Biochemical Journal

Research article

Human α-synemin interacts directly with vinculin and metavinculin

Ning Sun, David R. Critchley, Denise Paulin, Zhenlin Li, Richard M. Robson

Abstract

Synemin is a very large, unique member of the IF (intermediate filament) protein superfamily. Association of synemin with the major IF proteins, desmin and/or vimentin, within muscle cells forms heteropolymeric IFs. We have previously identified interactions of avian synemin with α-actinin and vinculin. Avian synemin, however, is expressed as only one form, whereas human synemin is expressed as two major splice variants, namely α- and β-synemins. The larger α-synemin contains an additional 312-amino-acid insert (termed SNTIII) located near the end of the long C-terminal tail domain. Whether α- and β-synemins have different cellular functions is unclear. In the present study we show, by in vitro protein–protein interaction assays, that SNTIII interacts directly with both vinculin and metavinculin. Furthermore, SNTIII interacts with vinculin in vivo, and this association is promoted by PtdIns(4,5)P2. SNTIII also specifically co-localizes with vinculin within focal adhesions when transiently expressed in mammalian cells. In contrast, other regions of synemin show distinct localization patterns in comparison with those of SNTIII, without labelling focal adhesions. Our results indicate that α-synemin, but not β-synemin, interacts with both vinculin and metavinculin, thereby linking the heteropolymeric IFs to adhesion-type junctions, such as the costameres located within human striated muscle cells.

  • costamere
  • focal adhesion
  • intermediate filament
  • metavinculin
  • synemin
  • vinculin

INTRODUCTION

IFs (intermediate filaments) are very long, ∼10-nm diameter, filamentous components that function in general as mechanical integrators of cellular space [1]. They help maintain both the morphology and integrity of animal cells [2]. Different types of IFs are composed of cell-type specific IF proteins that, in turn, give rise to the distinct properties of the filaments they form [3,4]. There are more than 65 IF genes encoding distinct IF proteins within the human genome [5]. All of the IF proteins share a common tripartite structural feature, with a central ∼310-amino-acid conserved rod domain, flanked by hypervariable N-terminal head and C-terminal tail domains [4,68].

Synemin is a very large, unique type VI member of the IF protein superfamily [9]. It is expressed in all three types (skeletal, cardiac and smooth) of muscle cells [911], as well as in some specific non-muscle tissues and cells [10,1218]. The human synemin gene encodes two major isoforms, α-synemin (molecular mass of 180 kDa by SDS/PAGE and 172.7 kDa from the sequence) and β-synemin (molecular mass of 150 kDa by SDS/PAGE and 140.1 kDa from the sequence), which are generated by alternative splicing [19]. The larger α-synemin contains an additional unique 312-amino-acid insert located near the end of the long C-terminal tail domain. This insert is absent from and not homologous with any of the sequence within the smaller β-synemin isoform [19]. In contrast with the human α- and β-synemin isoforms, avian synemin is expressed in only one form, which is considered to be the orthologue of human α-synemin. Interestingly, avian synemin and human α- and β-synemins all fail to assemble into homopolymeric filaments in cells, requiring the presence of the type III IF protein desmin and/or vimentin to form heteropolymeric IFs [19,20].

Several proteins have been identified that interact with synemin. It has been shown previously that the rod domain of both avian [20] and human [21] synemin interacts directly with desmin. The human synemin rod domain also interacts with α-dystrobrevin [21], and dystrophin and utrophin [22]. The avian SNT (synemin tail) domain was originally shown to interact with vinculin and α-actinin [23]. Also, a fragment within the human α-SNT domain, sharing part of the β-SNT sequence, was reported to bind to dystrophin and utrophin [22]. Interestingly, the proteins found to interact with the SNT domain are all actin-binding proteins known to be present within myofibrils and/or within the costameric structures located in striated muscle cells. The costameres represent specialized focal-adhesion-type structures located periodically along and immediately subjacent to the sarcolemma [2426]. Taken together, these findings suggest that one of the functions of synemin is to link the heteropolymeric IFs to the sarcolemma.

Because avian synemin shares very low sequence identity (∼35%) with the human synemins, and considering that the human synemin gene generates two major isoforms (α and β) resulting in different lengths of their tail domains, the protein interactions of human and avian synemins may differ. Thus it is important to characterize interactions of the human synemin isoforms with those proteins shown previously to bind to the avian SNT domain. Most importantly, these studies will help to ascertain whether the two human synemin isoforms have different functions.

The major purpose of the present study was to identify and clarify the interactions of human α- and β-synemins with both vinculin and its muscle-specific isoform, metavinculin. The cytoskeletal protein vinculin is a major component of cell–cell and cell–extracellular matrix adhesion junctions [2729]. It is composed of a 90 kDa N-terminal head domain [VH (vinculin head)], a proline-rich linker and a 27 kDa C-terminal tail domain [VT (vinculin tail)] [30]. Metavinculin contains an additional 68-amino-acid insert within the tail domain [31,32]. It was originally shown that the VT interacts with avian synemin [23]. Whether there are interactions between the MVT (metavinculin tail) and human synemin isoforms is unknown. Preliminary studies in our laboratory have showed no interactions of human β-synemin with vinculin and metavinculin (N. Sun, unpublished work). We hypothesize that the unique 312-amino-acid insert within α-synemin contains the binding site for vinculin and metavinculin, and therefore confers extra functions to α-synemin. In the present study we show that the specific 312-amino-acid sequence present within the tail domain of human α-synemin, which is absent in β-synemin, interacts directly with the tail domains of both human vinculin and metavinculin. Taken together, our results indicate that α- and β-synemins have different functions within muscle cells. Based on its direct interaction with vinculin and metavinculin, human α-synemin appears able to link the heteropolymeric IFs to adhesion-type junctions, such as the costameres within striated muscle cells, thereby fulfilling an overall very important cytoskeletal role within the muscle cell cytoskeleton.

MATERIALS AND METHODS

Generation of cDNA constructs

The full-length human α- and β-synemin cDNAs, which served as the templates for amplifying all of the human synemin regions by PCR, have been described previously [22]. All primers used for PCR were synthesized at the DNA Sequencing and Synthesis Facility, Iowa State University, Ames, IA, U.S.A. cDNAs encoding human SNT fragments (SNTIa, residues 321–579; SNTIb, residues 580–920; SNTIII, residues 1151–1462; and SNβTII, residues 921–1251) were amplified by PCR using appropriate primer pairs, and were subsequently cloned into pFLAG-ATS expression vectors (Sigma) at 5′-HindIII and 3′-BglII sites as FLAG-tagged constructs. In order to increase the solubility of the SNTless region (residues 1–329), it was cloned into the pMAL-C2X MBP (where MBP is maltose-binding protein) expression vector (New England Biolabs) at 5′-EcoRI and 3′-SalI sites as a MBP-tagged fusion protein (MBP–SNTless).

The SNTIII region was subsequently divided into three consecutive subfragments named SNTIIIa (residues 1151–1243), SNTIIIb (residues 1244–1358) and SNTIIIc (residues 1359–1462). The three subfragments and full-length SNTIII were PCR amplified using primer pairs containing 5′-BamHI and 3′-HindIII sites, and were subsequently cloned into pMAL-C2X vectors as MBP-tagged proteins.

SNTIII was also PCR amplified and subcloned into the pEGFPc3 [where EGFP is enhanced GFP (green fluorescent protein)] expression vector (BD Biosciences), at 5′-HindIII and 3′-BamHI sites. SNTless cDNA was digested out of the MBP–SNTless construct at the sites of 5′-EcoRI and 3′-SalI and then inserted into the pEGFPc2 vector (BD Biosciences). The full-length human β-SNT domain (SNβT, residues 321–1251) was PCR amplified and was subsequently cloned into the pEGFPc3 vector at 5′-HindIII and 3′-BamHI sites.

The cDNAs of the human VT (residues 877–1066) and MVT (residues 877–1134) were amplified by RT (reverse transcriptase)–PCR from the human skeletal muscle total RNA library (BD Biosciences). The amplified VT and MVT cDNAs were then cloned into the pGEX-4T2 GST (where GST is glutathione transferase) expression vector (GE Healthcare) at 5′-BamHI and 3′-XhoI sites.

The accuracy of all constructs was confirmed by automated sequencing in the DNA Sequencing and Synthesis Facility at Iowa State University, Ames, IA, U.S.A.

Expression and purification of recombinant proteins

Each expression construct was transformed into Escherichia coli BL21-Codon Plus (DE3) bacterial cells (Stratagene). Protein expression was induced by IPTG (isopropyl β-D-thiogalactoside). FLAG-tagged synemin fusion proteins were purified by affinity chromatography using anti-FLAG M2-agarose affinity gel (Sigma). GST–VT and GST–MVT were batch-purified using glutathione-agarose (Sigma). MBP-tagged synemin fusion proteins were batch-purified or purified by affinity chromatography using amylose resin (New England Biolabs).

Antibodies

Anti-synemin 2856 pAb (polyclonal antibody) has been described previously [20]. Anti-FLAG pAb, anti-vinculin mAb (monoclonal antibody) hVIN-1, anti-desmin mAb DEU-10, anti-α-actinin mAb BM7.5 and anti-vimentin mAb V9 were obtained from Sigma. Anti-GST mAb, anti-GFP pAb and HRP (horseradish peroxidase)-conjugated anti-GFP pAb were obtained from Santa Cruz Biotechnology. Anti-MBP pAb and HRP-conjugated anti-MBP mAb were obtained from New England Biolabs. Alexa Fluor® secondary antibodies and Alexa Fluor® 594 phalloidin were obtained from Invitrogen.

Western blot analysis and blot overlay assays

Western blot analysis and blot overlay assays were conducted as described previously [20]. For blot overlay assays, equal picomoles of purified GST–VT, GST–MVT, BSA and GST were subjected to SDS/PAGE, and then transferred on to nitrocellulose membranes, with BSA and GST serving as controls. The resulting blot containing the respective transferred proteins was overlaid with 10 μg/ml of purified synemin fusion protein in PBS [140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4 (pH 7.4)], containing 0.1% (v/v) Tween 20 and 1% (w/v) non-fat dried skimmed milk powder. As a negative control, a duplicate blot was overlaid with buffer containing no protein. Protein interactions were detected with anti-FLAG pAb or anti-MBP serum by chemiluminescence. Densitometric analysis from three independent experiments of the blot overlay assays was performed using Kodak one-dimensional image analysis software (Eastman Kodak Company). For Western blot analysis using human tissue protein lysates, human adult skeletal muscle and human uterine smooth muscle total protein lysates were purchased from Biochain.

GST pulldown assays

Reactions containing 50 μM of each GST fusion protein were incubated with a 20 μl bed volume of glutathione-agarose beads in binding buffer [50 mM Tris/HCl, 150 mM NaCl and 0.5% Triton-X 100 (pH 7.4)] for 1 h at 4 °C. Purified synemin fusion protein was then added to reach a final concentration of 50 μM and was incubated for 12 h at 4 °C. After extensive washing with PBST (PBS containing 0.2% Tween 20), the beads were eluted with 2× SDS sample buffer. Eluates were subjected to SDS/PAGE and analysed by Western blotting with an anti-FLAG pAb or anti-MBP serum.

Solid-phase binding assays by ELISA

Solid-phase binding assays were performed as previously described [33] with minor modifications. Assays were all performed in duplicate. Briefly, 96-well microtitre plates (Nunc) were coated overnight at 4 °C with 100 nM purified GST–VT or GST–MVT diluted in PBS. Wells were then blocked with PBS containing 1% (w/v) non-fat dried skimmed milk powder for 4 h at 37 °C. A series of dilutions of MBP-tagged synemin fusion protein or of MBP alone ranging from 800 nM to 0 nM (i.e. buffer only), were then added and incubated for 1 h at 25 °C. After extensive washing with PBS containing 0.1% (v/v) Tween 20, bound MBP fusion proteins were detected with the HRP-conjugated anti-MBP mAb and the HRP substrate ABTS [2,2′-azino-bis-(3-ethylbenzthiazoline-6-sulfonic acid)] (Southern Biotechnology), by measuring absorbance at 405 nm (A405). Data were then analysed using Microsoft Excel.

Cell culture

A-10 cells, a rat aorta vascular smooth muscle cell line, were purchased from American Type Culture Collection. NIH 3T3 cells were a gift of Dr Janice E. Buss (Iowa State University, Ames, IA, U.S.A.). The A-10 cells were maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) FBS (fetal bovine serum), 4.5 g/l glucose, 4 mM L-glutamine, 100 units/ml penicillin G and 100 units/ml streptomycin at 37 °C (95% air/5% CO2). The NIH 3T3 cells were cultured under the same conditions as the A-10 cells, except that 10% (v/v) calf serum was used in place of 10% FBS in the growth medium.

Immunoprecipitation assays

NIH 3T3 cells were grown to 50% confluence and then transfected with either EGFP–SNTIII plasmids or the EGFP vectors alone, using JetPEI transfection reagent (Polyplus Transfection). Cells were then lysed using M-PER mammalian cell extraction buffer (Pierce Biotechnology) at 24 h post-transfection. The resulting cell lysates were then incubated in the presence or absence of 100 μM PtdIns(4,5)P2 (Avanti Polar Lipids) micelles at 4 °C for 20 min. The PtdIns(4,5)P2 micelles were prepared as previously described [34]. The samples were then incubated for 4 h at 4 °C with an anti-GFP pAb coupled to Protein A–agarose (Santa Cruz Biotechnology). After extensive washing with PBST, the beads were eluted with 2× SDS sample buffer. The resulting eluates were subjected to SDS/PAGE and analysed by Western blotting with an anti-vinculin hVIN-1 mAb or an HRP-conjugated anti-GFP pAb.

Immunofluorescence, confocal microscopy and transfection studies

Adult human skeletal muscle tissue sections were purchased from Biochain. The tissue sections were rinsed with PBS, permeabilized by 0.2% Triton X-100 in PBS, and blocked with 10% BSA in PBS at 25 °C. The sections were then incubated with anti-synemin 2856 pAb plus anti-vinculin mAb hVIN-1, anti-synemin 2856 pAb plus anti-α-actinin mAb BM7.5 or with anti-synemin 2856 plus anti-desmin mAb DEU-10, followed by incubation with appropriate Alexa Fluor® secondary antibodies. After extensive washing with PBS, the sections were mounted with coverslips using Vectashield mounting medium (Vector Laboratories). For transfection studies, cells were seeded on to collagen-coated coverslips to reach approx. 50% confluence. Cells were then transfected with EGFP-tagged synemin constructs, or with the EGFP vectors alone, using JetPEI transfection reagent. At 24–36 h post-transfection, the cells were fixed, permeabilized and immunostained with anti-GFP pAb plus anti-vinculin hVIN-1, anti-GFP plus anti-vimentin mAb V9 or with anti-GFP pAb plus Alexa Fluor® 594 phalloidin following the procedures described above. Epifluorescence microscopy was conducted using a Leica DMIRE2 inverted microscope equipped with a CCD (charge-coupled-device) camera in the Hybridoma Facility at Iowa State University (Ames, IA, U.S.A.). Confocal microscopy was conducted using a Leica TCS NT confocal microscope in the Confocal Microscopy and Image Analysis Facility at Iowa State University (Ames, IA, U.S.A.).

RESULTS

Mapping the binding sites of vinculin and metavinculin in human synemin

How the large IF protein synemin contributes to the organization as well as functionality of the molecular contractile apparatus within mammalian muscle cells has long been a central interest of our studies. In order to examine the putative interactions of human synemin with vinculin and metavinculin, and to map the binding site(s) of vinculin and metavinculin within human synemin isoforms, polypeptides corresponding to specific regions spanning the entire amino acid sequences of human α- and β-synemins were generated as shown in Figure 1(A). SNTless, SNTIa, SNTIb and SNβTII are regions shared by both human α-synemin and β-synemin, whereas SNTIII is the additional 312-amino-acid insert present only in α-synemin. Regions covering the C-terminal tail (SNTIa, SNTIb, SNTIII and synβTII) were all cloned as FLAG-tagged fusion proteins. SNTless resulted from deleting the entire long C-terminal tail domain, and was expressed as a MBP-tagged fusion protein to increase its solubility during purification. These synemin fusion proteins were expressed in bacteria, affinity-purified and analysed by SDS/PAGE as shown in Figure 2(A). The identities of these bands were confirmed by Western blot analysis with anti-FLAG and anti-MBP pAbs (results not shown). In the present study, human VT and MVT were chosen because we had previously determined that only the VT interacts with avian synemin whereas the VH does not [23], and vinculin is highly conserved among different species, i.e. over 95% identity [35]. Human VT and MVT were expressed as GST-tagged fusion proteins (Figure 1B).

Figure 1 Schematic diagram of each protein and construct used in the in vitro binding studies

(A) Schematic diagram of human synemin and the synemin constructs. Human synemin is composed of a short, 10 amino acid head domain (H), a conserved central 310-amino-acid rod domain (Rod) and a long C-terminal tail domain (Tail). Human α- and β-synemin sequences are identical except that α-synemin contains an additional 312-amino-acid insert (SNTIII, ∼33 kDa) located near the end of the C-terminal tail. Human α- and β-synemins were divided as illustrated. The predicted molecular mass (without tags), and the relative position of each fragment in the amino acid sequence, are shown. Fragments spanning the human β-SNT domain (SNTIa, SNTIb and SNβTII) were cloned into the pFLAG-ATS vector as FLAG-tagged fusion proteins. SNTless and the three subfragments of SNTIII (SNTIIIa, SNTIIIb, and SNTIIIc) were cloned into pMALC2X vector as MBP fusion proteins. SNTIII was constructed as both a FLAG-tagged fusion protein and an MBP-tagged fusion protein. (B) Schematic diagram of vinculin/metavinculin and the tail constructs. Vinculin and metavinculin are composed of a globular head domain, a proline-rich linker and an elongated tail domain. Metavinculin contains an additional 68 amino acids (shaded oval) within the tail domain. VT and MVT domains were cloned as GST fusion proteins.

Figure 2 SNTIII interacts with both VT and MVT domains in vitro

(A) SDS/PAGE of purified synemin fusion proteins. (B) Blot overlays. Purified GST–VT, GST–MVT, BSA and GST alone (100 pM of each) were electrophoresed and transferred on to nitrocellulose membranes. After blocking, membranes were overlaid with or without 10 μg/ml purified synemin fusion proteins in 1% non-fat dried skimmed milk in PBS (pH 7.4). Bound synemin fusion proteins were detected with an anti-FLAG or anti-MBP pAb by chemiluminescence. Only SNTIII binds to both VT and MVT. (C) GST–VT or GST alone was immobilized on glutathione-agarose and incubated with each purified FLAG-tagged SNT fragment as described in the Materials and methods section. The resulting eluates were each divided into equal volumes and subjected to Western blot analysis with anti-FLAG pAb and anti-GST mAb respectively. (D) The same experimental design was followed as described in (C), with the substitution of GST–MVT for GST–VT. SNTIII is specifically precipitated by GST–VT and GST–MVT. IB, immunoblot.

Interactions of specific protein domains of human synemin with human VT and MVT were analysed by blot overlay assays. As shown in Figure 2(B), blots containing purified GST–VT, GST–MVT, BSA and GST were probed with each purified synemin fusion protein. Specific interactions of the synemin fusion proteins with GST–VT and/or GST–MVT were detected with anti-MBP and anti-FLAG pAbs. Only SNTIII, which is the 312-amino-acid insert present in α-synemin but absent in β-synemin, showed interactions with both VT and MVT. None of the other synemin fusion proteins showed any detectable interaction with VT or MVT (Figure 2B). These results indicate that α-synemin harbours the binding site(s) for both vinculin and metavinculin.

GST pulldown assays were also conducted to examine the possible interactions of synemin fusion proteins with VT or MVT, and to confirm the results of the blot overlay assays. The GST pulldown assays (Figures 2C and 2D) demonstrated that only SNTIII was specifically precipitated by both GST–VT (Figure 2C) and GST–MVT (Figure 2D). None of the other synemin fusion proteins were precipitated by either VT or MVT. GST alone did not precipitate any of the FLAG-tagged synemin fusion proteins. Interactions of MBP–SNTless with GST–VT and GST–MVT were also tested by GST pulldown assays. MBP–SNTless was not precipitated by either GST–VT or GST–MVT (results not shown). These results further indicated that SNTIII is the region that interacts directly with vinculin or metavinculin.

Binding of SNTIII to VT is regulated by the intramolecular head–tail association of vinculin

Vinculin cycles between ‘open’ and ‘closed’ conformations which is regulated by the intramolecular association between the head and tail domains [36,37]. The vinculin head–tail association is known to block most of the ligand-binding sites, such as those for talin [38] and α-actinin [39] in the head domain, and for F-actin [40] in the tail domain. To confirm the in vivo interaction of SNTIII with the VT, NIH 3T3 cells were transfected with EGFP-tagged SNTIII plasmids and subjected to immunoprecipitation with an anti-GFP pAb. Because PtdIns(4,5)P2 has been reported to induce a conformational change in vinculin that results in increased binding to its ligands [34,41,42], immunoprecipitation assays were also conducted in the presence or absence of exogenous PtdIns(4,5)P2. As shown in Figure 3, in the presence of PtdIns(4,5)P2, vinculin was co-precipitated with EGFP–SNTIII by the anti-GFP pAb, demonstrating an interaction between vinculin and SNTIII. Without exogenous PtdIns(4,5)P2, vinculin was not detected in the samples of EGFP–SNTIII precipitation, indicating that the association between vinculin and SNTIII is promoted by PtdIns(4,5)P2. These results are consistent with a model in which ligand binding can be regulated by the intramolecular head–tail association of the vinculin molecules.

Figure 3 Immunoprecipitation of vinculin with overexpressed SNTIII in NIH 3T3 cells

NIH 3T3 cells were transfected with EGFP–SNTIII plasmids or EGFP vector alone. At 24 h post-transfection, the cells were lysed, and the resulting cell lysates were incubated with an anti-GFP pAb immobilized on Protein A–agarose, in the presence or absence of 100 μM PtdIns(4,5)P2. Each resulting immunoprecipitate was divided into equal volumes and subjected to Western blot analysis with an anti-vinculin mAb hVIN-1 or HRP-conjugated anti-GFP pAb respectively. Mock, which served as a negative control, denotes the incubation of mock transfected cell lysates with anti-GFP immobilized on Protein A–agarose. Lys denotes the load of non-transfected NIH 3T3 cell lysates. Vinculin was immunoprecipitated by an anti-GFP pAb along with the EGFP–SNTIII in the presence of 100 μM PtdIns(4,5)P2.

The C-terminal 104-amino-acid sequence of SNTIII contains the primary binding site for vinculin and metavinculin

To further define the region within human α-synemin that mediates the interaction with VT and MVT, SNTIII was divided into three consecutive subfragments named SNTIIIa, SNTIIIb and SNTIIIc (Figure 1A). This division of SNTIII was done according to the regions within its cDNA sequence where optimal primer design was possible. The three subfragments were all cloned as MBP-tagged fusion proteins. Full-length SNTIII was also generated as an MBP-tagged fusion protein and was used in parallel as a positive control in subsequent experiments (Figure 4A). The specific interaction of each subfragment with VT and MVT was analysed by blot overlay assays. As shown in Figure 4(B), MBP-tagged full-length SNTIII demonstrated specific interactions with both VT and MVT, as had the FLAG-tagged SNTIII. This result indicates that the interactions are highly specific regardless of the nature of the attached recombinant tag. SNTIIIc, which is the C-terminal 104-amino-acid sequence of SNTIII, showed similar binding activities as the full-length SNTIII with VT and MVT (Figures 4B and 4C). Relatively weak binding of SNTIIIa, the N-terminal 93-amino-acid sequence of SNTIII, to VT/MVT was observed (Figures 4B and 4C). No interaction of SNTIIIb, the central 115-amino-acid sequence within SNTIII, with VT or MVT was observed in the blot overlay assays. These results indicate that the primary binding site for vinculin and metavinculin in human α-synemin is located in the C-terminal 104-amino-acid sequence within SNTIII.

Figure 4 The C-terminal 104-amino-acid sequence of SNTIII is the primary binding site for both vinculin and metavinculin

(A) SDS/PAGE of purified MBP–SNTIII and the three MBP-tagged subfragments. MW, molecular mass marker. (B) Blot overlays. As in Figure 2(B), 100 pM of purified GST–VT, GST–MVT, BSA and GST alone were subjected to SDS/PAGE and transferred on to nitrocellulose membranes. The membranes were then overlaid with or without 10 μg/ml purified MBP-tagged SNTIII sub-fragments. Bound proteins were detected with anti-MBP pAb. An overlay of MBP-tagged full-length SNTIII served as a positive control. (C) Densitometry analysis of three independent replicates of blot overlay assays. The signal intensity of background on each blot was set as zero. SNTIIIc showed a similar affinity with VT and MVT as the full-length SNTIII.

Interactions of VT and MVT with SNTIII/SNTIIIc are saturable and act in a competitive manner

To determine whether the interactions of SNTIII and SNTIIIc with VT or MVT are saturable and to provide quantitative assessments of the interactions, solid-phase binding assays by ELISA were conducted as previously described [33]. Increasing concentrations of MBP–SNTIII or MBP–SNTIIIc were added to 96-well plates coated with equal picomoles of GST–VT or GST–MVT per well. Duplicate plates incubated with increasing concentrations of purified MBP alone served as controls. As shown in Figure 5, the interactions of SNTIII with VT (Figure 5A) and MVT (Figure 5B) are both dose-dependent and saturable. Determination of the concentrations of MBP–SNTIII at the half-maximal A405 yielded Kd values of 25±4 nM and 53±7 nM for SNTIII binding to VT and MVT respectively. Similarly, interactions of SNTIIIc with VT (Figure 5A) and MVT (Figure 5B) were also dose-dependent and saturable, yielding Kd values of 55±9 nM and 70±11 nM. No specific interactions were observed between MBP alone and the VT or MVT (Figures 5A and 5B). Experiments using 96-well plates coated with GST alone also did not show any specific interactions of SNTIII or of SNTIIIc with GST itself (results not shown). These results provide further evidence that interactions of SNTIII or SNTIIIc with VT and MVT are highly specific.

Figure 5 Interactions of VT and MVT with SNTIII or with SNTIIIc are both saturable and act in a competitive manner

(A and B) 96-well plates were coated with 100 nM purified GST–VT (A) or GST–MVT (B) and were incubated with increasing concentrations of MBP–SNTIII (●), MBP–SNTIIIc (▲) or MBP alone (■) for 1 h at 25 °C. After extensive washing, bound MBP fusion proteins were detected with HRP-conjugated anti-MBP mAb and the HRP substrate ABTS by measurement of A405. Background A405 values of wells incubated with buffer only were subtracted. The resulting A405 values were plotted versus the concentration of incubated MBP fusion proteins. The values represent the means±S.D. from three independent experiments (the data were fitted to a moving average trendline with a period of 2). Interactions of VT and MVT with SNTIII, or with SNTIIIc, were both saturable. (C) Increasing concentrations of GST–VT, along with 300 nM purified MBP–SNTIIIc, were added to a 96-well plate coated with GST–MVT for 1 h at 25 °C. Bound SNTIIIc was then determined as described in (A) and (B). Values represent the means±S.D. from three independent experiments. The values were plotted versus the concentration of GST–VT, and fitted to a linear regression trendline. The VT competes with the MVT for the interaction with SNTIIIc.

We also determined whether VT and MVT compete with each other in binding to SNIIIc. Purified MBP–SNTIIIc at a previously determined saturation concentration (300 nM; Figure 5B) along with increasing concentrations of GST–VT were added to 96-well plates coated with GST–MVT. With an increase in concentration of VT, the amount of SNTIIIc bound with MVT decreased as reflected by the A405 (Figure 5C). This result indicates that VT competes with MVT for binding to SNTIIIc.

Synemin co-localizes with vinculin/metavinculin within costameres of human adult skeletal muscle cells

To examine the relationship of synemin and vinculin/metavinculin within a cellular context, such as mammalian muscle cells, subcellular locations of synemin and vinculin/metavinculin within human adult skeletal muscle cells were analysed by immunohistochemistry using confocal microscopy. Expression of both α- and β-synemins in human skeletal muscles was first confirmed by RT–PCR amplification of synemin cDNA fragments from a human skeletal muscle total RNA library. The primer pair of 5′-GAAAAAGAAATTAAAATACCCCACGAA-3′ and 5′-AAACCAATGCCCATCATTCTC-3′ was used in the PCR, which will amplify a ∼1 kb cDNA fragment from β-synemin mRNA and a ∼2 kb fragment from α-synemin mRNA. The results indicated that both α- and β-synemin mRNAs were present within human skeletal muscle cells (Figure 6A, left-hand panel). The two amplified fragments were also confirmed as synemin cDNAs by automated sequencing. An anti-synemin 2856 pAb was then used for immunolabelling of synemin within the human skeletal muscle cells. This antibody was previously shown to specifically recognize both human α- and β-synemins expressed within mammalian cells [43] as well as porcine α- and β-synemins [11]. We also analysed the specificity of this antibody for human synemin by Western blot analysis. The results showed that the anti-synemin 2856 pAb recognizes two specific bands at approx. 180 kDa and 150 kDa within both human skeletal muscle and human uterine smooth muscle total protein lysates (Figure 6A, right-hand panel). These molecular masses correspond in size to human α-synemin (180 kDa) and β-synemin (150 kDa). The relative expression levels of these two synemin isoforms are in agreement with previous studies reporting that the smaller β-synemin is expressed at a relatively higher amount than that of the larger α-synemin isoform in human skeletal muscle tissues, and is expressed at an approximately equal amount with that of α-synemin in human smooth muscle tissues [19,21,44].

Figure 6 Confocal microscopy studies indicate that synemin co-localizes with vinculin within human adult skeletal muscle cells

(A) Left-hand panel, RT–PCR amplifying a ∼1 kb cDNA fragment from β-synemin mRNA and a ∼2 kb cDNA fragment from α-synemin mRNA that present in a human skeletal muscle total RNA library. Right-hand panel, Western blotting of human adult skeletal muscle (SK) and uterine smooth muscle (UT) total protein lysates with anti-synemin 2856 pAb. A total of 50 μg of each total protein lysate was separated by SDS/PAGE and subjected to Western blotting using anti-synemin 2856. (BD) Confocal microscopy analysis of the localization of synemin and desmin within longitudinal sections of human skeletal muscle. Synemin was labelled green with anti-synemin 2856 pAb followed by staining with secondary Alexa Fluor® 488 goat anti-rabbit IgG. Desmin was labelled red with anti-desmin DEU-10 mAb, followed by staining with secondary Alexa Fluor® 594 goat anti-mouse IgG. Synemin co-localized with desmin at the myofibrillar Z-lines (D). (EL) Confocal microscopy analysis of synemin and vinculin within human skeletal muscle. Synemin was labelled green with anti-synemin 2856. Vinculin was labelled red with anti-vinculin hVIN-1 mAb. (F), (G) and (H) represent the boxed area in (E). (J), (K) and (L) represent the boxed area in (I). In cross-section (EH), synemin co-localized with vinculin along the sarcolemma (H, arrows). In longitudinal sections (IL), synemin co-localized with vinculin within the costameres (L, inset, arrows). (MP) Confocal microscopy analysis of synemin and α-actinin within human skeletal muscle. Synemin was labelled green with anti-synemin 2856. α-Actinin was labelled red with an anti-α-actinin mAb BM7.5. (N), (O) and (P) represent the boxed area in (M). Synemin co-localized with α-actinin at the myofibrillar Z-lines as well as along the sarcolemma (P, arrows). (Bars, 25 μm)

Double immunofluorescence staining of synemin and either desmin, α-actinin or vinculin, within human adult skeletal muscle cells was then performed using the antibodies anti-synemin 2856, anti-desmin DEU-10, anti α-actinin BM7.5 and anti-vinculin hVIN-1. Anti-vinculin hVIN-1 recognizes a specific epitope within the head domain of vinculin. Thus this antibody also labels any metavinculin expressed within human skeletal muscles. In longitudinal sections, synemin co-localized with desmin in striated patterns at the myofibrillar Z-lines (Figures 6B–6D). This result is in agreement with previous studies showing co-localization of synemin and desmin at the Z-lines within mammalian skeletal muscles [11,19,21], indicating the formation and presence of synemin/desmin heteropolymeric IFs.

In both longitudinal and cross-sections the labelling of vinculin was primarily restricted to along the sarcolemma (Figures 6G and 6K). In cross-sections, the labelling of synemin was not only within each muscle fibre, but also significantly co-localized with vinculin along the sarcolemma (Figures 6E–6H, arrows). In longitudinal sections, striations of synemin labelling ended on the sarcolemma that was labelled with vinculin (Figures 6I–6L), and co-localized with vinculin in dot-like structures consistent with the costameres present periodically along the sarcolemma (Figure 6L, inset, arrows). These confocal microscopy results indicate association of synemin with vinculin and metavinculin within mammalian (human) skeletal muscle cells.

We have previously reported the interaction of avian synemin with α-actinin [20,23]. Thus the localizations of human synemin and α-actinin within human skeletal muscle tissues were also examined. Synemin co-localized with α-actinin at the myofibrillar Z-lines as well as along the sarcolemma (Figures 6M–6P, arrows), indicating co-localization of synemin with α-actinin within mammalian (human) skeletal muscle cells.

SNTIII and SNTIIIc co-localize with vinculin at focal adhesion sites within A-10 cells

To further clarify whether SNTIII and SNTIIIc interact with vinculin and metavinculin within living cells, the subcellular localization of expressed SNTIII and SNTIIIc was examined by confocal microscopy. A-10 cells, a rat smooth muscle cell line, were transiently transfected with EGFP–SNTIII or EGFP–SNTIIIc plasmids and immunolabelled with anti-vinculin hVIN-1. Cells with a moderate expression level of GFPs were then examined. Transfection of the EGFP vector alone into A-10 cells served as a reference and showed classical non-specific localization throughout the cytoplasm (Figure 7B). Localization of vinculin was restricted primarily to both the large focal adhesions and the smaller focal complexes (Figures 7C, 7G and 7K). Both EGFP–SNTIII and EGFP–SNTIIIc clearly co-localized with vinculin within focal adhesions (Figures 7H and 7L), further supporting our findings obtained in the in vitro protein–protein binding assays. In addition, the distribution of SNTIII within the focal adhesions was clearly independent of vimentin IFs, as shown by confocal images of EGFP–SNTIII transfected cells immunolabelled with anti-vimentin V9 (Figures 7M–7P). Previous results have indicated that full-length human synemin associates with vimentin IFs within transfected mammalian cells [19]. Thus the results shown in the present study suggest that the interaction of SNTIII with vinculin helps link the synemin/vimentin heteropolymeric IFs directly to the focal adhesions.

Figure 7 SNTIII and SNTIIIc specifically localize within focal adhesions in A-10 cells

(AL) A-10 cells were transfected with EGFP, EGFP–SNTIII or EGFP–SNTIIIc plasmids. Vinculin was then labelled red 24 h post-transfection with anti-vinculin hVIN-1 plus secondary Alexa Fluor® 594 goat anti-mouse IgG. The last three pictures in each row represent enlarged views of the boxed area in the first picture. SNTIII and SNTIIIc both co-localized with vinculin within focal adhesions (H and L). (MP) A-10 cells were transfected with EGFP–SNTIII plasmids. Cells were then immunostained 24 h post-transfection with anti-vimentin V9 and secondary Alexa Fluor® 594 goat anti-mouse IgG. (NP) represents the boxed area in (M). SNTIII localized within focal adhesion-like structures at the cell periphery where vimentin was absent. (Bars, 25 μm)

SNβT and SNTless did not localize to focal adhesions

To ascertain whether only SNTIII and SNTIIIc co-localize with vinculin in focal adhesions, subcellular localizations of expressed protein domains of β-synemin that do not contain the SNTIII region were examined in A-10 cells using confocal microscopy. Human β-synemin was divided into two regions named SNβT and SNTless. SNβT is the full-length C-terminal tail domain of β-synemin, which does not contain the N-terminal head and rod domains. SNTless is the N-terminal head and rod domains of β-synemin. Both of the two regions were cloned into EGFP vectors and subsequently transfected into A-10 cells. Transient expression of EGFP–SNβT within the A-10 cells resulted in a diffuse cytoplasmic distribution without any specific focal adhesion-like structures (Figures 8A–8D). Transfection of EGFP–SNTless into the A-10 cells resulted in numerous perinuclear aggregates within the cytoplasm that did not co-localize with vinculin (Figures 8E–8G). The latter result agrees with the studies of ‘tail-less’ vimentin [45], desmin [46] and GFAP (glial fibrillary acidic protein) [47], in which all three showed formation of cytoplasmic aggregates within the transfected cells. Clearly, neither SNβT nor SNTless co-localized with vinculin within focal adhesion sites (Figures 8D and 8G). The latter set of results demonstrated that only regions from α-synemin (SNTIII and SNTIIIc) specifically co-localized with vinculin in focal adhesions.

Figure 8 SNβT and SNTless, which together cover the full-length amino acid sequence of β-synemin, do not co-localize with vinculin

(AD) Confocal microscopy of subcellular localization of SNβT within A-10 cells. A-10 cells were transfected with EGFP–SNβT plasmids and then immunostained with anti-vinculin hVIN-1. (BD) represents enlarged views of the boxed area in (A). (EG) Subcellular localization of EGFP–SNTless within A-10 cells. A-10 cells were transfected with EGFP–SNTless plasmids. Vinculin was then labelled red with anti-vinculin hVIN-1. Neither SNβT nor SNTless co-localized with vinculin within focal adhesions. (Bars, 25 μm)

Expression of SNTIII within A-10 cells did not disrupt normal cellular architecture

To examine whether exogenous expression of SNTIII may cause mislocalization of endogenous proteins, the overall appearances of focal adhesions, actin stress fibres and IFs within EGFP–SNTIII transfected A-10 cells were analysed using epifluorescence microscopy. As shown in Figure 9, the organization of the focal adhesions (Figure 9B), actin stress fibres (Figure 9D) and vimentin IFs (Figure 9F) within EGFP–SNTIII expressing cells (Figure 9, arrows) did not display significant changes in comparison with surrounding cells without EGFP–SNTIII expression. These results indicate that expression of SNTIII did not disrupt the normal cellular architecture within the transfected cells. Therefore co-localization of SNTIII with vinculin at focal adhesion sites, as shown in Figure 7, is not a result of non-specific distribution of SNTIII within the cells.

Figure 9 Expression of SNTIII did not result in significant changes in the cellular organization of focal adhesions, actin stress fibres and IFs

A-10 cells were transfected with EGFP–SNTIII plasmids followed by immunostaining of vinculin, F-actin and vimentin to label the focal adhesions (B), actin stress fibres (D) and IFs (F) respectively. (B), (D) and (F) represent the same microscope observation field as (A), (C) and (E) respectively. Arrows indicate the identical cell(s) in the same microscope observation field. In comparison with the surrounding cells without EGFP–SNTIII expression, cells that express EGFP–SNTIII did not display any significant difference in the appearances of the focal adhesions, actin stress fibres and IFs. (Bars, 15 μm).

DISCUSSION

In the present study we have defined the interactions of human α- and β-synemins with vinculin and metavinculin. Recombinant protein regions that include the entire amino acid sequence of human synemin were generated to map the vinculin/metavinculin binding site(s). Blot overlay assays showed that SNTIII, the 312-amino-acid insert present only in α-synemin, interacts with both VT and MVT (Figure 2B). These interactions are highly specific, as indicated by the absence of interactions of SNTIII with both BSA and GST, which served as negative controls (Figures 2B and 4B). Furthermore, both FLAG-tagged and MBP-tagged SNTIII showed specific interactions with VT and MVT (Figures 2B and 4B), demonstrating that the interactions are not dependent on the nature of the attached fusion tag. In vitro GST pulldown assays provided additional information regarding the interactions of different regions of human synemin with VT and MVT, also demonstrating specifically that only SNTIII present in α-synemin interacts with both VT and MVT (Figures 2C and 2D).

The primary binding site for VT/MVT within SNTIII was further mapped to its C-terminal 104-amino-acid sequence (SNTIIIc) (Figure 4B). As shown in the solid-phase binding assays (Figures 5A and 5B), SNTIIIc binds to VT and MVT with higher dissociation constants than those obtained with SNTIII. This indicates that the full-length SNTIII has higher affinities for VT and MVT than the truncated form (SNTIIIc). Also, SNTIIIa, albeit much weaker than that of SNTIIIc, has affinity for VT and MVT. Thus the full-length SNTIII may exhibit optimal interactions with VT and MVT in vitro and/or in vivo.

Interactions of VT with SNTIII and with SNTIIIc also showed lower dissociation constants than those of MVT, indicating a higher affinity of VT for α-synemin. The latter difference in the binding affinities could result from the presence of the additional 68-amino-acid insert within the MVT, which may interfere with its binding to SNTIIIc. VT and MVT also compete with each other for binding to SNTIIIc (Figure 5C), indicating that they bind to the same site on α-synemin. We have also examined for the first time the subcellular localization of different regions of human synemin within mammalian muscle cells by confocal microscopy. The subcellular localization pattern of proteins, or protein domains, often reflects their actual cellular functions. Using transfection studies, we demonstrated that SNTIII and SNTIIIc co-localized with vinculin at focal adhesion sites within mammalian smooth muscle cells (Figures 7E–7L). This provides additional evidence for the specific in vivo interactions of SNTIII and SNTIIIc with vinculin and metavinculin. SNβT and SNTless, which together cover the entire amino acid sequence of β-synemin, showed subcellular localizations distinct from those of SNTIII and SNTIIIc, by not labelling vinculin-containing focal adhesion sites (Figure 8). These results further demonstrate that it is SNTIII, which is absent in β-synemin, that specifically interacts with vinculin and metavinculin, and thereby is targeted to focal adhesions within mammalian muscle cells. The localization of SNTIII and SNTIIIc within focal adhesions were also independent of the vimentin IFs (Figures 7M–7P). It has been reported that both full-length human α- or β-synemin co-localized with vimentin when transiently expressed within mammalian cells [19]. These different localization patterns between SNTIII and full-length human synemin indicate that SNTIII is the region within human α-synemin that helps link synemin/vimentin heteropolymeric IFs to the focal adhesion sites. Confocal microscopy studies of subcellular localization of synemin and vinculin/metavinculin within human adult skeletal muscle cells also reveal that synemin co-localizes with vinculin within the costameres (Figure 6L), which are considered specialized focal adhesions located along and subjacent to the sarcolemma [26]. Thus α-synemin may directly link the synemin/desmin heteropolymeric IFs to the vinculin-rich costameric structures within the mammalian (human) skeletal muscle cells.

In the present study we have provided evidence that the interaction of α-synemin with VT may also be regulated by the molecular conformation of vinculin within a cellular context. Immunoprecipitation studies showed that vinculin was precipitated by SNTIII in the presence of exogenous PtdIns(4,5)P2 (Figure 3). This may suggest that the soluble pool of vinculin in the cell lysates was in the ‘closed’ form, which would block the binding site of SNTIII within the VT domain. It is known that PtdIns(4,5)P2 interferes with the ‘closed’ form of vinculin [41,42], which would thereby allow binding of SNTIII to the VT. It was previously shown that vinculin is in its activated (‘open’) form when recruited in focal adhesions [48]. Thus localization of SNTIII within focal adhesions may result from the direct interaction with the VT of activated vinculin. Previous studies have reported close associations of focal adhesions and IFs within mammalian cells [4951]. It has also previously been shown that focal adhesions are critical sites for the nascent assembly of vimentin [49] and keratin [50] IFs. However, direct interactions of IFs with specific components of focal adhesions were not identified in those studies. Our results provide the first in vitro/in vivo evidence of a direct interaction of IF proteins with components of focal adhesions. Whether the interaction of α-synemin with vinculin is able to direct the nascent formation of vimentin IFs at focal adhesions remains to be explored.

Overall, the results of the present study show that α-synemin, but not β-synemin, interacts directly with both vinculin and metavinculin. These findings indicate that the human α- and β-synemin isoforms have different cellular functions. The sequence of α-synemin is identical with β-synemin with the exception of the additional 312-amino-acid insert in its tail domain. Our results indicate that this additional insert confers extra functions for α-synemin. Interaction of α-synemin with vinculin and metavinculin may directly anchor the heteropolymeric IFs to adhesion-type junctions, such as the costameric regions within striated muscle cells. The latter interactions within costameres would strengthen the linkage of the synemin/desmin heteropolymeric IFs to the cell membrane, and thereby help maintain the structural integrity and functional stability of the muscle cell cytoskeleton during muscle contraction.

Acknowledgments

We thank Dr Ted W. Huiatt, Dr Steven Lonergan, and Dr Jo-Anne Powell Coffman for critical reading of this manuscript and their valuable suggestions. This project was supported by National Research Initiative Competitive Grant no. 2003-35206-12823 from the USDA Cooperative State Research, Education, and Extension Service. This journal paper of the Iowa Agriculture and Home Economics Experiment Station, Ames, Iowa, Project No. 6616, was supported by Hatch Act and State of Iowa funds.

Abbreviations: ABTS, 2,2′-azino-bis-(3-ethylbenzthiazoline-6-sulfonic acid); EGFP, enhanced green fluorescent protein; FBS, fetal bovine serum; GFP, green fluorescent protein; GST, glutathione transferase; HRP, horseradish peroxidase; IF, intermediate filament; mAb, monoclonal antibody; MBP, maltose-binding protein; MVT, metavinculin tail; pAb, polyclonal antibody; PBST, PBS containing 0.2% Tween 20; RT, reverse transcriptase; SNβT, β-synemin tail; SNT, synemin tail; VH, vinculin head; VT, vinculin tail

References

View Abstract