Research Article

The Kinase Domain of Titin Controls Muscle Gene Expression and Protein Turnover

See allHide authors and affiliations

Science  10 Jun 2005:
Vol. 308, Issue 5728, pp. 1599-1603
DOI: 10.1126/science.1110463

Abstract

The giant sarcomeric protein titin contains a protein kinase domain (TK) ideally positioned to sense mechanical load. We identified a signaling complex where TK interacts with the zinc-finger protein nbr1 through a mechanically inducible conformation. Nbr1 targets the ubiquitin-associated p62/SQSTM1 to sarcomeres, and p62 in turn interacts with MuRF2, a muscle-specific RING-B-box E3 ligase and ligand of the transactivation domain of the serum response transcription factor (SRF). Nuclear translocation of MuRF2 was induced by mechanical inactivity and caused reduction of nuclear SRF and repression of transcription. A human mutation in the titin protein kinase domain causes hereditary muscle disease by disrupting this pathway.

During muscle differentiation, a specific program of gene expression leads to the translation of myofibrillar proteins and their assembly into contractile units, the sarcomeres, which are constantly remodeled to adapt to changes in mechanical load. The giant protein titin (also known as connectin) acts as a molecular blueprint for sarcomere assembly by providing specific attachment sites for numerous sarcomeric proteins, as well as acting as a molecular spring (1, 2). Titin also contains a catalytic serine-threonine kinase domain (TK), which is inhibited by a specific dual mechanism (3). However, the upstream elements controlling TK activation, its range of cellular substrates, and particularly the role of TK in mature muscle are largely unknown. Spanning half sarcomeres from Z disk to M band, titin is in a unique position to sense mechanical strain along the sarcomere (1). The elastic properties of the titin molecule and the mechanical deformation of the M band during stretch and contraction (4) suggest that the signaling properties of TK might be modulated by mechanically induced conformational changes. Molecular dynamics simulations suggest that mechanical strain can induce a catalytically active conformation of TK (5).

The catalytic kinase domain of titin interacts with nbr1. We searched for further elements of a putative signaling pathway that might recognize mechanically induced conformational intermediates of titin's catalytic domain. In a systematic two-hybrid screening approach with various structure-based open states of the catalytic site [kin1, kin2, and kin3 (6)], we identified the zinc-finger protein nbr1 (7) as a TK ligand, which interacted via its N-terminal PB1 domain with the semiopened construct kin3 (Fig. 1, A and B). This interaction was also seen in precipitation experiments with nbr1 and TK-kin3 (fig. S1A). Kin1, where the complete regulatory domain closes the active site, and kin2, where the α helix R1 (3) is deleted, did not interact. Thus, αR1 was necessary but not sufficient for nbr1 binding, which also required a semiopened catalytic cleft. Such a conformational intermediate of TK activation is predicted to be mechanically inducible (5), suggesting that the nbr1 interaction with the regulatory domain of TK may channel into mechanically modulated signaling in muscle. PB1 domains provide protein-interaction modules in diverse signaling proteins, allowing the formation of large, heteromultimeric complexes (8). Because related zinc-finger proteins have been found to act as scaffolds for signalosome assembly (9), we searched for further nbr1-interacting proteins in muscle by yeast two-hybrid screens. We identified the nbr1-related zinc-finger protein p62 [also known as SQSTM1, ZIP, or ORCA (10, 11)] as a further ligand of nbr1 in cardiac and skeletal muscle libraries (Fig. 1B). Both nbr1 and p62 were in vitro substrates of TK with substrate sites in the N termini (Ser115 or Ser116 for nbr1), although p62 was a significantly poorer substrate (KM value is lower by ∼10) than telethonin (3) or nbr1 (fig. S3).

Fig. 1.

Interaction of the titin kinase domain with the nbr1 muscle signalosome. (A) The C-terminal regulatory tail of the titin kinase domain (TK-RD) with secondary structure elements highlighted; residue numbering according to the crystal structure (3, 34). Arg279 in αR1, which is mutated in the human myopathy, is highlighted. TK constructs used for yeast two-hybrid interaction analysis are shown below, and interaction with nbr1 and TK-RD on the right. Kin1 does not interact with the TK-RD, because the intramolecular interaction is favored in the autoinhibited form (3). (B) Nbr1 acts as a scaffold to target p62 and MuRF2 to TK. Protein interactions were identified and mapped in two-hybrid systems and biochemically. Interacting domains are connected by arrows. Nbr1 homodimerizes via the first coiled-coil domain (CC) and interacts via its PB1 domain with that of p62 and with TK. Both proteins contain a ZZ zinc-finger domain and a C-terminal UBA domain. The UBA domain of p62 interacts with the RING- and the B-box (BB) domain region of MuRF2, which can multimerize via a coiled-coil domain; in the cardiac isoform MuRF2p27, this domain is spliced out (16). MuRF2 in turn interacts with the last 54 residues of the SRF transactivation domain (TA).

The nbr1 ligand p62 interacts with the RING-B-box protein MuRF2. P62 is a multivalent scaffolding platform, which interacts with several kinase signaling pathways apart from that of TK (11). The ubiquitin-associated (UBA) domain of p62 is an important component of pathways controlling focal turnover in bone (12) as well as interacting with several signaling proteins in neurons (13).

We asked which interactions of p62 could channel TK signaling into muscle-specific responses. We identified the RING/B-Box protein MuRF2 as a ligand of the p62 UBA domain (Fig. 1B and fig. S2). Nbr1, p62, and MuRF2 proteins interacted in successive pairs in vitro and could be precipitated in a large complex from muscle tissue extracts, demonstrating their association in vivo (fig. S1B). MuRF2, a muscle-specific zinc-finger protein, is involved in primary myofibrillogenesis (1416) and can shuttle between the cytosol and the nucleus under atrophic conditions (16). The closely related MuRF1 has been implicated in ubiquitin-controlled protein turnover in atrophic muscle (17) but also interacts with the ubiquitin-like modifier SUMO-3 (18) and the SUMO E2 ligase Ubc9 (19).

Because the interactions identified here suggest a link between TK and diverse signaling pathways, we assessed protein localization in muscle cells. Nbr1 and p62 were both detected at the sarcomeric M band (Fig. 2, A and B), together with TK (20) and MuRF2 (16). Thus, nbr1 appears to act as a muscle cytoskeleton-associated kinase scaffolding protein to assemble large sarcomeric signalosomes via its interactions with multiple proteins, linking TK to p62 and MuRF2. When nbr1 and TK kin3 were cotransfected into neonatal rat cardiomyocytes (NRCs), both proteins colocalized in sarcomeres (Fig. 2C). Overexpression of nbr1 perturbed normal localization and sequestered p62 (and to a lesser extent MuRF2) from M lines into cytosolic aggregates (fig. S2C). Thus, in muscle, nbr1 associates with TK and is essential for the correct targeting of the complex of p62 and MuRF2 to the M band.

Fig. 2.

Sarcomeric association of nbr1 and p62. (A) Confocal microscopy of neonatal rat cardiomyocytes demonstrates that nbr1 is a sarcomere-associated protein that colocalizes with the titin M-band epitope T51 (20), resulting in yellow color in the overlay (left images). Occasional Z-disk association can be observed additionally in skeletal muscle. (B) P62 is localized at the sarcomeric M band, similar to nbr1; it can additionally be found in a more diffuse pattern as well as at intercalated disks. (C) Cotransfection of TK-kin3 and GFP-nbr1 in NRCs shows both exogenous proteins localized together in sarcomeres (arrows) as well as in cytoplasmic aggregates (arrowheads). (Inset) Images magnified 2.6 fold.

TK-associated protein localization is mechanically modulated and regulates muscle gene expression. The interaction of nbr1 with a mechanically inducible conformation of TK, mimicked in TK kin3, led us to test the possible mechanical modulation of the localization of these proteins in muscle cells. NRCs are terminally differentiated and mechanically active but respond to stimuli modulating muscle-specific gene expression. Mechanical arrest of NRCs by various agents (6) resulted in marked nuclear accumulation of MuRF2 (Fig. 3, A and B) and the dissociation of p62 from the sarcomere and its relocalization to intercalated disks (fig. S4), suggesting that mechanical signals contribute crucially to regulating MuRF2 nuclear localization. To detect possible nuclear ligands of MuRF2, we performed two-hybrid interaction screening for MuRF2 ligands and detected an interaction of the RING/B-box domains with the transactivation domain of the serum response factor SRF (Fig. 1B). Activation of SRF-driven transcription of immediate-early genes like c-fos plays a central role in the response of muscle to hypertrophic stimuli including mechanical stress (21, 22). SRF cooperates with other transcription factors in myogenic transcription [reviewed in (23)] and is crucial for heart development (24, 25) and postnatal hypertrophic growth (23). The interaction with SRF suggests that MuRF2 might modulate muscle gene transcription by influencing SRF. In NRCs, we found that nuclear MuRF2 was associated with a strong reduction of the nuclear concentrations of endogenous SRF (Fig. 3, A and B) and its cytoplasmic accumulation. Similarly, overexpression of the small cardiac isoform MuRF2p27 alone in untreated NRCs resulted in a similar cytoplasmic localization of SRF (fig. S5A). When we assayed SRF transcriptional activity by reporter gene assays with the SRF-dependent c-fos promoter, we found that overexpressed MuRF2p27 led to strong suppression of c-fos activity (fig. S5B). Mutation of two zinc-coordinating cysteines (Cys29 and Cys78) to alanine in the RING domain completely abolished nuclear localization of MuRF2. Nuclear MuRF2 therefore affects the nuclear pool of SRF in a RING domain–dependent way and represses transcriptional activity in muscle cells. Similar mechanically induced MuRF2 relocalization could also be observed in vivo, when the mechanical activity of skeletal muscle was arrested by denervation. MuRF2 appeared in the nuclei of sciatic denervated muscle fibers as early as 6 hours after mechanical arrest (up to 53% from under 4% in control fibers), suggesting this relocalization is an early event in denervation-induced atrophy (fig. S6).

Fig. 3.

The intracellular localization of MuRF2 is mechanically modulated. (A) In control NRCs, endogenous MuRF2 is associated with sarcomeres and SRF is nuclear. (B) Arrest of beating leads to nuclear translocation of MuRF2 and cytoplasmic accumulation of SRF; note reduced cell size. (Inset) Reduction of SRF in NRC nuclear extracts. Lane 1, control cells; lane 2, KCl-arrested cells. (C) In NRCs transfected with TK (arrows), MuRF2 concentrations are strongly reduced, and MuRF2 is excluded from the nucleus (asterisk, nontransfected cell). (Inset) Western blot (top row) confirms the reduced MuRF2 concentrations (bottom row, Ponceau Red–stained actin control band). C, mock-transfected control; TK, TK-transfected cells. (D) Transfection of TK in BHK-Bi cells stimulates transcription of myomesin (MYOM), BNP, or c-fos reporter genes. Mutation of the catalytic Asp127 to Ala127 (DA) abolishes activating effect. C, mock-transfected control; WT, wild-type TK. (E) MYOM reporter activity was assayed in NRCs cultivated in the presence of 20 μM phenylephrine (control, C). Transfected TK (WT) results in further stimulation of myomesin promoter activity, which is not observed for catalytically inactive TK (DA). (F) Influence of mechanical activity on muscle promoter activity. NRCs were cultured in presence of phenylephrine (mock-transfected controls, C). Arrest of contractile activity leads to reduction in myomesin promoter activity (arrested, A), which can be rescued by transfection of wild-type titin protein kinase (A+WT). Error bars indicate SEM.

Transfection of TK in NRCs leads to dissociation of MuRF2 from the sarcomere, nuclear exclusion, and strongly reduced MuRF2 protein amounts (Fig. 3C). Thus, TK activity can modulate MuRF2 protein concentrations as well as intracellular localization and causes opposite effects than mechanical arrest, leading us to test whether the overexpression of TK in myocytes would result in changes in muscle gene expression. We first tested this in the myogenic titin knock-out cell line BHK-Bi (26), which contains a homozygous deletion in the titin gene including the kinase domain, thus abolishing endogenous TK activity. The MYOM promoter drives expression of the M-band protein myomesin, a constitutive sarcomeric protein that is linked to myosin and titin with a constant stoichiometry to the thick filament proteins (4), serving as a reporter for sarcomere synthesis. The MYOM promoter contains various binding sites for both myogenic transcription factors like MyoD as well as for SRF, GATA, and NFκB (27) and therefore could sense the combined input of the putative transcriptional regulation by TK-nbr1-p62-MuRF2 in muscle. We also tested the activity of the c-fos and the brain natriuretic peptide (BNP) promoter, markers of strain-regulated hypertrophy (28). In the titin knockout cell line BHK-Bi, exogenous TK activity led to strong increases of MYOM promoter as well as increases of c-fos promoter and BNP promoter activities (Fig. 3D). Similarly, in NRCs, which contain endogenous titin protein kinase activity, activation of promoters including MYOM was observed (Fig. 3E) even in presence of phenylephrine, an α-adrenergic agonist inducing cardiac hypertrophy. This suggests that TK and phenylephrine-activated α-adrenergic receptors signal via distinct pathways. Mechanical arrest of NRCs resulted in strong reduction of transcriptional activity including the MYOM promoter, even under persistent α-adrenergic stimulation (Fig. 3F). Overexpression of TK could rescue the mechanical arrest–dependent depression of muscle gene expression (Fig. 3F). Thus, the TK-nbr1-p62-MuRF2 pathway may be involved in mechanical signaling in muscle, in which strain modulated TK conformation and activity is channeled into the nbr1-p62-MuRF2 complex. Mechanical arrest would result in the relocalization of MuRF2 to the nucleus and of p62 to intercalated disks, whereas TK activity would counteract these changes and concomitantly induce the release of the inhibitory action of nuclear MuRF2 on SRF-mediated muscle gene expression.

A human mutation in the titin kinase domain disrupts nbr1 binding and leads to hereditary muscle disease. Support for the concept of a TK pathway feeding into muscle turnover via the nbr1-p62-MuRF complex comes from the analysis of hereditary myopathy with early respiratory failure (HMERF), an autosomal dominant muscle disease with proximal weakness of the upper and lower extremities and early involvement of neck flexors and respiratory muscles, causing respiratory failure as a frequent cause of death (29). At the ultrastructural level, abnormal Z disks and actin aggregates, with dissolving Z disk, I band, and M band structures, and radical focal myofibrillar breakdown are observed (29).

A genome-wide screen on two large unrelated Swedish families mapped the disease locus to chromosome 2q24-31 (30). Titin was the strongest positional and functional candidate gene. Sequencing revealed a heterozygous CGG→TGG change (Fig. 4A), leading to the exchange of a completely conserved arginine to tryptophan at position 279 (R279W) in αR1 of the TK regulatory tail (Fig. 4B). This showed complete cosegregation with the disease, and with a likewise segregating core haplotype, in the two families (Fig. 4C). This change was not reported in single nucleotide polymorphism databases and was not found in 200 (400 chromosomes) normal Swedish controls. Another Swedish patient with identical phenotype but without known genealogical relation to anyone in the two original families was found to have the same mutation on the same haplotype, indicating a common ancestry.

Fig. 4.

A mutation in the human titin kinase domain causes a severe autosomal-dominant hereditary myopathy. (A) The single base pair exchange in αR1, leading to the exchange of R279 to W (34) as shown in the partial nucleotide sequences of heterozygous patient (W/M) and wild-type healthy control (W/W). The peptide sequence of αR1 is identical between the human, mouse, rat, and rabbit titin, arguing for crucially conserved interactions. Numbering of DNA sequence is arbitrary in these samples. (B) Ribbon diagram of the titin kinase domain with αR1 in red illustrates that the mutated residue R279 (side chain highlighted in blue) is a surface-exposed residue. (C) Pedigree of two Swedish families with HMERF; affected members are shown by solid symbols and unaffected by open symbols. (Bottom) Segregation of the titin protein kinase mutation is shown by Msp I restriction fragment length polymorphism analysis, and R279W heterozygous members are marked by red asterisks.

When recombinant R279W-mutant TK was assayed for calmodulin-stimulated catalytic activity, no significant change of activity in comparison to wild-type TK was observed (fig. S7). Because the R279W mutation in αR1 results in a drastic change of a surface-exposed basic to a nonpolar, bulky amino acid in the nbr1 binding site, we tested the interaction of TK with nbr1, which was dramatically reduced in the mutant TK (Fig. 5, A and B). In patient biopsies, nbr1 was localized abnormally diffusely in diseased muscle instead of being M band– and Z disk–associated, although in HMERF 50% of TK was expected to be wild type (Fig. 5, C and D). P62 accumulated in many diseased muscle fibers of the HMERF patients (Fig. 5, E and F). MuRF2 showed unusual nuclear localization in centralized nuclei of patient muscle fibers (Fig. 5G) not observed in peripheral nuclei of normal muscle or in three other myopathies with centralized nuclei (Duchenne muscular dystrophy, tibial muscular dystrophy, and myotubular myopathy). Furthermore, MuRF1, a close homolog of MuRF2, did not show nuclear localization in HMERF patients or in the other myopathies tested with centralized nuclei.

Fig. 5.

The interaction of the titin kinase domain with nbr1 is suppressed in the R279W mutant. (A) Yeast two-hybrid assays with wild-type TK (WT) and the R279W mutant (RW). Wild-type TK kin3 shows nbr1 interaction via activation of the HIS3 reporter gene by growth on histidine-free medium, as well as activation of GAL4 β-galactosidase activity (not shown). Both markers for protein interaction are repressed in TK-R279W. (B) Precipitation binding assays demonstrate that TK-R279W fails to interact with glutathione S-transferase (GST)–nbr1, in contrast to wild-type titin protein kinase. Lane 1, 10% of input cell lysate; lane 2, eluted fraction from nbr1 beads; lane3, control beads with GST alone. Muscle biopsies from controls and HMERF patients were analyzed by immunofluorescence for nbr1, p62, and MuRF2. In normal skeletal muscle, nbr1 is mostly M band–associated, with occasional weaker Z disk association (C). Note the amorphous localization of nbr1 in the HMERF sample with preserved staining of the Z-disk titin epitope T12 (D). P62 is sarcomere-associated in normal muscle (E) but forms large cytoplasmic aggregates in HMERF (F), partly colocalizing with degraded Z-disk material stained by titin T12 (arrows). MuRF2 is localized aberrantly in centralized nuclei (arrowheads) (G) in HMERF muscle fibers.

A knockout mouse model with a large deletion of M-band titin, encompassing TK and the binding sites for MuRF1/2/3, DRAL, and myomesin, is lethal when activated at embryonic stages and developed severe progressive muscle weakness at later stages of development (31). Similarly, a cell line with a targeted heterozygous deletion of the kinase exon leads to perturbed myofibril assembly (32). Both mutations affect numerous known protein binding sites in addition to TK itself, and their effects can therefore not be attributed to abolished kinase signaling alone. In HMERF, primary myofibril assembly seems not to be affected. The severe pathology starts in adulthood, indicating an activity-related turnover or maintenance defect. The importance of tight regulation of kinase function for accurate turnover control is suggested by the fact that in HMERF a dominant mutation is pathogenic, despite an otherwise catalytically active enzyme and intact M-band titin. All reported cardiomyopathy patients with titin mutations lack skeletal muscle weakness, and TMD/LGMD2J and HMERF patients with titin mutations suffer from skeletal muscle weakness and atrophy without clinically dominant cardiac involvement. This may be caused by the expression of tissue-specific isoforms of MuRF or other components of the M band (15, 16, 33) in cardiac and skeletal muscle (fig. S8). The HMERF mutation we report here supports the suggested in vivo functions of the titin protein kinase as a physiological link between sarcomere activity and transcriptional regulation. Thus, the TK-associated protein complex may act as a central switchboard where input from various pathways leading to hypertrophic or stress response and protein turnover links the sarcomere to mechanical modulation of muscle gene transcription, a feedback mechanism that may prove to be important for therapeutic intervention in skeletal myopathies and heart failure.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1110463/DC1

Materials and Methods

Figs. S1 to S8

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article