

FIG. 4. T is an excellent substrate for transglutaminase . Hanes-Woolf plot of data from the putrescine incorporation assay. Bo-vine T is comparable with DMC as a substrate for transglutami-nase (n = 6-7 separate experiments; representative datashown) . Km values are micromolar ; Vmax values are expressedas nanomoles per minute per milligram of transglutaminase .



FIG. 3. The transglutaminase-mediatedincorporation of putrescine is a specificreaction . Plot of initial velocity versus sub-strate concentration data from the putres-cine incorporation assay. Transglutami-nase substrates, like bovine T (9) andDMC (0), have long, linear stretches ofamino acids that include at least one ac-cessible glutamine residue (n = 6-7 sepa-rate experiments) . BSA(X) and myoglobin(0) are primarily globular in nature and arenot substrates of the enzyme (n = 3-4separate experiments) .

of enzyme/min . These values are very similar to thevalues obtained for DMC (5 .5 ± 0.5 uM and 35.8± 1 .5 nmol/mg of enzyme/min, Fig . 4), one of thebest transglutaminase substrates identified to date(Facchiano et al ., 1993) .

T isoform comparisonBecause the isoforms of bovine T appeared to be

differentially cross-linked by transglutaminase (Dudekand Johnson, 1993), the substrate specificity of severalrecombinant human T isoforms was examined usingthe [ 3H]putrescine incorporation assay describedabove . Isoforms T3, T4, and T4L were significantlybetter substrates for transglutaminase than purified bo-vine T (Fig . 5) . Furthermore, isoforms T4 and T4L,both of which contain four microtubule-binding do-mains were better substrates than T3, which has onlythree microtubule-binding domains . In addition, thehuman T construct T264, a 122 amino acid fragmentof T, was also a substrate of transglutaminase, althoughit was not as good a substrate as the isoforms or aspurified bovine ,7- (Fig . 5) . The stoichiometry of incor-poration (moles of putrescine incorporated per moleof T) for each isoform is indicated in the inset ofFig . 5 .Immunoblot analysis indicated that isoforms T3, T4,

and T4L are all cross-linked by transglutaminase (Fig.6A), as evidenced by the formation of dimers and ofhigh molecular weight polymers that appear at the topof the gel . This is in contrast to construct T264, whichapparently cannot be cross-linked by transglutaminase(Fig . 6B), even though it is a transglutaminase sub-strate in terms of putrescine incorporation (Fig . 5) . AT-immunoreactive band that migrates just ahead ofeach T isoform is also evident in the samples thatwere incubated with transglutaminase (Fig . 6A) . Thisimmunoreactive band most likely represents either across-linked product of smaller T fragments, or an in-tramolecular cross-linking of the isoform, which in-


FIG. 5. Transglutaminase differentially modifies the isoforms ofhuman T. Bar graph of data from the putrescine incorporationassay. Bovine T, the recombinant human isoforms T3, T4, andT4L, and recombinant human Tdeletion construct T264 (5 ^and myoglobin (20 ^were incubated with transglutaminaseand [3H]putrescine as described in Experimental Procedures .T3 is a better substrate for transglutaminase than purified bovineT (`p < 0.05), and T4 andT4L are better substrates than eitherbovine T or T3 (..p < 0.05) (n = 5 separate experiments) .Fragment T264 is also a transglutaminase substrate, althoughnot as good a substrate as purified bovine T (tp < 0.05) . Myo-globin is not a substrate of transglutaminase . (n = 5 for bovineT and isoforms ; n = 8 for T264 ; n = 4 for myoglobin.) Inset: Thestoichiometry of putrescine incorporation (moles of putrescineincorporated per mole of -r) .

duces a conformational change that allows T to migratecloser to its predicted molecular mass .

Proteolysis experimentsTo determine which segments of the T molecule

were susceptible to modification by transglutaminase,the human T isoforms T3 and T4 were radiolabeledwith [ 3H ]putrescine, as described above . These iso-forms were then partially proteolyzed by a-chymotryp-sin, separated by electrophoresis, and examined bywestern blot analysis and autoradiography . To moreclearly determine the cleavage sites of a-chymotryp-sin, the same procedure was repeated with unlabeledputrescine, and the resulting proteolytic profile of T3was determined by probing the blots with several dif-ferent T antibodies . The antibodies included Alz-50,Tau-2, T14, 5E2, and T46.1, whose epitopes span thelength of the r molecule (Fig . 7A) (Kosik et al ., 1988 ;Goedert et al ., 1991b; Watanabe et al ., 1992) .T that has been modified by the addition of putres-

cine has an increased mobility on SDS-polyacrylamidegels compared with unmodified T, and thus it migratescloser to the position expected from its predicted mo-lecular mass (data not shown) . Figure 7B shows thea-chymotrypsin proteolytic profile of putrescine-modi-fied T3, as seen by immunoblotting with antibodiesT14 and Alz-50 . Complete cleavage occurs within 5min of incubation with a-chymotrypsin, resulting indiscrete breakdown products that remain stable

1765

throughout the entire 30-min proteolysis (Fig . 7B) .One of the primary cleavage sites in T3 appears to beat a phenylalanine residue (Phe 8 ) in the middle of theAlz-50 epitope (Goedert et al ., 1991b), because theimmunoreactivity of this antibody is abolished within5 min after the addition of the protease (Fig . 7B) .However, as cleavage at this site alone is not enoughto explain the increased mobility of the major T bandon SDS-polyacrylamide gels (compare lanes 1 and 2,Fig . 7B), it is likely that a-chymotrypsin also cleaves

FIG. 6. Cross-linking of recombinant humanT isoforms by trans-glutaminase. A: Immunoblot (Tau-1, 1 :12,000) of human 7- iso-forms T3, T4, and T4L (0.05 pg) incubated with transglutaminasein the presence of 5 mM EGTA (-) or 5 mM CaCl2 (+) for 30min at 37°C . All isoforms are cross-linked by transglutaminaseto form high molecular weight multimers, including very highmolecular weight polymers that are visible at the top of blot . B:Colloidal gold total protein stain of the recombinant human Tfragment T264 (*) incubated with transglutaminase (TG, shownby itself in the first lane) in the presence of 5 mM EGTA or 5mM CaCl2 (Ca) (as indicated) for 30 min at 37°C . T264 is notcross-linked by transglutaminase, as dimeric and higher molecu-lar weight complexes cannot be detected . The two additionalbands, not marked with an asterisk in the second and third lanes,are protein impurities from the bacterial expression system andare not immunoreactive with T antibodies .


1766


FIG. 7. Proteolysis of putrescine-modified T . A: Schematic diagram of the human T4L isoform, indicating the locations of antibodyepitopes (underlined) (Kosik et al ., 1988 ; Goedert et al ., 1991 ; Watanabe et al ., 1992). Also shown are the microtubule-binding domains(shaded) and the location of the deletion construct T264 . B; Immunoblots probed with antibodies T14 and Alz-50, showing the a-chymotrypsin proteolysis of putrescine-modified T3 . Within 5 min of incubation with the protease, modified T3 is cleaved into a slightlylower molecular weight form and several other fragments that are apparently not cleaved further (lanes 2 and 3) . Production of theslightly lower molecular weight fragment probably involves cleavage at the extreme amino-terminus of the protein, including a cleavagewithin the Alz-50 epitope, which abolishes the immunoreactivity of this antibody within 5 min of incubation with a-chymotrypsin (lane5) . See text for details . C: Proteolytic profiles of [3H]putrescine-labeled T3, visualized by autoradiography and immunoblot analysis .T3 was radiolabeled with [ 3 H] putrescine by transglutaminase as described in Experimental Procedures . This radiolabeled T3 was thenproteolyzed by a-chymotrypsin for 30 min at 37°C. Autoradiographs of the sample indicate that transglutaminase (TG) is not labeledby this process (lane 1) and that only a few fragments of T3 (TAU) contain the radiolabel (lane 2) . The major proteolytic fragment(lane 2, arrow), which contains the majority of the radiolabel, is also labeled by immunoblot analysis with antibodies T46.1 and 5E2(lanes 3 and 4) but not with antibody T14 (lane 5), thus indicating a carboxy-terminal fragment (see A) .

T3 at either Tyr", Tyr z9 , or both, thus resulting in a5-8% reduction in the size of T3 (see Fig . 7B) .

After radiolabeling T3 with [3H]putrescine via thetransglutaminase reaction, and proteolyzing with a-chymotrypsin, a comparison of the resulting autoradio-graphs and western blots revealed regions of the Tmolecule that are potential sites of modification bytransglutaminase . Even though the T3 isoform contains14 glutamine residues that span the length of the mole-cule, only a few proteolytic fragments contain the ra-diolabel (Fig . 7C) . The largest and most prevalent ofthese T3 fragments is a carboxy-terminal segment thatcontains the epitopes for T46.1 and 5E2 but not for T14(Fig . 7C, arrows) . This has been verified by antibodydouble-labeling experiments with T46.1, 5E2, and T14(data not shown) . This same fragment was also labeledin the T4 isoform, which has an extra microtubule-binding domain in this region, and the resulting T4fragment migrated slightly slower than the correspond-ing T3 fragment, as expected (data not shown) . Theother radiolabeled fragments of T3 were more difficult

J. Neurochem., Vol. 65, No . 4, 1995

to identify ; however, preliminary experiments involv-ing the separation of radiolabeled T fragments on areverse-phase HPLC column indicate that one majorproteolytic fragment and two secondary fragmentscontain most of the radiolabeled putrescine (unpub-lished data) .

DISCUSSION

Tissue transglutaminase is an enzyme that can cata-lyze the cross-linking of substrate proteins into or-dered, high molecular weight complexes that are rigid,insoluble, and resistant to proteolysis (Folk and Finlay-son, 1977) . This is ofparticular interest in Alzheimer'sdisease, as neurofibrillary tangles have biochemicalproperties that are similar to transglutaminase reactionproducts (Selkoe et al ., 1982b) . As mentioned earlier,the idea that transglutaminase is involved in neuro-fibrillary tangle formation is not a new one . Selkoeet al . (1982a,b) suggested that neurofibrillary tanglescontain nondisulfide covalent bonds and proposed that


transglutaminase was involved in the cross-linking ofneurofilament proteins . Miller and Anderton (1986)also investigated the cross-linking of neurofilamentand microtubule proteins and noted that the microtu-bule-associated proteins were "especially rapidlycrosslinked ." Given that the core of the neurofibrillarytangle consists of the T protein (Kosik et al ., 1986;Wood et al., 1986 ; Goedert et al ., 1988 ; Lee et al .,1991), and that T from Alzheimer disease brains pres-ents as a smear similar to that observed when purifiedT is cross-linked by transglutaminase (Fig . 2), we in-vestigated the substrate properties of T with respect totissue transglutaminase . In this study, two separate butsimilar transglutaminase reactions were monitored,i.e., the incorporation of putrescine into substrate pro-teins and the cross-linking of substrate proteins (Fig .1) . Because cross-linking by transglutaminase canonly occur at glutamine residues that are modified bythe enzyme, it is necessary to establish which gluta-mines can be modified by the enzyme (via the putres-cine incorporation assay) before determining which ofthese sites are involved in the cross-linking of T invitro .

Purified bovine T is an excellent substrate of trans-glutaminase, as determined by its putrescine incorpora-tion (Fig . 3) . Other proteins such as BSA and myoglo-bin are not substrates . The higher background seen inthe BSA curve (Fig . 3) is probably due to a nonspecificprecipitation of unincorporated [3H]putrescine ratherthan incorporation into the BSA protein itself, as thelevel of radiolabel precipitated with this protein doesnot change with increasing concentrations of BSA, upto 120 jLM (data not shown) . The human T isoformsT3, T4, and T4L are even better substrates for transglu-taminase than purified bovine T (Fig . 5) . Although thereason for these differences is unknown, several factorscould be involved, including the following : (1) T3, T4,and T4L are purified constructs, whereas the bovine Tcontains all isoforms, several breakdown products, andpossible impurities that may add to the protein concen-tration without being modified by transglutaminase ;(2) sequence differences between human and bovineT may make human T an inherently better substrate oftransglutaminase than bovine T; and (3) the isoformsare bacterially expressed and thus do not contain post-translational modifications, which may or may not alterputrescine incorporation . In a comparison of the iso-forms, T4 and T4L, both of which contain four micro-tubule-binding domains, are better substrates than T3(Fig . 5), which has only three microtubule bindingdomains. This may be because the four-repeat isoformsare in a conformation that is more accessible to trans-glutaminase or because they contain one extra gluta-mine that may be modified by the enzyme .The human T construct T264 is also a substrate of

transglutaminase, albeit not as good a substrate as theisoforms or as purified bovine T (Fig . 5) . However,although T264 is a substrate of transglutaminase, itcannot be cross-linked by the enzyme (Fig . 6B) . This

1767

appeared somewhat surprising at first, especially con-sidering that T264 contains 16 lysine residues thatcould potentially react with the enzyme-substrate inter-mediate. However, this finding agrees with studies in-dicating that transglutaminase is remarkably selectivefor the lysines it uses in cross-linking reactions (Para-meswaran et al., 1990 ; Lorand et al ., 1992) . Thus,T264 provided an excellent control, demonstrating thatputrescine incorporation and cross-linking are indeedtwo distinct reactions ; i.e ., transglutaminase is able toincorporate putrescine into all of its substrates ; how-ever, only certain substrates can be cross-linked by theenzyme. It can therefore be postulated that to be cross-linked by transglutaminase, substrates must first be ori-ented in a specific conformation, wherein the lysineresidue of the "donor" protein is closely apposed tothe substrate glutamine residue of the "acceptor" pro-tein . Such an orientation is needed to ensure that thelysine is incorporated rather than another primaryamine . If this is the case, then only substrate proteinsthat are already "self-associated" in a specific arrange-ment would have the potential to be cross-linked bytransglutaminase . In agreement with this hypothesis,there is evidence to suggest that T readily self-associ-ates into dimers and higher order oligomers and canform soluble filamentous structures (Wille et al ., 1992 ;de Ancos et al ., 1993 ; Crowther et al ., 1994), a processthat requires the presence of the microtubule-bindingdomains of T (Wischik et al ., 1988 ; Jakes et al ., 1991 ;Ksiezak-Reding and Yen, 1991; Wille et al ., 1992) .Even under reducing conditions and in the absence oftransglutaminase, a small amount of T can often beseen in the dimeric form on an immunoblot (data notshown) . It is possible that this self-aggregation re-quires the first microtubule-binding domain of T, andas construct T264 lacks most of the first microtubule-binding domain, this could explain why it is not cross-linked by transglutaminase (Fig . 6B) . It must be noted,however, that even a protein that aggregates, but thatdoes not have a lysine next to a substrate glutamineresidue, will not likely be cross-linked by transglutami-nase, and this possibility also exists for T264 .

Predicting the ability of a given protein to be modi-fied by transglutaminase, based on primary amino acidsequence, is not straightforward . Neither the absolutenumber of glutamines, which are involved in the firstcatalytic step of the transglutaminase mechanism (Folkand Finlayson, 1977 ; Greenberg et al ., 1991), nor thepercentage of glutamines in a protein, appears to besufficient to qualify a protein as a transglutaminasesubstrate . Based on the number of glutamine residuesalone, it would be expected that T4L is a better sub-strate than T4; however, these isoforms are comparablesubstrates . Myoglobin contains five glutamine resi-dues, but it is not a substrate . The amino acid sequenceof T264 consists of 4.9% glutamine residues, whereasthe T4L sequence contains only 4.3% ; yet T4L is amuch better substrate than T264. Thus, because not allglutamine residues within a substrate protein can be

J. Neurochem ., Vol. 65, No. 4, 1995

1768

modified by transglutaminase, the differences in gluta-mine compositions among the isoforms are not suffi-cient to explain their differences in modification bythe enzyme . To further complicate matters, there ap-pears to be no particular primary amino acid consensussequence for transglutaminase, although a glutamineresidue that is modified by the enzyme is usually posi-tioned in either a highly flexible part of the protein orin a sequence of linear secondary structure, therebymaking it accessible to the enzyme (Coussons et al .,1992) . Also, certain charged amino acid residues lo-cated near the glutamine may enhance or inhibit bind-ing (Coussons et al ., 1992; Facchiano and Luini,1992) .To determine the relative locations of the glutamine

residues in T that are modified by transglutaminase,T3 and T4 were labeled with ['H] putrescine by trans-glutaminase, proteolyzed with a-chymotrypsin, andthe breakdown products were analyzed by autoradiog-raphy and immunoblotting . Results of these experi-ments indicate that transglutaminase modifies selec-tively only one or a few of the numerous glutamineresidues in the T sequence . Only three proteolytic frag-ments of T3, and only two to three fragments of T4,contained label (Fig . 7C and unpublished observa-tions) . The primary radiolabeled fragment from T3 isa carboxy-terminal fragment that contains the epitopesfor T46.1 and 5E2 (Fig . 7C, arrows), although thenumber of modified glutamines in this fragment couldnot be ascertained by our methods. It is speculated thatthis fragment represents a carboxyl segment of T3 thathas been cleaved at Tyr 197 . Such a cleavage wouldgenerate a carboxy-terminal breakdown product withan approximate molecular mass of 23,300 Da, whichwould include epitopes for T46.1 and 5E2 but not T14 .This agrees with theories that the PHF backbone con-sists of T arranged in an anti-parallel fashion, joinedsomewhere near the carboxyl end of the molecule(Kondo et al ., 1988 ; Wischik et al ., 1988 ; Ksiezak-Reding and Yen, 1991; Wille et al ., 1992) . The othertwo radiolabeled fragments of T3 were more difficultto identify through western blotting techniques ; how-ever, preliminary evidence from HPLC separation ofthe proteolytic fragments of T3 indicates that most ofthe radiolabeled peptides elute as one major and twosecondary peaks (unpublished data) .

It should be noted that transglutaminase is a cal-cium-activated enzyme (Folk and Finlayson, 1977) .Much evidence suggests that a disruption in calciumhomeostasis leads to increased cytosolic calcium inAlzheimer disease (for review, see Mattson, 1992) .For example, increased levels of calcium-binding pro-teins (Hof et al ., 1991 ; Sutherland et al ., 1993) andactivated calpain (Saito et al ., 1993) have been notedin Alzheimer disease brains, and increased concentra-tions of bound, cytosolic, and free calcium have beenfound in fibroblasts from Alzheimer patients (Petersonand Goldman, 1986) . Studies by Mattson (1992) haveshown that sustained or increased levels of intracellular



calcium result in the depolymerization of microtubulesin cultured cortical neurons . It has been proposed thatthis depolymerization of microtubules may lead to anincrease in the concentration of free, unbound T, whichis then available to self-aggregate (Mattson, 1992) .Because transglutaminase is also activated by calcium,pathological increases in calcium could provide amechanism whereby this aggregated T could be ex-posed to abnormal transglutaminase activity . It is inter-esting that Mattson (1992) noted that calcium influxled not only to the depolymerization of miçrotubules,but also to the formation of 8-15-nm straight filamentsand an increased immunoreactivity with the Alz-50antibody . Increased Alz-50 immunoreactivity is alsoseen when T is cross-linked by transglutaminase invitro (Dudek and Johnson, 1993) .Because of the unique properties of transglutami-

nase activation and catalysis, and the characteristics ofits interaction with T, it can perhaps be hypothesizedthat transglutaminase, or some similar cross-linkingenzyme, may be involved in converting T from itsnormal, soluble form to a polymerized, insoluble form,such as that found in the neurofibrillary tangle . If solu-ble PHFs, which contain T in an ordered, antiparallelstate (Ksiezak-Reding and Yen, 1991 ; Wille et al .,1992), were pathologically exposed to increased cal-cium levels and abnormal transglutaminase activity,the result could be ordered, insoluble aggregations ofT (Dudek and Johnson, 1993) such as that seen inneurofibrillary tangles . This type of stabilization is notunprecedented, as it occurs in transglutaminase-medi-ated clot formation ; wherein fibrin polymers that areinitially held together by noncovalent bonds becomestabilized by E-(y-glutamyl)lysine bonds (Folk andFinlayson, 1977) . A similar reaction may occur in thekeratinization of the epidermis, where soluble epider-mal structural proteins are cross-linked into insolublehigh molecular weight polymers (Folk, 1980) .

In summary, T is an excellent in vitro substrate oftissue transglutaminase, comparable with DMC, oneof the best substrates known (Facchiano et al ., 1993) .The human T isoforms are differentially cross-linked,as T4 and T4L are better transglutaminase substratesthan T3 . In the absence of polyamines, transglutami-nase cross-links T into dimers, trimers, and high molec-ular weight aggregates . This cross-linking of T is aspecific reaction, as a partial segment of T (T264),which lacks part of the first microtubule-binding do-main, is a substrate for transglutaminase but cannot becross-linked . Because transglutaminase-cross-linkedproteins contain e- (y-glutamyl) lysine bonds ratherthan disulfide bonds, transglutaminase-cross-linked Tis insoluble in detergents and reducing agents . Thismay be important in pathological conditions such asAlzheimer disease, in which T has been shown to formaggregated, insoluble polymers. T is modified by trans-glutaminase at only one or a few specific sites, includ-ing at least one site in the carboxyl half of the molecule .This agrees with previous findings that the cores of


soluble PHFs consist of the carboxyl 40% of the Tmolecule (Ksiezak-Reding and Yen, 1991) .

It must be emphasized, however, that there is as yetno direct evidence for the existence of c-(y-glutamyl)-lysine isopeptide bonds in neurofibrillary tangles . Asnoted previously (Selkoe et al ., 1982a), attempts todetect such bonds should be highly informative . Fur-ther research is required to elucidate the putative roleof transglutaminase in the formation of neurofibrillarytangles in Alzheimer disease . However, even if trans-glutaminase itself is not involved in the formation ofneurofibrillary tangles, it provides a useful model sys-tem for examining the properties of aggregated, insolu-ble T polymers, which are a hallmark of Alzheimerdisease .

Acknowledgment: Special thanks to K . Kosik, V . Lee,L. Binder, P. Davies, P . Seubert (Athena Neurosciences,Inc .), and P. Birckbichler for the antibodies 5E2, T46.1 andT14, Tau-1 and Tau-2, Alz-50, 7C11 and 8C11, and CUB7402 (respectively) ; and to C. Scott for the recombinanthuman -r isoforms and the construct T264 . This study wassupported by National Institutes of Health grants NS27538,AG12396, and AG06569 and a grant from the Ruth K. BroadBiomedical Research Foundation.

REFERENCESAchyuthan K . E . and Greenberg C . S . (1987) Identification of a

guanosine triphosphate-binding site on guinea pig liver trans-glutaminase : role ofGTP and calcium ions in modulating activ-ity . J. Biol. Chem. 262, 1901-1906.

Birckbichler P . J ., Upchurch H . F ., Patterson M . L. Jr ., and ConwayE . (1985) A monoclonal antibody to cellular transglutaminase .Hybridoma 4, 179-186.

Coussons P. J ., Price N . C ., Kelly S . M ., Smith B ., and Sawyer L .(1992) Factors that govern the specificity of transglutaminase-catalysed modification of proteins and peptides . Biochem. J.282, 929-930.

Crowther R . A ., Olesen O. F ., Jakes R ., and Goedert M. (1992)The microtubule binding repeats of tau protein assemble intofilaments like those found in Alzheimer's disease . FEBS Lett.309, 199-202 .

Crowther R . A ., Olesen O . F ., Smith M . J ., Jakes R ., and GoedertM . (1994) Assembly of Alzheimer-like filaments from full-length tau protein . FEBS Lett. 337, 135-138 .

de Ancos J . G ., Correas I ., and Avila J. (1993) Differences inmicrotubule binding and self-association abilities of bovinebrain tau isoforms . J. Biol. Chem . 268, 7976-7982 .

Drubin D. G . and Kirschner M . W . (1986) Tau protein function inliving cells . J. Cell Biol. 103, 2739-2746 .

Dudek S . M . and Johnson G . V . W . (1993) Transglutaminase cata-lyzes the formation of sodium dodecyl sulfate-insoluble, Alz-50-reactive polymers of r . J. Neurochem. 61, 1159-1162 .

El Alaoui S ., Mian S ., Lawry J ., Quash G., and Griffin M . (1992)Cell cycle kinetics, tissue transglutaminase and programmedcell death (apoptosis) . FEBS Lett. 311, 174-178 .

Facchiano F. and Luini A . (1992) Tetanus toxin potently stimulatestissue transglutaminase. A possiblemechanism ofneurotoxicity.J. Biol. Chem . 267, 13267-13271 .

Facchiano F., Benfenati F., Valtorta F ., and Luini A . (1993) Cova-lent modification of synapsin I by a tetanus toxin-activatedtransglutaminase . J. Biol. Chem . 268, 4588-4591 .

Fesus L ., Thomazy V ., and Falus A . (1987) Induction and activationof tissue transglutaminase during programmed cell death . FEBSLett. 224, 104-108 .

Fesus L ., Thomazy V., Autuori F., Cem M. P., Tarcsa E ., and

1769

Piacentini M. (1989) Apoptotic hepatocytes become insolublein detergents and chaotropic agents as a result of transglutami-nase action . FEBS Lett. 245, 150-154.

Folk J . E . (1980) Transglutaminases . Annu. Rev. Biochem . 49, 517-531 .

Folk J . E . and Finlayson J . S . (1977) The e-(y-glutamyl) lysinecrosslink and the catalytic role of transglutaminases . Adv . Pro-tein Chem. 31, 1-133 .

Gambetti P ., Autilio-Gambetti L ., and Papasozomenos S . C . (1981)Bodian's silver method stains neurofilament polypeptides. Sci-ence 213, 1521-1522 .

Gilad G . M . and Varon L . E . (1985) Transglutaminase activity inrat brain : characterization, distribution, and changes with age .J. Neurochem. 45, 1522-1526 .

Goedert M . (1993) Tau protein and the neurofibrillary pathology ofAlzheimer's disease. Trends Neurosci. 16, 460-465 .

Goedert M., Wischik C . M., Crowther R . A ., Walker J . E ., and KlugA . (1988) Cloning and sequencing of the cDNA encoding acore protein of the paired helical filament of Alzheimer disease :identification as the microtubule-associated protein tau . Proc.Natl. Acad. Sci. USA 85, 4051-4055 .

Goedert M., Spillantini M . G ., Jakes R ., Rutherford D ., and CrowtherR. A . (1989) Multiple isoforms of human microtubule-associ-ated protein tau : sequences and localization in neurofibrillarytangles of Alzheimer's disease . Neuron 3, 519-526 .

Goedert M., Sisodia S . S ., and Price D . L . (1991a) Neurofibrillarytangles and 6-amyloid deposits in Alzheimer's disease . Curr.Opin . Neurobiol. 1, 441-447 .

Goedert M., Spillantini M . G., and Jakes R. (1991b) Localizationof the Alz-50 epitope in recombinant human microtubule-asso-ciated protein tau . Neurosci. Lett. 126, 149-154 .

Goedert M., Spillantini M . G., Cairns N. J., and Crowther R. A .(1992) Tau proteins of Alzheimer paired helical filaments : ab-normal phosphorylation of all six brain isoforms . Neuron 8,159-168 .

Greenberg C . S ., Birckbichler P. J ., and Rice R . H . (1991) Transglu-taminases : multifunctional cross-linking enzymes that stabilizetissues . FASEB J. 5, 3071-3077.

Greenberg S. G. and Davies P . (1990) A preparation of Alzheimerpaired helical filaments that displays distinct T proteins by poly-acrylamide gel electrophoresis . Proc . Natl . Acad. Sci. USA 87,5827-5831 .

Himmler A . (1989) Structure of the bovine tau gene: alternativelyspliced transcripts generate a protein family . Mol. Cell. Biol . 9,1389-1396 .

Hof P . R ., Cox K., Young W . G ., Celio M . R., Rogers J ., andMorrison J . H . (1991) Parvalburnin-immunoreactive neuronsin the neocortex are resistant to degeneration in Alzheimer'sdisease . J. Neuropathol. Exp. Neurol . 50, 451-462.

Jakes R., Novak M., Davison M., and Wischik C. M . (1991) Identi-fication of 3- and 4-repeat tau isoforms within the PHF in Alz-heimer's disease . EMBO J. 10, 2725-2729 .

Johnson G . V . W., Jope R . S ., and Binder L . 1 . (1989) Proteolysisof tau by calpain . Biochem. Biophys. Res. Commun . 163,1505-1515 .

Knops J., Kosik K. S ., Lee G., Pardee J . D ., Cohen-Gould L., andMcConlogue L. (1991) Overexpression of tau in a nonneuronalcell induces long cellular processes. J. Cell Biol. 114, 725-733 .

Kondo J ., Honda T ., Mori H ., Hamada Y ., Miura R ., Ogawara M.,and Ihara Y . (1988) The carboxyl third of tau is tightly boundto paired helical filaments . Neuron 1, 827-834 .

Kosik K. S ., Joachim C. L., and Selkoe D. J . (1986) Microtubule-associated protein tau (T) is a major antigenic component ofpaired helical filaments in Alzheimer disease. Proc. Natl. Acad.Sci. USA 83, 4044-4048 .

Kosik K . S ., Orecchio L. D ., Binder L., Trojanowski J . Q ., LeeV. M.-Y., and Lee G . (1988) Epitopes that span the tau mole-cule are shared with paired helical filaments . Neuron 1, 817-825 .

Ksiezak-Reding H . and Yen S.-H . (1991) Structural stability of


1770

paired helical filaments requires microtubule-binding domainsof tau : a model for self-association . Neuron 6, 717-728 .

Lee V. M.-Y ., Balin B . J., Otvos L. Jr., and Trojanowski J . Q.(1991) A68 : a major subunit of paired helical filaments andderived forms of normal tau . Science 251, 675-678 .

Lorand L., Velasco P . T ., Murthy S . N . P., Wilson J ., and Parames-waran K . N . (1992) Isolation of transglutaminase-reactive se-quences from complex biological systems : a prominent lysinedonor sequence in bovine lens . Proc . Natl. Acad. Sci. USA 89,11161-11163 .

Lowry O . H ., Rosebrough N . J., Farr A . L ., and Randall R. J . (1951)Protein measurement with the Folin phenol reagent. J. Biol.Chem . 193, 265-275 .

Lu Q . and Wood J. G . (1993) Functional studies of Alzheimer'sdisease tau protein . J. Neurosci. 13, 508-515 .

Maccioni R . B . and Seeds N . W . (1986) Transglutaminase andneuronal differentiation . Mal . Cell. Biochem. 69, 161-168 .

Mattson M . P. (1992) Calcium as sculptor and destroyer of neuralcircuitry . Exp. Gerontol. 27, 29-49 .

Miller C . C . J . and Anderton B . H . (1986) Transglutaminase andthe neuronal cytoskeleton in Alzheimer's disease . J. Neuro-chem . 46, 1912-1922 .

Morishima-Kawashima M., Hasegawa M., Takio K ., Suzuki M.,Titani K ., and Ihara Y . (1993) Ubiquitin is conjugated withamino-terminally processed tau in paired helical filaments . Neu-ron 10, 1151-1160 .

Papasozomenos S . C. and Binder L. I . (1987) Phosphorylation deter-mines two distinct species of tau in the central nervous system.Cell Motil. Cytoskeleton 8, 210-226.

Parameswaran K . N ., Velasco P . T ., Wilson J ., and Lorand L . (1990)Labeling of e-lysine crosslinking sites in proteins with peptidesubstrates of factor XIIIa and transglutaminase . Proc. Natl.Acad. Sci. USA 87, 8472-8475 .

Peterson C . and Goldman J . E. (1986) Alterations in calcium contentand biochemical processes in cultured skin fibroblasts from agedand Alzheimer donors . Proc . Natl. Acad. Sci . USA 83, 2758-2762 .

Saito K.-I ., Elce J . S ., Hamos J . E ., and Nixon R . A . (1993) Wide-spread activation of calcium-activated neutral proteinase (cal-pain) in the brain in Alzheimer disease : a potential molecularbasis for neuronal degeneration. Proc. Nod. Acad. Sci . USA 90,2628-2632.

Scott C. W., Blowers D. P., Barth P . T ., Lo M . M . S ., Salama A . I .,


M. L . MILLER AND G. V. W. JOHNSON

and Caputo C . B . (1991) Differences in the abilities of humantau isoforms to promote microtubule assembly . J. Neurosci.Res. 30, 154-162 .

Scott C . W., Mika A . B ., Lo M . M. S ., Norris T . E ., and CaputoC . B . (1992) Tau protein induces bundling of microtubules invitro : comparison of different tau isoforms and a tau proteinfragment. J. Neurosci. Res . 33, 19-29.

Selkoe D . J ., Abraham C ., and Ihara Y . (1982a) Brain transglutami-nase : in vitro crosslinking of human neurofilament proteins intoinsoluble polymers . Proc. Natl . Acad. Sci. USA 79,6070-6074 .

Selkoe D . J ., Ihara Y., andSalazar F. J. (1982b) Alzheimer's disease:insolubility of partially purified paired helical filaments in so-dium dodecyl sulfate and urea . Science 215, 1243-1245 .

Sternberger N. H ., Sternberger L . A., and Ulrich J . (1985) Aberrantneurofilament phosphorylation in Alzheimer disease . Proc.Natl. Acad. Sci. USA 82, 4274-4276 .

Sutherland M. K., Wong L ., Somerville M. J., Yoong L. K . K.,Bergeron C ., Parmentier M., and McLachlan D . R . (1993) Re-duction of calbindin-28k mRNA levels in Alzheimer as com-pared to Huntington hippocampus . Mol. Brain Res. 18, 32-42 .

TowbinH ., Staehelin T ., and Gordon J . (1979) Electrophoretic trans-fer of protein from polyacrylamide gels to nitrocellulose sheets:procedure and some applications . Proc. Natl. Acad. Sci. USA76, 4350-4354.

Vigo-Pelfrey C., Seubert P ., Barbour R., Blomquist C ., Lee M ., LeeD ., Coria F., Chang L., Miller B ., Lieberburg I., and SchenkD . (1995) Elevation of microtubule-associated protein tau incerebral spinal fluid of Alzheimer's disease patients . Neurology45, 788-793 .

Watanabe N ., Takio K ., Hasegawa M., Arai T ., Titani K., and IharaY . (1992) Tau 2 : a probe for a Set conformation in the aminoterminus of T . J. Neurochem. 58, 960-966 .

Wille H ., Drewes G ., Biernat J., Mandelkow E.-M., and MandelkowE. (1992) Alzheimer-like paired helical filaments and antiparal-lel dimers formed from microtubule-associated protein tau invitro . J. Cell Biol. 118, 573-584 .

Wischik C . M ., Novak M., Edwards P. C ., Klug A ., Tichelaar W .,and Crowther R. A . (1988) Structural characterization of thecore of the paired helical filament of Alzheimer's disease . Proc .Natl. Acad. Sci. USA 85, 4884-4888 .

Wood J. G ., Mirra S . S ., Pollock N. J ., and Binder L. I. (1986)Neurofibrillary tangles of Alzheimer disease share antigenicdeterminants with the axonal microtubule-associated protein tau(-r) . Proc. Natl. Acad. Sci. USA 83, 4040-4043 .

transglutaminase cross-linking of the τ protein

Documents