Xin is a protein that is expressed during early developmental stages of cardiac and skeletal muscles. Immunolocalization studies indicated a peripheral localization in embryonic mouse heart, where Xin localizes with β-catenin and N-cadherin. In adult tissues, Xin is found primarily in the intercalated discs of cardiomyocytes and the myotendinous junctions of skeletal muscle cells, both specialized attachment sites of the myofibrillar ends to the sarcolemma. A large part of the Xin protein consists of unique 16 amino acid repeats with unknown function. We have investigated the characteristics of the Xin repeats by transfection experiments and actin-binding assays and ascertained that, upon expression in cultured cells, these repeats bind to and stabilize the actin-based cytoskeleton. In vitro co-sedimentation assays with skeletal muscle actin indicated that they not only directly bind actin filaments, but also have the capability of arranging microfilaments into networks that sediment upon low-speed centrifugation. Very similar repeats were also found in `Xin-repeat protein 2' (XIRP2), a novel protein that seems to be expressed mainly in striated muscles. Human XIRP2 contains 28 Xin repeats with properties identical to those of Xin. We conclude that the Xin repeats define a novel, repetitive actin-binding motif present in at least two different muscle proteins. These Xin-repeat proteins therefore constitute the first two members of a novel family of actin-binding proteins.
The actin-based cytoskeleton is involved in many kinds of motile activities as well as in organizing cellular shape. Actin itself is not only extraordinarily conserved but is also among the most abundant proteins in eukaryotes. It seems, therefore, only logical that actin has widely been used as a template for the evolution of a great variety of proteins that bind to it and that in turn influence actin's detailed function and three-dimensional organization in cells. Based on primary structures, it was possible to group the plethora of actin-binding proteins into 60 distinct classes but it is assumed that this inventory is far from being complete (Pollard, 1999). Interestingly, the vast majority of actin-binding regions in these proteins are essentially globular domains, like (for instance) the myosin motor domain (Ruppel and Spudich, 1996; Geeves and Holmes, 1999) or the frequently occurring ABD (actin-binding domain) consisting of a tandem of two globular CH (calponin homology) domains (Stradal et al., 1998; Carugo et al., 1997; Gimona and Winder, 1998).
By contrast, there are relatively few examples in which multiple lateral contacts with several actin subunits are formed by extended rod-shaped proteins or protein subunits. Tropomyosin is the `classical' example for this binding mechanism (Pittenger et al., 1994; Perry, 2001). In addition, the giant protein nebulin has been proposed to bind in a similar fashion along thin myofilaments, thereby providing an elegant mechanism for the regulation of thin-filament length in skeletal muscle (Kruger et al., 1991; Labeit et al., 1991; Trinick, 1994). More recently identified examples for repetitive actin binding modules are the calponin-like (CLIK23) repeats from calponin (Gimona and Mital, 1998; Burgstaller et al., 2002; Lener et al., 2004) and the Caenorhabditis elegans UNC-87 protein (Kranewitter et al., 2001) or the seven-repeat motif in cortactin (He et al., 1998; Weed et al., 2000).
In 1996, a novel protein (called Xin) was found, using differential display, to be specifically expressed during early stages of cardiac development (Wang et al., 1996a). The protein was detected in actin-based cell-cell contacts of chicken and murine cardiac-muscle cells using specific antibodies and was found in a complex together with N-cadherin and β-catenin upon immunoprecipitation (Wang et al., 1999; Sinn et al., 2002). Treatment of cultured chicken embryos with antisense Xin oligonucleotides resulted in an abnormal development of the heart, indicating an important role for Xin in cardiac morphogenesis (Wang et al., 1999). The N-terminal half of the murine protein mainly consists of 16 repeats of a novel motif with unknown function and binding partners. We have cloned the human homologue of Xin and characterized its actin-binding capacities. Furthermore, we have identified a second, novel protein containing highly similar repeats, which we have named Xin-repeat protein 2 (XIRP2). Biochemical and cell-biological data suggest that Xin repeats form a novel actin-binding motif and that they define a new family of proteins capable of stabilizing F-actin by lateral binding.
Materials and Methods
Construction of expression constructs and purification of recombinant proteins
Xin and XIRP2 fragments encoding the complete repeat region of both proteins (for a description of all constructs used in this study, see Figs 1A, 9A) were amplified by polymerase chain reaction (PCR) (Saiki et al., 1985) using the Extend long-template PCR system according to the instructions of the manufacturer (Roche Diagnostics, Mannheim, Germany) with human genomic DNA isolated from cultured human skeletal-muscle cells as a template. Primers contained the appropriate restriction sites to allow cloning into our expression vectors. Both products were cloned into the pET23-T7 vector (Obermann et al., 1997). The cloned Xin PCR product was used as a template to amplify truncated variants of the repeat region using Pfu polymerase (Promega, Mannheim, Germany), which were also cloned into this vector. Constructs were transformed to Escherichia coli BL21-CodonPlus cells (Stratagene, Amsterdam, The Netherlands). Because bacterially expressed polypeptides contain a C-terminal His6 tag, expression and purification of the recombinant proteins was performed using Ni2+-NTA agarose beads (Qiagen, Hilden, Germany) as described previously (Obermann et al., 1997). The purity of the preparations was verified by sodium-dodecyl-sulfate polyacrylamide-gel electrophoresis (SDS-PAGE), and the protein concentration was determined using the BCA assay (Interchim, Mannheim, Germany).
For expression in eukaryotic cells, full-length Xin and the above Xin- and XIRP2-encoding cDNA fragments were cloned into a pEGFP-N3 vector (BD Biosciences Clontech, Heidelberg, Germany) containing a modified multiple cloning site using standard procedures. A fusion between monomeric red fluorescent protein (mRFP) and actin was obtained by exchanging the enhanced green fluorescent protein (EGFP) sequence in EGFP-actin (BD Biosciences, Palo Alto, CA, USA) for mRFP (provided by R. Tsien, Howard Hughes Medical Institute, La Jolla, CA) (Campbell et al., 2002).
Circular dichroism spectroscopy
XR6-11 and XR1-16 (for construct denomination see Fig. 1A) were expressed and purified as described above. Elution fractions from the Ni2+-NTA column were dialysed against low-salt buffer (20 mM Tris-HCl, pH 7.5) and fractions containing the Xin polypeptide were applied to a Mono Q anion exchange column (Mono Q 10/10; Amersham, Freiburg, Germany) equilibrated in low-salt buffer. After extensive washing with low-salt buffer, protein was eluted with a salt gradient (0-1 M KCl in low-salt buffer). The Xin polypeptides eluted at a salt concentration of 0.25 M KCl. After dialysis against phosphate buffer [20 mM sodium phosphate buffer (NaPi), pH 7.4], they were concentrated by ultrafiltration using a Vivaspin Concentrator (Sartorius, Göttingen, Germany) and the protein concentration was determined as described (Pace et al., 1995). With the XR1-16 protein solution (concentration 0.7 mg ml–1), 20 circular dichroism (CD) spectra in the far ultraviolet (UV) range were recorded at 10°C using a J-715 spectropolarimeter (Jasco, Gross-Umstadt, Germany) and cells with a light path of 0.1 mm or 1 mm. Similarly, spectra were obtained with the native XR6-11 polypeptide, and with XR6-11 that was denatured by an overnight incubation in 4 M guanidine hydrochloride (GndHCl) at room temperature, at a final concentration of 2 μM. For both native and denatured polypeptides, CD spectra were accumulated with identical parameter settings. For all spectra, 15 scans were averaged and the mean molar residue ellipticity was plotted against the wavelength.
PtK2 cells and A7r5 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS), 4 mM L-glutamine, 1% non-essential amino acids (NEAA), 2 mM sodium pyruvate, 100 U/ml penicillin and 1 μg/ml streptomycin (all from Invitrogen, Karlsruhe, Germany). B16F1 cells were cultured as described elsewhere (Ballestrem et al., 1998).
Transient transfection experiments and immunofluorescence microscopy
PtK2 and A7r5 cells were seeded onto glass coverslips, grown to 75% confluence and transfected using FuGENE 6 according to the manufacturer's instructions (Roche Diagnostics, Mannheim, Germany). In most experiments, cells were transfected overnight with 2 μg DNA and 4 μl FuGENE 6. 24 hours and 48 hours after transfection, cells were fixed with 4% paraformaldehyde in PBS for 10 minutes and permeabilized with 0.5% Triton X-100 in PBS for 10 minutes. Fixed cells were incubated with coumarin phenyl isothiocyanate-(CPITC-) conjugated phalloidin (Sigma-Aldrich, Taufkirchen, Germany) to visualize F-actin before they were embedded in Mowiol (Calbiochem, Bad Soden/Ts, Germany). Cells were analysed and photographed using an Axiophot or Axiovert microscope (Carl Zeiss, Jena, Germany) equipped with a cooled charge-coupled device (CCD) camera.
B16F1 cells were transfected using Superfect (Qiagen, Hilden, Germany) according to the manufacturer's instructions. After 24 hours, cells were trypsinized and plated on coverslips coated with laminin (Roche Diagnostics, Mannheim, Germany). Approximately 3 hours later, live-cell imaging was performed with the Focht Chamber System 2 (Bioptechs, Butler, PA, USA) or an open chamber system from Warner Instruments (Hamden, CT, USA) using IPLab Spectrum (Scanalytics, Fairfax, VA, USA). For phalloidin staining, the cells were fixed on the microscope with a mixture of 4% formaldehyde and 0.25% glutaraldehyde in PBS for 20 minutes, extracted with 0.1% Triton X-100 in PBS for 1 minute and stained with Alexa-594-conjugated phalloidin (Molecular Probes, Leiden, the Netherlands).
For Latrunculin A (LatA) experiments, A7r5 cells were grown on glass coverslips and transfected with the construct encoding XR1-16-EGFP using Superfect according to the instructions of the manufacturer (Qiagen, Hilden, Germany). The cells were incubated with the Superfect-DNA mixture for 4 hours, and cells were grown for 48 hours in fresh culture medium. Subsequently, 0.5 μM LatA was added to the medium and cells were fixed and permeabilized with paraformaldehyde (PFA) and Triton X-100 as described above at several time points after the addition of LatA. To visualize focal adhesions and F-actin, cells were stained with, respectively, mouse monoclonal antibody (mAb) vin-11-5 (Sigma-Aldrich, Taufkirchen, Germany), which is specific for vinculin, and Texas-Red-conjugated goat anti-mouse IgG secondary antibody (Southern Biotech, Birmingham, AL, USA), and CPITC-conjugated phalloidin (Sigma-Aldrich, Taufkirchen, Germany). Cells were embedded and analysed by microscopy as described above.
Alternatively, A7r5 cells were double transfected with constructs encoding mRFP-actin and XR1-16-EGFP or an empty EGFP vector as control. Video analysis was performed to visualize the effect of the expression of Xin-EGFP on the stability of the actin cytoskeleton during LatA addition and wash out. Separate recordings of EGFP- and mRFP-tagged proteins expressed in the same cell were performed essentially as described (Steffen et al., 2004). Sequential image stacks recorded using IPLab 3.1 software were analysed and processed using Scion Image 1.62 (Scion Corporation, Frederick, MD).
Actin co-sedimentation assays
Muscle actin was purified from acetone powder prepared from chicken breast muscle using the protocol of Spudich and Watt (Spudich and Watt, 1971). Co-sedimentation assays were performed in F-actin buffer (100 μM ATP, 1 mM EGTA, 100 mM KCl, 2 mM MgCl2, 20 mM imidazole, pH 7.5) or G-actin buffer (200 μM ATP, 100 μM dithiothreitol, 200 μM CaCl2, 20 mM imidazole, pH 7.0). Ten microliters of a 50 μM G-actin solution were added to 70 μl of F-actin buffer, and the actin molecules were allowed to polymerize for 45 minutes at room temperature. F-actin buffer and purified Xin or XIRP2 fragments were added and the mixture (final volume 100 μl) was incubated for 45 minutes and centrifuged for 1 hour at 4°C at 100,000 g or for 45 minutes at 10,000 g, for high-speed and low-speed co-sedimentation assays, respectively. Pellets were washed with H2O and resuspended in 125 μl SDS sample buffer. 25 μl 5× SDS sample buffer was added to the supernatants. Equal quantities of the samples were analysed by SDSPAGE. Gels were stained with Coomassie Brilliant Blue and protein amounts were quantified by densitometry using QuantityOne software (BioRad, Munich, Germany). For competition assays, 10 μl from a 50 μM G-actin solution were added to 70 μl of F-actin buffer, and the actin molecules were allowed to polymerize as described above. Subsequently, pig-stomach tropomyosin (a kind gift from M. Gimona, Consortio Mario Negri Sud, Santa Maria Imbaro, Italy) was added to a final concentration of 1.6 μM. Following incubation at room temperature for 30 minutes, F-actin buffer and increasing amounts of XR1-11 were added to a final volume of 100 μl, the mixture was incubated for 45 minutes at room temperature and centrifuged for 45 minutes at 10,000 g. Pellets and supernatants were analysed as described above.
Mixtures of Xin repeat constructs and F-actin were directly applied to glow-discharged formvar-carbon coated 300-mesh copper grids. After 2 minutes, the solution was removed and filaments were negatively contrasted with 1% aqueous uranyl acetate. After air drying, grids were observed in a Philips CM12 transmission electron microscope operated at an accelerating voltage of 80 kV.
Human Xin contains novel 16 amino acid repeats
The recently discovered murine protein Xin has a length of 1132 amino acids and a predicted molecular mass of 124 kDa (accession number XM_125236). The N-terminal 60% of this protein consists mainly of 16 variations of a novel repeat motif with unknown function. The consensus sequence of these `Xin repeats' is GDVQXXRWLFETXPLD. The corresponding human protein (1121 amino acids, 122 kDa), deduced from our experimental data (accession number AJ626899) and the genomic sequence (accession number AC092053), also contains 16 Xin repeats (Fig. 1A,B) and shows 78% identity (81% identity within the repeat region) to the murine protein. The consensus sequences of human and murine Xin repeats are identical (Fig. 1B).
CD spectroscopy of recombinant Xin suggests an extended structure of the repeat region
The bacterially expressed Xin-repeat constructs XR6-11 and XR1-16 were purified, and molar extinction coefficients of ϵ=25,310 M–1 cm–1, and ϵ=67,515 M–1 cm–1 were calculated, respectively (Pace et al., 1995). For both constructs, CD spectra were recorded in the near- and far-UV ranges. Because the results for both constructs were highly similar, only those for XR6-11 are described here in detail. Both constructs contain several aromatic residues (tryptophan and phenylalanine in the repeats, and tyrosine in the linker sequences). Therefore, we expected a significant CD signal in the near-UV range (350-250 nm), which is known to depend on aromatic amino acids and their asymmetric electronic environment. Indeed, we obtained a clear CD signal, which implies the presence of some tertiary structure elements in the Xin-repeat region (Fig. 1C). Upon denaturation of XR6-11 with GndHCl, the signal in the near-UV range completely disappeared, obviously because of the loss of tertiary structure (Fig. 1C).
In the far-UV range, we obtained CD spectra with three characteristic elements: a maximum of –2×103 deg cm2 dmol–1 at 192 nm, a minimum of –11.8×103 deg cm2 dmol–1 at 202 nm, and a minimum of –5×103 deg cm2 dmol–1 at 222 nm (Fig. 1D). This CD spectrum is different from that of random-coil model peptides. Therefore, we conclude that Xin repeats in solution are not in a denatured state but rather adopt a more extended structure, with only few folded portions. A small α-helical input could be responsible for the ellipticity shift at 222 nm towards lower values in the spectrum, but the α-helical content at this point does not exceed 10%.
Xin is targeted to F-actin upon transient expression in non-muscle cells
To learn more about the cellular function of Xin and the repeat motifs, we generated an EGFP-tagged construct of full-length Xin (Fig. 1A) and investigated its subcellular localization and dynamics in both fixed and live cells. To do so, PtK2 cells were transfected with Xin-EGFP. Xin clearly exhibited targeting to F-actin-containing structures such as cell-cell contacts (Fig. 2A,C), focal adhesions (Fig. 2A,D) and stress fibres (Fig. 2A,E) that were identified by CPITC-conjugated phalloidin (Fig. 2B-E, blue).
To determine whether Xin repeats are responsible for the association of Xin with stress fibres, a construct encoding the Xin repeats (XR1-16) was transfected into A7r5 smooth-muscle cells and PtK2 cells. Because the obtained results were comparable in both cell types, we confine our description to A7r5 cells. XR1-16 targeted to F-actin-containing structures (such as stress fibres) and focal adhesions (Fig. 3A,B). This localization pattern was virtually identical to that of the complete Xin molecule (Fig. 2). Subsequently, a series of truncations of the repeat region was examined. Both XR1-11 (data not shown) and XR1-6 (Fig. 3C,D) showed targeting to focal adhesions and stress fibres. Further truncations down to three Xin repeats (i.e. constructs XR1-5, XR1-4, XR1-3 and XR4-6) bound preferentially to stress fibres (Fig. 3E,F). Interestingly, we observed a significantly stronger tendency for focal-adhesion targeting in the longer Xin-repeat constructs. By contrast, only a small proportion of the two domain constructs XR3-4 and XR10-11 appeared to be very weakly associated with stress fibres in fixed cells (not shown), whereas the vast majority was diffusely distributed throughout the cytoplasm and the nucleus. Because a weak artificial association of EGFP with stress fibres upon fixation was described earlier (Schmitz and Bereiter-Hahn, 2001), the distribution of the two domain constructs was also investigated in live cells. These experiments revealed no detectable amounts of XR3-4 associated with F-actin-containing structures (Fig. 3G-H), whereas XR10-11 [which contains an inter-repeat sequence (GEVLAHGSPSREEGTD) that is weakly homologous to the Xin-repeat consensus sequence] showed minimal binding to adhesion plaques but not to actin filaments (Fig. 3I,J). The specificity of the association with stress fibres of all constructs containing more than two repeats was confirmed using live microscopy.
To investigate whether Xin repeats bind all actin filaments or only a subset, we transfected highly motile B16F1 cells with XR1-16, seeded them on laminin and studied the distribution of the recombinant Xin fragment using live-cell imaging microscopy. Consistent with the localization of Xin repeats in PtK2 and A7r5 cells, XR1-16 was present along stress fibres and associated with focal adhesions (Fig. 4). Interestingly, this construct mainly targeted contractile actin structures, not the cell periphery, where lamellipodia and filopodia were formed. The virtual exclusion from the lamellipodial actin meshwork was confirmed by fixation and counterstaining of a cell expressing XR1-16 with phalloidin (Fig. 4I,J). Of note, B16F1 cells expressing high levels of XR1-16 were more contracted and displayed less-prominent lamellipodia than untransfected cells (not shown). These observations suggest that Xin repeats stabilize stress fibres, but not dynamic F-actin in cellular protrusions such as lamellipodia and filopodia. These results, in particular the continuous stress-fibre association of Xin (Fig. 2E, arrows, Fig. 3A,C,E, arrows) prompted us to investigate biochemically the potential of Xin repeats to bind directly to F-actin.
Xin repeats bind to F-actin in vitro
To examine for direct binding of the Xin repeats to F-actin, we performed actin-filament co-sedimentation assays using bacterially expressed, purified XR1-16. This polypeptide clearly bound to and sedimented with skeletal-muscle actin in a dose-dependent manner (Fig. 5A,B). Binding was saturable at an actin-to-Xin stoichiometry of ∼4:1 (Fig. 5C). Subsequently, actin-binding experiments were repeated with constructs containing fewer repeats. In these experiments, three different constructs containing six Xin repeats (XR1-6, XR6-11 and XR11-16; Fig. 1A) also bound F-actin in a fashion similar to XR1-16 (Fig. 5D-F, Table 1), but saturation was reached at an actin-to-Xin stoichiometry of ∼2:1. Actin-binding assays using truncated variants of XR1-6 showed that all constructs (Fig. 1) containing at least three repeats bound actin (Fig. 5, Table 1). Interestingly, saturation with the construct XR1-3 was reached at an actin-to-Xin stoichiometry of ∼1:1. This clearly indicates that the repeat region of Xin contains multiple independent actin-binding sites. The efficiency of actin-binding activity related directly to the number of repeats within the recombinant constructs. By contrast, two truncated recombinant proteins containing only two Xin repeats separated by a short 19 amino acids inter-repeat region (XR3-4, Fig. 5J) or a long 55 amino acid inter-repeat region (XR10-11; not shown) exhibited no detectable actin binding under these conditions.
In a subsequent series of experiments, we investigated whether Xin repeats were capable of inducing the formation of F-actin structures that can be sedimented at low centrifugal force. Indeed, XR1-11 (Fig. 6A,B) and all other constructs (not shown) that bind F-actin also caused significant cross-linking of actin filaments. Again, the efficiency of cross-linking was directly related to the number of repeats in the recombinant polypeptides. The results of these experiments are summarized in Table 1.
This assay allows two distinct interpretations: the sedimentable F-actin aggregates can be the result either of a bundling activity or of the formation of cross-linked gels. To distinguish between these possibilities, mixtures of recombinant XR1-11 and F-actin were investigated by electron microscopy. Negatively stained samples of preparations containing the Xin-repeat construct essentially ruled out the formation of bundles. Instead, the addition of XR1-11 to actin yielded a cross-linked F-actin meshwork (Fig. 6). Subsequent, falling-ball viscometry experiments using mixtures of F-actin and XR1-11 showed a linear increase in viscosity with increasing amounts of XR1-11, confirming the formation of a meshwork (P.V., D.O.F. and P.F.M.v.d.V., unpublished).
Xin repeats inhibit binding of tropomyosin to actin filaments
To investigate whether Xin and tropomyosin might compete for the same binding site on the actin filament, we mixed actin with a fixed amount of tropomyosin and various quantities of XR1-11. These samples were investigated after low-speed sedimentation. In the absence of the Xin repeats, no actin or tropomyosin was found in the pellet. The addition of XR1-11 to a concentration of only 0.5 μM resulted in significant cross-linking of actin filaments together with associated tropomyosin (Fig. 7). The addition of more XR1-11 caused a dose-dependent increase of the amount of F-actin and a reduction of the amount of tropomyosin in the pellet (Fig. 7), indicating that binding of tropomyosin to F-actin is inhibited by the association of Xin to these filaments. Similar results were obtained when XR1-6 was used instead of XR1-11 (results not shown).
Stress fibres are stabilized upon binding of Xin
To determine the consequences of Xin-repeat binding to F-actin in more detail, we examined the depolymerization dynamics upon treatment with LatA. LatA causes alterations in the structure of G-actin (Morton et al., 2000) and thus inhibits actin polymerization both in vitro (Coue et al., 1987) and in vivo (Spector et al., 1983).
To do this, we transfected A7r5 cells with XR1-16-EGFP and treated them with LatA. After different times, cells were fixed and the localization of the Xin fragment, actin organization (visualized by phalloidin-CPITC staining) and the focal adhesion pattern (by vinculin labelling) were determined and compared with those of untransfected cells. In cells transfected with Xin, stress fibres and focal adhesions seemed significantly more pronounced than in untransfected cells (Fig. 8A-E). The addition of 0.5 μM LatA to the culture medium resulted in focal disturbances of stress fibres after a 5-minute incubation in most untransfected cells (Fig. 8F,G), and the effects of the treatment with LatA significantly increased upon longer incubation times. By contrast, in transfected cells, the actin cytoskeleton remained relatively well preserved (Fig. 8H-J,N-P,S-U). In addition, focal adhesions displayed a more elongated appearance than those in untransfected cells (Fig. 8P,U). (For more details, see the legend to Fig. 8.) In cells showing only low XR1-16 expression levels, the first signs of stress-fibre disassembly were observed after 15 minutes (not shown). Taken together, these results indicate that binding of Xin repeats to actin filaments protects them from depolymerization originated by the G-actin-sequestration activity of LatA, in a dose-dependent manner.
Similar experiments performed with A7r5 cells co-expressing XR1-16-EGFP and mRFP-actin using live two-colour fluorescence video microscopy confirmed the above findings (see Movie 1 in supplementary material). In control cells expressing actin and EGFP alone, LatA induced rapid disassembly of the actin cytoskeleton, whereas, in cells transfected with XR1-16, the prominent stress fibres in the body of the cells were only minimally affected. Both control and XR1-16-expressing cells showed rapid disassembly of lamellipodia, the actin filaments of which do not bind Xin repeats (see also Fig. 4). Thus, these filaments were not protected from the depolymerizing effect of LatA. Upon washout of LatA, both transfected and untransfected cells recovered from the effects of the drug.
Identification and characterization of a novel Xin-related protein
Databases were searched for sequence similarities to Xin repeats. This revealed genomic clones, cDNAs and expressed sequence tags in all vertebrates, including mammals (man, mouse, rat, bull), birds (chicken), amphibians (Xenopus) and fish (Tetraodon), but no sequences in invertebrates such as Drosophila, C. elegans, Ciona intestinalis, yeast or bacteria. Most sequences were available in mammals, in which we discovered two similar but not identical sequences. Although one of them obviously encoded the human homologue of murine Xin, the second encoded a related protein with highly homologous repeats but distinct non-repeat regions. We tentatively named this protein Xin-repeat protein 2 (XIRP2). Recently, the expression of a gene called CMYA3 that seems to encode porcine XIRP2, was reported to be restricted to striated muscles in Sus scrofa (Pan et al., 2003). Partial human XIRP2-encoding sequences were found in a cDNA (accession number AK096430), a bacterial-artificial-chromosome clone (accession number AC093684) and several expressed sequence tags from heart, tongue and skeletal muscle. The contig assembled from these sequences contains 28 Xin repeats in a region spanning approximately 1000 residues (accession number AJ626901; for a schematic organization and an alignment, see Fig. 9A,B). The consensus sequence of the repeats in human XIRP2 is GDVKXXXWLFETQPLD and thus identical to that described for chicken Xin protein (Wang et al., 1999).
We repeated several of the above experiments with the XIRP2 repeats to compare their characteristics with those of Xin. Actin-binding assays performed with the complete bacterially expressed XIRP2 repeat region (XIRP2 1-28) showed that these repeats also directly bind and cross link skeletal-muscle F-actin (Fig. 9C,D). Analogous to the experiments with XR1-16, the complete repeat region of XIRP2 was expressed in A7r5 (Fig. 9E,F) and PtK2 (not shown) cells. In both cell types, XIRP2 repeats were clearly targeted to F-actin-containing structures with characteristics highly similar to Xin.
Xin is a protein initially identified in a differential display screen performed to isolate genes that are expressed during atrioventricular valve and septum formation in the chicken heart (Wang et al., 1996a). It is expressed exclusively in striated muscle tissues, and treatment of chick embryos with antisense Xin oligonucleotides resulted in aberrant cardiac morphogenesis indicating an important role for Xin in this process (Wang et al., 1999). In adult tissues, the protein was localized in intercalated discs in the heart and myotendinous junctions of skeletal muscle (Sinn et al., 2002). A conspicuous feature of Xin is the presence of multiple copies of 16 amino acid tandem repeats that are found only in Xin and the closely related XIRP2, which is described in this study for the first time. In this report, we show that Xin repeats directly bind actin filaments in vitro and cross link these filaments into loose networks. Furthermore, we provide evidence that stress fibres and focal adhesions in cultured cells are stabilized upon binding of Xin-repeat-containing polypeptides. Thus, we have identified the Xin repeats as a novel actin-binding motif, and the Xin/XIRP2 family as a novel group of actin-binding proteins.
At present, the precise protein structure of Xin repeats is not known. Most secondary-structure prediction programs assume a strong tendency for the formation of α-helices. Our CD spectroscopy data revealed that purified Xin repeats are able to adopt tertiary structures in solution that disappear upon denaturation of the polypeptide with GndHCl. However, the CD spectrum of the native protein in the far-UV range does not indicate that large portions of the molecule exhibit α-helical or other secondary structures. This is reminiscent of tropomodulin, which is rather unstructured in solution in the absence of tropomyosin and only forms stable α-helices upon binding to this major ligand (Kostyukova et al., 2000). Similarly, nebulin repeats have a strong tendency to fold as unstable, transient helices in solution, whereas their structure is stabilized by binding to actin filaments (Jin and Wang, 1991; Chen et al., 1993; Pfuhl et al., 1994; Wang, 1996b). Future studies of Xin-repeat-actin complexes might help to elucidate the structure of Xin repeats upon binding to actin filaments.
Our actin-binding studies suggest that three Xin repeats are sufficient for binding and for the formation of an actin-filament meshwork. The phenomenon that multiple copies of tandem repeats are needed to mediate actin binding, whereas a single copy is not sufficient, is also known from calponin and the related C. elegans protein UNC-87 (Kranewitter et al., 2001). Studies with recombinant polypeptides derived from the latter protein showed that at least three or four tandem repeats (dependent on which part of the repeat region was investigated) are required to bind and bundle actin filaments significantly. Addition of further repeats to the polypeptides leads to a proportional increase of actin binding and bundling capacity (Kranewitter et al., 2001). In calponin, all three tandem repeats are required for targeting to stress fibres in cultured cells (Gimona and Mital, 1998). Our data indicate that, in the case of Xin, three repeats are also necessary and sufficient to bind actin filaments. Although the high homology of the repeats suggests equally conserved interaction sites and a 1:1 stoichiometry, three repeat and six repeat constructs bound F-actin in an actin-to-Xin stoichiometry of 1:1 and 2:1, respectively, implying that three repeats bind one actin molecule. An alternative explanation would be the existence of multiple Xin-binding sites on each actin subunit, as demonstrated for nebulin, in which three identical recombinant fragments bound to actin subunits at three different sites (Lukoyanova et al., 2002).
Interestingly, tropomyosin binding to actin was affected by saturating concentrations of Xin, indicating the occupation of overlapping sites on the actin filament. By contrast, UNC-87 was reported not to displace tropomyosin and was capable of binding the actin filament simultaneously in almost stoichiometric amounts (Kranewitter et al., 2001). We therefore suggest that Xin, like tropomyosin and nebulin, is an actin filament side-binding protein, and that Xin and tropomyosin interact at similar regions along the filament. The saturation binding data (Figs 5, 9) support this idea and the estimated Kd for F-actin in the range of ∼0.45 μM is within the range of binding constants of most actin-binding proteins (Pollard, 1999). One major difference between nebulin repeats and Xin repeats is the strict succession of repeats with linkers essentially without length variability in the case of nebulin, whereas both Xin and XIRP2 contain spacers between 19 and 55 residues without sequence conservation. This arrangement of actin-binding motifs might provide the repeat region of both Xin-repeat proteins with sufficient flexibility to explain their actin-filament cross-linking activity observed in electron micrographs, low-speed co-sedimentation assays (Fig. 6, Fig. 9D) and falling-ball viscometry experiments. For a complete understanding of the Xin-repeat/actin interaction, their complexes will have to be analysed at the ultrastructural level using special methods of image analysis (see Lukoyanova et al., 2002).
We could not detect any actin polymerization or cross-linking activity of Xin repeats in low-speed co-sedimentation assays using G-actin. Furthermore, no association of Xin repeats with G-actin was detected using chemical cross-linking assays. The possibility that Xin repeats can dimerize in solution was excluded by chemical cross-linking experiments (M.H., D.O.F. and P.F.M.v.d.V., unpublished), indicating that the F-actin cross-linking activity of the Xin repeats is not a result of their ability to polymerize.
The in vitro actin-binding data in the first place clearly confirmed and explained the results of the transfections of Xin-repeat-containing constructs into cultured cells. The observed targeting of recombinant proteins containing three and more repeats to stress fibres and focal adhesions (Figs 2, 3, 4, 8, 9) are obviously, at least in part, a direct result of the capacity of the repeats to bind F-actin. Live-cell imaging in different cell types, however, revealed that binding was confined to less-dynamic actin structures, whereas the constructs were excluded from highly motile lamellipodial regions (Fig. 4). Moreover, some cells overexpressing Xin repeats at high levels appeared to be more contracted and less motile than untransfected cells (data not shown). In this respect, it is interesting that multiple copies of the short actin-binding repeat from human calponin or C. elegans UNC-87 were also reported to effectively inhibit cell motility and to function as a mechanical stabilizer of the actin cytoskeleton (Lener et al., 2004).
Our experiments with LatA also indicate that the association of Xin repeats with stress fibres stabilizes these structures. On the one hand, the actin-filament bundles in transfected cells are not only more pronounced but also considerably more resistant to the depolymerizing effects of this compound, indicating a specific stabilization of contractile stress fibres upon binding of Xin. On the other hand, focal adhesions in transfected cells are more pronounced and larger (Fig. 8C-E). It has previously been reported that focal adhesions (and actin filaments directly attached to these structures) are generally more resistant to LatA treatment (Cai et al., 2000). Their highly elongated morphology in transfected cells after LatA treatment is reminiscent of the sliding adhesions at the trailing ends of human foreskin fibroblasts (Zamir et al., 2000). This might be due to the remaining stress fibres in Xin-transfected cells, which are still capable of exerting contractile force.
The transfection of Xin constructs of different lengths yielded another interesting observation: although the three-repeat constructs only associated with stress fibres, the six-repeat constructs bound equally strongly to stress fibres and focal adhesions, and the construct comprising the entire Xin-repeat region showed a more pronounced tendency for focal adhesion targeting (Fig. 3). This indicates that not all the repeats or inter-repeat regions have identical properties. Instead, some of these motifs might participate in mediating interactions with focal contact proteins. The presence of conserved proline-rich regions in some of the inter-repeat regions (PTRPQP, amino acids 143-148 between XR2 and XR3; PSPDLIPPGP, amino acids 331-340, between XR7 and XR8; PQPEAPPK, amino acids 581-588, between XR12 and XR13) are indicative of such interactions.
Previous immunolocalization studies (Wang et al., 1999; Sinn et al., 2002) have indicated that, in cultured skeletal muscle cells, Xin is localized near cell-cell contacts at the periphery of the myotubes and to stress-fibre-like structures that are considered to be precursors of myofibrils (Dlugosz et al., 1984). In the adult animal, Xin was exclusively detected in the intercalated discs of cardiac muscle cells and at myotendinous junctions in skeletal-muscle fibres. Both of these structures are attachment sites of the actin cytoskeleton and of the contractile apparatus to the sarcolemma. At these sites, Xin seems to be located in complexes containing at least β-catenin and cadherin (Wang et al., 1999; Sinn et al., 2002). This distinctive localization and the structure of the mammalian Xin molecule, with highly conserved proline-rich regions, suggest that the protein has a function in linking unidentified ligands that reside within these specialized membrane-attachment sites to the subsarcolemmal and/or myofibrillar actin filaments. The identification of further binding partners will be crucial to elucidate completely the interaction of Xin repeats with the actin cytoskeleton and to characterize the precise cellular functions of Xin and XIRP2.
Supplemental material available online at http://jcs.biologists.org/cgi/content/full/117/22/5257/DC1
This work was in part supported by a grant from the Deutsche Forschungsgemeinschaft (D.O.F.). We thank A. Guhlan and B. Mai for excellent technical assistance, M. Gimona (Department of Cell Biology and Oncology, Consorzio Mario Negri Sud, Santa Maria Imbaro, Italy) for donation of tropomyosin, R. Tsien (Howard Hughes Medical Institute, La Jolla, CA, USA) for donation of the mRFP-encoding cDNA, and M. Walter (Department of Physical Biochemistry, University of Potsdam, Germany) for help with the CD-spectroscopy experiments. We thank M. Gimona, R. Seckler (Department of Physical Biochemistry, University of Potsdam, Germany) and J. Wehland (Department of Cell Biology, German Research Centre for Biotechnology, Braunschweig, Germany) for helpful discussions.