The Intermediate Filament Protein, Vimentin, in the Lens Is aTarget for Cross-linking by Transglutaminase*

(Received for publication, September 17, 1997, and in revised form, January 12, 1998)

Sophie Clément‡, Pauline T. Velasco‡, S. N. Prasanna Murthy‡, James H. Wilson‡†,Thomas J. Lukas§, Robert D. Goldman‡, and Laszlo Lorand‡¶

From the Departments of ‡Cell and Molecular Biology and §Molecular Pharmacology and Biological Chemistry,Northwestern University Medical School, Chicago, Illinois 60611

Mere addition of Ca21 to a lens cortical homogenate(bovine) generates a series of products composed of avariety of high molecular weight vimentin species. TheCa21-induced cross-linking of this cytoskeletal elementseems to be mediated by the intrinsic transglutaminaseof lens, because the reaction could be blocked at themonomeric state of vimentin by the inclusion of smallsynthetic substrates of the enzyme dansylcadaverine ordansyl-e-aminocaproyl-Gln-Gln-Ile-Val. These compoundsare known to compete against the Gln or Lys function-alities of proteins that would participate in forming theNe(g-glutamyl)lysine protein-to-protein cross-links. Thecytosolic transglutaminase-catalyzed reactions could bereproduced with purified bovine lens vimentin and alsowith recombinant human vimentin preparations. Em-ploying the latter system, we have titrated the transglu-taminase-reactive sites of vimentin and, by sequencingthe dansyl-tracer-labeled segments of the protein, wehave shown that residues Gln453 and Gln460 served asacceptor functionalities and Lys97, Lys104, Lys294, andLys439 as electron donor functionalities in vimentin. Thetransglutaminase-dependent reaction of this intermedi-ate filament protein might influence the shape and plas-ticity of the fiber cells, and the enzyme-catalyzed cross-linking of vimentin, in conjunction with other lensconstituents, may contribute to the process of cataractformation.

Significant amounts of the branched Ne(g-glutamyl)lysineisopeptide, usually a manifestation of the posttranslationalmodifications of proteins by TGase,1 were found in the proteo-lytic digests of high molecular weight polymers present in

human cataractous lens specimens. Complementary experi-ments, in which the intrinsic Ca21-activated TGase was al-lowed to promote the cross-linking of endogenous proteins,revealed that among the soluble constituents of the lens, pri-marily a small subset of crystallins was targeted by the enzyme(1). Enzyme-specific probes, such as dansylcadaverine, dansyl-e-aca-QQIV, or similar biotinylated analogues were synthe-sized for the identification and analysis of the substrates of theenzyme. These compounds compete against the TGase-cata-lyzed cross-linking of proteins, and by becoming incorporatedin the enzyme-driven reaction into the biological substratesthemselves, they could be employed for locating the Gln (oracceptor) and the Lys (or donor) functionalities exhibiting theunique specificities (1–8).

Based on experiments with human red blood cells (9, 10) andwith keratinocytes (11, 12), it seemed reasonable to assumethat in addition to the specific subsets of crystallins, somemembrane and cytoskeletal elements in the lens might alsobecome cross-linked by TGase. Therefore, we undertook to ex-amine changes occurring in the skeletal proteins of the lens asthe latent TGase becomes activated by Ca21 in the tissue. Thisline of research was greatly spurred by our recent finding thata TGase antigen became tightly bound to the IF network insome fibroblasts and keratinocytes (13, 14). The present reportfocuses on the intermediate filament protein vimentin as sub-strate for the cross-linking enzyme. Notwithstanding the mes-enchymal preference for the expression of the vimentin-encod-ing gene and the epithelial origin of the ocular lens (15, 16),vimentin is known to be actively synthesized in the lens epi-thelium and in the cortical fiber cells (17). We have now iden-tified the TGase-reactive Gln and Lys side chains of vimentin,which can act as acceptor and donor functionalities of theprotein in cross-linking onto itself or to other membrane andcytoskeletal elements.

MATERIALS AND METHODS

Experiments with Lens Homogenates—Frozen calf lenses (Pel-Freez,Rogers, AR) were thawed to room temperature and decapsulated, andthe cortex was separated from the nucleus. Cortical portions from twolenses were homogenized by hand in 3 ml of 50 mM Tris-HCl, pH 7.5,and 100 mM NaCl in a Potter-Elvehjem tissue grinder. Incubations werecarried out for 90 min and for 21 h at 37 °C in reaction volumes of 200ml containing about 80 mg of lens protein/ml, 20% glycerol, 50 mMTris-HCl, pH 7.5, 100 mM NaCl, 2 mM leupeptin (obtained through theUnited States-Japan Cooperative Cancer Research Program) and 1–8mM CaCl2 or 2 mM EDTA. To some mixtures, 2 mM dansylcadaverine(18, 19) or dansyl-e-aca-QQIV (6, 7) was also added. Following incuba-tion, the samples were centrifuged (Eppendorf microcentrifuge, Brink-mann, Westbury, NY; 16,000 3 g for 5 min.). Both supernatants andpellets were analyzed by SDS-PAGE. The supernatants were preparedfor electrophoresis by mixing 5-ml aliquots with 45 ml of solubilizationbuffer (50 mM Tris-HCl, pH 6.8, 9 M urea, 40 mM dithiothreitol, 2% SDS)at 37 °C for 30 min. The pellets were washed twice with 1 ml of 50 mMTris-HCl, pH 7.5, 100 mM NaCl and then solubilized in 65 ml of solubi-

* This research was supported by Grants EY-03942 and NIGMS-36806 and 30861 from the National Institutes of Health. Some aspectsof these investigations were presented at the meeting of the Associationfor Research in Vision and Ophthalmology (Lorand, L., Velasco, P. T.,Murthy, S. N. P., Clement, S., Quinlan, R., and Goldman, R. D. (1996)Invest. Ophthalmol. Visual Sci. 37, S600, Abstr. 2767–B612). The costsof publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked “advertise-ment” in accordance with 18 U.S.C. Section 1734 solely to indicate thisfact.

† This work is dedicated to the memory of our colleague, James H.Wilson, who died on June 20, 1997.

¶ To whom correspondence should be addressed: Dept. of Cell andMolecular Biology, Northwestern University Medical School, 303 EastChicago Ave., Chicago, IL 60611-3008. Tel.: 312-503-0591; Fax:312-503-0590

1 The abbreviations used are: TGase, transglutaminase; dansylcadav-erine, N-(5-aminopentyl)-5-dimethylaminonaphthalene-1-sulfonamide;dansyl, 5-dimethylaminonaphthalene-1-sulfonyl; dansyl-e-aca-QQIV,dansyl-e-aminocaproyl-Gln-Gln-Ile-Val; HPLC, high performance liq-uid chromatography; PAGE, polyacrylamide gel electrophoresis; IF,intermediate filament.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 13, Issue of March 27, pp. 7604–7609, 1998© 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org7604

This is an Open Access article under the CC BY license.

http://creativecommons.org/licenses/by/4.0/

lization buffer at 37 °C for 30 min and clarified by centrifugation at16,000 3 g for 5 min.

Electrophoretic Analysis and Western Blotting—SDS-PAGE was car-ried out in 10% acrylamide gel (20), and the proteins (35 mg) wereelectroblotted to nitrocellulose (21) and immunostained with a rabbitantiserum to baby hamster kidney cell vimentin (22) (1:10,000 dilutionin 2% milk/phosphate-buffered saline solution) or with monoclonal IgGagainst dansyl (23) (1:300,000 dilution in 2% milk/phosphate-bufferedsaline solution) overnight at room temperature. After three washeswith 2% milk/phosphate-buffered saline solution, the blots were treatedwith alkaline phosphatase-conjugated secondary antibody (diluted1:5000 in 2% milk/phosphate-buffered saline solution) against eitherrabbit IgG (Promega Corporation, Madison, WI) or mouse IgG (wholemolecule; Sigma) for 2 h at room temperature. Alkaline phosphataseactivity was developed with 5-bromo-4-chloro-indolyl phosphate (Sig-ma, 300 mg/ml) as substrate and nitro blue tetrazolium (Sigma, 150mg/ml) as electron acceptor in 100 mM Tris-HCl, pH 9.5, 100 mM NaCl,5 mM MgCl2.

TGase-mediated Cross-linking of Purified Lens Vimentin and Inhi-bition by Dansyl-e-aca-QQIV and Dansylcadaverine—The reaction ofpurified bovine lens vimentin (24, 25) and purified human recombinantvimentin expressed in Escherichia coli (26) with a cytosolic TGase (27,28) was examined in 40-ml mixtures comprising 5 mM Tris-HCl, pH 7.5,10 mM NaCl, 10 mM dithiothreitol, 10 mg of vimentin, 10 mg/ml TGase,and 5 mM of either CaCl2 or EDTA. Some mixtures contained theTGase-specific probes of dansylcadaverine (18, 19) or dansyl-e-aca-QQIV (6, 7). After 2 h at 37 °C, the reactions were stopped by additionof 3 ml of 100 mM EDTA. Electrophoresis was carried out as describedabove.

Identification of the TGase-reactive Sites in Vimentin—Recombinanthuman vimentin (2 mg) expressed in E. coli (26) was allowed to reactwith the cytosolic TGase (80 mg/ml) in a 1.2-ml solution of 5 mM Tris-HCl, pH 7.5, 10 mM NaCl, 10 mM dithiothreitol for 2 h at 37 °C witheither 2 mM dansylcadaverine or dansyl-e-aca-QQIV in the presence of5 mM CaCl2. At the end of the incubation period, the proteins wereprecipitated with 7% trichloroacetic acid and sedimented by centrifu-gation (5 min at 14,000 3 g), and the pellets were extracted repeatedly(8 times) either with 1 ml of ethanol:ether (1:1, v/v) to remove theunreacted dansylcadaverine or with 1 ml of N,N-dimethylformamidecontaining 1% N-methylmorpholine and 5% H2O for removal of theunreacted dansyl-e-aca-QQIV probe.

The washed pellets were resuspended in 1 ml of 5 mM Tris-HCl, pH7.5, 10 mM NaCl and digested with 200 mg of trypsin treated withL-(tosylamido-2-phenyl)ethylchloromethylketone (Worthington Bio-chemicals, Freehold, NY) at 37 °C for 48 h. Digestion was terminated byaddition of 400 mg of phenylmethylsulfonyl fluoride.

Dansyl-labeled peptides were isolated on a monoclonal anti-dansylAffi-Gel 10 column (6, 19) and concentrated by lyophilization. Separa-tion was carried out by HPLC on a Ultrasphere C8 column (BeckmanInstruments) with a gradient of acetonitrile (5–40% for separating thedansylcadaverine-labeled peptides or 10–60% for the dansyl-e-aca-QQIV-labeled peptides) in H2O, 0.1% trifluoroacetic acid over 90 min ata flow rate of 1 ml/min. Effluents were monitored by absorbance (220nm) and by fluorescence (excitation, 338 nm; emission, 500 nm). Aminoacid sequencing was carried out in an Applied Biosystems Procisesequencer.

RESULTS AND DISCUSSION

Experiments with Lens Homogenates—Examination of pro-teins by SDS-PAGE showed that in the control lens specimen,vimentin partitioned mostly into the insoluble portion of thehomogenate (compare bands on the immunoblots of the 45–60-kDa region (Fig. 1A, lane 4), representing the supernatant,with those representing the sediment fraction (Fig. 1B, lane 4)),whereas the crystallins (21–32 kDa) (Fig. 1, A and B, lane 1,stained with Amido Black) were present mostly in the solublephase. Except for a faint band, seemingly corresponding to atrimeric or tetrameric species of vimentin, in Fig. 1B, lane 4, nohigh molecular weight forms of the intermediate filament pro-tein were seen in this bovine lens preparation. However, themere addition of Ca21 to the homogenate generated significantamounts of a series of vimentins with Mr . 60,000 (Fig. 1, Aand B, lanes 5 and 6, X). The production of these higher vimen-tin species seemed to be somewhat enhanced if leupeptin, acalpain inhibitor that is known to suppress the proteolytic

degradation of vimentin in rabbit lens (29), was included in theincubation mixture (Fig. 1, A and B, lane 6). The higher mo-lecular weight vimentin product separated mainly with theinsoluble phase of the homogenate (Fig. 1B, lanes 5 and 6),whereas some of the lower multiples of vimentin (Mr #200,000) appeared also in the supernatant (Fig. 1A, lanes 5 and6). The endogenous cross-linking of vimentin could also bedocumented at lower concentrations of Ca21 (Fig. 1C) by incu-bating the lens homogenate over a longer time period (21 h).These SDS-PAGE experiments, carried out after reducing theproteins with dithiothreitol in urea, demonstrated that thecross-linked vimentin species were produced in the lens tissueby covalent bonds other than disulfides. The additional find-ings presented in Fig. 2 provided clear proof for the involve-ment of the Ca21-activated, endogenous TGase in the observedreaction. As a primary amine, dansylcadaverine is known tocompete against the Lys residues of proteins in the TGase-mediated cross-linking reaction (18, 19), whereas dansyl-e-aca-QQIV competes against the enzyme-reactive Gln residues incross-linking (6, 7). The presence of either of these compounds(2 mM) during treatment of the homogenate with Ca21 reducedthe amount of cross-linked vimentin products, as is best seen inFig. 2A, lanes 5 and 6. Probing the transblots with a mono-clonal antibody against the dansyl moiety (Fig. 2, lanes 7 and 8)visualized monomeric vimentin at about 55 kDa, as well as theother TGase-reactive proteins in the lens homogenate.

Reactions of the Purified Bovine Lens Vimentin and Recom-

FIG. 1. Ca21-dependent cross-linking of vimentin in the lenshomogenate. Calf lens homogenate was incubated for 90 min at 37 °Cin the presence of 2 mM leupeptin (lanes 3 and 6) with 2 mM EDTA(lanes 1 and 4) or 8 mM CaCl2 (lanes 2, 3, 5, and 6). The supernatants(A) and solubilized pellets (B), separated by centrifugation, were ana-lyzed by SDS-PAGE and then electroblotted to nitrocellulose. Blotswere either stained for protein with Amido Black (lanes 1–3) or devel-oped with a specific antiserum to vimentin (lanes 4–6). For experimen-tal details, see “Materials and Methods.” The position of the parentvimentin at 55 kDa (V) and the cross-linked products of vimentin (X)are marked. C illustrates the production of cross-linked vimentin (par-titioned into the pellet as in B) upon the incubation of lens homogenateat Ca21 concentrations of 1 (lanes 1 and 6), 2 (lanes 2 and 7), 4 (lanes 3and 8) and 8 (lanes 4 and 9) mM for 21 h at 37 °C. Lanes 5 and 10 arethe controls with EDTA (2 mM). Lanes 1–5 show the Amido Black-stained blots, whereas lanes 6–10 are immunoblots developed with theantibody to vimentin.

Vimentin Is a Transglutaminase Substrate 7605

binant Human Vimentin with Cytosolic TGase—Two well-de-fined preparations of vimentin were employed to demonstratethat the intermediate filament component was capable of act-ing as a bifunctional substrate for TGase, that its cross-linkingcould be blocked essentially at the monomeric stage with theinclusion of either synthetic substrate of TGase (dansylcadav-erine or dansyl-e-aca-QQIV) in the reaction mixture, and thatthe dansyl-containing tracers could be employed for labelingpredominantly the monomeric parent vimentin. As illustratedby the experiments in Fig. 3, all criteria for TGase catalysiswere realized with the purified bovine lens vimentin (A), aswell as with the recombinant human vimentin as substrates(B). Under the experimental conditions employed (TGase:vi-mentin ratio of 1:25; pH 7.5; 37 °C), in the absence of dansylca-daverine, only a trace amount of monomeric vimentin re-mained at 2 h (Fig. 3, A and B, sets a, lanes 2), and the largecross-linked products of the enzymatic reaction could barelyenter into the stacking gel during SDS-PAGE. However, withthe inclusion of 2 mM dansylcadaverine (sets b and c) or dansyl-e-aca-QQIV (sets d and e) in the incubation mixture, cross-linking of the protein could be blocked essentially at the mo-nomeric stage. Lower concentrations of the inhibitors (0.5 and1 mM) still allowed for the formation of some vimentin productsof intermediate size. Photographs of the gels taken under UVillumination showed (sets c for dansylcadaverine and sets e fordansyl-e-aca-QQIV) that with 2 mM of either blocking agent inthe mixture, the fluorescent label was found almost exclusivelyin monomeric vimentin. The incorporation of the dansyl-e-aca-QQIV tracer, apparently by labeling more than a single donorsite in the parent vimentin molecule, caused a noticeable up-ward shift from the original molecular mass value of 55 kDa ofthe protein in SDS-PAGE (sets d and e).

Identification of the TGase-reactive Gln and Lys Residues ofVimentin—The experiments described in Fig. 3 were used as aguide for identifying the TGase-reactive acceptor and donorfunctionalities of vimentin. Details are given under “Materialsand Methods,” and Fig. 4 is a summary of the protocol em-ployed for analyzing the reaction of recombinant vimentin withcytosolic TGase. A key feature of the procedure was the use ofan anti-dansyl affinity column (6, 19) for separating the fluo-rescent peptides that carried the dansyl hapten from all theother fragments in the tryptic digest of the enzyme-modifiedvimentin samples. The HPLC profile of the dansylcadaverine-labeled peptides is shown in Fig. 5A, and that of the fragmentsobtained from the reaction with dansyl-e-aca-QQIV is pre-

sented in Fig. 5B. Edman degradation (12 cycles) of the twomajor fluorescent peaks marked I and II in Fig. 5A, eluting atapproximately 58 and 67 min, respectively, showed that thesepeptides were derived from the same segment of the C-terminaltail of the vimentin molecule, corresponding to the DGXVI-NETSXHH sequence from residue 451 through 462 (30). In thesequencing of the peak II material, no known amino acid (X)was recovered in the 3rd and 10th Edman cycles where Gln453

and Gln460 would be expected, and this was taken to indicatethat both Gln residues carried the dansylcadaverine adduct. Bycontrast, sequencing of the peak I material comprising thesame 451–462 segment of vimentin yielded Gln in the 3rd andthe 10th cycles, although only in amounts of about half of whatwould have been expected without derivatization by dansylca-daverine. The data are consistent with the idea that the peak Imaterial, actually appearing almost as a double peak on HPLC,represented a mixture of two peptides of the same sequence,one in which Gln453 and one in which Gln460 was decoratedwith the dansylcadaverine label DGQVINETSXHH and DGX-VINETSQHH. Altogether, it was quite remarkable to find thatTGase reacted with high selectivity with only 2 of the 33 Glnresidues present in vimentin. It was equally interesting thatthe two TGase-titratable acceptor side chains of this interme-diate filament protein (Gln453 and Gln460) were located close to

FIG. 2. The inclusion of synthetic substrates for TGase, i.e.dansylcadaverine or dansyl-e-aminocaproyl-QQIV, in the reac-tion of the calf lens homogenate with Ca21 inhibits the forma-tion of the high molecular weight vimentin species. Experimentswere carried out as in Fig. 1 in the presence of 2 mM leupeptin, 8 mMCaCl2 with either 2 mM dansylcadaverine (lanes 2, 5, and 7) or dansyl-e-aminocaproyl-QQIV (lanes 3, 6, and 8) added. After centrifugation,the supernatants (A) and pellets (B) were subjected to SDS-PAGE,followed by transfer to nitrocellulose. Blots were stained for proteinwith Amido Black (lanes 1–3) and were immunostained with antibodyto vimentin (lanes 4–6) or to dansyl (lanes 7 and 8). For experimentaldetails, see “Materials and Methods.” The position of vimentin at 55kDa (V) is indicated.

FIG. 3. TGase-catalyzed incorporation of dansylcadaverineand dansyl-e-aminocaproyl-QQIV into vimentin. Reactions werecarried out for 2 h at 37 °C in mixtures containing either purified bovinelens vimentin (0.25 mg/ml; A) or recombinant human vimentin (0.25mg/ml; B), guinea pig liver TGase (10 mg/ml), 5 mM EDTA (set a, lane 1)or 5 mM CaCl2 (for all other lanes). Dansylcadaverine (sets b and c) ordansyl-e-aminocaproyl-QQIV (sets d and e) was included at variousconcentrations (0.5–2 mM) as shown. Set a, which did not contain eitherof these compounds, served as control. The samples were subjected toSDS-PAGE and photographed under UV light (sets c and e) prior tostaining the corresponding sets (a, b, and d) with Coomassie Blue. Forexperimental details, see “Materials and Methods.”

Vimentin Is a Transglutaminase Substrate7606

each other in the primary sequence in the C-terminal unfoldeddomain of the molecule.

The TGase-mediated titration of Lys side chains with dansyl-e-aminocaproyl-QQIV provided a more complex picture. Fivefluorescent peaks (marked I–V in Fig. 5B) were collected fol-lowing the HPLC separation of the tryptic digest of labeledvimentin. Each of these was rechromatographed and submittedfor sequencing. High confidence sequences were obtained forthe peak II, III, and IV materials, identifying Lys97 and Lys104

unambiguously as residues reacting with TGase (Table I). Thepeak I material appeared to be a mixture of several labeledfragments, sequences for which could not be resolved by Edmandegradation. Peak V turned out to be a mixture of two labeledpeptides, present in essentially equal amounts; reliable readingof the double sequence, however, could be accomplished byknowing the amino acid sequence of vimentin itself. The twopeptides were derived from protein segments starting withSer293 and Glu425, respectively, providing evidence that resi-dues Lys294 and Lys439 could also act as TGase-reactive donorside chains. As illustrated in Fig. 6, in the vimentin moleculeLys294 is located near the interface of the helical 2B subdomain(121 residues) with the L2 linker segment (8 residues), whereasLys439, like the two TGase-reactive acceptor side chains Gln453

and Gln460, is in the 55 residue-length non-helical C-terminaltail. On the other hand, the donor residues Lys97 and Lys104,which were identified with high confidence from sequencingthe peak II, III, and IV peptides (Fig. 5B and Table I), straddlethe zone of transition from the non-helical N-terminal H1 seg-ment (18 residues) to the helical rod of vimentin, designated as1A (35 residues).

The Gln and Lys residues highlighted in Fig. 6 are potentialsites for various posttranslational possibilities by TGase. Theseinclude the hydrolysis of Gln residues, i.e. a Gln3Glu transi-tion; the incorporation of low molecular weight primaryamines; cross-linking of vimentin to vimentin or to other pro-teins by formation of Ne(g-glutamyl)lysine isopeptides or by apolyamine bridge (spermine, spermidine, or putrescine) be-tween two reactive Gln residues. It is clear from our data thatvimentin could act as a bifunctional substrate cross-linkingonto itself, carrying TGase-reactive acceptor (Gln) as well as

donor (Lys) residues. This may be contrasted with the sub-strate qualities of involucrin in human keratinocytes, the aniontransporter band 3 in human red cells or aB crystallin inbovine lens, each of which is known to act only as a monofunc-tional substrate for the enzyme. Involucrin and the cytoplasmicdomain of band 3 contain just a single enzyme-reactive Gln intheir sequences: Gln496, located in the C-terminal region ofinvolucrin (31), and Gln30, in the N-terminal portion of band 3(19); bovine aB crystallin was reported (5, 6) to carry only anenzyme-reactive Lys, Lys175, at the C-terminal end of the pro-tein (although there is evidence that in the lens tissue, underthe influence of the intrinsic TGase, the penultimate Lys174 ofaB crystallin also reacts2).

It is interesting to note that the TGase-reactive sites of thevimentin molecule (Fig. 6) are present in regions of the proteinthat are also known to be susceptible to other posttranslationalmodifications, such as phosphorylation and proteolysis (32–37).In general, in terms of the architecture of intermediate fila-

2 S. N. P. Murthy and L. Lorand, unpublished data.

FIG. 4. Outline of the analytical scheme for identifying the Glnand Lys residues of human vimentin, which serve as specifictargets for TGase.

FIG. 5. Identification of transglutaminase-reactive residues invimentin. A, isolation of dansylcadaverine-labeled peptides from thetrypsin digest of TGase-modified vimentin. The fluorescent peaksmarked I and II, emerging from the C8 HPLC column (see “Materialsand Methods”) at 58 and 67 min, respectively, were collected, concen-trated, and subjected to amino acid sequencing. B, isolation of dansyl-e-aminocaproyl-QQIV-labeled peptides from the trypsin digest ofTGase-derivatized vimentin. The fluorescent peaks marked I–V werecollected, concentrated, rechromatographed, and submitted for aminoacid sequencing.

Vimentin Is a Transglutaminase Substrate 7607

ment proteins, TGase seems to act with high selectivity atfunctionally important areas of vimentin. The tail domain(comprising the TGase-reactive Gln453, Gln460, and Lys439 res-idues; see Fig. 6), by interacting with the a helical rod, is knownto stabilize the correct filamentous assembly of vimentin invitro as well as in vivo (38), and was also shown to bind withactin structures (39). Tail-less vimentin mutants display anunusual tendency for abnormal type of aggregation (38, 40, 41).A nine-amino acid sequence b-motif, present in the tail do-mains of all type III IF proteins, including vimentin, has beenshown to be essential for normal IF assembly. Remarkably, asingle G3V mutation at position 452 (adjacent to the TGase-reactive Gln453 residue) is known to interfere with normalassembly (42). The area comprising two of the TGase-reactiveLys donor sites (Lys97 and Lys104), located at the beginning ofthe rod domain (Fig. 6), is highly conserved across several typesof intermediate filament proteins. This structural domain isconsidered to be critical for the assembly of the keratin units(43); a single point mutation in this segment can disrupt fila-ment formation in vitro as well as in vivo (44–50). The Lysresidues of keratins involved in cross-linking to two of the cellenvelope proteins, loricrin and involucrin, are all located in thehead domains (cross-linking to loricrin utilizes Lys73 in keratin1, Lys70 in keratin 2e, and Lys9 in keratin 10 (51); cross-linkingto involucrin occurs at Lys73 with keratin 1, Lys69 with keratin2e, and Lys71 with keratin 5 (52)). It is interesting that thisN-terminal domain is missing in the type III IF vimentinmolecule (53).

Intermediate filaments of 10 nm in diameter were longthought to be static cytoskeletal elements of cells. Recently,however, it has been shown that their polymerization state in

vivo is governed by an equilibrium between soluble vimentins,V, and the polymerized form of this protein: nV u Vn (54).TGase activity could conceivably influence the normal distri-bution and partitioning of vimentin in the cell. For example, ifthe enzyme modified mainly the soluble vimentin species, mV3 mV9 (e.g. by deamidation or cross-linking, where m , n),there might be a disruption of the IF network by depletion ofthe soluble vimentin pool available for normal polymerization.If, on the other hand, TGase acted by preferentially cross-linking (xl) the assembled subunits of vimentin in situ ((V)n3xl(V)n), the dynamic nature of the network could become com-promised because the covalent fusion of subunits would pre-clude the dissociation of the IF ensemble. Either of these pos-sibilities could have major consequences for cell function andsurvival. The change in filament stability may be important inregulating changes in cell shape that occur during develop-ment. In support of this idea, it has been shown that interme-diate filaments play major roles in cell shape and the mechan-ical properties of cytoplasm (55). Close analogy for some of thechanges that might occur in the IF network is the stabilizationof fibrin clots by activated Factor XIII, a member of the TGasefamily of enzymes. Dissection of the biochemical steps in thelast stages of blood coagulation (56) showed that the reversibleself-assembly of fibrin molecules into the clot, n(fibrin) u (fi-brin)n, is followed by the enzyme-catalyzed cross-linking of thefibrin network: (fibrin)n3 xl(fibrin)n. The ordered assembly ofmolecules into the (fibrin)n structure was shown to accelerategreatly the rate of the enzymatic reaction (57, 58), and theintroduction of a few Ne(g-glutamyl)lysine covalent side chainbridges at strategic locations into the clot results in an approx-imately 5-fold increase in clot stiffness (59, 60).

TABLE ISequencing the TGase-reactive, dansyl-«-aminocaproyl-QQIV-labeled Lys residues of human vimentin

The superscript denotes residue numbers in the primary sequence of the protein. Boldface X indicates an Edman cycle without recovery of aknown amino acid and assigned to be an enzyme-derivatized Lys.

Peak number Sequences

IIa Thr94 Glu Phe X97 Asn Thr ArgIIIa Thr101 Asn Glu X104 Val Glu Leu Gln Gln LysIVa Thr101 Asn Glu X104 Val Glu Leu GlnVb Ser293 X294 Phe Ala Asp Leu Ser Glu Ala Ala Asn

Glu425 Thr Asn Leu Asp Ser Leu Pro Leu Val Asp Thr436

a High confidence sequence.b Near equimolar mixture of two peptides, sequenced simultaneously. Identification of the sequence beginning with Glu425 was taken as evidence

that Lys439 downstream must have been derivatized with the dansyl-containing tracer by TGase.

FIG. 6. Location of TGase-reactiveGln and Lys residues of vimentin inrelation to the architecture of this in-termediate filament protein. A, theamino acid sequence for human vimentin(27) with the bold letters highlighting theLys (97, 104, 294, and 439) and Gln (435and 460) residues of the protein, which wehave shown to incorporate dansyl-e-aca-QQIV or dansylcadaverine in the TGase-directed reaction. Residues identified inthe sequence with highest confidence (i.e.,Lys-97, Lys-104, Gln-435, and Gln-460)are marked with asterisks. B, organiza-tion of subdomains in vimentin accordingto Steinert and Parry (53). The number ofamino acid residues assigned to the vari-ous segments are indicated in italics, andthe TGase-reactive Gln and Lys residuesare marked by arrows.

Vimentin Is a Transglutaminase Substrate7608

It remains to be seen when and how TGase becomes acti-vated in the lens. We have recently shown3 that the cross-linking activity of the purified lens enzyme is inhibited by GTP.In fact, the sensitivity of the lens TGase to GTP (which prob-ably reflects on the affinity of the enzyme for binding thenucleotide) is about an order of magnitude greater than that ofthe guinea pig liver enzyme. Thus, like the liver TGase, whichwas shown to function as a G protein in signal transduction(61–65), the lens enzyme seems to represent a gene productpotentially with a dual role in the lens cell. Under the appro-priate signal, it may mediate the activation of phospholipase C.Otherwise, it could function in remodeling reactions, as de-scribed in this report for the modification of vimentin. Theconcentration of GTP would affect the cross-linking potential ofthe lens enzyme, and a significant drop in the concentration ofthe nucleoside triphosphate, as it may occur in cataract forma-tion, would greatly favor the expression of this activity.

Acknowledgment—We thank Dr. Ying-Hao Chou for helpfuldiscussion.

REFERENCES

1. Lorand, L., Hsu, L. K. H., Siefring, G. E., Jr., and Rafferty, N. S. (1981) Proc.Natl. Acad. Sci. U. S. A. 78, 1356–1360

2. Berbers, G. A., Bentlage, H. C., Brans, A. M., Bloemendal, H., and de Jong,W. W. (1983) Euro. J. Biochem. 135, 315–320

3. Parameswaran, K. N., Velasco, P. T., Wilson, J., and Lorand, L. (1990) Proc.Natl. Acad. Sci. U. S. A. 87, 8472–8475

4. Lorand, L., Parameswaran, K. N., and Velasco, P. T. (1991) Proc. Natl. Acad.Sci. U. S. A. 88, 82–83

5. Groenen, P. J., Bloemendal, H., and de Jong, W. W. (1992) Eur. J. Biochem.205, 671–674

6. Lorand, L., Velasco, P. T., Murthy, S. N. P., Wilson, J., and Parameswaran,K. N. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 11161–11163

7. Lorand, L., Parameswaran, K. N., Velasco, P. T., and Murthy, S. N. P. (1992)Bioconjugate Chem. 3, 37–41

8. Groenen, P. J., Seccia, M., Smulders, R. H., Gravela, E., Cheeseman, K. H.,Bloemendal, H., and de Jong, W. W. (1993) Biochem. J. 295, 399–404

9. Siefring, G. E., Apostol, A. B., Velasco, P. T., and Lorand, L. (1978)Biochemistry 17, 2598–2604

10. Bjerrum, O. J., Hawkins, M., Swanson, P., Griffin, M., and Lorand, L. (1981)J. Supramol. Struct. Cell Biochem. 16, 289–301

11. Rice, R. H., and Green, H. (1979) Cell 18, 681–69412. Steinert, P. M., and Idler, W. W. (1979) Biochemistry 18, 5664–566913. Trejo-Skalli, A. V., Velasco, P. T., Murthy, S. N. P., Lorand, L., and Goldman,

R. D. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 8940–894414. Clément, S., Trejo-Skalli, A. V., Gu, L., Velasco, P. T., Lorand, L., and

Goldman, R. D. (1997) J. Invest. Dermatol. 109, 778–78215. Sax, C. M., Farrell, F. X., Zehner, Z. E., and Piatigorsky, J. (1990) Dev. Biol.

139, 56–6416. Sax. C. M., Stover, D. M., Ilagan, J. C., Zehner, Z. E., and Piatigorsky, J. (1993)

Gene 130, 277–28117. Bagchi, M., Caporale, M. J., Wechter, R. S., and Maisel, H. (1985) Exp. Eye Res.

40, 385–39218. Lorand, L., Rule, N. G., Ong, H. H., Furlanetto, R., Jacobsen, A., Downey, J.,

Oner, N., and Bruner-Lorand, J. (1968) Biochemistry 7, 1214–122319. Murthy, S. N. P., Wilson, J., Zhang, Y., and Lorand, L. (1994) J. Biol. Chem.

269, 22907–2291120. Laemmli, U. K. (1970) Nature 227, 680–68521. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. U. S. A.

76, 4350–435422. Yang, H. Y., Lieska, N., Goldman, A. E., and Goldman, R. D. (1985) J. Cell Biol.

100, 620–63123. Velasco, P. T., Karush, F., and Lorand, L. (1988) Biochem. Biophys. Res.

Commun. 152, 505–51124. Lieska, N., Chen, J., Maisel, H., and Romero-Herrera, A. E. (1980) Biochim.

Biophys. Acta 626, 136–15325. Lehto, V. P., and Virtanen, I. (1983) J. Cell Biol. 96, 703–71626. Chou, Y. H., Opal, P., Quinlan, R. A., and Goldman, R. D. (1996) J. Cell Science

109, 817–82627. Croall, D. E., and DeMartino, G. N. (1986) Cell Calcium 7, 29–3928. Jeong, J. M., Murthy, S. N. P., Radek, J. T., and Lorand, L. (1995) J. Biol.

Chem. 270, 5654–565829. Lorand, L., Conrad, S. M., and Velasco, P. T. (1985) Biochemistry 24,

1525–153130. Ferrari, S., Battini, R., Kaczmarek, L., Rittling, S., Calabretta, B., de Riel,

J. K., Philiponis, V., Wei, J. F., and Baserga, R. (1986) Mol. Cell. Biol. 11,3614–3620

31. Simon, M., and Green, H. (1988) J. Biol. Chem. 263, 18093–1809832. Nelson, W. J., and Traub, P. Mol. Cell. Biol. 3, 1146–115633. Geisler, N., and Weber, K. (1988) EMBO J. 7, 15–2034. Ando, S., Tanabe, K., Gonda, Y., Sato, C., and Inagaki, M. (1989) Biochemistry

28, 2974–297935. Evans, R. M. (1989) J. Cell Biol. 108, 67–7836. Geisler, N., Hatzfeld, M., and Weber, K. (1989) Eur. J. Biochem. 183, 441–44737. Chou, Y. H., Bischoff, J. R., Beach, D., and Goldman, R. D. (1990) Cell 62,

1063–107138. McCormick, M. B., Kouklis, P., Syder, A., and Fuchs, E. (1993) J. Cell Biol.

122, 395–40739. Klymkowsky, R., Evans, R., Domingo, A., and Backhus, J. (1994) J. Cell Sci.

107, 1609–162240. Eckelt, A., Herrmann, H., and Franke, W. W. (1992) Eur. J. Cell Biol. 58,

319–33041. Rogers, K. R., Eckelt, A., Nimmrich, V., Janssen, K. P., Schliwa, M.,

Herrmann, H., and Franke, W. W. (1995) Eur. J. Cell Biol. 66, 136–15042. Kouklis, P. D., Hatzfeld, M., Brunkener, M., Weber, K., and Georgatos, S. D.

(1993) J. Cell Sci. 106, 919–92843. Coulombe, P. A., Chan, Y. M., Albers, K., and Fuchs, E. (1990) J. Cell Biol. 111,

3049–306444. Coulombe, P. A., Hutton, M. E., Letai, A., Hebert, A., Paller, A. S., and Fuchs,

E. (1991) Cell 66, 1301–131145. Lane, E. B., Rugg, E. L., Navsaria, H., Leigh, I. M., Heagerty, A. H., Ishida-

Yamamoto, A., and Eady, R. A. (1992) Nature 356, 244–24646. Hovnanian, A., Pollack, E., Hilal, L., Rochat, A., Prost, C., Barrandon, Y., and

Goossens, M. (1993) Nat. Genet. 3, 327–33247. Sybert, V. P., and Stephens, K. (1994) J. Invest. Dermatol. 102, 82248. Chen, H., Bonifas, J. M., Matsumura, K., Ikeda, S., Leyden, W. A., and

Epstein, E. H., Jr. (1995) J. Invest. Dermatol. 105, 629–63249. Chan, Y. M., Cheng, J., Gedde-Dahl, T., Jr., Niemi, K. M., and Fuchs, E. (1996)

J. Invest. Dermatol. 106, 327–33450. Fuchs, E. (1997) Mol. Biol. Cell 8, 189–20351. Steinert, P. M., and Marekov, L. N. (1995) J. Biol. Chem. 270, 17702–1771152. Steinert, P. M., and Marekov, L. N. (1997) J. Biol. Chem. 272, 2021–203053. Steinert, P. M., and Parry, D. A. (1985) Annu. Rev. Cell Biol. 1, 41–6554. Vikstrom, K. L., Lim, S. S., Goldman, R. D., and Borisy, G. G. (1992) J. Cell

Biol. 118, 121–12955. Goldman, R. D., Khuon, S., Chou, Y. H., Opal, P., and Steinert, P. M. (1996)

J. Cell Biol. 134, 971–98356. Mann, K. G., and Lorand, L. (1993) Methods Enzymol. 222, 1–1057. Samokhin, G. P., and Lorand, L. (1995) J. Biol. Chem. 270, 21827–2183258. Lorand, L., Parameswaran, K. N., and Murthy, S. N. P. (1998) Proc. Natl.

Acad. Sci., U. S. A. 95, 537–54159. Mockros, L. F. Roberts, W. W., and Lorand, L. (1974) Biophys. Chem. 2,

164–16960. Shen, L., and Lorand, L. (1983) J. Clin. Invest. 71, 1336–134161. Im, M.-J., and Graham, R. M. (1990) J. Biol. Chem. 265, 18944–1895162. Im, M.-J., Riek, R. P., and Graham, R. M. (1990) J. Biol. Chem. 265,

18952–1896063. Im, M.-J., Gray, C., and Rim, A. J. (1992) J. Biol. Chem. 267, 8887–889464. Nakaoka, H., Perez, D. M., Back, K. J., Das, T., Husain, A., Misono, K., Im,

M.-J., and Graham, R. M. (1994) Science 264, 1593–159665. Iisma, S. E., Chung, L., Wu, M-J., Teller, D. C., Yee, V. C., and Graham, R. M.

(1997) Biochemistry 36, 11655–116643 S. N. P. Murthy, P. T. Velasco, and L. Lorand, unpublished data.

Vimentin Is a Transglutaminase Substrate 7609

The Intermediate Filament Protein, Vimentin, in the Lens Is a Target for Cross-linking by Transglutaminase*MATERIALS AND METHODSRESULTS AND DISCUSSIONACKNOWLEDGEMENTREFERENCES