Báo cáo khoa học: Actin-binding domain of mouse plectin Crystal structure and binding to vimentin

Orthorhombic crystals in the P212121 space group with only one molecule in the asymmetric unit (form II) were prepared by the same method combined with seeding. Drops (6 lL) were prepared by mixing equal volumes of protein solution (see above) and precipitant solution [10% (w/v) PEG 8000, 0.1 M cacodylate buffer, pH 6.5, 0.2 M calcium acetate]. Drops were collected after 24 h of equili- bration and the precipitate was removed by centrifugation

a (A˚ ) b (A˚ ) c (A˚ ) b ((cid:3)) Mosaicity Solvent content (%) Completeness (%) R(I)merge I/r(I) 55.31 108.92 63.75 115.25 0.3 60 96.2 (71.7) 6.0 (40.0) 22.2 (2.1) 32.52 51.23 144.72 90 0.4 50 97.6 (86.3) 3.8 (31.0) 33.9 (3.5)

1876 J. Sˇ evcˇ ı´ k et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Table 2. Refinement statistics.

structure I structure II Parameter

30.0–2.0 42 811/2286 15.1 19.4 15.3 0.12 1919/1919 188 32.6 50.0–2.0 15 889/848 19.7 29.9 20.2 0.21 1884 58 33.8 Resolution limits (A˚ ) No. of reflections all/Rfree set Rwork (%) Rfree (%) Rall (%) ESU based on Rfree (A˚ ) Protein atoms, molecule A/B Water molecules Wilson plot B factor Average B (A˚ 2)

Amino acid sequence identities of the ABDs of utrophin, dystrophin and fimbrin with pleABD/2a, determined by the program FASTA [34], are 48, 47 and 23%, respectively. Because of differences in the relative orientation of CH1 and CH2 subdomains in the structures of utrophin, dystrophin and fimbrin, not the whole ABD, but the CH1 and CH2 subdomains of utrophin were used as two model structures. PleABD/2a structure I was determined by molecular replacement using the program MOLREP [35] four times, twice with CH1 and twice with CH2. After each MOLREP session the solution was subjected to five cycles of refine- ment with the program REFMAC5 [36] to improve the model, and the resulting PDB file was fixed in the next MOLREP session. Using this procedure, R factors in the four MOLREP sessions were 58, 52, 45 and 38%, and correlation coeffi- cients 25, 39, 54 and 68%. MOLREP unambiguously showed that there were two protein molecules in the asymmetric unit (hereafter referred to as molecules A and B). For build- ing the model, the program ARP/WARP in the mode WARPNTRACE [37] was used. The program automatically built a model consisting of 378 out of 490 residues (molecules A and B) with a connectivity index of 0.92. The remaining residues, including those forming the loop connecting the CH1 and CH2 subdomains, were built manually using the program O [38] running on a Silicon Graphics Station.

main-chain A molecule B molecule side-chain A molecule B molecule 42.2 39.3 46.6 44.2 46.2 36.3 – 39.8 – 38.9

Structure I was refined with the program REFMAC5. Sparse matrix was used as the method of minimization. Refinement of the structure was altered with correcting the amino acid sequence and building the parts which were different from those of utrophin and which were not built by WARPNTRACE. The structure was refined against 95% of the data, the remaining 5% (randomly excluded from the full data set) were used for crossvalidation in which Rfree was calculated to follow the progress of refinement [39]. After each refinement cycle, ARP, an automated refinement procedure [40], was applied for modeling and updating the solvent structure. After the R factor had fallen to about 20%, refinement continued with anisotropic temperature factors and hydrogen atoms generated in standard geometries.

the CH1 subdomain and the remaining three the CH2 subdomain. The mouse plectin ABD isoform analyzed here, pleABD/2a, is unique, as it contains a five amino acid-long sequence (Fig. 1) inserted by differential splicing of a short exon (exon 2a) between the first two exons of the ABD [6]. Comprising 245 residues, the analyzed recombinant protein contained pleABD/2a as a 237 residues-long fragment (amino acids 181–417; EMBL accession number AF 188008) flanked by six amino-terminal (GSHMEF) and two carboxy-terminal (EF) residues (added as cloning requirement). The amino acid residues in the structures are numbered from 1 to 245 according to the sequence of the recombinant protein, i.e. amino acids with numbers 7–243 in the structure correspond to 181–417 in the AF 188008 sequence.

Structure II was determined by molecular replacement using molecule A from structure I as a model. Solution was straightforward giving an R factor and correlation coeffi- cient of 39 and 63%, respectively. Refinement was per- formed in the same way as that of structure I. Refinement statistics of both structures, average temperature factors, and target deviations against stereochemical restraints are given in Table 2. For visualization and rebuilding of the structures O and XTALVIEW programs were used [38,41].

Other methods

MALDI-TOF MS as well as ESI MS were performed at the mass spectrometry unit, Vienna Biocenter, Vienna, Austria.

0.04 (0.02) 2.59 (1.94) 0.21 (0.20) 0.02 (0.02) 2.38 (1.50) 3.57 (2.00) 5.27 (3.00) 7.26 (4.50) 0.03 (0.02) 2.02 (1.94) 0.22 (0.20) 0.01 (0.02) 2.16 (1.50) 3.44 (2.00) 4.70 (3.00) 7.20 (4.50) Waters Rms deviation from ideal geometry (target values are given in parentheses) Bond distances (A˚ ) Bond angles ((cid:3)) Chiral centers (A˚ 3) Planar groups (A˚ ) Main-chain bond B-values (A˚ 2) Main-chain angle B-values (A˚ 2) Side-chain bond B-values (A˚ 2) Side-chain angle B-values (A˚ 2)

Results

Description of the structures

Two crystal structures of pleABD/2a were determined. One of them, structure I was derived from a monoclinic crystal containing two protein molecules (A and B) in the asymmetric unit, the other, structure II, from an orthorhom- bic crystal, containing one molecule in the asymmetric unit (for details see Materials and methods). As found in structure I, molecule A, pleABD/2a is an a-protein consisting of 11 helices: a1 (residues 8–25), a2 (48–58), a3 (70–86), a4 (96– 100), a5 (104–118), a6 (134–145), a7 (165–172), a8 (181–186), a9 (189–204), a10 (212–215), and a11 (222–235) (Figs 1 and 2A). Helices a1–a5 form the CH1 and helices a6–a11 the CH2 subdomain. The subdomains are connected by a flexible 15 residues-long loop (119–133). For five N-terminal (GSHME) and eight C-terminal residues (RVPGAQEF) of both molecules in structure I there was no electron density observed. In structure II there was no electron density for the first seven (N-terminal) nor for the last eight (C-terminal)

Like other ABDs of the canonical CH1-CH2 type, the ABD of plectin is encoded by seven exons, the first four encoding

Structure and vimentin-binding of plectin ABD (Eur. J. Biochem. 271) 1877

(cid:1) FEBS 2004

residues. It is reasonable to conclude that N- and C-terminal residues formed highly disordered tails. The molecular mass of the recombinant protein was 28 631 Da, as determined by mass spectrometry. This was in a good agreement with its theoretical mass of 28 656 Da, indicating that the structure indeed contained all of the 245 encoded residues.

The contacts between molecules in the crystal lattice apparently did not change the orientation of the CH1 with respect to the CH2 subdomain in spite of the fact that the loop connecting the two subdomains theoretically could including those found in allow various orientations utrophin. It can be concluded that the conformation of pleABD/2a as found in structures I and II is stable and not subject to conformational changes due to different crystal packing.

Quality of protein structure models

The final R and Rfree factors for structure I were 15.3 and 19.4% (Table 2). There were two protein molecules (A and B) and 188 solvent molecules in the asymmetric unit. The Ramachandran plot [43] calculated by the program PRO- CHECK [44] showed that 92.1 and 94.4% of the residues of molecules A and B, respectively, were in the most favored regions. The remainder was in additionally allowed regions except for Thr158, which in both molecules had the same dihedral angles being just at the outer border of the additionally allowed region. Thr158 was localized in a surface loop and there was no doubt about its conforma- tion, as the electron density in this region was clear.

ABDs can adopt open (e.g. utrophin) or closed (e.g. fimbrin) conformations. PleABD/2a was in a closed confor- mation in which the CH1 and CH2 subdomains were facing each other through helices a1 and a5, and helices a10 (including the subsequent loop) and a11. Molecules A and B in the asymmetric unit of the pleABD/2a structure I formed a crystallographic dimer (Fig. 2B). In the contact area of the two molecules there were two hydrogen bonds formed between side chains of Glu125(A)/Asn88(B) and Lys130(A)/ Glu131(B) and one bond mediated by a water molecule Asn151(A)/W66/Asp150(B). The solvent-accessible surface buried at the plectin dimer interface was 760 A˚ 2, corres- ponding to (cid:1) 3% (380 A˚ 2) of the surface of each isolated molecule (11 800 A˚ 2). This was far below the minimum of 9% required for classification of a dimer as a protein complex [42], suggesting that the crystallographic dimer hardly could exist in solution. Moreover, structure II has confirmed that pleABD/2a can exist as a monomer in solution.

Fig. 1. Amino acid sequence alignment and structural features of ABDs from mouse and human plectin, utrophin, dystrophin and fimbrin. Boxes indicate helices (a1–a11) as given by the program PROCHECK [44]. Connecting segments between the CH1 and CH2 subdomains are in red. The three amino acid residues differing in the human and mouse plectin sequence are in gray italics (*). Note that in the PDB structure 1MB8 only two differing amino acid residues were reported. The sequence encoded by exon 2a is double underlined. Amino acid residues of the identified actin- binding sites of fimbrin and dystrophin are shaded. Numbers on the right correspond to amino acid positions. Dashes in sequences indicate gaps introduced to allow maximum alignment. PLM, mouse plectin ABD (EMBL accession no. AF188008); PLH, human plectin ABD (EMBL accession no. U53204); UTR, utrophin ABD (PDB code 1QAG); FIM, L-fimbrin ABD1 (PDB code 1AOA); DYS, dystrophin ABD (PDB code 1DXX).

1878 J. Sˇ evcˇ ı´ k et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

of the pleABD/2a molecule; temperature factors were highest in the loop connecting the two subdomains.

A diagram of the average main-chain temperature factors of molecules A and B as a function of residue number (Fig. 3A) revealed an amazing similarity of their tempera- ture factor profiles, confirming the conformational stability

For structure II, the final R and Rfree factors were 20.2 and 29.9% (Table 2). The structure contained one protein

Fig. 2. Ribbon representation of the pleABD/ 2a structure I and comparison with utrophin and fimbrin ABDs. (A) Stereo view of pleABD/2a. Individual helices are numbered. (B) Crystal- lographic dimer as seen in the asymmetric unit. The views are related by 90(cid:3) rotation around a horizontal axis. Molecules A and B are shown in red and blue, respectively. (C) Overlap (stereo view) of CH1 and CH2 subdomains of pleABD/2a molecule A with the CH1 subdomain of utrophin molecule A and the CH2 subdomain of utrophin molecule B. Utrophin molecules are colored in light (molecule A) and dark blue (molecule B). The plectin molecule is in red. (D) Overlap (stereo view) of pleABD/2a (red) with fimbrin ABD (green). Figures were generated using the program MOLSCRIPT [52].

Structure and vimentin-binding of plectin ABD (Eur. J. Biochem. 271) 1879

(cid:1) FEBS 2004

differences in Ca positions (up to 2.5 A˚ ) were found in the surface loop region, which was not unexpected considering that each molecule has a different environment in the crystal.

and 58 solvent molecules in the asymmetric unit and in the Ramachandran plot there were 88.2% of the residues in the most favored region and 11.8% in additionally allowed regions. Fluctuation of B-values of structure II was similar to that of structure I, with average B-values of structure II being slightly lower (Fig. 3A), which could be due to the fact that structure II data were collected at cryogenic tempera- ture. Refinement parameters of structure II were slightly worse compared to those of structure I, which was unexpected considering that the crystal data seemed to be better for structure II.

Comparison of ABDs from mouse plectin, human plectin, fimbrin, utrophin and dystrophin

Least squares superpositions of mouse plectin with the corresponding subdomains of human plectin, utrophin, dystrophin and fimbrin are summarized in Table 3. In these superpositions only the CH1 (8–119) and CH2 (134–236) subdomains from structure I molecule A were used; the connecting segment was excluded as its conformation differs substantially among the various proteins compared (helix in utrophin and dystrophin, helix–loop in fimbrin, and loop in plectin). Furthermore, as the ABDs of utrophin and dystrophin adopt an open conformation (contrary to the plectin ABD), the CH1 domain from utrophin molecule A and the CH2 domain from molecule B were used in the overlap. The different numbers of Ca atoms involved in

Fig. 3. Comparisons of mouse and human ABDs. Average main chain B factor values of mouse ABDs (A), and differences between Ca atoms in superpositioned structures (B) are plotted as a function of residue numbers. IA, structure I molecule A; IB structure I molecule B; II, structure II; HP, structure of human plectin.

The major difference in the tertiary structures of human and mouse plectin was found at their N termini, where the first a-helical segment (a1) of human plectin exceeded that from mouse by nine amino acid residues. However, since six of these additional residues were encoded by one of the first exons (1c) preceding the actual ABD-encoding exons 2–8, this does not reflect a structural difference of mouse and human ABDs per se.

Table 3. Superposition of ABD domains (without segment connecting subdomains). IA(B), structure I, molecule A(B); II, structure II; HP, human plectin.

rmsd (A˚ ) No. of Ca atomsa ABD

a Ca atoms separated by more than 1.75 A˚ were omitted. b CH1 is from utrophin (dystrophin) molecule A, CH2 from utrophin (dystrophin) molecule B (see Figs 2C,D, and relevant text).

To document minor structural differences between human and mouse plectin ADBs, structure I molecule B (IB), structure II (II), and the structure of human plectin (HP) were superimposed on structure I molecule A (IA). Figure 3B shows the deviations of Ca atoms observed in these superpositions as a function of residue number (with the exception of the first N- and the last C-terminal residues, which differed by more than 5 A˚ ). Excluding the residues for which distances between corresponding Ca atoms exceeded 1.75 A˚ (Fig. 3B, dashed line), the rms displace- ment values were 0.25 (IB/IA, 232 Ca atoms), 0.48 (II/IA, 225 atoms), and 0.62 A˚ (HP/IA, 224 atoms). The largest

IA/IB IA/II IB/II IA/HP IA/utrophinb IA/dystrophinb IA/fimbrin 218 211 212 203 192 176 78 0.23 0.44 0.48 0.55 0.73 0.82 1.12

1880 J. Sˇ evcˇ ı´ k et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

superpositions (reflecting the degree of structural similarit- ies) clearly showed highest similarity of mouse plectin with human plectin and lowest with fimbrin (Table 3). The nearly identical conformations of the pleABD/2a structure and the corresponding subdomains of the A and B molecules of utrophin and dystrophin probably have not arisen by chance and may have significance for functional properties of these ABDs.

The ABD of plectin binds to vimentin

The IF protein vimentin has been reported to specifically interact with the CH1 subdomain of the first (N-terminal) of fimbrin’s two ABDs [20]. In light of the extensive structural resemblance of fimbrin and plectin ABDs it was therefore of interest to assess IF-binding activity of the plectin ABD. Moreover, a second IF protein interaction domain at the N terminus in addition to its C-terminal IF-binding site [26] would raise plectin’s functional versatility, particularly as a cytoskeletal crosslinking element. To assess the plectin ABD–vimentin interaction, in vitro overlay assays were performed. Ple1aABD, an ABD version of plectin preceded by a sequence encoded by exon 1a[6], pleABD/2a and truncated versions of the plectin ABD missing half of the CH1 domain (ple4–8), or the complete CH1 domain (ple6– 9) were immobilized on nitrocellulose membranes and overlaid with full-length vimentin. In agreement with previously reported findings [20] all proteins, except the one missing the entire CH1 domain (ple6–9) and the negative control (BSA), showed binding to vimentin, similar to the positive control (recombinant plectin repeat 5) (Fig. 4). Thus only the CH1, but not the CH2, domain of plectin’s ABD was found capable of binding to vimentin.

To specify the subdomain of vimentin that bound to plectin’s ABD, several truncated versions of vimentin were expressed in Escherichia coli and subjected to overlay assays. The ABD version used in these experiments contained the preceding 66 amino acid residues-long sequence specific for plectin isoform 1c (ple1cABD/2a), enabling detection of bound protein via isoform 1c-specific antibodies [31]. This fragment showed strong binding to full-length vimentin and to VimNR, a truncated version of vimentin containing its N-terminal domain and central rod domain, but lacking the C-terminal domain (Fig. 5). Only very weak interactions were observed with vimentin fragments corresponding to the N-terminal (VimN), or rod domains alone (VimR), or to the rod in conjunction with the C-terminal domain (Vim- RC). These data suggested that the plectin ABD-binding site of vimentin was contained in the N-terminal part of the molecule comprising the head domain and possibly parts of the rod.

The head domain of vimentin had previously been shown to harbor the binding site(s) for the CH1 subdomain of fimbrin’s N-terminal ABD [20], whereas desmin, an IF protein of similar type, was found to interact with the corresponding domain of calponin via the N-terminal part of its rod domain [19]. Interestingly, fimbrin and calponin have been found to interact with tetrameric forms of soluble vimentin and desmin. As we were unable to cosediment pleABD/2a with filamentous vimentin in sedimentation assays (data not shown) it seems that the plectin ABD interacts with vimentin in the same way.

Fig. 4. Overlay of various plectin ABD versions with full-length vimen- tin. Recombinant versions of the plectin ABD starting with exon 1a- encoded sequences (ple1aABD), or starting with exon 2-encoded sequences and containing 2a-encoded sequences (pleADB/2a), or lacking part of (ple4–8), or the whole CH1 domain (ple6–9), as well as a fragment corresponding to the repeat 5 domain of plectin (positive control), and BSA (negative control) were subjected, in duplicate, to 12.5% SDS/PAGE. Proteins on one gel were blotted onto a nitrocel- lulose membrane and overlaid with recombinant full-length mouse vimentin (B), proteins on a second gel were stained with Coomassie Blue (A). All proteins, except for ple6–9 and BSA, showed significant binding to vimentin.

Structure and vimentin-binding of plectin ABD (Eur. J. Biochem. 271) 1881

(cid:1) FEBS 2004

Elution and SDS/PAGE of bound proteins revealed that a single major fragment of (cid:1) 18 kDa was retained on the column (Fig. 6). This fragment could be mapped to (cid:2)Coil 1(cid:3), an N-terminal segment of the vimentin rod domain (amino acid residue positions 124–276 of mouse vimentin, EMBL accession no. M26251), using MALDI-TOF mass spectro- metric sequence analysis.

Discussion

Because of its expression in the form of several distinct isoforms, the ABD of plectin is unique among those of other actin-binding protein family members [6]. The isoform crystallized and analyzed here contains an extra five amino acids (HWRAE; positions 28–32) encoded by exon 2a, a differentially spliced exon inserted between the common exons 2 and 3. This insertion makes the connecting loop of helices a1 and a2 longer. Amino acid residues contained in this loop are located closely to one (the first) of three segments identified as direct docking sites for actin within dystrophin’s ABD [22] (Fig. 1). This insertion in pleABD/ 2a may result in higher flexibility of this region and the surface exposure and amino acid composition of this segment (facilitating charge–charge as well as hydrophobic interactions) may improve binding properties of the ABD. In fact, we have previously shown that this isoform exhibits a higher affinity to actin than several other isoforms without the 2a sequence [6].

Our analysis revealed that both, mouse and human pleABD/2a are structurally similar to the fimbrin ABD, although these domains share only little sequence identity. The fimbrin and plectin ABDs have a similar closed conformation, differing from the open conformation des- cribed for the ABDs of dystrophin and utrophin [21–24]. It is interesting to note that until now, sequences correspond- ing to exons 2a and 3a of mouse plectin have been identified neither in other family members, nor in human plectin. In insertion of exon 2a into the sequence of this view, human pleABD/2a as reported [24] seems without rational explanation.

Although the ABDs of fimbrin, utrophin and a-actinin are related, they appear to have different effects on F-actin conformation upon binding and thus may use different mechanisms of association [25,45,46]. Up to now a consen-

To confirm the specificity of the plectin ABD–vimentin interaction and to more precisely map vimentin’s plectin ABD-binding site, vimentin purified in urea was kept in its soluble (tetrameric) form by dialysis into solution D (see Material and methods). The protein was than subjected to limited chymotryptic digestion and fragments generated were applied to a pleABD/2a-Sepharose affinity column.

Fig. 5. Overlay of recombinant vimentin fragments with plectin ABD. Recombinant versions of vimentin subdomains were immobilized on nitrocellulose membranes as described in Fig. 4, and overlaid with the ple1cABD/2a. Vim, full-length vimentin; VimN, N-terminal domain; VimR, rod domain; VimRC, vimentin without N-terminal domain; VimNR, vimentin without C-terminal domain. To detect vimentin- bound plectin ABD, plectin isoform 1c-specific antibodies [31] were used. Note the strong binding of plectin ABD to full-length vimentin and VimNR, but only weak or no binding to other vimentin fragments.

Fig. 6. Affinity-binding of proteolytically derived fragments of vimentin to pleABD/2a. Partial chymotryptic digestion of vimentin and affinity- chromatography of fragments on a pleABD2a-Sepharose column was carried out as described in the text. SDS/PAGE of eluted fractions is shown. Lanes 1–11, wash fractions; 12–27, salt-gradient elution of bound proteins. Co, sample loaded onto column. The molecular mass of size markers run in left-most lane is indicated. The18 kDa fragment of vimentin binding to pleABD/2a mapped to the N-terminal part of the vimentin rod domain, as determined by MALDI-TOF mass spectrometry.

1882 J. Sˇ evcˇ ı´ k et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

to cosediment with vimentin filaments (data not shown), suggesting that it, too, interacted with soluble IF protein subunits rather than with their filamentous polymers. Binding assays revealed that the major vimentin-binding interface was localized within the CH1 subdomain of plectin. Since ple4–8 (pleABD/2a lacking half of the CH1 domain) bound to vimentin, sequences preceding those encoded by exon 4 apparently did not contribute to this interaction. However, as fragments corresponding to the first half of plectin’s CH1 domain (exons 2–4) have not been examined in our studies, a role of this part in vimentin- binding can not be fully excluded.

sus on the mode of binding (open or closed conformation of ABD) of utrophin and dystrophin to actin filaments has not been reached [47,48]. It has been reported that fimbrin [25] and human plectin ABDs [24] bind to actin filaments in their closed conformation. In view of these findings it is expected that pleABD/2a could adopt a closed as well as an open conformation in binding to actin, as the connecting segment between the CH1 and CH2 domains is flexible and long enough to allow both conformations (as proposed also for human plectin [24]). The notion of pleABD/2a adopting an open conformation is supported by similar characteristics of contact areas in the CH1 and CH2 structures of pleABD/ 2a, utrophin, and dystrophin. ABD of fimbrin is expected to bind only in its closed conformation, in agreement with the finding that 72% of its contact area (excluding the connecting segment) is hydrophobic and there is no hydrogen bond shorter than 3.25 A˚ . On the other hand, only 59% of the corresponding contact areas of pleABD/2a and utrophin are hydrophobic and there are five and six hydrogen bonds, respectively. Moreover, the secondary structure prediction of pleABD/2a suggested a helix to exist between the CH1 and CH2 subdomains. The sites of pleABD/2a responsible for binding to actin have not been identified so far. Assuming that they are similar to those reported for dystrophin [22], the amino acid sequences 13–22, 88–116 and 131–146 might be involved, slightly differing from the binding segments predicted for fimbrin [25] (Fig. 1).

The ABDs of dystrophin and utrophin crystallized as antiparallel dimers [21,22], whereas fimbrin and human plectin ABDs crystallized in monomeric form [23,24]. When pleABD/2a was crystallized at pH 9, we also found two molecules in the asymmetric unit [28]. However, conditions at pH 6.5 led to the formation of crystals with only one pleABD/2a molecule in the asymmetric unit. The assess- ment of the contact area between molecules A and B in pleABD/2a crystals obtained at pH 9 (crystal form I) showed that these molecules do not interact strongly enough to form dimers which could exist also in solution. Therefore, the crystallographic dimer observed in the asymmetric unit was probably an artefact of crystallization rather than a dimerization product.

Although CH subdomains forming a canonical ABD structurally seem to be highly conserved, data are accumu- lating that show functional diversity of CH1 and CH2 subdomains [8]. It was reported that proteins containing either single CH subdomains (calponin), or more complex ABDs (fimbrin) can interact with IF proteins. Calponin binds to desmin, the major IF protein in smooth and skeletal muscle [17–19] and fimbrin interacts with and colocalizes with vimentin in filopodia, retraction fibres, and podosomes at the ventral surface of cultured macro- phages [20]. In both cases there is evidence that binding occurred to IF subunit proteins in their nonfilamentous state [17–20]. Fimbrin was unable to bind to polymerized vimentin in cosedimentation assays and binding occurred at a stoichiometry of 1 : 4, suggesting that the IF protein was in its tetrameric form [20].

Using an overlay assay we found strong binding of plectin’s ABD to a fragment of vimentin comprising the rod and its N-terminal domains, but when examined individu- ally, both of these domains showed only weak binding. Using a similar method, it had been shown that a vimentin fragment lacking the C terminus (N410/VimNR) was capable of binding to fimbrin, contrary to a fragment lacking the initial 102 amino acid residues (102C/VimRC), suggesting that the fimbrin-binding site was located in the N-terminal head domain of vimentin [20]. The N-terminal vimentin fragment (VimN) and the rod fragment (VimR) used in our experiments ended and started, respectively, at Phe95. Consequently, the rod-containing fragment used in our experiments and the N-terminal fragment N410/Vim- NR used in [20] overlapped by a few amino acid residues. With this consideration, the results of our overlay assay were consistent with the findings reported in [20]. Using as an alternative method affinity-binding of proteolytic frag- ments of vimentin in combination with mass spectrometry, we found that plectin bound to a (cid:1) 18 kDa fragment corresponding to (cid:2)Coil 1(cid:3) [49], the N-terminal part of the a- helical rod domain of vimentin. This is in agreement with similar experiments in which the calponin-binding site was restricted to the N-terminal part of the desmin rod domain [19]. However, in the overlay assay, the rod domain of vimentin (VimR) alone showed little binding to plectin. As fragments obtained by partial proteolytic digestion of properly folded full-length vimentin are more likely to preserve the structure of the native protein than recombi- nantly prepared fragments, we assume the true plectin ABD-binding site of vimentin to be localized in the N- terminal part of its rod domain. Binding to the evolutionary highly conserved CH domains could be a common feature of IF proteins in general. Likewise, considering that the a- helical coiled-coil structure of vimentin’s rod domain is highly conserved in all IF protein family members, plectin’s ABD may bind also to other IF proteins, such as desmin. The functional significance of the plectin ABD–vimentin interaction remains elusive. The primary activity of the ABD supposedly is actin-binding, a function demonstrated for both fimbrin and plectin [14,50]. The CH domain of calponin, on the other hand, doesn’t seem to be required for actin-binding as calponin interacts with actin via its C-terminal domain [51]. Thus, with an IF protein-binding site residing in their N-terminal CH domains in addition to their genuine actin-binding activities, fimbrin, plectin and calponin, might influence the assembly and organization of both actin and IF cytoskeletal networks. The interaction of fimbrin’s ABD with vimentin has been shown to be adhesion dependent and it has been suggested that this complex is a

As one may expect on the basis of their structural similarity, the ABDs of fimbrin and plectin apparently also have a number of functions in common, including binding to vimentin. Similar to fimbrin [20] the ABD of plectin failed

Structure and vimentin-binding of plectin ABD (Eur. J. Biochem. 271) 1883

(cid:1) FEBS 2004

11. Jarrett, H.W. & Foster, J.L. (1995) Alternate binding of actin and calmodulin to multiple sites on dystrophin. J. Biol. Chem. 270, 5578–5586.

12. Winder, S.J., Hemmings, L., Bolton, S.J., Maciver, S.K., Tinsley, J.M., Davies, K.E., Critchley, D.R. & Kendrick-Jones, J. (1995) Calmodulin regulation of utrophin actin binding. Biochem. Soc. Trans. 23, 397S.

for 13. Fukami, K., Furuhashi, K., Inagaki, M., Endo, T., Hatano, S. & (1992) Requirement of phosphatidylinositol alpha-actinin function. Nature 359, Takenawa, T. 4,5-bisphosphate 150–152.

transient structure involved in early cell adhesion [20]. In smooth muscle cells calponin may be an integral component of desmin IFs in the vicinity of dense bodies [18]. The interaction between IF proteins, such as desmin or vimentin, and proteins containing one or more CH domains could represent a highly conserved function related to the estab- lishment of cell adhesion structures. By sequestering soluble vimentin at certain sites, such as focal adhesion contacts, plectin may favor and/or initiate IF network formation at these sites by locally increasing IF protein concentrations. Once filament assembly has been initiated, plectin may stabilize and anchor the filaments at these sites via its alternative C-terminal IF-binding site. Future experiments should show whether this model can be verified.

14. Andra¨ , K., Nikolic, B., Sto¨ cher, M., Drenckhahn, D. & Wiche, G. (1998) Not just scaffolding: plectin regulates actin dynamics in cultured cells. Genes Dev. 12, 3442–3451.

Acknowledgements

15. Rezniczek, G.A., de Pereda, J.M., Reipert, S. & Wiche, G. (1998) Linking integrin a6b4-based cell adhesion to the inter- mediate filament cytoskeleton: direct interaction between the b4 subunit and plectin at multiple molecular sites. J. Cell Biol. 141, 209–225.

16. Geerts, D., Fontao, L., Nievers, M.G., Schaapveld, R.Q., Purkis, P.E., Wheeler, G.N., Lane, E.B., Leigh, I.M. & Sonnenberg, A. (1999) Binding of integrin a6b4 to plectin prevents plectin association with F-actin but does not interfere with intermediate filament binding. J. Cell Biol. 147, 417–434. 17. Wang, P. & Gusev, N.B. (1996) Interaction of smooth muscle calponin and desmin. FEBS Lett. 392, 255–258.

We thank our colleagues Kamaran Abdoulrahman, Branislav Nikolic, Gernot Putz, Gu¨ nther Rezniczek, Daniel Spazierer and Ferdinand Steinbo¨ ck, for providing various reagents and for valuable discussions. We are grateful to the EMBL Hamburg team for providing us with synchrotron facilities and for help in data collection. We would also like to thank the European Community for supporting J.S. and L.U. through the Access to Research Infrastructure Action of the Improving Human Potential Programme to the EMBL Hamburg Outstation (contract HPRI-CT-1999–00017). This work was supported by the Slovak Academy of Sciences Grant 2/1018/21 (J.S. and L.U.), Austrian Science Research Fund Grant P14520 (G.W.), and an Austrian Federal Ministry of Education, Science, and Culture Research Contract (G.W.). 18. Mabuchi, K., Li, B., Ip, W. & Tao, T. (1997) Association of calponin with desmin intermediate filaments. Bundle formation of smooth muscle desmin intermediate filaments by calponin and its binding site on the desmin molecule. J. Biol. Chem. 272, 22662– 22666.

References

1. Wiche, G. (1998) Role of plectin in cytoskeleton organization and 19. Fujii, T., Takagi, H., Arimoto, M., Ootani, H. & Ueeda, T. (2000) Bundle formation of smooth muscle desmin intermediate fila- ments by calponin and its binding site on the desmin molecule. J. Biochem. 127, 457–465. dynamics. J. Cell Sci. 111, 2477–2486.

2. Uitto, J., Pulkkinen, L., Smith, F.J.D. & McLean, W.H.I. (1996) Plectin and human genetic disorders of the skin and muscle. Exp. Dermatol. 5, 237–246. 20. Correia, I., Chu, D., Chou, Y.H., Goldman, R.D. & Matsudaira, P. (1999) Integrating the actin and vimentin cytoskeletons. adhe- sion-dependent formation of fimbrin-vimentin complexes in macrophages. J. Cell Biol. 146, 831–842.

21. Keep, N.H., Winder, S.J., Moores, C.A., Walke, S., Norwood, F.L.M. & Kendrick-Jones, J. (1999) Crystal structure of the actin- binding region of utrophin reveals a head-to-tail dimer. Structure 7, 1539–1546.

3. Andra¨ , K., Lassmann, H., Bittner, R., Shorny, S., Fa¨ ssler, R., Propst, F. & Wiche, G. (1997) Targeted inactivation of plectin reveals essential function in maintaining the integrity of skin, muscle and heart cytoarchitecture. Genes Dev. 11, 3143–3156. 4. Ruhrberg, C. & Watt, F.M. (1997) The plakin family: versatile organizers of cytoskeletal architecture. Curr. Opin. Genet. Dev. 7, 392–397. 5. Leung, L.C., Green, K.J. & Liem, R.K.H. (2002) Plakins: a family of versatile cytolinker proteins. Trends Cell Biol. 12, 37–45. 22. Norwood, F.L., Sutherland-Smith, A.J., Keep, N.H. & Kendrick- Jones, J. (2000) The structure of the N-terminal actin-binding domain of human dystrophin and how mutations in this domain may cause Duchenne or Becker muscular dystrophy. Structure 8, 481–491.

23. Goldsmith, S.C., Pokala, N., Shen, W., Fedorov, A.A., Mat- sudaira, P. & Almo, S.C. (1997) The structure of an actin- crosslinking domain from human fimbrin. Nat. Struct. Biol. 4, 708–712. 6. Fuchs, P., Zo¨ rer, M., Rezniczek, G.A., Spazierer, D., Oehler, S., Castan˜ o´ n, M.J., Hauptmann, R. & Wiche, G. (1999) Unusual 5¢transcript complexity of plectin isoforms: novel tissue-specific exons modulate actin binding activity. Hum. Mol. Genet. 8, 2461– 2472.

24. Garcia-Alvarez, B., Bobkov, A., Sonnenberg, A. & De Pereda, J.M. (2003) Structural and functional analysis of the actin binding domain of plectin suggests alternative mechanisms for binding to F-actin and integrin beta 4. Structure 11, 615–625.

7. Rezniczek, G.A., Abrahamsberg, C., Fuchs, P., Spazierer, D. & Wiche, G. (2003) Plectin 5¢-transcript diversity: short alternative sequences determine stability of gene products, initiation of translation and subcellular localization of isoforms. Hum. Mol. Genet. 12, 3181–3194.

8. Gimona, M., Djinovic-Carugo, K., Kranewitter, W.J. & Winder, S.J. (2002) Functional plasticity of CH domains. FEBS Lett. 513, 98–106. 25. Hanein, D., Volkmann, N., Goldsmith, S., Michon, A.M., Leh- man, W., Craig, R., DeRosier, D., Almo, S. & Matsudaira, P. (1998) An atomic model of fimbrin binding to F-actin and its implications for filament crosslinking and regulation. Nat. Struct. Biol. 5, 787–792.

9. Winder, S.J., Hemmings, L., Maciver, S.K., Bolton, S.J., Tinsley, J.M., Davies, K.E., Critchley, D.R. & Kendrick-Jones, J. (1995) Utrophin actin binding domain: analysis of actin binding and cellular targeting. J. Cell Sci. 108, 63–71. 26. Nikolic, B., MacNulty, E., Mir, B. & Wiche, G. (1996) Basic amino acid residue cluster within nuclear targeting sequence motif is essential for cytoplasmic plectin-vimentin network junctions. J. Cell Biol. 134, 1455–1467. 10. Gimona, M. & Winder, S.J. (1998) Single calponin homology domains are not actin-binding domains. Curr. Biol. 8, 674–675.

1884 J. Sˇ evcˇ ı´ k et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

40. Lamzin, V.S. & Wilson, K.S. (1997) Automated refinement for protein crystallography. Methods Enzymol. 277, 269–305. 41. McRee, D.E.

(1999) XtalView/Xfit-A versatile program for manipulating atomic coordinates and electron density. J. Struct. Biol. 125, 156–165. 27. Steinbo¨ ck, F.A., Nikolic, B., Coulombe, P.A., Fuchs, E., Traub, P. & Wiche, G. (2000) Dose-dependent linkage, assembly inhibi- tion and disassembly of vimentin and cytokeratin 5/14 filaments through plectin’s intermediate filament-binding domain. J. Cell Sci. 113, 483–491.

42. Janin, J., Miller, S. & Chothia, C. (1988) Surface, subunit inter- faces and interior of oligomeric proteins. J. Mol. Biol. 204, 155–164. 28. Urbanikova, L., Janda, L., Popov, A., Wiche, G. & Sevcik, J. (2002) Purification, crystallization and preliminary X-ray analysis of the plectin actin-binding domain. Acta Crystallogr. D 58, 1368– 1370.

43. Ramakrishnan, C. & Ramachandran, G.N. (1965) Stereochemical criteria for polypeptide and protein chain conformations. II. Allowed conformations for a pair of peptide units. Biophys. J. 5, 909–933. 29. Nagai, K. & Thogersen, H.C. (1987) Synthesis and sequence- specific proteolysis of hybrid proteins produced in Escherichia coli. Methods Enzymol. 153, 461–481.

44. Morris, A.L., MacArthur, M.W., Hutchinson, E.G. & Thornton, (1992) Stereochemical quality of protein structure J.M. coordinates. Proteins 12, 345–364. 30. Giese, G. & Traub, P. (1986) Induction of vimentin synthesis in mouse myeloma cells MPC-11 by 12–0-tetradecanoylphorbol-13- acetate. Eur. J. Cell Biol. 40, 266–274.

45. Moores, C.A., Keep, N.H. & Kendrick-Jones, J. (2000) Structure of the utrophin binding domain bound to F-actin reveals binding by an induced fit mechanism. J. Mol. Biol. 297, 465–480. 31. Andra¨ , K., Kornacker, I., Jo¨ rgl, A., Zo¨ rer, M., Spazierer, D., Fuchs, P., Fischer, I. & Wiche, G. (2003) Plectin-isoform-specific rescue of hemidesmosomal defects in plectin (-/-) keratinocytes. J. Invest. Dermatol. 120, 189–197.

32. Popov, A.N. & Bourenkov, G.P. (2003) Choice of data-collection parameters based on statistic modelling. Acta Crystallogr. D 59, 1145–1153. 33. Matthews, B.W. (1968) Solvent content of protein crystals. J. Mol. Biol. 33, 491–497.

34. Pearson, W.R. & Lipman, D.J. (1988) Improved tools for biological sequence comparison. Proc. Natl. Acad. Sci. USA 85, 2444–2448.

35. Vagin, A. & Teplyakov, A. (1997) Molrep: An automated program for molecular replacement. J. Appl. Crystallogr. 30, 1022–1025.

36. Murshudov, G.N., Vagin, A.A. & Dodson, E.J. (1997) Refine- ment of macromolecular structures by the maximum likelihood methods. Acta Crystallogr. D 53, 240–255.

46. McGough, A., Way, M. & DeRosier, D. (1994) Determination of the a-actinin binding site on actin filaments by cryoelectron microscopy and image analysis. J. Cell Biol. 126, 433–443. 47. Galkin, V.E., Orlova, A., VanLoock, M.S., Rybakova, N.I., Ervasti, J.M. & Egelman, E.H. (2002) The utrophin actin-binding domain binds F-actin in two different modes: implications for spectrin superfamily of proteins. J. Cell Biol. 157, 243–251. 48. Sutherland-Smith, A.J., Moores, C.A., Norwood, F.L.M., Hatch, V., Craig, R., Kendrick-Jones, J. & Lehman, W. (2003) An atomic model for actin binding by the CH domains and spectrin-repeat modules of utrophin and dystrophin. J. Mol. Biol. 329, 15–33. 49. Strelkov, S.V., Herrmann, H. & Aebi, U. (2003) Molecular architecture of intermediate filaments. Bioessays 25, 243–251. 50. Glenney, J.R., Kaulfus, P., Matsudaira, P. & Weber, K. (1981) F-actin binding and bundling properties of fimbrin, a major cytoskeletal protein of microvillus core filaments. J. Biol. Chem. 256, 9283–9288. 37. Morris, R.J., Perrakis, A. & Lamzin, V.S. (2002) ARP/wARP’s model-building algorithms. I. The main chain. Acta Crystallogr. D 58, 968–975.

51. Gimona, M. & Mital, R. (1998) The single CH domain of calponin is neither sufficient nor necessary for F-actin binding. J. Cell Sci. 111, 1813–1821.

38. Jones, T.A. (1978) A graphics model building and refinement system for macromolecules. J. Appl. Crystallogr. 11, 268–272. 39. Bru¨ nger, A.T. (1993) Assessment of phase accurancy by cross- the Rfree-value. Methods and applications. Acta 52. Kraulis, P.J. (1991) MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystalogr. 24, 946–950. validation: Crystallogr. D 49, 24–36.