Journal of Science 113, 483-491 (2000) 483 Printed in Great Britain © The Company of Biologists Limited 2000 JCS0847

Dose-dependent linkage, assembly inhibition and disassembly of and 5/14 filaments through plectin’s intermediate filament-binding domain

Ferdinand A. Steinböck1, Branislav Nikolic1, Pierre A. Coulombe2, Elaine Fuchs3, Peter Traub4 and Gerhard Wiche1,* 1Institute of Biochemistry and Molecular Cell Biology, University of Vienna, Vienna Biocenter, 1030 Vienna, Austria 2Departments of Biological Chemistry and Dermatology, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA 3Howard Hughes Medical Institute, Departments of Molecular Genetics and Cell Biology and of Biochemistry and Molecular Biology, The University of Chicago, Chicago, IL 60637, USA 4Max-Planck-Institut für Zellbiologie, D-68526 Ladenburg, Germany *Author for correspondence (e-mail: [email protected])

Accepted 26 November 1999; published on WWW 19 January 2000

SUMMARY

Plectin, the largest and most versatile member of the became more and more crosslinked upon incubation with cytolinker/ family of characterized to date, increasing concentrations of plectin repeat 5. However, at has a tripartite structure comprising a central 200 nm-long high proportions of plectin to IF proteins, disassembly of α-helical rod domain flanked by large globular domains. filaments occurred. Again, vimentin filaments proved The C-terminal domain comprises a short tail region considerably more sensitive towards disassembly than preceded by six highly conserved repeats (each 28-39 kDa), those composed of 5 and 14. In general, IFs one of which (repeat 5) contains plectin’s intermediate formed from recombinant proteins were found to be filament (IF)-binding site. We used recombinant and native slightly more responsive towards plectin influences than proteins to assess the effects of plectin repeat 5-binding to their native counterparts. A dose-dependent plectin- IF proteins of different types. Quantitative Eu3+-based inflicted collapse and putative disruption of IFs was also overlay assays showed that plectin’s repeat 5 domain bound observed in vivo after ectopic expression of vimentin and to type III IF proteins (vimentin) with preference over type plectin’s repeat 5 domain in cotransfected vimentin- I and II cytokeratins 5 and 14. The ability of both types of deficient SW13 (vim−) cells. Our results suggest an IF proteins to self-assemble into filaments in vitro was involvement of plectin not only in crosslinking and impaired by plectin’s repeat 5 domain in a concentration- stabilization of cytoskeletal IF networks, but also in dependent manner, as revealed by negative staining and regulation of their dynamics. rotary shadowing electron microscopy. This effect was much more pronounced in the case of vimentin compared Key words: Plectin, Cytolinker , Intermediate filament, to cytokeratins 5/14. Preassembled filaments of both types Vimentin, Cytokeratin

INTRODUCTION neurofilament proteins and B), and its interaction with the , components of the subplasma The intermediate filament-associated protein plectin (Wiche et membrane skeleton (α-/fodrin), transmembrane al., 1982) is abundantly expressed in many different tissues and receptors, and high molecular mass cell types. Based on the phenotypic analyses of epidermolysis associated proteins (for recent reviews, see Fuchs and bullosa simplex (EBS)-MD patients, who carry autosomal Cleveland, 1998; Wiche, 1998; Steinböck and Wiche, 1999). recessive mutations in the plectin gene (Gache et al., 1996; Together with the neuronal and epithelial isoforms of BPAG1 McLean et al., 1996; Smith et al., 1996), as well as of plectin- (Stanley et al., 1981), /BPAG1n (Brown et al., deficient mice (Andrä et al., 1997), the protein has been 1995a,b; Yang et al., 1996) and BPAG1e (Sawamura et al., proposed to provide cells with mechanical strength by acting 1991), (Ruhrberg et al., 1997), as a linker element of the cytoskeleton. This was supported by (Ruhrberg et al., 1996) and (Green et al., 1990; several previous studies showing the binding of plectin to 1992), plectin forms a family of structurally related proteins, cytoplasmic as well as nuclear intermediate filament (IF) referred to as the plakin or cytolinker protein family (Uitto et subunit proteins (vimentin, , cytokeratins, GFAP, al., 1996; Wiche, 1998). 484 F. A. Steinböck and others

With a molecular mass of >500 kDa, plectin is the largest urea, 50 mM Tris, pH 8.1, 2 mM DTT and 0.3 mg/ml PMSF. Further known cytolinker protein. The molecule has a dumbbell-like purification of vimentin and cytokeratins was done by ion-exchange shape with two globular domains flanking a long α-helical column chromatography using Pharmacia DEAE Sepharose CL-6B. coiled-coil rod structure (Foisner and Wiche, 1987; Wiche et Proteins were eluted by a gradient of 0-0.2 M guanidine-HCl in − al., 1991). Transient transfection of plectin cDNA expression solubilizing solution and stored at 80¡C. The molecular masses of constructs into monkey (COS) and rat kangaroo (PtK2) cell the purified recombinant versions of vimentin and cytokeratins 5 and 14 were indistinguishable from their native counterparts upon analysis lines, both of which contain dense networks of vimentin as well by SDS-PAGE (data not shown). The purity of protein preparations as cytokeratin filaments, indicated that only the carboxy- was as documented previously (Coulombe and Fuchs, 1990; Nikolic terminal globular domain, and not the amino-terminal globular et al., 1996). or the central rod domain of the molecule, was indispensable for IF association (Wiche et al., 1993). In subsequent studies Purification of IF proteins from cells and tissues the IF-binding site of plectin could be mapped to a stretch of Vimentin was purified from rat glioma C6 cells following the protocol approximately 50 amino acid residues residing within repeat 5, of Foisner et al. (1988). Samples of vimentin were stored frozen in 6 the penultimate of six homologous repeats constituting M urea at −80ûC. Cytokeratins (K5, K6, K14, K17 and small amounts plectin’s carboxy-terminal globular domain (Nikolic et al., of K16) were isolated from human skin and purified by Mono Q 1996). This stretch is located just downstream of a highly column according to Wawersik et al. (1997). conserved central region approximately 200 amino acids long In vitro assembly of IFs (consisting of tandem repeats of a 19-amino-acid motif), which Prior to assembly, samples of recombinant or native vimentin were is common to all six carboxy-terminal repeats (Wiche et al., dialyzed against 5 mM Tris, pH 8.5, 1 mM EDTA and 1 mM DTT 1991; Liu et al., 1996). (overnight at 4°C), and centrifuged at 200000 g for 10 minutes at 4°C, To investigate the binding and networking capacities of using a Beckman benchtop ultracentrifuge. After addition of 0.1 plectin for different types of IFs we have carried out volumes of 200 mM Tris, pH 7.0, and 1.6 M NaCl, assembly mixtures biochemical and ultrastructural studies using native and/or were incubated for 1 hour at 37¡C. Cytokeratins 5 and 14 were bacterially expressed recombinant forms of vimentin (type III coassembled by dialysis first against 5 M urea in 50 mM Tris, pH 8.1, IF protein) and cytokeratins 5 and 14 (type I and II IF proteins). 2 mM DTT, 0.3 mg/ml PMSF, for 2 hours at room temperature, then We report here that plectin’s repeat 5 domain was able to bind against 2.5 M urea (in the same solution) for 2 hours at room temperature, and finally against 2 mM Tris, pH 9.0, overnight at 4¡C. to and crosslink all types of IF proteins, but showed higher Filament assembly was induced by addition of 0.1 volumes of 100 affinity for type III IF proteins. Interestingly, we found that this mM Tris, pH 7.0, followed by incubation for 2 hours at room domain could also inhibit filament assembly in vitro in a temperature. concentration-dependent manner, with distinct efficiencies depending on the type of IF subunit protein, and that it could Electron microscopy cause the disruption of assembled IFs. For negative staining, 6 µl of in vitro assembly mixtures pipetted onto Formvar/carbon-coated and glow-discharged 400 mesh copper grids, were stained with 10 drops of 1% uranyl acetate for approximately 90 MATERIALS AND METHODS seconds. For rotary shadowing, samples (100 µl, mixed 1:1 with glycerol) were sprayed onto freshly cleaved mica and prepared for Expression plasmids electron microscopy essentially as described by Tyler and Branton For bacterial expression, full-length mouse vimentin cDNA (GenBank (1980). Shadowing with platinum (0.74 nm) occurred at an angle of accession number M26251), including the stop codon at position bp 7¡, with carbon (10 nm) at 90¡. Specimens were visualized in a JEOL 1879-1881, was excised from mammalian expression vector pMC- 1210 electron microscope operated at 80 kV. V21 (Ming Chen, 1992) and subcloned in several steps into a modified version of plasmid pET23a (Novagen), yielding plasmid pFS129. Tag- Europium overlay binding assay less vimentin encoded by pFS129 was used in all experiments. Urea-solubilized IF proteins were dialyzed against 50 mM NaHCO3, pET15b-derived plasmids pBN135 and pBN132 (encoding pH 8.5, and labeled with Eu3+ overnight at room temperature, using plectin’s repeat domains 4 and 5, respectively) have been described in 10 µl Eu3+-labeling reagent per 100 µl protein (0.5-1.5 mg/ml) Nikolic et al. (1996), and plasmids pET-K5 and pET-K14 (encoding according to the protocol of the manufacturer (Wallac, Turku, human cytokeratins K5 and K14, respectively) in Coulombe and Finland). Fuchs (1990). For binding assays, 96-well microtiter plates were coated For transfection of cultured cells, we used plasmids pBN36 and (overnight at 4¡C) with 100 µl of recombinant plectin proteins (100 pBN47, both derived from mammalian expression vector pAD29 nM) or BSA type H1 (Gerbu, Gaiberg, Germany), all in 25 mM (Nikolic et al., 1996), encoding amino acid residues leu4069-leu4367 Na2B4O7, pH 9.3. Coating was followed by blocking with 4% BSA (repeat 5) and tyr3780-thr4024 (repeat 4), respectively. in PBS, pH 7.5, for at least 1 hour at room temperature. After washing with PBS, pH 7.5, plates were overlayed with dilutions of Eu3+- Recombinant proteins labeled proteins (10-500 nM) in 100 µl of PBS, pH 7.5, 1 mM EGTA, Recombinant proteins corresponding to the repeat 4 (pBN135) or 2 mM MgCl2, 1 mM DTT, 0.1% Tween 20, for 90 minutes at room repeat 5 (pBN132) domains of rat plectin, mouse vimentin (pFS129) temperature. Plates were washed at least six times with PBS, pH 7.5, and human cytokeratins 5 (pET-K5) and 14 (pET-K14), were and protein bound was then determined by releasing the complexed expressed in E. coli BL21(DE3). His-tagged versions of recombinant Eu3+ with enhancement solution and measuring fluorescence with a proteins (plectin mutant proteins) were purified by affinity Delfia time-resolved fluorometer (Wallac). Counts were converted to chromatography on Ni2+ columns, following the protocol of the concentrations by comparison with a 1 nM Eu3+ standard. manufacturer (Novagen, pET system manual). Mouse vimentin and human cytokeratins 5 and 14 (untagged) were isolated from lysed Cell culture, DNA transfection and immunofluorescence bacterial pellets following the inclusion body preparation procedure microscopy of Nagai and Thogersen (1987). Final pellets were solubilized in 6 M Vimentin-deficient human adenocarcinoma SW13 (vim−) cells (Sarria Effects of plectin on IF assembly 485 et al., 1994) were grown attached to glass slides (in plastic dishes) in recombinant forms of cytokeratins K6 and K16 (data not DMEM supplemented with 2 mM L-glutamine, 50 U/ml penicillin, shown). This confirmed earlier observations that suggested a 50 µg/ml streptomycin and 10% heat-inactivated FCS. Subconfluent preferred interaction of plectin’s repeat 5 domain with vimentin cell cultures were transiently transfected with 30 µg plasmid DNA/10 rather than cytokeratin filaments upon overexpression in PtK2 cm plate using the calcium phosphate precipitation method (Graham cells (Nikolic et al., 1996). and van der Erb, 1973). 16-64 hours post-transfection cells were fixed − ° in methanol at 20 C and processed for immunofluorescence Plectin’s repeat 5 domain affects self-assembly of IF microscopy as previously described (Wiche et al., 1993). The proteins in vitro following immunoreagents were used: affinity-purified goat anti- mouse vimentin IgG (Giese and Traub, 1986) and anti-myc mAb 1- Vimentin filaments assembled in vitro became packed into 9E10.2 (American Type Culture Collection) as primary antibodies; dense aggregates upon incubation with the IF-binding repeat 5 Texas Red-conjugated AffiniPure donkey anti-mouse IgG (H+L) and domain of plectin (Nikolic et al., 1996). To assess whether this fluorescein (DTAF)-conjugated AffiniPure donkey anti-goat IgG domain could also influence self-assembly of IF subunit (H+L), both with minimal species crossreactivity (Accurate, proteins, purified and urea-solubilized vimentin from rat Westbury, NY), as secondary antibodies. Specimens were viewed glioma C6 cells were freed of urea and exposed to in vitro using a Zeiss Axiophot fluorescence microscope (Carl Zeiss, assembly conditions in the presence of increasing Oberkochen, Germany) equipped with a BioRad MRC600 confocal concentrations of the plectin mutant protein. At a molar ratio laser scanning system (BioRad Laboratories, Richmond, CA). Digital images were processed using the software package Adobe Photoshop of vimentin:plectin repeat 5 of 10:1, numerous filaments 5.0. decorated with globular structures were observed by negative staining electron microscopy (Fig. 2, row A). Frequently we observed long undecorated filaments of uniform width and RESULTS occasionally a few disassembled and/or unravelled filaments. Changing the ratio to 5:1 by increasing the concentration of Plectin’s repeat 5 domain binds to vimentin in plectin led to a decrease in the number and length of assembled preference to cytokeratins filaments (Fig. 2, row B). Consistent with this trend, at a 1:1 In a previous study a unique IF binding site comprising ratio of vimentin:repeat 5 only very few and short assembled approximately 50 amino acid residues and containing a filaments were observed (Fig. 2, row C). Finally, at a bipartite nuclear localization signal was mapped to the proportion of 1:2 (vimentin:repeat 5), filaments could not be carboxy-terminal repeat 5 domain of plectin. Recombinant visualized anymore. Instead, the grids were covered with mutant proteins containing this domain, but not those lacking irregularly shaped structures forming aggregates of various it, were found to bind to vimentin in vitro and to associate with sizes (Fig. 2, row D). In control experiments carried out with IFs in vivo upon expression in cultured cells (Nikolic et al., a recombinant version of plectin’s repeat 4 instead of the repeat 1996). To assess the relative binding affinities of this domain to distinct IF subunit proteins, rat glioma C6 cell (native) A B vimentin and/or recombinant forms of 4 4 vimentin and cytokeratins 5 and 14 were 3 3 Eu3+-labeled and overlayed in increasing concentrations onto plectin’s repeat 5 2 2 domain immobilized on microtiter plates. 1 1 The binding data of native vimentin (Fig. protein bound (pmol) 1A) were close to linear over the protein bound (pmol) 0 0 concentration range tested (0-500 nM) 0 200 400 600 0 200 400 600 enabling a KD estimate of approximately C6 cell vimentin (nM) recombinant vimentin (nM) 1.0 µM. Using plectin’s repeat 4 domain, which lacks an IF binding site (Nikolic et C D al., 1996), instead of repeat 5 domain, no 4 4 significant binding was observed (Fig. 1A). 3 3 Data obtained with bacterially expressed mouse vimentin were very similar to those 2 2 of native vimentin (Fig. 1B). 1 1 Recombinant versions of human 0 0 cytokeratin proteins showed lower, protein bound (pmol) protein bound (pmol) 0 200 400 600 0 200 400 600 though still significant, binding to plectin’s repeat 5 domain (Fig. 1C,D). Assayed 5 (nM) (nM) individually, cytokeratins K5 and K14 3+ bound at levels of approximately 25-40%, Fig. 1. Concentration-dependent binding of various Eu -labeled IF proteins to immobilized plectin’s repeat 5 domain. Purified His-tagged recombinant versions of compared to vimentin. Binding data plectin’s repeat 5 (closed circles) and repeat 4 (open circles) domains were coated onto obtained using equimolar mixtures of microtiter plates at concentrations of 100 nM, and overlaid with Eu3+-labeled rat glioma cytokeratins 5 and 14 were intermediate C6 cell vimentin (A), recombinant mouse vimentin (B), human cytokeratin 5 (C) and between those of each one alone (data not human cytokeratin 14 (D), at the concentrations indicated. Results were obtained from at shown). Similar results were obtained using least three independent assays, each carried out with triplicate sets of data. 486 F. A. Steinböck and others

5 domain, filament assembly of vimentin occurred without To assess effects on cytokeratin filament assembly, similar detectable interference (data not shown). experiments were carried out using plectin mutant proteins Negatively stained IFs assembled from recombinant and equimolar mixtures of bacterially expressed cytokeratins mouse vimentin ultrastructurally appeared similar to filaments K5 and K14, or a mixture of epidermal human cytokeratins formed from native vimentin (Fig. 3A). When assembled containing K5, K6, K14, K17 and small amounts of K16 in the presence of increasing concentrations of plectin’s repeat 5 domain, recombinant vimentin formed lesser and shorter filaments, until at sufficiently high ratios of plectin to vimentin, filament formation was totally supressed (Fig. 3B). Again, plectin’s repeat 4 domain did not show this effect (data not shown). When visualized by rotary shadowing electron microscopy, filaments assembled from recombinant vimentin displayed a beaded substructure (Fig. 3D), typical of filaments assembled from non-recombinant vimentin (Henderson et al., 1982; Foisner et al., 1988). Under these conditions the inhibitory effects of plectin mutant proteins on filament formation became apparent not only for the repeat 5 domain (Fig. 3F), but to a certain extent also for the repeat 4 domain. Unlike negatively stained specimens (data not shown), rotary- shadowed filaments assembled in the presence of a 2:1 molar excess of plectin’s repeat 4 domain over vimentin, appeared fragmented and less compact (Fig. 3E). In addition, a relatively high number of globular structures with or without extending filamentous protrusions were observed under these conditions. This indicated very low affinity binding of plectin’s repeat 4 domain to IFs, which was hardly detectable using the Eu3+ binding assay (see Fig. 1). Also, hardly any of the nonfilamentous aggregates observed after incubation of vimentin with plectin’s repeat 5 domain in negatively stained specimens (Fig. 2, row D) were detectable by the rotary shadowing technique (Fig. 3F). These differences in specimen visualization were likely due to increased mechanical shear forces Fig. 2. Electron microscopy of uranyl acetate-stained C6 cell vimentin assembled in the presence of during sample application, in increasing concentrations of plectin’s repeat 5 domain. Micrographs of three independent particular spraying onto grids for experiments (rows I-III) are shown. The following molar ratios of C6 cell vimentin:plectin’s repeat 5 rotary shadowing, compared to domain were used: (A) 10:1; (B) 5:1; (C) 1:1; (D) 1:2. Note the decrease in number and length of negative staining. assembled filaments in A-C, and the absence of filaments in D. Bar, 200 nm. Effects of plectin on IF assembly 487

Fig. 3. Electron microscopy of uranyl acetate-stained (A-C) and rotary shadowed (D-F) filaments assembled from recombinant vimentin in the presence or absence of plectin mutant proteins or plectin’s repeat 5 alone (C). (A,B) Recombinant vimentin assembled alone (A), or in the presence of a twofold molar excess of plectin’s repeat 5 domain (B). Note the lack of filaments in B. (C) Plectin’s repeat 5 domain subjected to filament assembly conditions at a concentration similar to that used in B. (D-F) Recombinant vimentin assembled alone (D), or in the presence of a twofold molar excess of plectin’s repeat 4 (E) or repeat 5 domain (F). Note the shorter length and partial disintegration of filaments, as well as more frequent globular structures in E, compared to D. A few filaments in D show signs of disintegration, usually not seen in equivalent samples of C6 cell (native) vimentin (data not shown). Bars, 200 nm (A,B-F).

(Fig. 4A). Analysis by rotary shadowing electron microscopy revealed hardly any difference between filaments assembled in the absence of plectin’s repeat 5 domain (Fig. 4B) and those assembled from a 1:2 mixture (cytokeratins:repeat 5) (Fig. 4C). At a corresponding ratio of 1:4 very few filaments were visible (Fig. 4D), and none at a ratio of 1:8 (Fig. 4E). In the case of epidermal (native) cytokeratins (Fig. 4F-H), even at a molar ratio of IF proteins to plectin’s repeat 5 domain of 1:8, a few short filaments were visualized by rotary shadowing electron microscopy (Fig. 4H). Thus, as in the case of vimentin, effects of plectin mutant proteins on cytokeratin filament assembly were less evident using native compared to recombinant protein subunits, or negative staining compared to rotary shadowing. Filaments preformed from recombinant vimentin disassemble upon incubation with plectin’s repeat 5 domain When incubated with plectin mutant proteins, filaments preassembled from recombinant vimentin exhibited slightly different properties compared to the ones assembled from native (rat glioma C6 cell) vimentin. Using negative staining electron microscopy, filaments reconstituted from recombinant (but not native) vimentin were found to partially disassemble upon incubation with plectin’s repeat 5 domain at a twofold molar excess over vimentin (Fig. 5A). Under similar conditions plectin’s repeat 4 domain showed no detectable effects on either type of filaments (data not shown). In contrast, when specimens were viewed after rotary shadowing, a decrease in number, length and apparent compactness of the reconstituted filaments also became noticeable with plectin’s repeat 4 domain (compare Fig. 5B,C), although the more drastic effect of plectin’s repeat 5 domain was uncontested. Filaments assembled from recombinant vimentin no longer existed after an incubation with the repeat 5 domain at equimolar ratios (Fig. 5D). At a fivefold molar excess of the plectin protein we observed the formation of globular structures (Fig. 5E), also shown in Fig. 4E,H, where ctyokeratins 5 and 14 were (Wawersik et al., 1997). Consistent with the weaker affinity assembled in the presence of an eightfold molar excess of of plectin’s repeat 5 domain to cytokeratins compared to repeat 5. When viewed by rotary shadowing, even filaments vimentin, as revealed by the microtiter plate binding assay consisting of native vimentin showed partial decomposition (Fig. 1), we found that higher concentrations of plectin upon incubation with equimolar ratios of repeat 5 (Fig. 5F). mutant protein were required to inhibit filament formation Taken together these results suggested that filaments assembled (Fig. 4). As revealed by negative staining electron from recombinant vimentin were less resistant towards plectin microscopy, assembly of filaments at molar ratios of mutant protein-induced disassembly than their native recombinant cytokeratins K5 and K14:repeat 5 of 1:8 still led counterparts. In contrast, filaments formed from recombinant to the formation of filaments, although at reduced amounts versions of cytokeratins K5 and K14 showed no effects upon 488 F. A. Steinböck and others

Fig. 4. Electron microscopy of filaments assembled from recombinant (A-E) or native cytokeratins 5 and 14 (F-H) in the presence or absence of plectin mutant proteins. Uranyl acetate- stained (A) and rotary shadowed (B-H) specimens are shown. (A) Recombinant cytokeratins assembled in the presence of eightfold molar excess of plectin’s repeat 5 domain. (B-E) Recombinant cytokeratins assembled alone (B), or in the presence of twofold (C), fourfold (D), or eightfold (E) molar excess of plectin’s repeat 5 domain. (F-H) Native cytokeratins assembled alone (F), or in the presence of eightfold molar excess of plectin’s repeat 4 (G), or repeat 5 domain (H). Bar, 100 nm (A); 200 nm (B-H).

Fig. 5. Disassembly of preassembled vimentin filaments upon incubation with plectin mutant proteins. Electron micrographs of uranyl acetate- stained (A) and rotary shadowed specimens (B-F) are shown. (A-E) Filaments assembled from recombinant vimentin were incubated with a twofold molar excess of plectin’s repeat 5 domain (A), or without plectin (B), or with a twofold molar excess of plectin’s repeat 4 domain (C), or with an equimolar amount (D) or fivefold molar excess of plectin’s repeat 5 domain (E). (F) Filaments assembled from C6 cell vimentin were incubated with equimolar amount of plectin’s repeat 5 domain. Bar, 100 nm (A,F); 200 nm (B-E). Effects of plectin on IF assembly 489 incubation with plectin mutant proteins (repeats 4 or 5) under similar conditions (data not shown). Plectin’s repeat 5 domain induces aggregation of ectopically expressed vimentin in vimentin-deficient SW13 cells Plectin mutant proteins containing the IF-binding domain have previously been shown to align with and cause the collapse of endogenous IF networks upon transient expression in cultured cells (Wiche et al., 1993; Nikolic et al., 1996). To study the effects of plectin mutant proteins on filament and IF network formation in a cellular environment free of preformed IFs, we transiently cotransfected cDNA expression constructs encoding full-length mouse vimentin and a myc-tagged plectin mutant protein (pBN36) comprising most of the carboxy-terminal repeat 5 domain, including the IF-binding site, into the vimentin-deficient human adenocarcinoma cell line SW13 (vim−). The localization of both ectopically expressed proteins was then assessed by double immunofluorescence microscopy, using anti-vimentin and anti-myc antibodies, at several time points post-transfection. Already at 16 hours post-transfection we observed colocalization of the plectin mutant protein with small, dense, vimentin-positive aggregates of unknown structure (Fig. 6A,B). At 36 hours, larger aggregates became prominent, and additional filamentous vimentin structures extending from these aggregates, or forming independently at more peripheral areas of the cells, were visible (Fig. 6C,D). 64 hours post-transfection many of the cells had died, apparently as a result of the overexpression of repeat 5. In the very few transfected cells visualized, vimentin was found in the form of dense aggregates, but hardly in filamentous form. Plectin’s repeat 5 domain was found colocalized with these aggregates as well as diffusely distributed throughout the whole cells (Fig. 6E,F). Note that images shown in Fig. 6A-F are representative of approximately 90% of the cell population at each time point (data not shown). Due to differences in protein expression levels various transition states were observed in the rest of the cells at all time points. Consistent with the in vitro data obtained using purified proteins (see above), the phenotypes observed in transfected SW13 cells could be most easily explained by assuming that the extent of filament aggregation caused by the plectin mutant protein was linked to the relative expression levels of both proteins Fig. 6. Coexpression of vimentin and plectin carboxy-terminal domains in the cells. In single-transfected control cells a diffuse (repeats 4 and 5) in vimentin-deficient SW13 cells. cDNA expression distribution of pBN36-encoded plectin mutant protein plasmids pBN36, pBN47 and pMC-V21, encoding the carboxy-terminal rat was observed (Fig. 6G, and data not shown) at all time plectin sequences L4069-L4367 (repeat 5), Y3780-T4024 (repeat 4) (SWISS- points. Single-transfection of cells with the vimentin PROT P30427) and full-length mouse vimentin, respectively, were expression plasmid pMC-V21, on the other hand, led to cotransfected into cells growing on glass coverslips. Post-transfection cells the formation of typical IFs, increasing in number and were processed for double immunofluorescence microscopy using antibodies length over the time period examined, without any to c-myc-tagged plectin (A,C,E,G,I; Texas Red) and vimentin (B,D,F,H,J; DTAF). (A-F) Cells coexpressing myc-tagged repeat 5 and vimentin at 16 evidence for patching or collapse. The final stages of hours (A,B), 36 hours (C,D) and 64 hours (E,F) post-transfection. vimentin expressed were well spread networks (Fig. (G,H) Single-transfected cell expressing repeat 5 (G) or vimentin (H). 6H), in agreement with results by Leung et al. (1999). (I,J) Double-transfected cells, expressing repeat 4 (I) and vimentin (J). In any event, similar to the situation in double- (G-J) 64 hours post-transfection. Note that colocalization occurs transfected cells, the extent of filament network predominantly in areas of bundled or aggregated vimentin. 490 F. A. Steinböck and others formation in individual cells varied, most likely due to different time-course experiments with double-transfected SW13 levels of protein expression. Similar to our in vitro results, the (vim−) cells indicated that plectin bound to unpolymerized expression of plectin’s repeat 4 domain (pBN47) did not cause forms of IF proteins as well as to assembled intact filaments. the aggregation of vimentin in double-transfected SW13 cells. In particular, aggregates of vimentin and plectin’s repeat 5 were Here, the plectin mutant protein was found uniformly distributed seen at early time points after transfection, indicated binding over the (Fig. 6I), while the vimentin network to the unpolymerized form of the protein. Whether the protein appeared unaffected (Fig. 6J). has a higher affinity to unassembled or assembled IF protein species remains to be shown. When IF subunit proteins were assembled in the presence of DISCUSSION repeat 5 protein, the latter was mostly observed at branching points or ends of filaments (see also Nikolic et al., 1996), We show here that plectin’s repeat 5 domain inhibits self suggesting that binding of repeat 5 inhibited filament assembly of vimentin and cytokeratin K5/K14 IF subunit elongation. Thus, inhibition of assembly could be a direct proteins in vitro in a concentration-dependent manner and, at consequence of repeat 5-binding to IF proteins, thereby sufficiently high concentrations, causes disassembly of blocking a site indispensable for elongation. Alternatively, preassembled IFs. Interestingly K5/K14 filaments were found plectin-binding may lead to a conformational change of IF to be more resistant towards these effects of plectin than subunit proteins that impairs their self assembly. The vimentin filaments. A likely explanation for this difference importance of amino- and carboxy-terminal regions of IF could lie in the lower binding affinity of cytokeratins to plectin, proteins for filament assembly and/or stability is well known compared to vimentin (see Fig. 1). Consistent with this, (for a review, see Fuchs and Weber, 1994). Considering the plectin’s carboxy-terminal domain overexpressed in transfected differences in their primary structure it seems very likely that PtK2 cells preferentially bound to the vimentin filament specific secondary and tertiary structures exist within these network and, only after the collapse of this network did it bind non-helical domains. Differences in the ability of plectin to to cytokeratin filaments, finally causing their collapse also interact with these variable regions of IF proteins could be the (Nikolic et al., 1996). Another explanation for this difference reason for the protein’s differential effects on IF assembly. could be that oligomers of cytokeratins, for whatever reason, Similar mechanisms are likely to be involved in the are more stable than those of other IF proteins. This would be disassembly of in vitro assembled filaments upon incubation in agreement with the results of Coulombe and Fuchs (1990), with repeat 5 domain. The binding of this fragment to the who demonstrated, using gel filtration chromatography and assembled filaments may cause conformational changes of chemical cross-linking, that lateral and end-to-end packing of subunit proteins followed by destruction of filaments. Another cytokeratins still occurred in 4 M urea, and the formation of possibility would be that repeat 5 preferentially binds to soluble cytokeratin heterodimers and tetramers occurred even in 9 M IF subunit proteins, which are then inefficiently, or not at all, urea and in the presence of reducing agent. K5/K14 filaments, incorporated into filaments, or cause filament instability. The fact in particular, have been shown by Franke et al. (1983) to be that IFs are not simply static but dynamic structures is well more stable than those composed of most other cytokeratins, as known based on several studies in which a number of different analyzed by melting curves of keratin complexes in urea. It is approaches were used to demonstrate a presumably continuous also possible that binding of plectin to vimentin and cytokeratin exchange of subunits taking place at many sites within entire IF proteins takes place at different sites on these proteins. Thus, filament networks (Soellner et al., 1985; Albers and Fuchs, 1987; the interaction with plectin could influence a region that is Vikstrom et al., 1989). Furthermore, rapid disassembly of IFs crucial for filament assembly in one case (vimentin), but is less upon exposure to synthetic peptides has been observed (Hatzfeld important in the other (cytokeratins). The localization of the and Weber, 1992; Goldman et al., 1996), as well as disruption plectin-binding site(s) on the various IF subunit proteins may of existing networks after transient expression of truncated clarify this question. cytokeratins (Albers and Fuchs, 1987) or microinjection of The precise mechanism of plectin-IF interaction is currently antibodies to IF subunit proteins (Klymkowsky et al., 1983). unknown. Plectin’s carboxy-terminal globular domain consists Our finding of dose-dependent inhibition of filament of six repeat regions and a short tail region. All six repeats assembly and disintegration of filaments through plectin is contain a core region of close to 200 amino acid residues, particularly intriguing in view of a wealth of data which display a striking homology. Insertions of sequences of characterizing the protein as a prototype cytolinker providing different lengths between the carboxy-terminal end of these cells with mechanical strength. The putative ability of plectin core regions and the beginning of the following subdomain to either stabilize or disassemble IF networks would be in line (repeat) are probably responsible for different functions and/or with its involvement in regulating the cytoskeleton, in properties of the repeats. Most likely these linking segments, particular IF dynamics, a role that would go beyond the forming looplike structures exposed on the surface of the integration and scaffolding of cytoplasmic space. In fact, a carboxy-terminal globular domain of the molecule, enable function of plectin in regulating cytoskeleton dynamics and contacts to binding partners including other plectin molecules. cellular signaling was suggested in our recent study (Andrä et Recombinant repeat 5, similar to the full-length protein, al., 1998), indicating an involvement of plectin in the control exhibited self-interaction, as it occurred in solution as a of actin filament dynamics. multimer under all conditions tested (this study, and data not shown). Aggregation due to self-interaction is probably part of We thank Irmgard Fischer and Birgit Mir (both Vienna-Biocenter) the function of the core regions of plectin repeats. Both our for excellent technical assistance with the electron microscopy and biochemical binding assays using purified proteins and the preparation of purified recombinant plectin protein fragments, and Effects of plectin on IF assembly 491

Linglei Ma (John Hopkins University) for preparation of the native McLean, W. H. I., Pulkkinen, L., Smith, F. D. J., Rugg, E. L., Lane, E. B., human keratin protein sample. This work was supported by a grant Bullrich, F., Burgeson, R. E., Amano, S., Hudson, D. L., Owaribe, K., from the Austrian Science Research Fund (P12389). McGrath, J. A., McMillan, J. R., Eady, R. A. J., Leigh, I. M., Christiano, A. M. and Uitto, J. (1996). Loss of plectin causes epidermolysis bullosa with muscular dystrophy: cDNA cloning and genomic organisation. Genes Dev. 10, 1724-1735. REFERENCES Ming Chen (1992). Studies on the relationship between the structure and function of the intermediate filament protein vimentin by site-directed Andrä, K., Lassmann, H., Bittner, R., Shorny, S., Fässler, R., Propst, F. mutagenesis and expression in mammalian cells. PhD dissertation, and Wiche, G. (1997). Targeted inactivation of plectin reveals essential University of Heidelberg, Germany. function in maintaining the integrity of skin, muscle and heart Nagai, K. and Thorgersen, H.-C. (1987). Synthesis and sequence-specific cytoarchitecture. Genes Dev. 11, 3143-3156. proteolysis of hybrid proteins produced in Escherichia coli. Methods Andrä, K., Nikolic, B., Stöcher, M., Drenckhahn, D. and Wiche, G. (1998). Enzymol. 153, 461-481. Not just scaffolding: plectin regulates actin dynamics in cultured cells. Nikolic, B., MacNulty, E., Mir, B. and Wiche, G. (1996). Basic amino acid Genes Dev. 12, 3442-3451. residue cluster within nuclear targeting sequence motif is essential for Albers, K. and Fuchs, E. (1987). The expression of mutant epidermal keratin cytoplasmic plectin-vimentin network junctions. J. Cell Biol. 134 (6), 1-14. cDNAs transfected in simple epithelial and squamous cell carcinoma lines. Ruhrberg, C., Hajibagheri, M. A., Simon, M., Dooley, T. P. and Watt, F. J. Cell Biol. 105, 791-806. M. (1996). Envoplakin, a novel precursor of cornified envelope that has Brown, A., Bernier, G., Mathieu, N., Rossant, J. and Kothary, R. (1995a). homology to desmoplakin. J. Cell Biol. 134, 715-729. The mouse dystonia musculorum gene is a neural isoform of bullous Ruhrberg, C., Hajibagheri, M. A., Parry, D. A. and Watt, F. M. (1997). pemphigoid antigen 1. Nature Genet. 10, 301-306. Periplakin, a novel component of cornified envelopes and that Brown, A., Dalpé, G., Mathieu, N. and Kothary, R. (1995b). Cloning and belongs to the plakin family and forms complexes with envoplakin. J. Cell characterization of the neural isoform of human dystonin. Genomics 29, Biol. 139, 1835-1849. 777-780. Sarria, A. J., Lieber, J. G., Nordeen, S. K. and Evans, R. M. (1994). The Coulombe, P. A. and Fuchs, E. (1990). Elucidating the early stages of keratin presence of a vimentin-type intermediate filament network affects the shape filament assembly. J. Cell Biol. 111, 153-169. of the nucleus in human SW13 cells. J. Cell Sci. 107, 1593-1607. Foisner, R. and Wiche, G. (1987). Structure and hydrodynamic properties of Sawamura, D., Li, K., Chou, M.-L. and Uitto, J. (1991). Human bullous plectin molecules. J. Mol. Biol. 198, 515-531. pemphigoid antigen (BPAG1). Amino acid sequences deduced from cloned Foisner, R., Leichtfried, F. E., Herrmann, H., Small, J. V., Lawson, D. and cDNAs predict biologically important peptide segments and protein Wiche, G. (1988). Cytoskeleton-associated plectin: in situ localization, in domains. J. Biol. Chem. 266, 17784-17790. vitro reconstitution, and binding to immobilized intermediate filament Smith, F. J. D., Eady, R. A. J., Leigh, I. M., McMillan, J. R., Rugg, E. L., proteins. J. Cell Biol. 106, 723-733. Kelsell, D. P., Bryant, S. P., Spurr, N. K., Geddes, J. F., Kirtschig, G., Franke, W. W., Schiller, D. L., Hatzfeld, M. and Winter, S. (1983). Protein Milana, G., de Bono, A. G., Owaribe, K., Wiche, G., Pulkkinen, L., complexes of intermediate-sized filaments: melting of cytokeratin Uitto, J., McLean, W. H. I. and Lane, E. B. (1996). Plectin deficiency complexes in urea reveals different polypeptide separation characteristics. results in muscular dystrophy with Epidermolysis bullosa. Nature Genet. 13, Proc. Natl. Acad. Sci. USA 80, 7113-7117. 450-457. Fuchs, E. and Weber, K. (1994). Intermediate filaments: structure, dynamics, Soellner, P., Quinlan, R. A. and Franke, W.W. (1985). Identification of a function, and disease. Annu. Rev. Biochem. 63, 345-382. distinct soluble subunit of an intermediate filament protein: tetrameric Fuchs, E. and Cleveland, D. W. (1998). A structural scaffolding of vimentin from living cells. Proc. Natl. Acad. Sci. USA 82, 7929-7933. intermediate filaments in health and disease. Science 279, 514-519. Stanley, J. R., Hawley-Nelson, P., Yuspa, S. H., Shevach, E. M. and Katz, Gache, Y., Chavanas, S., Lacour, J. P., Wiche, G., Meneguzzi, G. and S. I. (1981). Characterization of bullous pemphigoid antigen: a unique Ortonne, J. P. (1996). Defective expression of plectin in Epidermolysis basement of stratified squamous epithelia. Cell 24, 897- Bullosa Simplex with muscular dystrophy. J. Clin. Invest. 97, 2289-2298. 903. Giese, G. and Traub, P. (1986). Induction of vimentin synthesis in mouse Steinböck, F. A. and Wiche, G. (1999). Plectin: a cytolinker by design. Biol. myeloma cells MPC-11 by 12-O-tetradecanoylphorbol-13-acetate. Eur. J. Chem. 380, 151-158. Cell Biol. 40, 266-274. Tyler, J. M. and Branton, D. (1980). Rotary shadowing of extended Goldman, R. D., Khuon, S., Chou, Y. H., Opal, P. and Steinert, P. M. molecules dried from glycerol. J. Ultrastruct. Res. 71, 95-102. (1996). The function of intermediate filaments in cell shape and cytoskeletal Uitto, J., Pulkkinen, L., Smith, F. D. J. and McLean, W. H. I. (1996). Plectin integrity. J. Cell Biol. 134, 971-983. and human genetic disorders of the skin and muscle. Exp. Dermatol. 5, 237- Graham, F. L. and van der Erb, E. (1973). A new technique for the assay 246. of infectivity of human adenovirus 5 DNA. Virology 52, 456-467. Vikstrom, K. L., Borisy, G. G. and Goldman, R. D. (1989). Dynamic aspects Green, K., Parry, D. A. D., Steinert, P. M., Virata, M. L. A., Wagner, R. of intermediate filament networks in BHK-21 cells. Proc. Natl. Acad. Sci. M., Angst, B. D. and Nilles, L. A. (1990). Structure of the human USA 86, 549-553. . Implications for function in the desmosomal plaque. J. Biol. Wawersik, M., Paladini, R. D., Noensie, E. and Coulombe P. A. (1997). A Chem. 265, 2603-2612. proline residue in the α-helical rod domain of type I destabilizes Green, K., Virata, M. L. A., Elgart, G. W., Stanley, J. R. and Parry, D. A. keratin heterotetramers and influences incorporation into filaments. J. Biol. D. (1992). Comparative structural analysis of desmoplakin, bullous Chem. 272, 32557-65. pemphigoid antigen and plectin: members of a new gene family involved in Wiche, G., Herrmann, H., Leichtfried, F. and Pytela, R. (1982). Plectin: a organization of intermediate filaments. Int. J. Biol. Macromol. 14, 145-153. high-molecular-weight cytoskeletal polypeptide component that copurifies Hatzfeld, M. and Weber, K. (1992). A synthetic peptide representing the with intermediate filaments of the vimentin type. Cold Spring Harbor Symp. consensus sequence motif at the carboxy-terminal end of the rod domain Quant. Biol. 46, 475-482. inhibits intermediate filament assembly and disassembles preformed Wiche, G., Becker, B., Luber, K., Weitzer, G., Castañón, M. J., Hauptmann, filaments. J. Cell Biol. 116, 157-166. R., Stratowa, C. and Stewart, M. (1991). Cloning and sequenzing of rat Henderson, D., Geisler, N. and Weber, K. (1982). A periodic ultrastructure plectin indicates a 466-kD polypeptid chain with a three-domain structure in intermediate filaments. J. Mol. Biol. 155, 173-176. based on a central α-helical . J. Cell Biol. 114, 83-99. Klymkowsky, M. W., Miller, R. H. and Lane, E. B. (1983). Morphology, Wiche, G., Gromov, D., Donovan, A., Castañón, M. J. and Fuchs, E. behaviour, and interaction of cultured epithelial cells after the antibody- (1993). Expression of plectin mutant cDNA in cultured cells indicate a role induced disruption of keratin filament organization. J. Cell Biol. 96, 494-509. of C-terminal domain in intermediate filament association. J. Cell Biol. 121, Leung, C. L., Sun, D. and Liem, R. K. H. (1999). The intermediate filament 607-619. protein is the specific interaction partner of mouse BPAG1-n Wiche, G. (1998). Role of plectin in cytoskeleton organisation and dynamics. (dystonin) in neurons. J. Cell Biol. 144, 435-446. J. Cell Sci. 111, 2477-2486. Liu, C.-G., Maercker, C., Castañón, M. J., Hauptmann, R. and Wiche, G. Yang, Y., Dowling, J., Yu, Q.-C., Kouklis, P., Cleveland, D. W. and Fuchs, (1996). Human plectin: organization of the gene, sequence analysis, and E. (1996). An essential cytoskeletal linker protein connecting actin location (8q24). Proc. Natl. Acad. Sci. USA 93, 4278-4283. microfilaments to intermediate filaments. Cell 86, 655-665.