The giant muscle protein titin, the “backbone” of the sarcomere, harbors a complex molecular spring whose stiffness is variably tuned in health and disease. Titin is increasingly recognized as a crucial integrator of diverse myocyte signaling pathways. The titin-associated signalosome includes hotspots of protein-protein interactions important for the regulation of protein quality-control mechanisms, hypertrophic gene activation, and mechanosensing.
The forces developed by striated muscles are usually classified into an “active” component generated by the contractile proteins and a “passive” component generated in the stretched muscle by elastic elements. Both components are essential for muscle function. Mainly responsible for the passive force component are two types of structure: the network of collagen fibrils in the connective tissue surrounding individual myocytes (as well as cardiomyocytes), fascicles, and whole muscles, and the titin filaments in the sarcomeres (19, 25, 32, 60, 63, 87, 109, 110), the contractile units of striated muscle cells. Titin is the largest protein in the human body and can be considered the backbone of the (half-)sarcomere, with its NH2 terminus anchored at the Z-disk, the elastic spring segment running through the I band, and its longest portion bound to the thick filament, reaching all the way to the M band (FIGURE 1A). The >1-μm-long titin filament is best known as a molecular spring responsible for myocyte passive tension, stiffness, viscoelasticity, and passive elastic recoil (3, 18, 25, 32, 65, 69, 72, 80, 109). Titin is also well recognized for its A-band centering function in the sarcomere (33), which helps reduce sarcomere length (SL) inhomogeneities and for its role as a scaffolding protein governing myofibrillar assembly (for recent reviews, see Refs. 9, 23, 52, 102). Apart from these “classical” properties, titin has important signaling functions that have emerged only recently. In this brief review, we begin by summarizing current ideas on titin's mechanical function in skeletal muscle and then focus on novel evidence suggesting titin integrates diverse signaling pathways in both skeletal and cardiac myocytes. We also cover recent work demonstrating how titin stiffness can be adjusted through isoform switching or acutely by phosphorylation and how these mechanisms may be impaired in the failing heart.
The Titin Spring in Skeletal Muscle
Titin, first described as connectin (68, 111), is expressed in different isoforms (molecular weight of 3,000–3,700 kDa) generated by alternative splicing from the transcript of a single titin gene (49). Almost all of the differential splicing occurs in the (half) I-band segment, which harbors the molecular spring (FIGURE 1A). Skeletal-muscle sarcomeres express so-called N2A-titin isoforms [∼3,300–3,700 kDa (15, 87)], in which the elastic segment comprises tandem-immunoglobulin domain (Ig) regions (proximal, variable-length, and distal) flanking a unique sequence rich in proline, glutamic acid, valine, and lysine, the PEVK-domain (FIGURE 1A). Additionally, an “N2-A” segment containing four Ig domains and a few intervening sequences separates the PEVK domain from the variable-length Ig region. The segments in which most of the alternative splicing occurs are the variable-length Ig region and the PEVK domain (FIGURE 2) (49). The extensive splicing of I-band exons results in great diversity of titin-domain composition in the spring segment and is the reason for the large variability of titin-based stiffness among different skeletal muscles (87, 109, 110).
Long titin isoforms give rise to low myocyte passive tension, short isoforms to higher passive tension (8, 32, 60, 109). Initially, titin stiffness was thought to be related to muscle type in that stiff titin is expressed in fast-twitch muscles and compliant titin in slow-twitch muscles (49). However, when the titin size was measured in ∼40 different adult rabbit skeletal muscles, it was found to be diverse (3.3–3.7 MDa), with slow muscles indeed expressing long titin isoforms, but fast muscles expressing short, intermediate-size, or long titin isoforms (87). Overall, there was no or very low correlation between titin size and muscle fiber type or myosin-isoform composition in these muscles. As expected, titin size correlated inversely with titin-based passive tension. Thus slow muscles generally had low titin-based passive tension, but fast muscles showed largely diverse titin-based tension levels (87). In summary, it is now established that titin size varies considerably among different skeletal muscles, producing highly variable myocyte (titin-based) passive stiffness. A correlation of these parameters with muscle type, at best, is very low.
Titin and Collagen as Determinants of Passive Tension in Skeletal Muscle
Earlier evidence suggested that a large contribution to the passive stiffness of a muscle comes from structures outside the myocytes, particularly collagen (11, 22, 42). These observations were confirmed and extended recently by demonstrating that the contribution of collagen to total passive tension, relative to that of titin, varies greatly among different skeletal muscles (87). Some muscles, like rabbit soleus, had high collagen-based passive stiffness and low titin-borne stiffness, but high total passive stiffness. Other muscles, like rabbit psoas, had lower total passive stiffness than the soleus, low collagen-based stiffness, but high titin-borne stiffness. All in all, titin and collagen were about equally important as determinants of passive stiffness (87).
Titin-Based Passive Tension in Human Skeletal Muscle
Many human physiology textbooks depict a “physiological working range” for skeletal muscles and state that these muscles operate at or near the plateau of the bell-shaped active sarcomere length-tension curve. At this SL, passive tension is considered very low. But how low (or not) really is passive tension in working human muscles? Only very recently could the live SL of human skeletal muscles be measured directly. This was done using a minimally invasive high-speed imaging technique, an optical microendoscopy method observing “second-harmonic frequencies of light” generated in the muscle fibers, largely by myosin (64). This breakthrough study demonstrated that the live, working SL of human wrist muscles is between ∼3.0 and 3.4 μm. These relatively long SLs suggest the muscles work on the beginning descending limb of the active SL-tension curve, as one can induce from the thin-filament length in human muscles, which is ∼1.3 μm, and from the SL where maximum stress develops, which is ∼2.8 μm (79, 85). Passive force measurements on single human muscle fibers or isolated myofibrils (79) demonstrated that at 3.0- to 3.4-μm SL, titin contributes significantly to passive tension (FIGURE 1, B AND D): in human vastus lateralis, titin-based passive tension was >10 mN/mm2 at 3.0-μm SL, increasing to nearly 30 mN/mm2 at 3.4-μm SL (FIGURE 1, B–D). Similar passive-tension levels were reported for single human soleus fibers (101). This compares to a developed tension of typically 100–300 mN/mm2 at the optimum SL of 2.8 μm (79, 85). Thus, tentatively assuming that the live SL range for human soleus or vastus muscle is similar to that of wrist muscle, titin-based passive tension in the physiological working range is considerable and reaches ∼10% of the maximum active tension.
Molecular Mechanism of Titin Elasticity
The mechanical properties of titin have been investigated both in the environment of the sarcomere (for reviews, see Refs. 23, 56) and in vitro at the single-molecule level (reviewed in Ref. 58). Pioneering work on titin mechanics using atomic force microscopy (AFM) (67, 92) or laser optical trapping (38, 103) laid the groundwork for an impressive number of studies demonstrating that titin Ig domains readily unfold under a stretch force in vitro. The Ig domains can also refold in vitro under forces of up to ∼25 pN (7, 21). Furthermore, in silico experiments using the power of steered molecular dynamics simulations have provided us with atomic details of Ig-domain unfolding (98). However, despite a huge body of useful information on titin Ig-domain mechanics (reviewed in Ref. 58), it remains unknown, owing to methodological difficulty in measuring the phenomenon, whether these domains unfold under physiological conditions in the sarcomere.
The currently established titin-extensibility model (FIGURE 1, A AND B) depicts a scenario whereby titin extends at low stretch forces mainly by straightening interdomain linkers in the Ig regions (35, 60, 62, 101). Crystallographic evidence suggests the elasticity of the Ig regions can be defined by segmental flexibility of regular patterns of Ig super-motifs (107). The portion of titin's force-extension curve during which the Ig regions straighten can be modeled using entropic elasticity theory (55, 57, 113); most frequently, the wormlike-chain model has been used (for review, see Ref. 58). This model predicts that the stretch force (F) is related to the fractional extension (x/L) of a polymer chain through where A is the persistence length, a measure of the chain's bending rigidity, kB is the Boltzmann constant, T is absolute temperature, x is extension, and L is the contour length.
At higher forces, the chief contribution to titin extension arises from elongation of the PEVK-domain (FIGURE 1, A AND B). This largely unstructured region is thought to act as an entropic-enthalpic spring (14, 59, 101, 108), wherein the non-entropic factors may be electrostatic and hydrophobic interactions (reviewed in Ref. 58). Stretching the PEVK domain requires larger forces than stretching the Ig regions, mainly because of the difference in contour-length to persistence-length (L:A) ratio: this ratio is much higher for the PEVK domain (very flexible chain) than for the Ig regions (semi-flexible chain) (55, 57, 59, 62, 101). Hence, the titin I-band segment acts as a dual-stage molecular spring (FIGURE 1, A AND B).
Massive Ig-domain unfolding has been rejected as a general mechanism of extensibility (72, 100). In the distal Ig region, homotypic binding presumably of six parallel titin molecules (34) may hinder both extension and Ig-domain unfolding. However, Ig domains could unfold in overstretched muscle and, preferentially in the proximal and variable-length Ig regions, even at physiological SLs (72) (FIGURE 1A). Considering the in vitro mechanical properties of titin Ig domains, a possibility is that in stretched myofibrils the proximal/variable-length Ig domains exist in dynamic equilibrium between folded and unfolded states. The physiological function of dynamic Ig-domain unfolding-refolding in sarcomeres could be that of a shock absorber to quickly lower the potentially damaging impact of high stretch forces. Moreover, stretch-induced alterations to the dynamic steady-state between Ig-domain folding and unfolding can potentially modulate titin-based signaling in the myocyte (see below). However, as stated above, it remains to be shown whether titin Ig-domains really unfold in the sarcomere.
Emerging Evidence Linking Titin to Protein Quality-Control Mechanisms, Hypertrophic Signaling, and Mechanosensing
Next to the elastic spring function, a scaffolding role has been established for titin. The protein interacts with many structural proteins, such as α-actinin, nebulin, and γ-filamin in the Z disk, myosin heavy chain and myosin-binding protein-C in the A band, as well as myomesin and obscurin in the M band (FIGURE 2) (for original citations, see Ref. 56). Titin appears early in myocyte development and is required for sarcomere assembly (9, 23, 52, 94, 102). The focus of this review will be on exciting new evidence implicating titin in diverse myocyte signaling pathways (FIGURE 2). The sites on titin where binding to structural, contractile, and signaling molecules occurs have been reviewed in detail elsewhere (47, 52, 56). Here, we put forward a perspective whereby titin in the myocyte is a multiple-signal integrator linked to three main signaling functions: protein quality control, hypertrophic gene activation, and mechanosensing (FIGURE 2).
Titin and Protein Quality Control
The two main components of the protein quality-control system are the chaperones, which protect proteins from misfolding and help protein assembly, and the ubiquitin-proteasome system (UPS), which works together with proteases, the calpains, and recognizes specific proteins to target them for degradation (117). The small heat shock proteins αB-crystallin and HSP27 bind along the titin spring, but the exact binding sites on the N2A-titin isoform are not known, although binding to Ig domains is likely (6, 104, 123). However, an established binding site for αB-crystallin is in a cardiac-specific titin domain (see below). Interactions of the chaperones with titin may have protective effects on the myofibrils and help prevent stress-induced sarcomere degradation.
Titin is further connected to protein quality-control mechanisms through direct interactions with ubiquitous calpain-1 at the NH2 terminus of the I-band segment (10, 91) and with skeletal muscle-specific calpain-3 at both the N2-A domain and the M-band (78) (FIGURE 2). An indirect link exists between titin at the extreme NH2 terminus and an E3 ubiquitin ligase, MDM2, through their mutual binding partner, telethonin (also called T-cap) (99) (FIGURE 2). MDM2 targets both itself and the tumor suppressor protein p53 for degradation by the proteasome. Moreover, titin binds near the M band to other E3 ubiquitin ligases, the muscle-specific RING finger proteins (MURFs), specifically, MURF1 and MURF2 (28, 73, 118). There is an important connection between the calpain system and the UPS in that calpain (especially calpain-1) appears to be required for ubiquitin ligases to reach sarcomere proteins and initiate their degradation by the UPS. The concerted effort by calpain and ubiquitin ligases is relevant, e.g., during muscle wasting: in the presence of calpain inhibitors, E3 enzymes like MURF1 remain highly expressed, but sarcomere degradation is inhibited and muscle atrophy is suppressed (13). Moreover, calpain-1 is necessary for the regular turnover of aggregated proteins to prevent excessive autophagy, a process involving the degradation of a cell's own components through the lysosomal machinery (20, 117).
Skeletal muscle-specific calpain-3 (p94) is protected from autolytic activation and disassembly by interacting with the titin N2-A domain (FIGURE 2) and also with sites in the PEVK domain (30, 78). In human patients, a loss-of-function mutation in the calpain-3 gene causes limb-girdle muscular dystrophy type-2A (LGMD2A) (for review, see Ref. 12). If the binding of calpain-3 to titin is inhibited, as in the LGMD2A mutation, the activated protease severs titin within the PEVK-segment and the Z-disk and M-band regions. In summary, several lines of evidence link titin directly to the main elements of the protein quality-control machinery: chaperones, calpain proteases, and the UPS.
Titin and Hypertrophic Signaling
Links between titin and hypertrophic signaling mechanisms are manifold, again involving Z-disk, I-band, and M-band titin domains (FIGURE 2). Binding of the extreme NH2-terminal titin Ig domains, Z1/Z2, to telethonin (124) also recruits a telethonin-ligand, muscle LIM protein (MLP), to the Z disk (39, 40). In skeletal muscle, MLP appears to be bound preferentially at the Z disk. However, MLP has also been detected in the I band (1), at costameres, and abundantly in the cytosol, as well as in the nucleus. Shuttling of MLP to the nucleus (4) can activate transcriptional regulators and may enhance protein expression. MLP also binds to calcineurin, a protein phosphatase dephosphorylating nuclear factor of activated T cells (NFAT), which can thus translocate to the nucleus and induce a hypertrophic gene program (93). The MLP-calcineurin-NFAT hypertrophic signaling pathway is negatively regulated by another MLP-binding partner, protein kinase C interacting cousin of thioredoxin (PICOT) (37), which receives input from the PKC signaling pathway. This hypertrophic pathway is thought to be activated by stress or strain imposed onto the Z disk, but the exact mechanism of action and the role of titin's NH2 terminus in it remain obscure. In conclusion, Z-disk titin appears to be part of a macromolecular machinery acting as a node for hypertrophic signaling.
Ig domains at titin's N2-A-domain interact with the three homologous muscle-ankyrin-repeat proteins (MARPs), cardiac-ankyrin-repeat protein (CARP), diabetes-related ankyrin-repeat protein (DARP), and ankyrin-repeat-domain protein-2 (Ankrd2) (71, 119), which in turn bind to myopalladin (2), an important actin-regulating protein (82) (FIGURE 2). Since members of the MARP family also associate with transcription factors (41), a role for MARPs as nuclear regulators of transcription is likely. Thus, via MARP-binding, the N2-A-domain of titin could be involved in hypertrophic signaling mechanisms.
M-band titin has various links to pathways of muscle-growth regulation, particularly through the interaction with MURFs (FIGURE 2), proteins that can shuttle to the nucleus to alter muscle gene expression. If both MURF1 and MURF2 are knocked down in mice, mild skeletal muscle and extreme cardiac hypertrophy results (120), suggesting an inhibitory role of the MURFs, particularly MURF1 (116), in hypertrophic signaling. Another important signaling domain at the M band is the titin kinase, which during myofibrillogenesis is activated by tyrosine phosphorylation and subsequent binding of Ca2+/calmodulin (70). The titin-kinase domain controls muscle gene expression and protein turnover via association with the neighbor-of-BRCA1 gene-1 (nbr1) protein, which in turn signals to MURF2 via binding to p62 (FIGURE 2); MURF2 activates hypertrophic genes in the nucleus, such as serum response factor (53). Knockdown of the titin-kinase region disrupts sarcomere assembly and disturbs muscle growth (74, 86, 114). Mouse skeletal muscles deficient in titin's M-band region show reduced contractility (83). These findings point to a crucial role of the titin-kinase complex as a platform for hypertrophic signaling. Furthermore, a unique sequence of M-band titin is linked to regulatory pathways of muscle growth through binding to four-and-a-half LIM-domain protein-2 (FHL2) (FIGURE 2). This protein has numerous other interaction partners, including metabolic enzymes (51), and appears to be a transcriptional co-activator. Note that FHL2 also associates with a cardiac-specific titin domain in the I-band segment (see below) (51). Finally, an important additional aspect is that most of the human titin mutations reported to date, many of which result in disturbed muscle growth and the clinical phenotype of muscular dystrophy, were found in the molecule's M-band portion (reviewed in Refs. 26, 47). All these lines of evidence have established the M-band region of titin as a hotspot of hypertrophic signaling.
Titin and Mechanosensing
Emerging evidence implicates titin in stress-sensing pathways converging onto different sarcomeric regions (FIGURE 2). The exact mechanisms of mechanosensing in the myocyte are largely unknown, but it is obvious that there must be cross talk to downstream signaling events of the hypertrophic gene program. Accordingly, potential stress-sensor pathways linked to titin include the telethonin-MLP-calcineurin-NFAT axis associated with the Z1/Z2 Ig domains in the Z disk (37, 40), the MARP-Myopalladin-complex associated with the N2-A-domain in the I band (71), and the macromolecular assembly at the titin-kinase domain (FIGURE 2). Arguably best understood is the mechanosensitive function of the titin-kinase domain. Strain activates the titin-kinase domain and allows ATP binding, which triggers autophosphorylation and further kinase activity (88, 89). Thus the titin kinase may be a biological force sensor helping myocytes to adapt to changes in loading. Taken together, evidence for the involvement of titin in mechanosensing is still scattered but promising. The layout of titin in the sarcomere suggests the protein is well suited to act as a strain sensor and integrator at the crossroads of myocyte stress signaling.
Unique Properties of Cardiac Titin
In heart muscle, titin is expressed in two main isoforms: the N2B-isoform (3,000 kDa), which contains a short, stiff spring segment, and (variable) N2BA-isoforms (∼3,200–3,700 kDa), which contain longer springs and thus are more compliant (15) (FIGURE 3). The interactions and the connectivity described for skeletal-muscle titin (FIGURE 2) in principle also apply to titin in the cardiac myocyte (FIGURE 3). However, cardiac titin has some unique properties that arise from the co-expression of N2BA and N2B isoforms in the (half-)sarcomere and from the presence of a cardiac-specific region, the N2-B domain in the middle of the spring segment.
Cardiomyocyte-Specific Interactions at Titin's N2-B Domain
The human titin N2-B domain, encoded by exon 49, contains three Ig domains and a unique intervening sequence of 572 amino acids (N2-Bus). The N2-Bus is an additional extensible element in the cardiac titin spring, next to the Ig regions and the PEVK segment (61), but it is also involved in protein-protein interactions (FIGURE 3). The N2-Bus binds two isoforms of the four-and-a-half-LIM-domain protein, FHL1 (96) and FHL2 (51). Both FHL1 and FHL2 are transcriptional co-activators and interact with effector mitogen-activated protein kinases (MAPKs). FHL1 bound to the N2-Bus associates with ERK2 and MEK1/2, as well as Raf1, which is activated via the G-protein-coupled receptor (GPCR) pathway (96) (FIGURE 3). Knockout of FHL1 in mice increases myofibrillar compliance but reduces hypertrophic signaling and causes a blunted response to pressure overload-induced cardiac hypertrophy (96). Thus the macromolecular complex at titin's N2-Bus could be an I-band-based stress sensor in cardiomyocytes.
MAPK members also activate the small heat-shock protein, αB-crystallin, which in the ischemic heart associates with titin at the N2-Bus (FIGURE 3) and at Ig domains, where it appears to increase domain stability (6, 123). The binding to αB-crystallin provides a cardiac-specific link for titin to the protein quality-control system (see above). Deletion of the N2-B domain in mouse hearts resulted in reduced slack SL, elevated titin stiffness, atrophy, and severe diastolic dysfunction (90) (FIGURE 3). FHL2, but not αB-crystallin, was downregulated. This contrasted with a knockout of the constitutively expressed PEVK-segment (titin exons 219–225), which triggered diastolic dysfunction through cardiac hypertrophy (FIGURE 3), presumably by increasing the binding of FHL1 to the N2-Bus, thereby activating the N2-Bus-associated stress sensor (24). Taken together, the N2-B domain is an emerging hotspot for protein-protein interactions and likely responsible for cardiac-specific signaling properties of titin.
Titin Isoform Switching in Normal and Failing Heart
Compared with human donor hearts, the N2BA:N2B isoform expression ratio is increased in eccentrically remodeled explanted hearts from ischemic cardiomyopathy (76) or dilated cardiomyopathy (DCM) (66, 75) patients [“systolic heart failure” (SHF) patients]. This titin-isoform switch in end-stage human heart failure decreases cardiomyocyte passive stiffness (FIGURE 4A), possibly as a compensatory mechanism initiated in a globally stiffened (fibrotic) environment (for more detailed reviews, see Refs. 16, 54, 56). These results on human hearts contrast with those obtained from pacing tachycardia dog models (DCM models), which consistently revealed substantially decreased cardiac N2BA-to-N2B ratios in the left ventricle (36, 121), as well as the three other cardiac chambers (36). A reason for the different directions of titin-isoform switching in humans vs. pacing dog models is not known. As regards titin alterations, these dog models obviously do not reflect the patient situation well.
Not yet established is whether (and in which direction) titin-isoform transitions take place in human diastolic heart failure (DHF), a disease typically associated with decreased left ventricular (LV) distensibility and increased LV stiffness. Increased passive stiffness was found in enzymatically skinned cardiomyocytes obtained from biopsy samples of DHF compared with SHF (DCM) patients (105), consistent with a decreased N2BA-to-N2B titin expression ratio. However, myocardial samples from individual patients with DHF revealed an increased N2BA-to-N2B ratio but elevated passive stiffness at the level of the skinned isolated cardiomyocyte, compared with nonfailing controls (5). In aortic stenosis, another condition characterized by concentric LV remodeling, one study reported increased (5) and another somewhat decreased (115) cardiac N2BA-to-N2B titin ratio. In a patient-mimicking, old-dog model of DHF, the cardiac N2BA-to-N2B ratio was slightly but significantly lower than in normal old dogs (95).
What is required now is the study of larger numbers of human DHF samples (which for ethical reasons are more rarely obtained than human SHF samples). Moreover, titin-isoform expression in the heart varies with the location the tissue is procured from, especially in larger mammals, including humans: differences in N2BA-to-N2B ratio are found between the atria and the ventricles (the latter have a lower N2BA-to-N2B ratio) and the left and right ventricles (the latter have a higher N2BA-to-N2B ratio); there is transmural variability and variability from the heart's base to the apex [the latter has a lower N2BA:N2B ratio (8, 77)]. Therefore, location-matched tissue samples are desirable to unequivocally demonstrate which role titin-isoform switching and corresponding titin stiffness changes have in DHF. Regardless of current study limitations, it is clear that titin-isoform switching in failing hearts is an established mechanism to adjust myocardial passive stiffness over a time course of days to weeks (FIGURE 4A).
Toward a Mechanism of Titin-Isoform Switching
Triggers for titin-isoform transitions in failing hearts are currently unknown, but hints have come from a study on developing rat cardiomyocytes in culture (48). A characteristic feature of developing heart tissue, both in vitro (48) and in vivo (44, 50, 81, 112), is the dramatic reduction in titin size and change in titin-isoform composition before or around the time of birth (44) (FIGURE 4A, INSET). This isoform transition causes greatly increased titin-based passive tension in adult compared with fetal hearts (FIGURE 4A). Fetal cardiac titin is a very large N2BA-isoform (3.7 MDa), which is replaced pre- or perinatally by smaller-size N2BA titins and the N2B isoform, which in many adult mammals becomes the predominant cardiac titin isoform, especially in rodents and other small animals with a fast heart beat (adult rat hearts have >90% N2B and <10% N2BA), but as well in humans [normal adult human hearts express ∼65% N2B and ∼35% N2BA (76)] (FIGURE 4A, INSET). Skeletal muscles also show a down-sizing of (N2A) titin isoforms during development owing to differential splicing (84).
Addressing the mechanism of titin-isoform switching, we reported that, in cultured rat cardiomyocytes grown in thyroid hormone (T3) enriched serum, the perinatal isoform transition toward N2B titin was faster and more complete than in serum lacking T3 (48). Similar effects could be triggered by insulin (43). Notably, the switching toward N2B could be stalled by inhibiting phosphorylation of AKT through pharmacological blockade of phosphatidyl-inositol-3-kinase (PI3K) (48) or mammalian target of rapamycin (mTOR) (43). Elsewhere, a rat model fed with propylthiouracil to induce hypothyroidism showed reexpression of large cardiac N2BA-titin isoforms with high compliance (122). We have thus suggested that various hormones, including T3 and insulin or growth-regulating hormones acting via receptor tyrosine kinases, can induce titin-isoform switching through a signaling cascade converging onto PI3K/AKT (47) (FIGURE 3). The AKT target, mTOR, which is crucial for the regulation of protein synthesis, also appears to be a component of the molecular machinery controlling titin-isoform composition. This hypothesis needs to be probed further, and possible downstream (splice?) factors need to be identified. A useful model to study titin splicing factors could be an unusual rat strain with much-delayed developmental titin splicing; these rats live throughout their adult life with a giant fetal cardiac N2BA titin of ∼3.9 MDa (27). Developmental splicing of troponin T was not affected by the (yet unknown) mutation in this rat strain, implying that titin splicing is regulated separately. Finally, an important clinical implication is that, since pathological alterations in T3 and insulin signaling can modify PI3K/AKT effector pathways (FIGURE 3), metabolic heart disease, such as diabetic cardiomyopathy, is expected to be associated with altered titin-based passive stiffness.
Acute Adjustment of Titin Stiffness by Phosphorylation in Normal and Diseased Heart
Although titin-isoform switching is a confirmed mechanism for adjusting myocardial passive stiffness, recent studies suggested that increased passive stiffness in chronic human heart disease can arise from alterations in the phosphorylation state of titin (45). Protein kinases PKA and PKG were found to both phosphorylate titin at serine S469 in the cardiac-specific N2-Bus (45) (FIGURE 3). This posttranslational modification reduces titin-based myocardial stiffness (5, 17, 45, 46, 105) (FIGURE 4B). Interestingly, administration of insulin to cultured developing rat cardiomyocytes also increased the phosphorylation state of cardiac titin N2B and N2BA isoforms greatly, perhaps via activation of the well known “survival” pathway acting via PI3K, nitric oxide synthase, and PKG (43). In LV samples of human DCM patients, a titin phosphorylation deficit was detected (45), which may act to increase passive stiffness in heart failure patients (FIGURE 4B). A titin-phosphorylation deficit in human heart failure was confirmed in another study and shown to affect the N2B isoform more than N2BA (5). Mechanical experiments on skinned cardiomyocytes suggested that increased titin-based passive stiffness in heart failure can be reduced to near-normal values by PKA or PKG administration; the effect was more obvious with cardiomyocytes from DHF than SHF patients (5, 105). Whether titin phosphorylation is altered also in human DHF still needs to be unequivocally shown. If a phosphorylation deficit at the N2-Bus existed in DHF, one would expect this defect to stiffen the affected hearts and impair diastolic function. It will be intriguing to test whether pharmacological interventions affecting beta-adrenergic signaling or, perhaps more promising, cGMP-PKG signaling (e.g., through PDE5 inhibition or activation of natriuretic peptides) can benefit diastolic function especially in DHF through correction of a titin-phosphorylation deficit.
Two other sites in the titin spring are phosphorylated by PKCα in mouse and porcine hearts, specifically, serines S26 and S170 of the constitutively expressed PEVK domain encoded by titin exons 219–225 (31) (FIGURE 3). (Note that the PKCα-dependent phosphorylation of titin can take place in skeletal myocytes too.) Unlike phosphorylation of serine 469 in the human cardiac N2-Bus, phosphorylation of the PEVK sites increased passive cardiomyocyte stiffness (FIGURE 4C). The two phosphorylatable serines in the PEVK-domain are evolutionary well conserved, and it will be interesting to test whether phosphorylation at these sites may be altered in heart disease. Since PKCα activity is increased in heart failure (97), increased phosphorylation of the PEVK-phosphosites could contribute to increased passive diastolic stiffness in failing myocardium, in SHF and DHF. Obviously, one needs to sort out to which degree increased titin phosphorylation at the PEVK domain (FIGURE 4C) or decreased titin phosphorylation at the N2-Bus (FIGURE 4B), or both, contribute to elevated diastolic stiffness in heart failure. This task is challenging given the huge size of the titin molecule but could be accomplished, e.g., by generating phospho-specific antibodies against the respective phosphosites to be tested on heart tissue samples.
Possible Impact of Oxidative Stress on Cardiac Titin Stiffness
Metabolic risk-related oxidative stress impairs diastolic LV function (106). Interestingly, oxidative stress-induced formation of disulfide bridges within the cardiac N2-Bus reduces the contour length of the N2-Bus, stiffens the whole titin molecule, and increases myofibrillar passive tension (29) (FIGURE 4D). Hence, redox-dependent modifications of titin could be an important contributor to diastolic LV dysfunction. Since oxidative modification of titin increases titin stiffness and since elevated titin stiffness could modify gene expression through altered stress signaling at the N2-Bus (96), oxidative modification of titin might be an early event in myocardial hypertrophy.
The giant muscle protein titin has well established roles in the myocyte, which include passive force generation, scaffolding, and signal transduction. The molecular mechanisms of titin elasticity have been elucidated in detail, suggesting titin is mainly an entropic spring, but the issue of titin Ig-domain unfolding in sarcomeres remains unresolved. Emerging evidence links titin to fundamental signaling pathways, such as those regulating protein quality control, hypertrophic gene expression, and stress sensing. Titin can thus be viewed as a crucial integrating element at the crossroads of myocyte signaling. The mechanical and mechano-signaling functions of the titin springs are variably tuned in health and disease, particularly in the heart by altering passive stiffness through titin-isoform switching, phosphorylation, and oxidative stress-dependent intramolecular cross-linking (FIGURE 4). The titin filaments represent dynamic components of the sarcomeric cytoskeleton adjusting myofibrillar performance to the prevailing conditions in an ever-changing cellular environment.
This work was supported by grant Li 690/7-2 (KFO 155) from the German Research Foundation.
- ©2010 Int. Union Physiol. Sci./Am. Physiol. Soc.