Titin is a giant filamentous elastic protein that spans from the Z-disk to M-band regions of the sarcomere. The I-band region of titin is extensible and develops passive force in stretched sarcomeres. This force has been implicated as a factor involved in regulating cardiac contraction. To better understand the adaptation in the extensible region of titin, we report the sequence and annotation of the chicken and mouse titin genes and compare them to the human titin gene. Our results reveal a high degree of conservation within the genomic region encoding the A-band segment of titin, consistent with the structural similarity of vertebrate A-bands. In contrast, the genomic region encoding the Z-disk and I-band segments is highly divergent. This is most prominent within the central I-band segment, where chicken titin has fewer but larger PEVK exons (up to 1,992 bp). Furthermore, in mouse titin we found two LINE repeats that are inserted in the Z-disk and I-band regions, the regions that account for most of the splice isoform diversity. Transcript studies show that a group of 55 I-band exons is differentially expressed in chicken titin. Consistent with a large degree of titin isoform plasticity and variation in PEVK content, chicken skeletal titins range in size from ∼3,000 to ∼3,700 kDa and vary greatly in passive mechanical properties. Low-angle X-ray diffraction experiments reveal significant differences in myofilament lattice spacing that correlate with titin isoform expression. We conclude that titin splice diversity regulates structure and biomechanics of the sarcomere.
- comparative genomics
- isoform expression
- exon-intron structure
titin [also known as connectin (46)] constitutes the third myofilament of muscle, with a single titin polypeptide spanning the full distance from the Z-disk to M-band regions of the striated-muscle sarcomere. Titin's NH2-terminal region is embedded in the Z-disk and contains binding sites for proteins that are likely to play roles in Z-disk assembly and signaling (1, 23, 29, 52, 53, 58, 74). The COOH-terminal ∼2,000 kDa of titin is integrated into the A-band lattice of the sarcomere. cDNA sequencing studies have demonstrated that titin's A-band region is composed of regular arrays of immunoglobulin (Ig) and fibronectin type 3 (Fn3) modules that form so-called super repeats (35, 73); based on the correlation of these super repeats with thick-filament structure and their provision of a regular series of binding sites for myosin and myosin binding protein C, it has been suggested that the A-band portion of titin is a molecular ruler that regulates the assembly and length of the thick filament (59, 70). Titin's ∼200-kDa COOH-terminal region is an integral part of the M-band region. Similar to titin's NH2 terminus, the M-line region of titin has been proposed to perform structural roles and has been implicated in signal transduction through its kinase domain (19, 39, 47, 55). In conclusion, the NH2-terminal ∼100 kDa and COOH-terminal ∼2,000 kDa of the titin filament are integral parts of the sarcomeric Z-disk and A-band lattices.
In contrast to titin's Z-disk and A-band regions, titin's I-band region extends as the sarcomere is stretched (immunoelectron microscopy shows that on sarcomere stretch, epitopes in this region of the molecule increase their distance from both the Z-disk and A-band), and in doing so titin develops passive force (22, 43, 62). An important determinant of the passive force level is the PEVK sequence (36), so named because ∼70% of its residues are comprised of proline (P), glutamate (E), valine (V), and lysine (K). Passive force is responsible for restoring sarcomere length (SL) when a stretched muscle is released, maintaining the central location of the A-band in the sarcomere and determining the filling characteristics of the heart (21, 22, 31, 72). Thus titin plays a number of important roles in striated muscle structure and function.
The myocardium expresses titin isoforms that are generated by distinct splice pathways (2). Fetal and neonatal myocardium express a highly compliant fetal cardiac titin (28, 38, 54, 66), and the adult myocardium expresses the stiff N2B isoform and the more compliant N2BA isoform (4, 10, 51). To understand their contribution to the diastolic properties of the myocardium, several studies have related species-specific differences in titin N2B and N2BA isoform expression to altered mechanical properties of the heart (4, 51). Importantly, these isoforms are coexpressed at the level of the half-sarcomere, and the passive properties of the sarcomere can be tuned by varying their expression ratios (60, 63). Adaptations in splicing occur during various heart disease states in animal models as well as in humans (for recent reviews, see Refs. 24 and 41).
Several lines of evidence indicate that titin not only develops passive force but also regulates active force generation, possibly via an effect of titin on thick-filament structure or via modulation of interfilament lattice spacing (6, 7, 12–15, 30). The reduction of interfilament lattice spacing (the distance between thick and thin filaments) with an increase of SL is thought to contribute to the length-dependent changes in Ca2+ sensitivity in cardiac and skeletal muscle by increasing the likelihood of myosin attachment to actin and thereby increasing Ca2+ sensitivity (11, 45, 48). A role for titin in determining the lattice spacing and thereby modulating Ca2+ sensitivity has been suggested by recent low-angle X-ray diffraction work (for details, see Refs. 7 and 15).
As discussed above for the N2B and N2BA isoforms, differential expression of titin during muscle development as well as during disease results in titin isoforms with highly variable extensible regions. Until recently, the characterization of titin isoforms has focused on the analysis of cDNA sequences (2, 10, 36, 52, 53). With several genome projects nearing completion, additional titin sequences are now available. However, gaps are still present in the contigs, and the plethora of short unique exons, such as the PEVK exons, impose problems on automated annotation so that these exons are frequently missed. Hence, we have used the completed human titin cDNA sequence to identify genomic contigs in the mouse and chicken genomes. By closing the remaining gaps as well as identifying novel exons by homology comparison to the human sequence, we have annotated and analyzed the titin sequences encoding mouse and chicken titin. Large sections of the titin gene are highly conserved in these three species with regards to their primary structure and exon-intron organization. Comparison of human, chicken, and mouse genes revealed conserved charge patterns and coiled-coil potentials within titin's I-band region. However, a number of significant differences were also revealed that can generate a surprising diversity of titin isoforms that vary greatly in the PEVK region of the titin molecule. Sequence diversity in the transcript of two chicken muscles and the biomechanical and structural characterization of their fibers indicates that differential splicing of the extensible I-band region of titin is an important regulator of the structure and function of the striated muscle sarcomere.
MATERIALS AND METHODS
Sequencing and Annotation of Chicken and Mouse Genomic Titin
A sequence homology search using the human titin cDNA sequence (AJ277892) was used to derive the chicken and mouse genomic titin sequences from the Ensembl database (http://www.ensembl.org) and the Celera database, respectively. The mouse Celera sequence is a consensus of the strains 129X1/SvJ, DBA/2J, and A/J. The 10 gaps included in the mouse titin assembly were closed by genomic PCR by using 129SvJ DNA as a template. The sequence was verified by sequence comparison with the Mus musculus whole genome shotgun supercontig available at the NCBI mouse genome resources (NW_000176.1 Mm2_WIFeb01_27), which covers the complete titin locus. The chicken sequence (chromosome 7, bp 15,940,000 to 1,620,000) included 12 gaps, which were also closed by PCR and sequencing. The Artemis software package (http://www.sanger.ac.uk/Software/Artemis/) was used to enter exon-intron boundaries. To facilitate comparison between genomes, we used an exon-numbering system based on that of the human titin gene. The accession numbers are BN001113 and BN001114 for the chicken and mouse sequences, respectively.
We used 10- to 12-wk-old chickens that were killed and exsanguinated in accordance with National Institutes of Health and Washington State University Institutional Animal Care and Use Committee guidelines. Pectoralis and biventer muscles were dissected and either skinned (26) for use in mechanical studies (see Mechanics, Gel Electrophoresis, and X-ray Diffraction) or quick frozen for isolation of total RNA and conversion to biotinylated cDNA (38). We used a 50-mer oligonucleotide array containing Z-disk, I-band, and selected A-band exons of chicken titin and normalization, positive, and negative controls. Oligonucleotide probes were selected to obtain a melting temperature of 80 (±5), to obtain a GC content of 50% (±5), and to minimize dimer and hairpin formation. A biotin probe was used as a positive control for detection chemistries. A 5-bp mismatch for titin exon 304 (304MM) was included as a negative control.
Biotinylated target was mixed with 2× hybridization buffer (1 M NaH2PO4, 2 mM EDTA, 2% SDS, and 2% BSA, pH 4.8), heat denatured, and hybridized to the oligonucleotide array in a humidified chamber at 55°C for 18 h. Detection chemistries and signal amplification were achieved by using the TSA biotin system (PerkinElmer, Boston, MA). The fluorescence intensity was measured by using an arrayWoRxe auto biochip reader (Applied Precision, Issaquah, WA). The data were analyzed with a softWoRx tracker (Applied Precision) and GeneSifter (VizX Labs, Seattle, WA). All probe intensities were normalized to a constitutively expressed titin exon to account for variation in efficiency of conversion to biotinylated cDNA, variation in target loading, and overall upregulation of transcription. The mismatch probe was used for evaluation of array specificity. The median intensity within the spot area was determined and normalized to the median intensity value of the chosen constitutively expressed titin exon. Because each exon probe is printed in duplicate, the median intensity for each exon was averaged.
We performed three independent experiments per muscle type. Results from these experiments were then used to determine the mean ± SE for each exon. Exons identified as up- or downregulated had fivefold or greater change of their mean values. Results from different muscle types were compared with a two-tailed t-test for samples of unequal variance, and P < 0.05 was used as the criterion for statistical significance. For additional details, including the analyses, see Ref. 38. For RT-PCR studies, commercially available total RNAs (Stratagene, La Jolla, CA) were reverse-transcribed as described previously (37).
Mechanics, Gel Electrophoresis, and X-Ray Diffraction
Single fibers were dissected and then mechanically skinned in relaxing solution (for composition, see Ref. 72). Clips were attached to the ends of the fibers, and these were used to mount the fiber at one end to a force transducer (Akers 801E; Horton, Oslo, Norway) and at the other end to a high-speed servomotor (308B; Aurora Scientific, Richmond Hill, ON, Canada). Measured force was divided by the cross-sectional area of the fibers (determined from the width and thickness measured at the slack length and assuming an elliptical cross-sectional area) to obtain tension. Sarcomere length was measured online by using a laser diffraction system (26). While in the relaxing solution, fibers were stretched at a speed of 10% per second until a predetermined sarcomere length was reached, after which the fibers were released again. Preliminary experiments revealed that, to ensure reversible mechanical behavior, the maximal sarcomere length of pectoralis fibers had to be kept below ∼2.6 μm and those of the biventer muscle below ∼3.5 μm. The exact same stretch-release protocols were repeated after thick and thin filament had been extracted from the fibers with relaxing solution containing high concentrations of KCl and high KI, to remove titin's anchors in the sarcomere (see Refs. 64 and 72 for details). The reduction in passive tension was assumed to be derived from titin and the remaining tension from intermediate filaments (cf. Refs. 20, 26, 64, and 65). The intermediate filament-based tension was <10% of the titin-based tension. Because of our focus on titin, in this work we only report the titin-based tension.
For protein analysis, muscle samples were quick frozen, pulverized to a fine powder, rapidly solubilized, and then analyzed by SDS-agarose gel electrophoresis. The methods that were followed were as previously published (67); gels were stained with Coomassie blue.
The overall experimental arrangement for X-ray diffraction experiments has been described in detail previously (32). Briefly, experiments were performed on the BioCAT undulator-based beamline at the Advanced Photon Source, Argonne National Laboratory. The beam size at the sample position was collimated to ∼0.3 × 0.8 mm and at the detector was ∼0.040 × 1 mm (vertical × horizontal), and it contained a maximum incident flux of ∼3 × 1012 photons/s. Exposure times were 1 s. The small-angle camera had a 3 m sample-to-detector distance with a wavelength of 1.03 Å. For the X-ray studies, chemically skinned fiber bundles (consisting of ∼20 fibers) were mounted between a force transducer and a servomotor in a small trough that allowed simultaneous collection of the X-ray patterns and viewing of the striation pattern by using a long-working-distance objective of an inverted microscope equipped with a charge-coupled device (CCD) video camera. During the experiment, relaxing solution was continuously pumped through the chamber by using a peristaltic pump. Low-angle X-ray diffraction patterns were collected on a CCD-based X-ray detector. The intensities of the 1,0 and 1,1 reflections were determined from one-dimensional projections of reflection intensities along the equator, and the spacings between the reflections were measured by using the program FIT2D (32); spacings between the 1,0 equatorial reflections in the diffraction pattern were converted to d10 lattice spacings (32). Sarcomere lengths were determined from the video image as described previously (32) and were checked both before and immediately after the X-ray exposure.
Exon-Intron Organization of the Mouse and Chicken Titin Genes
We identified genomic titin sequences in the Celera and Sanger public data libraries by BLAST searches with the 5′ 1-kb human titin cDNA sequences. For mouse, this identified a 0.3 Mb contig from chromosome 2 (in a region where in situ hybridizations have previously localized the mouse titin gene; see Ref. 56). This mouse contig is likely to contain the complete titin gene, because both titin's 5′-end and 3′-end sequences match with ∼90% identity to the human contig. Inspection of the contig revealed 10 gaps. We determined the sequence of these gaps by PCR amplification using primers to flanking fragments. Similarly, BLAST searches of the chicken genome database (available on http://www.ensembl.org) identified a 0.25-Mb contig on chromosome 7 that is likely to span the entire chicken titin gene. A total of 12 gaps were closed by PCR, and this provided an additional 6 kb of genomic data.
We annotated both the mouse and chicken titin genes by comparison with the human titin gene (AJ277892) and by visual inspection of the translated sequence in Artemis. The comparison of the exon-intron organization is shown in Fig. 1. Although introns are often of different lengths and in general tend to be shorter in the chicken gene, the exon-intron organization is remarkably similar within the gene segment from exon 225 (end of PEVK region) up to exon 358 (titin kinase-encoding exon). For example, exon 326 and its boundaries coding for the largest known exon (16 kb) are conserved in all three species. [Note that since different exon numbers are present at the 5′- and 3′-ends of the 225–358 exon region, the numbering of exons in this region of the chicken and mouse gene is based on the human titin gene (h225–h358).] This high degree of conservation of exon-intron boundaries suggests that associated functions of this region are also conserved. Unlike exons h225–h358, the exon organization in segment h112–h224, coding for the PEVK region, is more diverse. Because of the short lengths of many PEVK exons (∼80 bp), analysis of this region is challenging and PEVK exons are generally missed by automatic annotation programs. Hence, we performed annotation of the PEVK region manually, by inspecting potential frames for their PEVK content (mouse), or PA + ED + VIL + KR content (chicken; see also Sequence composition of PEVK segment). Our results indicate that although human titin codes for 114 PEVK exons, 97 PEVK exons are present in mouse and 83 in chicken. In summary, the human, chicken, and mouse titin genes are highly conserved in their A/I junction-A-band encoding segment, spanning from h225 to h358. In contrast, titins have significantly diverged during evolution within their PEVK-encoding segment h112–h224. This suggests high structural and functional constraints on A/I junction and A-band titin but fewer constrains on PEVK titin.
Comparison of Human, Chicken, and Mouse I-Band Sequences
Sequence composition of PEVK segment.
Most of the PEVK segment of the human titin is organized into small exons that encode ∼28-residue PEVK repeats (2), also referred to as PPAK repeats when using a different phasing for alignment (27). Similarly, the mammalian PEVK sequences from dog, rat, and rabbit also contain PEVK/PPAK repeats (27). The now-available chicken titin sequence allows an evolutionary more distant vertebrate titin to be analyzed. Chicken titin has fewer (83 vs. 97 in mouse and 114 in human) but larger (up to 1,992 bp) PEVK exons (Fig. 2A). The PEVK coding potential in chicken, mouse, and human is 3,764, 3,504, and 3,303 amino acid residues, respectively. The chicken PEVK segment also contains many small exons (49 out of 83 encode between 25 and 30 residues), 14 of which end in a PPAK sequence. The overall primary structure of the PEVK is different from mammalian titins. Whereas for mammalians the PEVK segment can be defined as the region of titin with a ∼70% content of P + E + V + K, in chicken titin this value is 59%. Broadening this definition to content of PA (alanine) + ED (aspartic acid) + VI (isoleucine) L (leucine) + KR (arginine) increases the percentage to 77%.
We recently reported that a subset of exons in human titin are E rich and that they are conserved in human and mouse (34). We studied whether the same is true in the chicken PEVK segment. We found that in chicken the E content of exons varies between ∼10 and ∼50%, with 10 exons containing >30% glutamates (Fig. 2B). It is interesting to note that the distribution of E-rich exons is relatively uniform along the PEVK segment in chicken but that in the mouse and human they are segregated toward the ends, due to the insertion of E-poor exons in the middle of the segment (see Fig. 2B). This suggests that in mammals the PEVK segment has nonuniform properties (as implied by Ref. 50) and that in chicken this is not the case. Furthermore, by splicing in of PEVK exon groups with either many or few E-rich exons, titin isoforms with distinct mechanical properties can be obtained in mammals, providing an increased level of mechanical tuning.
Coiled-coil potentials in novex III and N2A exons.
The previous sequence analysis of the human titin gene has resulted in the discovery of three novel I-band exons (called novex I–III) that were missed during earlier cDNA-sequencing studies (these exons are found in low-abundant titin isoforms; see Ref. 2). Analysis of the mouse and chicken titin genes reveals that these species also contain novex I–III exons (see Fig. 1). Interestingly, the novex III exon codes for a sequence predicted to contain segments that form coiled-coil multimers (Fig. 2C). We also analyzed the entire I-band sequences from human, mouse, and chicken titin for the presence of coiled-coil potentials. This indicated that in addition to the coiled-coil potential of novex III sequences, the N2A segment of the conventional titins is also predicted to have regions with coiled-coil potential. This feature is found in all three species (Fig. 2D). We also used the MultiCoils program (71) to determine the predicted probability of forming dimers vs. trimers. This showed that dimer formation has the highest probability for both the novex III and N2A sequences.
LINE Repeats and Differential Splicing
A possible mechanism causing variation in the number of PEVK exons is based on duplication and/or deletion of exon groups. During the analysis of the human titin gene, it was previously noted that a group of seven PEVK exons appeared to have been triplicated (2). We postulated that this triplication was caused by a LINE-repeat insertion that was present between the repeated segments (2). Therefore, we analyzed the PEVK/LINE repeat region in mouse and chicken titin and found that in both species the LINE repeats as well as the flanking duplicated PEVK repeats are missing. This species comparison supports the idea that during evolution LINE-repeat insertions in the PEVK region duplicated whole groups of exons, thereby modifying the biomechanical properties of titin.
We examined all additional introns in the mouse titin gene for LINE repeats. This identified three LINE repeats located between exons 2 and 3, 13 and 14, and 50 and 51 (Fig. 1, gray boxes marked by L). Recently, we reported that titins expressed in striated and smooth muscles have an alternative 5′-end, as a result of differential splicing to exon 2 (37), i.e., close to one of the LINE-repeat insertions in mouse titin. We investigated by RT-PCR whether the LINE repeat insertions between exons 13 and 14 and between 50 and 51 in mice correlate with differential splicing. This indeed demonstrated in both cases a multitude of differential splice events (Fig. 3). Sequencing of amplified fragments suggests that joining of exons 10 and 11 and of 50 and 51 is incomplete and that therefore multiple upstream and downstream alternative acceptors can be used (for example, joining exons 10 and 14 or 50 and 72).
Exons 8–14 code for titin Z repeats (18, 58), whereas exons 50–72 locate within the central I-band region. To test if the expression/splicing of different-length versions in different titin regions are correlated (i.e., whether a large number of Z-disk exons is accompanied by a large number of I-band exons), we compared two different tissues that are known to express a long I-band titin version (quadriceps skeletal muscle) and a short I-band version (cardiac titin). Results indicate that cardiac titin expresses a larger number of Z-disk exons than quadriceps muscle (Fig. 3). indicating that differential splicing cascades are not positively correlated.
Impact of Differential Splicing on Biomechanics and Structure
We have recently adopted an array approach that allows surveying of a large number of exons in a single experiment (37, 38). For the present work, we made a chicken titin exon microarray that includes all Z-disk and I-band exons as well as representative A-band exons. In preliminary protein work, we had noted large differences in the mobility of titin from pectoralis (breast) and biventer cervicis (neck) muscles, and our microarray analysis therefore focused on these two muscle types (Fig. 4). We found that 57 exons were greater than fivefold upregulated (P < 0.01) in the chicken biventer transcript. Of these 57 upregulated exons, all except 2 (Z-disk exons 8 and 9) code for Ig-like domains and PEVK exons in titin's extensible region. Among these exons, three E-rich PEVK exons were found. We conclude that differential splicing accounts for the extensive diversity within the extensible region of chicken titin. We also performed protein gel electrophoresis (Fig. 5A). This showed that titin of chicken biventer comigrates with human soleus muscle (Mr ∼ 3.7 M), whereas pectoralis chicken titin comigrates with cardiac titin (estimated Mr ∼ 3.0 M). The much smaller titin expressed by pectoralis compared with biventer is consistent with the microarray study.
To gain insights into the functional significance of the different titin sizes, we measured passive tension behavior and performed low-angle X-ray diffraction studies. The passive tension data obtained on single skinned muscle fibers are shown in Fig. 5B. Results indicate that biventer fibers are considerably more compliant than pectoralis muscle fibers. For example at a SL of 2.4 μm, the biventer fibers develop 2.6 mN/mm2 and the pectoralis develop 21.3 mN/mm2. These differences are consistent with the much longer extensible region (due to the additional Ig and PEVK domains) of the biventer titin. A longer extensible region results at a given sarcomere length in a reduced fractional extension of the extensible region of titin and hence a reduced passive tension level (for details, see Refs. 20, 33, and 43).
Finally, we performed low-angle X-ray diffraction experiments to establish whether the different passive tension levels of the biventer and pectoralis fibers result in a difference in myofilament lattice spacing. Fibers were mounted in a small trough with windows for collection of X-ray patterns and viewing of striations to measure SL. Representative X-ray patterns are shown at the top of Fig. 5C. The separations of the intensity maxima of the 1,0 and 1,1 equatorial reflections in the diffraction patterns were measured and, by using Bragg's law, converted to d1,0 values (d1,0 is related to the thick-thin filament spacing as d1,0 × 2/3; see Ref. 49). We found that d1,0 lattice spacings are significantly less in pectoralis fibers than biventer fibers (Fig. 5C). If a linear relationship is assumed, the slope of the pectoralis data set is significantly steeper (P < 0.05). Below we discuss the significance of these findings.
Searching the genomic segments that code for titin-like cDNA sequences in the mouse and chicken genomes identified in both species only one genomic region that matches the set of human titin cDNAs with high homology: a 0.3-Mb region from the mouse chromosome 2 and a 0.25-Mb region from the chicken chromosome 7, respectively. The mechanisms that generate a diverse family of titin isoforms from a single locus include usage of different promoters and initiation sites in a tissue-specific fashion (37), internal termination and polyadenylation (2), and differential splicing (e.g., during development; see Ref. 38). Interestingly, this is reminiscent of the sls/d-titin locus in Drosophila melanogaster, where a coordinated set of differential initiation and termination sites, as well as differential splicing cascades, expresses the three different proteins kettin (with nebulin-related functions), zormin, and ∼2,000-kDa titin-like protein (3). We conclude that during evolution of birds and mammals, extensive variety in titin isoforms, as required for the diverse biomechanical properties of muscles, has not been achieved by generating multiple genes but rather from remodeling a single gene.
We found that segments of both the novex III and the N2A elements are predicted to have coiled-coil potential in all three examined species (Fig. 2, C and D), indicating that this is a conserved feature of titin. It is interesting to note that a titin-like protein in Caenorhabditis elegans also contains domains with coiled-coil potential (9) and thus that the ability to form multimers might be a basic property of the titin family of giant proteins. The novex III exon is found at the COOH terminus of novex-3 titin, a truncated titin isoform (Z1–Z2 to novex III) found at low levels in striated muscle (2). The NH2 terminus of novex-3 titin integrates into the Z-disk, but its COOH terminus is too short to reach the A-band (2). The function of novex-3 titin is not well understood. It has been speculated that coexpression of novex-3 titin and full-length titins may adjust the titin filament system to both three- and twofold symmetries of thick and thin filaments, respectively (2). Furthermore, novex-3 titin interacts with obscurin, another large protein (∼800 kDa) with multiple signaling domains, and it has been suggested that together novex-3 titin and obscurin function as an elastic signaling complex (2). The finding of coiled-coil potential suggests that the COOH terminus of novex-3 titin might multimerize, and this could be an important aspect of how novex-3 titin is integrated into the I-band lattice and possibly how it interacts with obscurin. The second site with coiled-coil potential, the N2A element, is found in the central I-band region of full-length titin molecules. This location coincides with the so-called junction line that was previously described by Funatsu and colleagues (16), who provided ultrastructural evidence that the junction line is a site where titin filaments interact. We propose that the junction line is due to multimerization of titin molecules that results from the coiled-coil potential of the N2A element.
The extensible behavior of the PEVK segment has been measured in single-molecule studies, which have shown that it has a low persistence length (measure of the bending rigidity) and that it can be extended to near its maximal contour length without evidence for structural transitions (40, 42, 50, 57, 68, 69). This has resulted in the notion that the PEVK segment behaves as an unfolded polypeptide (random coil), consistent with the preponderance of proline residues and charge clusters along the PEVK sequence that are likely to prevent the formation of stable structures (36). Recent circular dichroism studies support that the PEVK can be classified as intrinsically disordered (8). Although the chicken PEVK segment is more complex and is rich in PA + ED + VIL + KR sequences, we consider it nevertheless likely that it behaves mechanically similar to the mammalian PEVK segment (i.e., as a random coil with low persistence length). Whether this concept is correct remains to be experimentally tested. It is also important to note that E-rich PEVK exons are found in all three species, suggesting that their functions are conserved. Previous single-molecule studies have shown that the E-rich exons of human are responsible for lowering the persistence length in the presence of calcium (34), suggesting that this function is conserved in chicken and mammals (mouse and human).
The studies on chicken biventer and pectoralis skeletal muscles strengthen the concept that differential splicing of spring elements forms the molecular basis for tuning passive biomechanics. The findings demonstrate that titin isoforms expressed in these muscle types have extremely different I-band titin compositions with a group of ∼40 PEVK that are differentially expressed (Fig 4). Thus, in addition to N2B and N2BA titins (accounting for the stiffness diversity of cardiac titins in mammals), skeletal tissues can also have a highly variable PEVK composition. Consistent with this, we find that fibers from biventer and pectoralis muscles have extremely different passive tension levels (Fig. 5B).
Analysis of the available data provides several lines of evidence that the PEVK regions have been remodeled during recent evolution by the insertion of LINE repeats. In humans, two LINE repeats are associated with a triplicated PEVK region of 21 highly similar PEVK exons (2). This segment is missing in the mouse titin gene. A LINE repeat insertion has been identified in the titin I-band region of the mdm mouse mutant that affects PEVK splice patterns (17). In addition to modifying the gene structure, LINE-repeat insertions are also associated with differential splicing, presumably by modifying local pre-mRNA structures, causing partial blockage of adjacent splice donor/acceptors. Our RT-PCR studies suggest an association of LINE-repeat insertions with differential splicing: the three LINE repeats integrated into the wild-type mouse titin gene localize between, or adjacent to, differentially spliced exons (Fig. 3). To test whether the expression/splicing of different-length versions in different titin regions are correlated, we compared two different tissues that are known to express a long I-band titin version (quadriceps skeletal muscle) and a short I-band version (cardiac titin). Results indicate that cardiac titin expresses a larger number of Z-disk exons than quadriceps muscle, suggesting that differential splicing cascades are not positively correlated. Together, our data suggest that evolutionary adaptations of myofibrillar stiffness may have occurred via LINE-repeat insertions that can cause PEVK exon duplications or changes in differential splicing. During evolution, LINE-repeat insertions may have remodeled titin more rapidly than might be achieved by single mutations in, for example, a splice acceptor or donor site.
We found not only that passive tension varies but also that myofilament lattice spacing does, with a much larger spacing in biventer than pectoralis muscle (Fig. 5C). The layout of titin in the sarcomere suggests a possible mechanism for this finding. The segment of titin near the Z-disk binds strongly to the thin filament (61), and in the A-band titin attaches to the thick filament (62). Thus the elastic region of titin runs obliquely to the thin and thick filaments, and titin is expected to develop a longitudinal force (FL) and radial force (Fr), the latter of which compresses the lattice. This compressive force (as well as the longitudinal force) will be highest in the pectoralis fibers (because of titin's shorter extensible region in this fiber type), and this could underlie the lower lattice spacings of pectoralis fibers. Other mechanisms that play a role in determining lattice spacing may be involved as well. It has been proposed that the interfilament lattice spacing is one of the main determinants of the probability of actomyosin interaction at a given calcium concentration (11, 48). If so, it is possible that the effect of titin-based tension on lattice spacing affects the length dependence of active force development. Consistent with this idea, passive tension due to titin has been reported to increase the length dependence of calcium sensitivity of cardiac myocytes and skinned myocardium (5–7, 14, 15). In addition, it is possible that titin influences active force via an earlier proposed mechanism (12, 15, 25) in which the likelihood of cross-bridge interaction is enhanced by passive force-induced thick-filament strain. Further work is required to establish and distinguish between these possible mechanisms. Regardless of which mechanism is operative, one would predict that the stiffer titin isoforms in the pectoralis would show greater length-dependent activation, i.e., changes in calcium sensitivity with stretch, than the more compliant biventer.
In summary, our comparison of the sequence diversity in selected chicken muscles and their biomechanical and structural characterization indicates that differential splicing of the extensible I-band region of titin is an important regulator of the structure and function of the striated muscle sarcomere. Our data suggest an important mechanism for evolutionary adaptation of myofibrillar stiffness based on exon duplication and exon skipping in the PEVK region of the titin gene. The complete mouse and chicken titin gene sequences will be highly valuable tools for future molecular titin studies.
This work was supported by grants from the National Heart, Lung, and Blood Institute (HL-69008 and HL-61497), the Alexander von Humboldt Foundation, the Deutsche Forschungsgemeinschaft (La668/9-1), and the American Heart Association. Henk Granzier was supported by a Helmholtz-Humboldt-Fellowship from the Alexander von Humboldt Foundation.
We are grateful to Dr. Sumiko Kimura for communication of partial chicken connectin/titin cDNA data before publication. We acknowledge Honghui Zhou for excellent microarray work and Beate Goldbrich and Xiuju Luo for expert technical assistance.
Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Basic Energy Sciences, Office of Energy Research, under Contract No. W-31-109-ENG-38. BioCAT is a U.S. National Institutes of Health-Supported Research Center RR08630.
Present address for Y. Wu: Department of Internal Medicine, University of Iowa, Iowa City, IA 52242.
- Copyright © 2007 the American Physiological Society