Desmin intermediate filaments intimately surround myofibrils in vertebrate muscle forming a mesh-like filament network. Desmin attaches to sarcomeres through its high-affinity association with nebulin, a giant F-actin binding protein that co-extends along the length of actin thin filaments. Here, we further investigated the functional significance of the association of desmin and nebulin in cultured primary myocytes to address the hypothesis that this association is key in integrating myofibrils to the intermediate filament network. Surprisingly, we identified eight peptides along the length of desmin that are capable of binding to C-terminal modules 160–170 in nebulin. In this study, we identified a targeted mutation (K190A) in the desmin coil 1B region that results in its reduced binding with the nebulin C-terminal modules. Using immunofluorescence microscopy and quantitative analysis, we demonstrate that expression of the mutant desmin K190A in primary myocytes results in a significant reduction in assembled endogenous nebulin and desmin at the Z-disc. Non-uniform actin filaments were markedly prevalent in myocytes expressing GFP-tagged desmin K190A, suggesting that the near-crystalline organization of actin filaments in striated muscle depends on a stable interaction between desmin and nebulin. All together, these data are consistent with a model in which Z-disc-associated nebulin interacts with desmin through multiple sites to provide efficient stability to satisfy the dynamic contractile activity of myocytes.
Intermediate filaments (IFs) traditionally have important roles as force mechanosensors and mechanotranducers in cells subject to high levels of mechanical stress. IFs and their associated proteins comprise a dynamic cytoskeletal infrastructure that physically links and transmits signals from the cell membrane to cytoplasmic organelles and nuclear envelope (for reviews, see Goldman et al., 2008; Green et al., 2005; Hong et al., 2011; Toivola et al., 2005). The desmin IF network is responsible for laterally aligning myofibrils at the Z-discs and M-lines, forming a intermyofibrillar system that is thought to transmit synchronous force between adjacent myofibrils and relay it to neighboring myocytes (Kouloumenta et al., 2007; Meyer et al., 2010; Milner et al., 1996; Tonino et al., 2010) (for reviews, see Capetanaki, 2002; Green et al., 2005).
Desmin is a highly conserved fibrous type III IF protein that is expressed in all muscle types; it is structurally organized following a typical IF blueprint, except that it contains a pre-coiled domain preceding the central rod domain, a common feature of all type III IFs. Structurally, desmin has a central α-helical coil rod subdivided into coil 1A, linker 1 (L1), coil 1B, linker 12 (L12) and coil 2, which is flanked by non-helical head and tail domains (Herrmann and Aebi, 2004; Nicolet et al., 2010; Parry et al., 2007). Desmin filaments also have an abundant distribution at the intercalated discs and as a highly interwoven network of desmin filaments extending along myofibers (Georgatos et al., 1987; Kartenbeck et al., 1983; Meng et al., 1997; Nag and Huffaker, 1998; O'Neill et al., 2002; Price, 1984; Tokuyasu et al., 1985; Wang and Ramirez-Mitchell, 1983).
An early clue on how sarcomeres directly coordinate their contractile function with transmission of lateral force between neighboring myofibrils came when desmin was identified as a binding partner for nebulin modules M160–M183 in a yeast two-hybrid screen (Bang et al., 2002). Nebulin, a giant (~600–900 kDa) actin-binding protein, comprises ~3% of the total myofibrillar proteins in skeletal muscles, but considerably lower amounts are present in cardiac muscle (Bang et al., 2006; Fock and Hinssen, 2002; Kazmierski et al., 2003). The molecular properties of nebulin facilitate its important role in the regulation of thin filament length. First, its N-terminus extends towards the pointed end of the thin filaments in the middle of the sarcomere, and its C-terminus is anchored in the Z-disc; it interacts along its length with the major thin filament components, actin and tropomyosin (Kruger et al., 1991; Wright et al., 1993). Second, nebulin possesses a unique molecular layout formed from a series of ~35 amino acid tandem repeat modules (e.g. 185 modules in human nebulin), each of which is believed to interact with a single actin monomer (Jin and Wang, 1991; Labeit and Kolmerer, 1995; Pfuhl et al., 1996; Wang et al., 1996). Third, the molecular weight (i.e. lengths) of nebulin correlates with the type of skeletal muscle it is expressed in, as a result of alternative splicing (Kruger et al., 1991; Labeit et al., 1991). Based on these and other properties, nebulin was originally thought to function as a strict molecular ruler; that is, the length of nebulin specified thin filaments lengths (for reviews, see McElhinny et al., 2005; Trinick, 1994). However, new studies reveal that the main role of nebulin in thin filament length regulation probably involves a stabilization mechanism, whereby nebulin stabilizes the thin filaments even if it does not necessarily extend to the pointed ends of the filaments (Castillo et al., 2009; Pappas et al., 2010). Investigation of nebulin-null mice has revealed that the protein is clearly multi-functional: it is a key factor contributing to link cycling kinetics, isometric force production, Z-disc width specification and myofibrillar connectivity (Labeit et al., 2011; Ottenheijm and Granzier, 2010). Despite these findings, the mechanisms by which nebulin participates in myofibrillar connectivity through its interactions with desmin, and particularly, how the nebulin–desmin linkage is involved in muscle disease are incompletely understood.
Desminopathy is a human disease that affects cardiac and skeletal muscle and is often accompanied by respiratory insufficiency and neurological symptoms (van Spaendonck-Zwarts et al., 2010). Desminopathy is a distinct form of myofibrillar myopathy, a disease caused by missense or deletion mutations in desmin or αB-crystalline (Goldfarb et al., 1998; Wang et al., 2001). Although de novo sporadic and recessive mutations have been reported, desminopathy has a variable onset and is mostly inherited in an autosomal dominant pattern (Pinol-Ripoll et al., 2009). Patients with inherited desmin mutations present heterogeneous clinical symptoms with debilitating muscle weakness, with cardiomyopathy being the most common; muscle biopsies are characterized by a preponderance of cytoplasmic aggregates containing misfolded proteins (Arbustini et al., 2006; Maloyan and Robbins, 2010). Although these aggregates are mainly composed of desmin, many sarcomere proteins such as actin, titin, myosin and nebulin have also been detected (Schroder et al., 2007). Interestingly, although mutations have been identified in all domains of desmin, no obvious correlation between the localization of an individual mutation within desmin and its effect on protein assembly has been recognized (Bar et al., 2005; Hong et al., 2011; Taylor et al., 2007; Vernengo et al., 2010).
Several models have been proposed to explain the molecular mechanisms resulting in desminopathy. One model proposes that mutations in desmin directly interfere with the assembly dynamics of the desmin network and simultaneously in the formation of misfolded desmin that is segregated into distinctive amorphous cytoplasmic aggregates (Goldfarb et al., 2004). A second model, based on the analysis of disease-associated desmin tail mutants, suggests that the mutations severely impair the ability of the desmin network to respond to applied strain, supporting the proposal that desmin mutants do not share a common disease pathway (Bar et al., 2011; Schopferer et al., 2009). A third model predicts that desminopathy arises from alterations in the association between desmin and its interacting partners (IF-associated proteins), such as those observed between desmin and nebulin, suggesting that connections to the contractile apparatus underlie some of the symptoms related to progressive muscle weakness and cardiac dysfunction (Bar et al., 2005; Conover et al., 2009).
This study aims to define the functional significance of the association of desmin with the C-terminal end of nebulin at the Z-discs. We provide the first biochemical evidence that several regions scattered throughout desmin contribute to the interaction between these proteins. Additionally, this work corroborates the importance of the coil 1B region of desmin for the association of desmin and nebulin and gives further evidence that this linkage is essential for thin filament architecture. We report the identification of a targeted desmin mutation within the coil 1B region (K190A) that diminishes the interaction of desmin with nebulin in vitro. When desmin K190A is expressed in myocytes, it significantly reduces the ability of endogenous desmin and nebulin to assemble at the Z-discs. We also show that expression of this mutant desmin results in markedly disorganized actin thin filaments. Based on these results, we speculate that desmin filaments contact nebulin through its numerous domains to ensure that cells optimally regulate force transmission and signaling between neighboring myocytes according to the physiological needs of different muscle types. Furthermore, our findings are consistent with a model for a key role of desmin–nebulin linkage in establishing actin organization by facilitating regularly spaced, well-aligned Z-discs.
Peptide blot overlay identifies multiple binding sites within desmin for nebulin
Based on previous findings that nebulin modules 160–170 interact with endogenous desmin, and on the observation that several desmin domains contribute to its Z-disc assembly (Conover et al., 2009), we sought to define the binding sites of desmin for the C-terminal region of nebulin (Figs 1, 2). To identify the regions within desmin that interact with nebulin, the complete sequence of mouse desmin was displayed on membranes as spots of 13-residue peptides, overlapping by 5 residues. The membrane was incubated with GST-tagged nebulin M160–M170 and those peptides binding nebulin were revealed using anti-GST–HRP antibodies. Unexpectedly, eight nebulin-binding desmin peptides (P1–P8) were identified; these were not restricted to a particular region, but rather were found along the length of the molecule (Fig. 2). Specifically, these desmin peptides included: a 29mer (P1, SLGSPLSSPVFPRAGFGTK) and a 21mer (P2, SRTSGGAGGLGSLRSSRLGTT) in the head domain; a 13mer (P3, AEVNRLKGREPTR) in the coil 1A-L1 region; two 13mers (P4, ELRRQVEVLTNQR and P5, DLQRLKAKLQEEI) in the coil 1B region; two 21mers (P6, SKVSDLTQAANKNNDALRQAK and P7, LKGTNDSLMRQMRELEDRFAS) in the coil 2 region; and a 13mer (P8, TKKTVMIKTIETR) in the tail domain.
To ensure that our results were specific to nebulin binding, and not to the GST moiety we also probed the membrane with recombinant GST protein alone. No binding between desmin and GST was detected (supplementary material Fig. S1A). Identical concentrations of peptides (1 μM/cm) in each spot contribute to the reliability of this method (Bocquet et al., 2009; Weiser et al., 2005). Other less-intense incomplete spots were detected, which probably represent non-specific binding. Furthermore, we confirmed the specificity of our findings using a pull-down assay (Fig. 2B). In this assay, specific binding was detected between His-nebulin M160–M164 and desmin head, coil 1B, coil 2B and tail regions, whereas no significant binding of nebulin M160–M164 or full-length desmin to GST protein alone was detected (supplementary material Fig. S1B). All together, our results indicate that C-terminal nebulin contains multiple binding sites within desmin.
Identification of desmin Lys190, which when replaced with alanine, diminishes the interaction with nebulin M160–M170
Based on previous findings that GFP-tagged desmin coil 1B localizes primarily at the Z-discs in skeletal and cardiac myocytes (Conover et al., 2009), and that this region was identified in a yeast-two hybrid assay to be important for its interaction with nebulin (Bang et al., 2002), we focused our studies on this region of desmin. To further define the critical residues within desmin coil 1B that are responsible for its interaction with nebulin, and to determine whether it was possible to disrupt the binding of the two molecules in vitro, we introduced single alanine substitution mutations within desmin fragments P4 and P5 in a second custom SPOTs membrane. The substitutions included: D164A, E166A, N170A and R172A in P4 and K185A, Q187A, K190A and E195A in P5. Using this approach (Fig. 2A), we identified that substituting an alanine for a lysine at residue 190 within desmin resulted in a significantly reduced interaction with GST–nebulin M160–M170 (Fig. 3A). Densitometry analysis revealed that the signal was reduced by ~90% compared with the signal obtained with the non-mutated P5 peptide. Other less-prominent perturbations in binding were detected with other mutants; however, they were less convincing and they were not studied further. To address whether the mouse Lys190 residue within the desmin P5 is conserved in different mammalian species such as human, bovine, rat, rabbit and other more divergent species such as frog, fish and chick, we performed an online basic local alignment analysis (BLAST). Our analysis revealed that Lys190 is 100% conserved, and the desmin P5 peptide is ⩾90% conserved (identities) in all of the analyzed species (Fig. 3B). In conclusion, desmin Lys190 is highly conserved across divergent species and appears to be a crucial residue in modulating the interaction of desmin with nebulin.
Desmin K190A mutation has reduced binding capacity to both nebulin M160–M170 and nebulin M160–M164
To quantify the relative binding ability of WT and K190A desmin to C-terminal nebulin, we performed solid-phase binding assays (ELISAs). The relative binding profiles for nebulin M160–M170 or nebulin M160–M164 to WT or K190A desmin coil1b was compared. Our data show that all of the interactions tested were saturable. Using this approach, the dissociation constant (Kd) and binding capacity (Bmax) for WT desmin coil 1B with nebulin M160–M170 (Kd≈35 nM and Bmax≈3.5) and with nebulin M160–M164 (Kd≈14 nM and Bmax≈3.3) were obtained. In contrast to these findings, a lowered binding affinity for desmin coil 1B K190A with nebulin M160–M170 (Kd≈88 nM and Bmax≈2.2) and with nebulin M160–M164 (Kd ≈ 34 nM and Bmax≈1.9) was obtained. Our results show that desmin coil 1B K190A displays a ~1.6-times lower binding capacity for nebulin M160–M170, and ~1.7-times reduced binding capacity for nebulin M160–M164. Similarly, a ~2.5-times and ~2.6-times lowered binding affinity was obtained for each of the nebulin fragments, respectively, with desmin coil 1B K190A, indicating that the major binding site for this mutant likely lies within nebulin M160–M164 (Fig. 4B). Furthermore, representation of these data as Scatchard plots indicates a non-cooperative binding for coil 1B versus coil 1B K190A to each nebulin C-terminal fragment (supplementary material Fig. S2), and probably represents differences in stoichiometry. These data suggest that the number of WT desmin molecules binding to C-terminal nebulin is greater than the number of mutant desmin (K190A) molecules binding to nebulin.
The addition of the K190A mutation reduces desmin Z-disc targeting
To assess the assembly competency of desmin containing the K190A mutation and to determine whether the assembly of the mutant into the sarcomere would affect Z-disc structure, we expressed the desmin mutant in cardiomyocytes and stained for the Z-disc marker, α-actinin. Using this expression approach, we attempted to mimic the heterozygous dominant genotype present in the majority of desminopathy patients with genomic mutations in desmin. Cells expressing GFP alone, GFP–WT-desmin or GFP-tagged desmin coil 1B showed relatively unaffected distribution of α-actinin (Fig. 5). As expected, GFP-tagged desmin coil 1B localized at the Z-disc (Conover et al., 2009). Furthermore, our quantification of assembly revealed that the addition of the K190A mutation resulted in a significant decrease in Z-disc assembly of desmin containing this mutation with a concomitant increase in the number of desmin-containing aggregates in the transfected myocytes (Fig. 5A). Specifically, three distribution patterns were observed: (1) primarily ‘Z-disc’ (i.e. colocalized in striations with α-actinin), (2) primarily ‘aggregates’, and (3) ‘mixed’ assembly with both prominent Z-disc and aggregate distribution (Fig. 5).
In a previous study, we observed that aggregates containing desmin and nebulin were abundant and frequently found in myocytes expressing GFP–desmin containing a desminopathy-linked E245D mutation (Conover et al., 2009). The presence of cytoplamic aggregates is a key feature in human desminopathy and recently it was reported that the aggregates contain both mutant and WT desmin (Clemen et al., 2009). Similarly, the aggregates formed as a result of expressing desmin K190A, although not as prevalent, resembled the aggregates observed in desminopathy muscle biopsies, which contain desmin and other sarcomere-associated proteins (Schroder et al., 2007). Intriguingly, the different-sized and -shaped aggregates were frequently observed to localize at or near Z-discs (Fig. 5Bd,j). Together, these data revealed that formation of desmin-containing aggregates does not dramatically influence Z-disc (α-actinin) structure.
Expression of desmin K190A perturbs the endogenous Z-disc desmin filament network
Our observation that the presence of the K190A mutation significantly reduced the binding affinity of desmin coil 1B to C-terminal nebulin in vitro (Fig. 4), raised the question of whether the expression of this mutant would affect the distribution of the endogenous desmin network in cardiomyocytes (Fig. 6). As expected, we observed a reduction (i.e. competition) in the distribution of endogenous desmin at the Z-discs in cells expressing GFP–desmin. Notably, this effect was exacerbated by the expression of full-length GFP–desmin-K190A. Similarly, we found that expression of GFP-tagged desmin coil 1B K190A also resulted in a significant decrease in the Z-disc distribution of endogenous desmin.
The fact that the mutant desmin was more efficient at displacing endogenous desmin was surprising. One possible explanation for this observation is that the endogenous desmin filaments that incorporate GFP–desmin-K190A might be less efficient at localizing to the Z-discs than the endogenous desmin filaments that incorporate WT GFP–desmin because some of the filaments now have a defective major nebulin-binding site. In summary, our results suggest that nebulin facilitates the recruitment of endogenous desmin to the Z-discs.
GFP-tagged desmin coil 1B K190A assembled less efficiently to the Z-disc compared with the non-mutated form (Fig. 5A). Interestingly, as we reported previously, GFP-tagged desmin coil 1B assembles about twice as efficiently at the Z-disc as the WT GFP–desmin, probably because this domain contains a high-affinity nebulin-binding site (and potentially is missing binding sites for other proteins present in the full-length desmin molecule) (Conover et al., 2009). A preponderance of aggregates or dot-like structures containing both endogenous desmin and GFP–desmin-K190A were observed in cells expressing GFP-tagged desmin K190A or coil 1B K190A (Fig. 5Bd,j). The presence of endogenous desmin in the aggregates suggests that endogenous desmin filaments interact with mutant desmin filaments; these interactions also probably contribute to the reduction of endogenous desmin at the Z-disc (i.e. endogenous desmin is probably sequestered from the Z-disc in cells expressing GFP–desmin-K190A).
Perturbation of nebulin at the Z-discs upon expression of desmin K190A in skeletal myocytes
Because nebulin is much more highly expressed in skeletal muscle than in cardiac muscle, we addressed the question of whether expression of GFP-tagged coil 1B K190A would perturb the distribution of Z-disc nebulin in primary cultures of chick skeletal myocytes. Indeed, our results show that expression of this mutant desmin dramatically altered the ability of C-terminal nebulin to target to the Z-discs. We found that only ~3% of skeletal myocytes expressing the mutant contained assembled C-terminal (Z-disc) nebulin, compared with ~73% of cells expressing WT coil 1B or ~87% of cells expressing GFP alone (Fig. 7). As reported previously, we were unable to detect significant Z-disc assembly of full-length GFP–desmin in skeletal myocytes (Conover et al., 2009): <1% of the cells exhibited Z-disc-associated desmin assembly. Instead, GFP–desmin exhibited a dot-like, aggregated distribution pattern in these cells. Similarly, when we evaluated the effect of GFP–desmin expression on the distribution of endogenous nebulin, no Z-disc nebulin assembly was observed. The observed non-striated staining pattern of nebulin filaments suggests a disorganization of the filaments, and a failure to be enriched at the Z-discs. Our analysis revealed that only ~3% of skeletal myocytes expressing mutant coil 1B desmin K190A incorporated into the endogenous Z-disc associated desmin network as compared to ~67% of cells expressing coil 1B. These results indicate that expression of desmin K190A prevents endogenous desmin localization at the Z-disc with a significant impact on nebulin Z-disc assembly. Together, these results provide additional evidence in favor of a model in which the extra-sarcomeric desmin filament network anchors C-terminal nebulin to the Z-discs.
Perturbation of nebulin at the Z-discs upon expression of desmin K190A in cardiomyocytes
Cardiac nebulin has been characterized in primitive lamprey and hagfish hearts, and several studies have estimated that it is expressed at low levels in the vertebrate heart ranging from 0.4–3% of that found in skeletal muscle (Bang et al., 2006; Fock and Hinssen, 2002; Kazmierski et al., 2003). We next investigated the effect of desmin K190A on the distribution of endogenous nebulin in cardiomyocytes. Using an antibody against nebulin modules M176–M181 our data show that expression of GFP-tagged desmin coil 1B K190A leads to a noticeable reduction in localization of C-terminal nebulin at the Z-discs (Fig. 8). Specifically, we measured a ~36% reduction in C-terminal nebulin assembly in cardiomyocytes expressing GFP-tagged coil 1B K190A compared with GFP–coil-1B, whereas a ~28% reduction in nebulin Z-disc assembly occurred in cells expressing GFP–coil-1B compared with GFP alone. In cells expressing WT GFP–desmin, we detected a ~53% reduction in nebulin Z-disc assembly as compared with that in cells expressing GFP alone. An ~82% reduction in nebulin Z-disc assembly was observed when GFP–desmin-K190A was compared with GFP alone, whereas a ~62% reduction was observed when compared with GFP–desmin (Fig. 8A). GFP-tagged desmin coil 1B assembled in a mature striated pattern that localized in close proximity to Z-disc-associated nebulin (Fig. 8Bd–f). By contrast, GFP–desmin-K190A incorporated into desmin filaments, with little accumulation at the Z discs, whereas nebulin staining appeared less intense, diffuse and mostly non-striated in the identical myocytes (Fig. 8Bg–i).
Non-uniform actin thin filaments are prominent in cardiomyocytes expressing desmin K190A
Studies analyzing nebulin-knockout mouse models, isolated myocytes with nebulin levels reduced by siRNA treatment, and muscles from nebulin-null mice with nemaline rod accumulations (caused by mutations in nebulin) highlight the importance of nebulin in maintaining thin filament architecture (Bang et al., 2006; Ottenheijm et al., 2010; Pappas et al., 2008; Witt et al., 2006). Shorter thin filament lengths are thought to be one causative factor for the muscle weakness observed in the nebulin-knockout mice because they are predicted to reduce the amount of actin–myosin overlap, thereby decreasing isometric force and maximal active tension (Bang et al., 2009; Bang et al., 2006; Chandra et al., 2009; Gokhin et al., 2009; Witt et al., 2006). To address the effects of the K190A desmin mutation on actin filament organization, and thereby assess the contribution of reduced assembly of nebulin on Z-disc spacing and myofibrillar connectivity, we analyzed actin filament distribution in cardiomyocytes that expressed normal and mutated desmin. Our analysis shows a lack of regular gaps in phalloidin staining (F-actin) in the center of the sarcomeres (H-zones) in cells expressing GFP–desmin-K190A, in comparison to when WT GFP–desmin is expressed (Fig. 9e,n, compare arrowheads with arrows). Additionally, we also observed a preponderance of narrower actin filament bundles in cells expressing GFP–desmin-K190A. The majority of mutant desmin assembled in dot-like aggregates and lacked visible striated Z-disc staining (Fig. 9i–l). In these cells, F-actin appeared disorganized; most filament bundles lacked the predicted spacing and appeared non-uniform. These observations suggest that the absence of desmin filaments surrounding the Z-discs renders nebulin unable to stabilize thin filaments (Fig. 9b,f,j). Analysis of pixel intensity profiles of adjacent sarcomeres from these cells revealed noticeably shallower gaps in phalloidin staining at the H-zones compared with the staining in cells expressing GFP alone or WT GFP–desmin (Fig. 10).
Strikingly, in cells displaying GFP-tagged coil 1B K190A that assembled in striations at the Z-discs, the distribution of actin filaments appeared unperturbed and was similar to that observed in cells expressing GFP alone or WT GFP–coil-1B (Fig. 9b,n,r). In these cells, phalloidin staining with visible H-zone gaps and brightly stained Z-discs were observed (Fig. 9q–t). However, in ~25% of cells where GFP-tagged coil 1B K190A failed to assemble at the Z-discs, but instead assembled in dot-like aggregates (Conover et al., 2009) (Fig. 9q,m), pixel intensity profiles revealed the presence of some continuous F-actin staining at the M-lines, and slightly misaligned Z-discs, which is similar to that observed in cells expressing GFP–desmin-K190A (Fig. 10, and data not shown). We conclude that a mesh-like integrated Z-disc desmin network is crucial for organizing thin filaments in cardiomyocytes, and that its association with nebulin might have an important role in anchoring adjacent thin filaments to the Z-discs.
Cytoskeletal filament systems integrate intracellular and extracellular signals to promote the delivery of force during muscle contraction. In this study, we further investigated the significance of the association of the desmin IF network with sarcomeres. We focused our investigation on the association of desmin with nebulin, an actin-binding thin-filament-associated protein. Using biochemical approaches, we identified (1) multiple desmin binding sites throughout the molecule that interact with the C-terminus of nebulin; (2) a diminished interaction between desmin and nebulin when introducing a targeted mutation K190A in desmin in vitro; and (3) a significant perturbation in the assembly or organization of endogenous desmin, nebulin and filamentous actin in cultured cardiac and skeletal myocytes expressing K190A mutant desmin.
Using SPOTs blot overlay analyses we showed that desmin associates with nebulin M160–M170 by at least six peptides located within in its rod domain (three regions), head (two regions) and tail domains (one region). These results are consistent with the original yeast two-hybrid study that retrieved a majority of full-length desmin clones that bound with C-terminal nebulin modules M160–M183 (as bait) (Bang et al., 2002), and support a model in which desmin forms a rope-like network surrounding Z-discs with multiple surfaces available for nebulin to bind, thereby giving flexibility and extensibility to both filament systems.
Multiple interfaces involved in protein–protein interactions are not unique to nebulin and desmin. For example, talin a large adaptor protein binds to vinculin via ~11 binding sites and remodels the actin cytoskeleton by stabilizing the initial linkage between integrins and F-actin in focal adhesions (Gingras et al., 2005). In muscle cells, titin is known to associate with α-actinin through multiple binding sites, to actin through two binding sites and to tropomyosin through two binding sites, which offer distinct molecular functions (Gregorio et al., 1998; Linke et al., 2002; Raynaud et al., 2004; Sorimachi et al., 1997; Young et al., 1998). Furthermore, computational modeling based on electron micrographs predicts that a nebulin fragment spanning modules M170–173 associates as an extended structure with F-actin using three binding sites on each actin monomer (Lukoyanova et al., 2002).
Nebulin is the largest member of an evolutionary conserved tandem-repeat protein family in vertebrates (Bjorklund et al., 2010). Members of this family include nebulin, N-RAP, LASP-1, LASP-2 and nebulette (Kontrogianni-Konstantopoulos et al., 2009; Labeit et al., 2011; Pappas et al., 2011). These proteins share a similar domain structure, undergo alternative splicing, and are expressed in diverse tissues and cell types. Their common structure contains a unique N-terminal LIM domain, a central nebulin-repeat region and a C-terminal SH3 domain. The conserved desmin peptides that we identified to interact with nebulin might potentially facilitate interactions of other type III IFs with nebulin or other nebulin family members in muscle and non-muscle cells. Interestingly, nebulin modules M160–M164, which contain high-affinity binding sites for desmin, exhibit ~64% similar residues with the C-terminal modules of nebulette and ~55% similar residues with the N-terminal modules of the nebulin-related-anchoring protein (NRAP). Future studies will be needed to determine whether nebulette or NRAP associate with desmin.
Questions remain about whether nebulin family members aid desmin attachment to the Z-discs in cardiac and skeletal muscle, although emerging evidence supports nebulin-specific roles pertaining to Z-disc-width specification and thin filament length regulation. We cannot rule out completely the possibility that in cardiomyocytes, our anti-C-terminal nebulin antibodies could also partially detect nebulette. However, the fact that we found significantly reduced binding of mutant desmin K190A to C-terminal nebulin fragments in vitro, reduced assembly of C-terminal nebulin (using antibodies against modules M160–M164) at the Z-discs in skeletal myocytes, and diminished assembly of C-terminal nebulin (using antibodies against modules M178–181) in cardiomyocytes expressing this desmin K190A indicates that even a moderate disruption of the link with endogenous desmin has detrimental consequences for Z-disc nebulin or nebulette assembly and thereby their downstream functions.
Studies in cardiomyocytes have consistently localized C-terminal nebulin near the Z-discs; however, studies mapping nebulin epitopes in the interior of the Z-disc are missing (Fock and Hinssen, 2002; Kazmierski et al., 2003; McElhinny et al., 2005). The exact conformation of nebulin filaments in skeletal myocytes within the interior of the Z-disc also remains unclear. Studies suggest that nebulin filaments are arranged in a linear conformation along the actin filaments partially extending into the Z-disc (Millevoi et al., 1998; Wright et al., 1993); alternatively, another study proposes that nebulin crossbridges thin filaments in the Z-disc from opposite filaments (Pappas et al., 2008). Because desmin and CapZ, a protein that binds and regulates the barbed end of the actin filament, both have high-affinity binding sites within nebulin modules M160–M164, it is predicted that these modules are localized at the margin of the Z-disc (Caldwell et al., 1989; Conover et al., 2009; Pappas et al., 2008). We propose that desmin filaments serve as flexible anchors for nebulin filaments to orient thin filaments within the Z-disc. Thus, when the anchor is weakened (i.e. unstable IF network) this results in detachment of nebulin modules M160–M164 from the barbed ends of the thin filaments and consequent thin filament disarray. Future studies will define the precise conformation desmin filaments adopt to bind nebulin modules M160–M164.
Although one nebulin-knockout mouse model reported normal desmin distribution at the Z-discs by immunofluorescence labeling (Bang et al., 2006), a recent analysis of a different knockout model, using immunoelectron microscopy detected low staining of desmin in the intermyofibrillar spaces around Z-discs in skeletal muscle (Tonino et al., 2010). These findings support the idea that nebulin might be stabilizing desmin filaments surrounding the Z-discs. Our results show significantly decreased Z-disc localization of endogenous desmin when we expressed a mutant desmin with a diminished binding capacity to nebulin. Thus both studies suggest that a regulated desmin and nebulin association is required to preserve interfibrillar connectivity at the Z-discs. Our studies suggest that mixed desmin filaments containing mutant desmin K190A, might be inherently unable to target to the Z-disc as efficiently, therefore decreasing the number of binding sites available to firmly attach nebulin to the Z-discs. Future studies will discern the contributions of the nebulin–desmin complex in relation to human muscle disorders (e.g. desminopathy and nemaline myopathy). In this regard, the sarcomere ultrastructure in desmin-null muscles is not dramatically affected (Li et al., 1996; Milner et al., 1996); however, studies on gain- or loss-of-function missense desmin mutations might be directly relevant for deciphering the molecular basis of desminopathy.
Elongated intermediate filaments form from a rapid lateral alignment of tetramer complexes by end-to-end annealing of unit length filaments, in stark contrast to the assembly mechanism used by microtubules and microfilaments that polymerize and depolymerize by addition and loss of subunits at their ends (Coleman and Lazarides, 1992; Herrmann and Aebi, 2004; Kirmse et al., 2007). Furthermore, observations from living cells suggest that IFs incorporate and exchange subunits from the soluble pool along their length (Colakoglu and Brown, 2009). Because we saw an increased Z-disc localization of endogenous desmin in cells expressing GFP-tagged desmin coil 1B, we predict that GFP–coil 1B (which assembles at the Z-disc better than WT and other fragments tested) co-assembles with endogenous desmin, recruiting it to the Z-disc through an intercalated exchange mechanism.
Our previous study demonstrated that expression of a desminopathy-associated mutant desmin E245D in myocytes leads to uniform variations in actin thin filament lengths (Conover et al., 2009). In this study, we also observed alterations in actin filament architecture, albeit severe disorganization (with or without alterations in actin filament length) in cardiomyocytes expressing mutant desmin K190A. A key difference between our former and present study is that mutant desmin E245D bound to nebulin fragments with higher binding affinity, whereas the mutant desmin K190A bound nebulin with lowered binding affinity compared with unmutated desmin; however, expression of either mutant results in striking perturbations in actin filament architecture. Thus, both studies support the notion that the proper assembly of desmin IFs at the Z-discs through its interaction with nebulin influence actin thin filament architecture in myocytes. In summary, we propose a model that predicts that proper stoichiometry and positioning of Z-disc nebulin and desmin are crucial factors required for stability and proper spacing of adjacent thin filaments within the Z-discs, required to adequately transmit lateral force from Z-disc to Z-disc.
Materials and Methods
Desmin peptide array
A custom SPOTs cellulose membrane array containing 58 consecutive mouse desmin peptides (13mers with 5 aa overlap) was from Sigma Genosys. The membrane was blocked with 5% milk in binding buffer (20 mM HEPES, pH 7.4, 80 mM KCl, 2 mM MgCl2, 0.05% Tween) for 1 hour and probed with 0.5 μg/ml recombinant GST-nebulin M160–M170 for 3 hours. After washes, the membrane was incubated with 0.2 μg/ml HRP-conjugated anti-GST antibodies (GE Healthcare) for 1 hour. Desmin peptides that bound to nebulin M160–M170 were detected upon addition of SuperSignal chemiluminescent substrate (Thermo Scientific).
To identify specific desmin residues crucial for the interaction with nebulin, a second SPOTs membrane was designed that contained the desmin coil 1B mouse peptides P4, LRRAVEVLTNQRA (desmin residues 161–173) and P5, DLQRLKAKLQEEI (desmin residues 185–191). Single missense mutations were introduced (D164A, E166A, N170A, R172A) in P4 and (K185A, Q187A, K190A, E195A) in P5 and tested for binding to GST–nebulin M160–M170 using the protocol as described above. Densitometry analysis using ImageJ was used to quantify the relative intensity of the signal generated by the interaction of GST–nebulin M160–M170 with desmin non-mutated P4 or P4 K190A.
Cloning and site-directed mutagenesis
Primers used in this study are listed in supplementary material Table S1. To generate N-terminal GST-tagged desmin head, coil 1B, coil 2B and tail recombinant proteins, mouse heart cDNA (XM_130232) were amplified by RTPCR and fragments were cloned into pGEX4T-1 (GE Healthcare) using BamHI and EcoRI restriction enzymes. Details on the strategies used to generate the WT versions of these plasmids and N-terminal GST-tagged nebulin M160–M170 are published (Conover et al., 2009). The K190A mutation was introduced by PCR-based site-directed mutagenesis (QuikChange kit, Stratagene) into pGST-coil-1B and pGST-desmin constructs. Briefly, PCR products were digested with the DpnI restriction enzyme, and transformed into XL1-blue super competent E. coli (Stratagene). All plasmids were verified by DNA sequencing.
GST-pull down assays
Recombinant GST-tagged desmin and His-tagged nebulin M160–M164 proteins were purified from E. coli BL21 using glutathione beads (GE Healthcare) and the Ni-NTA system (Qiagen), respectively (Conover et al., 2009). The desmin fragments (head, coil 1B, coil 2B, tail) or GST alone attached to the glutathione beads were washed in binding buffer (see above) containing 0.1% Triton and then incubated with 0.3 μg/ml His–nebulin M160–M164 for 1 hour. After incubation, the beads were washed with binding buffer plus 0.1% Triton X-100 and the samples were analyzed on a 12% SDS-PAGE gel. Western blots on identical samples were performed as described previously (Conover et al., 2009) using affinity-purified anti-nebulin M160–M164 antibodies (5 μg/ml).
Sequence alignments were performed with ClustalW using the Gonnet matrix, open gap penalty 10 and extend gap penalty 0.1. Graphical output of particular sequences was performed using MacVector 11.0.4 software.
These assays were performed as described previously (Conover et al., 2009). Dissociation constants were determined from nonlinear regression curves fitted using a one-site binding equation, Y=BmaxX/(Kd+X), and Scatchard of saturation data plots were computed using Prism5 software. Values shown are the average of triplicate wells ± s.d. Each assay was performed at least three times.
Primary myocyte culture and transfection
Primary cultures of skeletal myocytes were isolated from pectoralis muscle of day 11 chick embryos (Ojima et al., 1999). Myoblasts were plated at ~4 × 105 cells per 35 mm culture dish and maintained for 24 hours in growth medium after isolation. Rat cardiomyocytes were isolated from E18 hearts, plated at a density of ~5 × 105 cells per 35 mm culture dish and maintained in DMEM supplemented with 10% FBS plus 1% penicillin and streptomycin, as described (Gustafson et al., 1987). Myocytes were transfected with the Effectene (Qiagen) following the manufacturer's instructions. 18 hours after transfection, the skeletal cultures were incubated in differentiation medium for 3 days and the cardiac cultures were incubated for an additional 6 days before fixation in 3% paraformaldehyde in relaxing buffer (150 mM KCl, 5 mM MgCl2, 10 mM MOPS, pH 7.4, 1mM EGTA and 4 mM ATP) for 15 minutes. To avoid potential non-specific effects, only cells that had moderate to low levels of GFP expression were analyzed.
Fixed cells were permeabilized with 0.2% Triton X-100 in PBS and blocked with 2% BSA and 1% normal donkey serum in PBS. Myocytes were stained with primary and fluorescently tagged secondary antibodies, as described (Conover et al., 2009). Stained samples were analyzed using a deconvolution Deltavision RT system (Applied Precision) with an inverted microscope (IX70 Olympus), a 100 × NA 1.3 objective and a charge-coupled device camera (CoolSNAP HQ: Photometrics). Images were deconvolved using SoftWRx 3.5.1 software and processed using Adobe Photoshop. ImageJ was used to generate the intensity profiles.
We thank Syerra Lea for help with the immunofluorescence analysis, Joseph Bahl and Verena Koenning for technical assistance with myocyte culture, and David Gillis for protein purification.
This work was funded by the National Institutes of Health [grant numbers HL57461 Minority Supplement to G.M.C., HL57461 and HL460831 to C.C.G.]; and the American Heart Association [grant numbers 2110057 to G.M.C., 0655637 to C.C.G.]. Deposited in PMC for release after 12 months.