Identification of a Novel Tropomodulin Isoform, Skeletal Tropomodulin, That Caps Actin Filament Pointed Ends in Fast Skeletal Muscle*

Tropomodulin (E-Tmod) is an actin filament pointed end capping protein that maintains the length of the sarcomeric actin filaments in striated muscle. Here, we describe the identification and characterization of a novel tropomodulin isoform, skeletal tropomodulin (Sk-Tmod) from chickens. Sk-Tmod is 62% identical in amino acid sequence to the previously described chicken E-Tmod and is the product of a different gene. Sk-Tmod isoform sequences are highly conserved across vertebrates and constitute an independent group in the tropomodulin family. In vitro, chicken Sk-Tmod caps actin and tropomyosin-actin filament pointed ends to the same extent as does chicken E-Tmod. However, E- and Sk-Tmods differ in their tissue distribution; Sk-Tmod predominates in fast skeletal muscle fibers, lens, and erythrocytes, while E-Tmod is found in heart and slow skeletal muscle fibers. Additionally, their expression is developmentally regulated during chicken breast muscle differentiation with Sk-Tmod replacing E-Tmod after hatching. Finally, in skeletal muscle fibers that coexpress both Sk- and E-Tmod, they are recruited to different actin filament-containing cytoskeletal structures within the cell: myofibrils and costameres, respectively. All together, these observations support the hypothesis that vertebrates have acquired different tropomodulin isoforms that play distinct roles in vivo.

One of the most spectacular examples of cytoskeletal architecture found in cells is the sarcomere, the contractile unit of striated muscle. Antiparallel polarized actin filaments (thin filaments), bipolar myosin filaments (thick filaments), titin, and nebulin filaments interdigitate and interact with each other to create sarcomeres, which support the contraction of muscle cells (1). Despite the similarity in the basic organization of the filaments in all sarcomeres, a closer examination reveals some important and interesting variations in sarcomeric archi-tecture in different types of muscle. Most strikingly, while thick filament lengths are remarkably constant among different vertebrate muscles (2)(3)(4), thin filament lengths exhibit considerable variability among physiologically distinct types of muscles (for review, see Ref. 5). For example, in the rabbit psoas muscle, a fast twitch skeletal muscle, thin filament lengths fall into a narrow range of 1.11 Ϯ 0.03 m (6). In contrast, in rat cardiac atrial muscle, thin filament lengths vary from 0.6 to 1.1 m (7). Differences in thin filament length between fast and slow skeletal muscles have also been observed in the fish Perca fluviaris (8). This variation in thin filament length can be an important determinant of the length-tension relationship during contraction and thus can influence the physiological properties of the muscles (i.e. cardiac, fast skeletal, and slow skeletal) (7,8). Nevertheless, the mechanisms underlying the regulation of variations in thin filament lengths are not yet understood.
An excellent candidate for regulating the length of the thin filaments is tropomodulin, the capping protein for the thin filament pointed ends (5,9). When embryonic chick cardiac myocytes are microinjected with an antibody that blocks tropomodulin's ability to cap actin filament pointed ends in vitro, actin elongates from the thin filament pointed ends (10). Furthermore, the cells are no longer able to beat, demonstrating that maintenance of thin filament length by tropomodulin is critical for normal contraction (10). The levels of tropomodulin also appear to be important for the stabilization and proper organization of thin filaments in myofibrils, based on results from studies employing sense and antisense erythrocyte tropomodulin (E-Tmod) 1 mRNAs in adenovirus vectors in neonatal rat cardiac myocytes (11). In transgenic mice, overexpression of tropomodulin in the heart after birth results in dilated cardiomyopathy, presumably due to alterations in normal myofibril organization (12).
Many myofibrillar components exist as families of multiple isoforms, which are differentially expressed in distinct types of muscle (for review, see Ref. 13). The variable expression of protein isoforms among muscles of different types is a major determinant of the contractile properties of each muscle. Our previous studies and those of others had identified so far only one tropomodulin isoform, E-Tmod, which is expressed in both skeletal and cardiac muscle (14 -17). In this study, we have identified and cloned a novel tropomodulin isoform from vertebrates, Sk-Tmod, which is highly expressed in fast skeletal muscle and is not present in slow skeletal or cardiac muscle.
The high conservation of Sk-Tmod sequences among vertebrates and its different tissue and subcellular distribution as compared with E-Tmod suggest that the Sk-Tmod isoform may be functionally distinct from the E-Tmod isoform.

EXPERIMENTAL PROCEDURES
Antibodies-A panel of 19 monoclonal anti-tropomodulin antibodies were generated using recombinant chicken E-Tmod as an antigen (16) (Chris W. Grant, Custom Monoclonals, West Sacramento, CA). The antibodies were purified from the hybridoma culture supernatants over a Protein G-Sepharose fast performance liquid chromatography column (Amersham Pharmacia Biotech, Uppsala, Sweden) and eluted with 0.2 M glycine-HCl, pH 2.8. The specificity of the antibodies was characterized by two-dimensional gel electrophoresis, followed by Western blotting (see below) and by immunofluorescence. Monoclonal antibodies (mAbs) 17, 23, and 47 recognize a single tropomodulin spot by Western blots in both heart and breast muscle (pectoralis major, PM) ( Fig. 1, left  panel). In contrast, mAbs 9 and 95 only recognize a tropomodulin spot in heart and none in breast muscle (PM) (Fig. 1, right panel). Mixing experiments demonstrate that tropomodulin spots from heart and breast muscle (PM) do not comigrate (Fig. 1, bottom left panel). By immunofluorescence, mAb 95 specifically recognizes E-Tmod, as demonstrated by staining of embryonic breast muscle myofibrils, which contain E-Tmod (17), and its inability to stain adult chicken breast muscle myofibrils, which contain Sk-Tmod (data not shown and see below). In contrast, although mAb 17 recognizes both E-and Sk-Tmod isoforms on Western blots, it preferentially recognizes Sk-Tmod by immunofluorescence since it does not stain E-Tmod in the myofibrils of embryonic breast muscle but stains the pointed ends of thin filaments in isolated myofibrils from adult chicken breast muscle, which contain Sk-Tmod (data not shown and see below).
Polyclonal antibodies to recombinant chicken Sk-Tmod (see below) were generated in rabbits (R3577) and affinity-purified over a column of recombinant chicken Sk-Tmod coupled to cyanogen bromide-activated Sepharose 4B (Sigma) by standard procedures. The monoclonal antisarcomeric ␣-actinin antibody (clone 9A2B8) (18) was generously provided by Drs. S. and H. Holtzer (University of Pennsylvania, Philadelphia, PA). The monoclonal anti-chicken fast C-protein antibody (clone MF1) (19) developed by Dr. D. A. Fischman was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. Polyclonal rabbit antibodies generated against bovine brain ␣-spectrin were prepared as described previously (21). Rabbit anti-mouse IgG ϩ IgM was purchased from Pierce. Bodipy-labeled phallacidin, tetramethylrhodamine B isothiocyanate-conjugated donkey anti-mouse IgG, and rhodamine-conjugated goat anti-rabbit antibodies were purchased from Molecular Probes, Inc. (Eugene, OR), Accurate Chemical and Scientific Corp. (Westbury, NY), and Roche Molecular Biochemicals, respectively.
Comparison of Tropomodulin Isoforms by Two-dimensional Electrophoresis-Breast muscle (PM) from 12-day chicken embryos and breast, heart biceps femoris (BF), and posterior latissimus dorsi muscle from adult chickens (about 2 months old) were dissected and washed with ice-cold phosphate-buffered saline plus protease inhibitors (17). Muscles were minced and then homogenized in a Polytron homogenizer with 10 volumes of 9.2 M urea and 1% (v/v) 2-␤-mercaptoethanol, and their tropomodulin isoform composition was characterized by two-dimensional gel electrophoresis followed by Western blotting (17). The migration position of muscle ␣-actin (stained with Ponceau S) was used as a positional marker to compare the mobility of tropomodulin isoforms electrophoresed on different gels.
For comparison of tropomodulins from adult chicken PM, lens, and erythrocyte membranes, proteins were solubilized in 9.2 M urea, 1% 2-␤-mercaptoethanol, and 2% Nonidet P-40 to improve the resolution of the first dimension. Membranes were prepared from chicken erythrocytes according to Granger (22). Tropomodulin was immunoprecipitated from adult chicken lens extracts using rabbit polyclonal antibodies (R745) (20), as described previously for rat lenses (23). Prior to twodimensional gel electrophoresis, purified rabbit skeletal muscle ␣-actin was added to the immunoprecipitated lens tropomodulin and the erythrocyte membrane samples so that the mobility of tropomodulin in the erythrocytes and lens could be compared with that of the other muscle tissues.
Isolation of Chicken Sk-Tmod cDNAs-Tropomodulin from adult chicken breast muscle was immunoprecipitated as described (24) using anti-tropomodulin mAb 17. Samples were electrophoresed on 7.5-15% acrylamide linear gradient SDS-polyacrylamide gels at pH 9.1 (20,25), and polypeptides were transferred to polyvinylidene difluoride membrane (0.45 m; Millipore Corp., Bedford, MA) in CAPS buffer (10 mM CAPS, pH 11, 10% methanol). The membrane was stained briefly in 0.1% Ponceau S in 1% acetic acid and the region of the membrane containing the tropomodulin band was excised and subjected to tryptic digestion and microsequencing. The sequences of four peptides were obtained (Dr. John Leszyk, University of Massachusetts Medical School, Shrewsbury, MA) (underlined sequences in Fig. 2).
Degenerate primers to amplify Sk-Tmod cDNA were designed based on the peptides TLQSLNIESNFITSAGMMSVIK and YKPVPDEP-PNPTNVEETLR (see below, Fig. 2). These primers were used to amplify a 300-bp cDNA fragment by polymerase chain reaction using Taq polymerase (Promega, Madison, WI) from a Zap adult chicken skeletal muscle cDNA library (Stratagene, La Jolla, CA). This fragment was labeled with [␣-32 P]dCTP using a Random Primer kit (Stratagene) and then used to screen the same library. Approximately 1 ϫ 10 6 colonies were screened, and phage containing putative Sk-Tmod cDNAs were isolated following the manufacturer's instructions. Clones were analyzed by restriction mapping, and selected clones were sequenced to identify overlapping fragments from the same cDNA sequence. Both strands from two of the longest clones (1.3 kb) were sequenced.
Identification of Additional Sk-Tmod Sequences-The available tropomodulin sequences at GenBank together with the new chicken Sk-Tmod cDNA sequence reported in this study, were used to screen the expressed sequence tag data base (dbEST). A number of overlapping ESTs from human, mouse, and zebrafish were identified that showed higher levels of sequence similarity to the chicken Sk-Tmod than to any other tropomodulin sequences. To determine the cDNA sequences of clones corresponding to these ESTs, several cDNA clones were obtained from the IMAGE Consortium clone collection (Genome Systems, St. Louis, MO) and sequenced. The longest mouse cDNA corresponded to amino acids 63-348 of chicken Sk-Tmod with a 95-bp 3Ј-UTR and the longest human cDNA corresponded to amino acids 20 -348 together with a 122-bp 3Ј-UTR. The complete coding sequence of the human Sk-Tmod was reconstructed from the sequences of five additional human ESTs, which overlapped by 200 -500 bp with the 1.1-kb human Sk-Tmod cDNA sequence and extended ϳ120 bp upstream to nucleotide Ϫ66 before the start codon. A zebrafish cDNA clone was also purchased from Genome Systems (St. Louis, MO), and it contained an apparently full-length Sk-Tmod cDNA of 1.4 kb with 5Ј-and 3Ј-UTRs of 62 and 328 bp, respectively, and encoded a protein of 343 amino acids.
Computational Analysis of Tropomodulin Sequences-For construction of the phylogenetic trees, the available ϳ40-kDa tropomodulin sequences from GenBank, the chicken Sk-Tmod, and the human, mouse, and zebrafish Sk-Tmods, were aligned with the program CLUSTAL (26). A human EST encoding a putative N-Tmod with ϳ82% identity to rat N-Tmod (27) was also included in the tree. A tropomodulin homolog encoding a predicted protein of 324 aa present in Caenorhabditis elegans was used to root the sequences of vertebrate tropomodulins. Other larger, more distantly related proteins such as the 64-kDa autoantigen D1 and a 684-amino acid protein in C. elegans were not included in this analysis (28). To optimize the alignment procedure, three different sets of gap opening and extension penalties were used (10 -0.05, 5-0.02, and 2-0.02). Each alignment was used independently to calculate the phylogenetic relationship between the aligned sequences. The distance method implemented in the programs PROT-DIST and NEIGHBOR (PHYLIP) (29) was used to infer the phylogeny of Sk-Tmod sequences. The bootstrap frequencies (100 cycles) were used to assess the reliability of the different branching nodes of the final tree (30). Identical branching arrangements were obtained with the three different alignment conditions. The bootstrap frequencies shown in Fig.  4 correspond to the alignment done with standard CLUSTAL parameters (10 -0.05).
Northern Analysis-Total mRNA from different adult chicken tissues was extracted using Trizol reagent (Life Technologies, Inc.) following the manufacturer's instructions. mRNAs were resolved by electrophoresis on a 1.2% agarose, 1.2 M formaldehyde gel and then transferred to a nylon Zeta-probe membrane (Bio-Rad). Blots were hybridized with [␣-32 P]dCTP full-length Sk-Tmod cDNA in hybridiza-tion solution following standard protocols (32). Blots were stripped following the manufacturer's instructions prior to incubating with other probes. For detection of E-Tmod mRNA, a fragment containing the E-Tmod coding sequence was obtained by the polymerase chain reaction and labeled with [␣-32 P]dCTP using the random primer kit (Stratagene). A 60-bp [␣-32 P]dCTP mouse 5.8 S ribosomal RNA fragment, kindly provided by Dr. Vincent Mauro (Scripps Research Institute, La Jolla, CA) was used as a probe to control for loading and integrity of the mRNA. The migrations of the 28 S and 18 S ribosomal RNAs were used to estimate the size of the tropomodulin mRNAs.
To obtain muscle cryosections, small pieces of adult chicken biceps femoris (BF) muscle were dissected, embedded in Tissue-Tek O.C.T. (Sakura Finetek, Torrance, CA) in a cryomold (Miles, Elkhart, IN) and immersed in liquid N 2 . Serial cross-sections of approximately 10 m thickness were obtained at Ϫ16°C on a Cryocut 1800 (Leica, Heidelberg, West Germany) and collected on a glass coverslip. Sections were processed for immunofluorescence as described above. Myofibrils and muscle sections were observed on a Zeiss Axioskop using a Zeiss 63ϫ Plan-Apochromat objective lens (1.4 numeric aperture) and a Zeiss 25ϫ multi-immersion objective (0.8 numeric aperture). Images were acquired with a cooled CCD-1300Y camera equipped with a Sony interline chip (Roper Scientific, Trenton, NJ) using IP laboratory software and then processed using Adobe Photoshop (Adobe Systems Inc. San Jose, CA). In Fig. 9, images were recorded using TriX (400 ASA) film.
Expression and Purification of Recombinant Sk-Tmod-The fulllength Sk-Tmod cDNA was excised from the pBluescript plasmid at EcoRI sites and subcloned in frame into the plasmid pGEX-KG (16) to express a glutathione S-transferase (GST) fusion protein in E. coli. Proper orientation of all inserts was confirmed by restriction mapping. For high yields, cells were grown to an A 605 of approximately 1 before inducing with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h, and the GST fusion protein was purified as described for E-Tmod (16) with some modifications. After affinity isolation of the GST fusion protein on a glutathione column and release from the GST by thrombin cleavage, Sk-Tmod was further purified by sequential anion exchange chromatography on a Resource Q column (Amersham Pharmacia Biotech), followed by a Mono Q column (Amersham Pharmacia Biotech). The purified protein was concentrated and dialyzed by vacuum dialysis into 20 mM HEPES, pH 7.3, 80 mM KCl, 1 mM dithiothreitol, 0.02% sodium azide, 20 g/ml phenylmethylsulfonyl fluoride and stored frozen at Ϫ80°C. The concentration of Sk-Tmod was determined by light absorption, using ⑀ 280 ϭ 10.24 mM Ϫ1 cm Ϫ1 . The extinction coefficient was calculated from the amino acid composition as described (34) using the ProtParam tool available at the EXPASY web site. The amino acid sequence of the GST linker that remains at the N-terminal end of Sk-Tmod is GSPGISGGGGG, which is directly followed by the Sk-Tmod sequence.
Actin Elongation Assays-Measurements of actin elongation were carried out as described previously (35,36), 2 using pyrenyl-labeled rabbit skeletal muscle actin (10% pyrenyl-labeled) and short gelsolincapped actin filaments as nuclei for polymerization. Actin polymerization was followed continuously for 5-7 h and measured again the next morning. The fluorescence changes (excitation 366.5 and emission 407 nm) were standardized against a Raman excitation peak and measured in a photon counting fluorimeter (Photon Technology International, Princeton, NJ). All experiments were carried out at 20°C with Mg 2ϩactin (converted from Ca 2ϩ -actin as described previously (36) in a medium containing 10 mM imidazole buffer, pH 7.0, 0.1 M KCl, 2 mM MgCl 2 , 1 mM azide, 1 mM dithiothreitol, 0.5 mM ATP, and 0.1 mM CaCl 2 (polymerization medium)). Gelsolin-capped actin filaments used as nuclei for polymerization were obtained by copolymerizing actin (usually 10 M) with gelsolin in the presence of calcium. Average sizes for filaments are given in the figure legend as the ratios of actin:gelsolin. The K d of Sk-Tmod for capping the pointed ends of pure actin filaments was estimated from the relative proportions of free and capped pointed ends when F-actin was half-maximally polymerized as described. 2 For the preparation of gelsolin-capped, tropomyosin-actin filaments as nuclei, rabbit skeletal muscle tropomyosin (37) in excess over that necessary for filament saturation (excess of about 1.0 M) was mixed with G-actin before copolymerization was started by the addition of salt. In addition, 1.0 M tropomyosin was added directly to the assay medium to ensure that elongating actin filaments were always saturated with tropomyosin. 2

RESULTS
Isolation of Sk-Tmod, a New Tropomodulin Isoform-While screening different muscles with a panel of mAbs generated against E-Tmod, we observed a spot in chicken breast muscle that migrated differently from E-Tmod on two-dimensional gels (Fig. 1). To determine whether this new tropomodulin spot was a novel isoform of tropomodulin, sequences of four peptides were obtained from tropomodulin immunoprecipitated from adult breast muscle. Comparison of the peptide sequences to the chicken E-Tmod amino acid sequence (16) demonstrated that all of the peptides were homologous but not identical to sequences in the C-terminal half of E-Tmod (Figs. 2 and 3). Using degenerate primers from these peptides, we isolated a polymerase chain reaction product, which was then used to isolate a full-length Sk-Tmod cDNA from an adult chicken skeletal muscle cDNA library. The complete nucleotide sequence of two of the longest clones (ϳ1.3 kb) was determined and is shown in Fig. 2. The sequence contains a single open reading frame of 1044 bp starting at nucleotide 56 (ATG) with a stop codon at position 1098 (TGA), followed by 131 bp of a 3Ј-untranslated region (3Ј-UTR), containing a polyadenylation signal (Fig. 2, bold) (38) and a short poly(A) tail. All four peptides obtained from the immunoprecipitated protein are present in the predicted amino acid sequence of the Sk-Tmod cDNA (Fig. 2, underlined).
We have designated this new isoform Sk-Tmod since, as we describe later, adult skeletal muscle is a major site of expression for this isoform. Comparison of the cDNA sequences of chicken Sk-and E-Tmods demonstrates that they are 64% identical along their entire length, which indicates that these sequences are unlikely to be alternatively spliced products from the same gene. In agreement with this conclusion, the chicken E-Tmod and Sk-Tmod genes are located on different chromosomes: E-Tmod is located on the chicken Z (sex) chro-2 A. Weber, C. Pennise, and V. M. Fowler, submitted for publication. mosome and Sk-Tmod is located on a tiny microchromosome. 3 A striking difference between the Sk-Tmod and E-Tmod cDNA sequences is the difference in size of their 3Ј-UTRs. The chicken E-Tmod cDNA has a long 3Ј-UTR of 2.5 kb (16), whereas the chicken Sk-Tmod cDNA has a short 3Ј-UTR of 131 bp (Fig. 2).
Sk-Tmod Sequences Are Highly Conserved among Vertebrates-The chicken Sk-Tmod cDNA encodes a protein of 348 amino acids which is 62% identical and 75% similar to chicken E-Tmod (Fig. 3). The calculated molecular mass and pI of Sk-Tmod are 39.2 kDa and 4.72, respectively, which is slightly smaller and more acidic than E-Tmod (40.3 kDa and pI of 5), and is in agreement with the different mobilities of the two isoforms observed on two-dimensional gels (Fig. 1). Interestingly, Sk-Tmod is 11 amino acids shorter than E-Tmod at the C-terminal end (Fig. 3), a region proposed to be important for pointed end capping activity (10). Notably, the recently de-scribed rat N-Tmod is also shorter than E-Tmod at the Cterminal end (351 versus 359 aa, respectively) (27).
We have identified ESTs and obtained cDNA clones encoding putative Sk-Tmods from several species, including mouse, human, and zebrafish. Alignment of these sequences with other known tropomodulin sequences allowed us to construct a phylogenetic tree using the 324-aa tropomodulin homolog of C. elegans as a root. This revealed that Sk-Tmods form an independent conserved group, which is similar but not identical to other vertebrate tropomodulins (Fig. 4) (14 -16, 27, 28).
Sk-Tmod Caps Thin Filament Pointed Ends-E-Tmod is the capping protein for the pointed ends of muscle thin filaments (5,9,28). To determine the location of Sk-Tmod in chicken skeletal muscle, we used polyclonal antibodies generated against Sk-Tmod to stain isolated myofibrils from adult chicken breast muscle and compared the patterns for Sk-Tmod staining and phalloidin staining for F-actin. In relaxed myofibrils, phalloidin binds along the length of the thin filaments 3 Dr. I. Nanda, personal communication. and appears as a broad banding pattern with narrow gaps corresponding to the H zones (gaps in the middle of the sarcomere with no actin filaments) (Fig. 5A). Sk-Tmod staining appears as a periodic doublet flanking the H zone in the middle of the sarcomere, consistent with a localization at the thin filament pointed ends ( Fig. 5B; merged image in C). In contrast, when myofibrils are stained for F-actin and ␣-actinin (Fig. 5, D and E) the merged image clearly shows that ␣-actinin staining is located at the Z line in the middle of the thin filament I-Z-I array (Fig. 5F).
Among all actin capping proteins, E-Tmod is distinguished by its weak capping activity for pointed ends of pure actin filaments (K d ϳ 0.1-0.4 M) and its strong capping activity for pointed ends of tropomyosin-actin filaments (K d Ͻ 1 nM) (36). This is a consequence of the ability of E-Tmod to bind tropomyosin and actin (16,36,39). To determine whether Sk-Tmod can cap actin filament pointed ends and whether its actin capping activity is enhanced by tropomyosin, we expressed chicken Sk-Tmod in E. coli and purified it to homogeneity (see "Experimental Procedures"). The pointed end capping activity of recombinant Sk-Tmod was evaluated from its ability to inhibit elongation of actin filaments capped at their barbed ends by gelsolin, using the fluorescence increase of pyrenyl actin to measure actin polymerization (36). In the absence of tropomyosin, Sk-Tmod inhibits the initial rate of elongation at the pointed filament end with an estimated K d of ϳ0.2 M (Fig. 6A). This indicates that Sk-Tmod can bind directly to pure actin filaments at their pointed ends with about the same affinity as E-Tmod (ϳ0.1-0.4 M) (36). Furthermore, Sk-Tmod also increases the critical concentration at the pointed end as demonstrated by a decrease in the total F-actin at steady state (i.e. lower end point of polymerization) (Fig. 6A), as previously observed for E-Tmod (36).
In the presence of skeletal muscle tropomyosin, nanomolar concentrations of Sk-Tmod completely block elongation from the pointed ends of the actin filaments and do not increase the critical concentration (Fig. 6B), similarly to E-Tmod (36). 2 The K d for Sk-Tmod capping of tropomyosin-actin filaments is estimated to be less than ϳ7 nM, from this experiment (Fig. 6B). The much higher capping activity of Sk-Tmod for a tropomyosin-actin filament than for pure actin is similar to E-Tmod and suggests that Sk-Tmod also binds tropomyosin in addition to actin at the pointed filament end. In fact, a tropomyosin blot overlay assay (16,24) demonstrates that skeletal muscle tropomyosin binds directly to Sk-Tmod (data not shown).
Tissue Expression Pattern of Sk-Tmod-A Sk-Tmod probe detects a single band of approximately 1.3 kb in adult breast muscle (PM) but not in heart, brain, cerebellum, spleen, lung, gizzard, liver, intestine, ovary, or thymus (Fig. 7, A and B). In contrast, an E-Tmod probe detects three bands of approximately 1.4, 2.1, and 2.6 kb in heart but not in breast muscle (PM) (Fig. 7A, right panel, asterisks) or in the other tissues (data not shown). Similar sizes for E-Tmod messages have been reported previously and are most likely due to utilization of different polyadenylation signals (15,16). Although Sk-Tmod is the predominant tropomodulin isoform in adult chicken breast muscle, E-Tmod is also present in breast muscle, albeit in low amounts, and can be detected by Northern blots of poly (A) ϩ RNA (16) or by grossly overexposing the Western blots of twodimensional gels (data not shown).
E-Tmod was originally identified in human erythrocytes and later shown to be in the rat lens (data not shown, and Refs. 23 and 39). The tropomodulin isoform composition of chicken lens fiber cells and erythrocytes was characterized by two-dimensional gel electrophoresis followed by Western blotting, because the very low amounts of mRNA contained in these biosynthetically inactive tissues make Northern blot analysis difficult. Fig.  7C shows that a single tropomodulin spot is detected in both adult chicken lens and erythrocytes (RBC) (Fig. 7C, left panel). Mixing experiments of lens or erythrocytes (RBC) with adult breast (PM) muscle (Fig. 7C, right panel) demonstrate that the tropomodulin spots in these tissues comigrate, indicating that Sk-Tmod rather than E-Tmod predominates in chicken lens and erythrocytes.
Sk-Tmod Replaces E-Tmod during Development of Chicken Skeletal Breast Muscle-Muscle growth and an increase in  The amino acid sequences from the ϳ40-kDa tropomodulins available in the data base were compared as described under "Experimental Procedures" to infer their phylogenetic relationships. The numbers at branch nodes correspond to bootstrap frequencies for each particular branching relationship. The 324-aa tropomodulin sequence from C. elegans was used to root the branches containing vertebrate sequences. muscle activity (as seen in flapping movement) are remarkable during the postnatal development of chicken breast muscle. The changes in the physiological properties of the muscle during development are paralleled by shifts in the expression of distinct isoforms for each sarcomeric component (13). To determine whether tropomodulin isoform expression changes during breast muscle development in vivo, we first compared message levels of E-and Sk-Tmod by Northern blotting. Sk-Tmod messages are detected exclusively in adult breast muscle and not in either day 12 or day 18 embryonic breast muscle (Fig. 8A, left  panel). When the same blot is reprobed with the E-Tmod specific probe, the three characteristic E-Tmod messages of approximately 1.4, 2.1, and 2.6 kb are detected in both day 12 or 18 day embryonic breast muscle but not in adult breast muscle (Fig. 8A, right panel).
In agreement with the Northern analysis, Western blots of two-dimensional gels show that E-Tmod is the most abundant tropomodulin isoform detected in breast muscle from day 12 embryos (Fig. 8B) and from day 20 embryos (just before hatching) (data not shown). Soon after hatching (2 days), the more acidic Sk-Tmod isoform starts to be detected (Fig. 8B), and, as development proceeds, the E-Tmod spot disappears and is replaced by the Sk-Tmod isoform. Thus, in breast muscle from chickens 1 week old or older, the Sk-Tmod isoform is the predominant tropomodulin isoform detected (Fig. 8B).

Sk-Tmod Is Enriched in Adult Fast Skeletal Muscle Fibers, Whereas E-Tmod Is Enriched in Slow Skeletal Muscle Fibers-
Adult breast muscle consists mainly of fast muscle fiber types, while embryonic breast muscle resembles adult slow fiber types in its physiological properties (40). Studies of isoform expression for other sarcomeric proteins have shown that the embryonic isoform often predominates in adult chicken slow fiber types as well as in cardiac muscle, while a different isoform is expressed in adult fast fibers (e.g. myosin light chain or Cprotein) (13). To further characterize the expression of Sk-Tmod and E-Tmod in different fiber types, we immunolocalized tropomodulins in serial cross sections of frozen adult chicken BF muscle which contains both fast and slow muscle fiber types. A monoclonal antibody specific for fast twitch C-protein (MF1) was used as a marker for fast muscle fiber types (19) and its staining pattern was compared with that of mAb 95, which specifically recognizes E-Tmod, and mAb 17, which preferentially recognizes Sk-Tmod by immunofluorescence (see "Experimental Procedures").
The staining pattern obtained with the fast C-protein antibody reveals the heterogeneity in the muscle fibers in BF (Fig.  9B). Most fibers stain brightly for fast C-protein (Fig. 9B,  double arrowhead), a subset of fibers appear moderately stained (Fig. 9B, short arrow), while in a few fibers fast Cprotein is hardly or not at all detected (Fig. 9B, long arrow). Overall, the intensity of the fiber staining for E-Tmod and fast C-protein is complementary to each other. For instance, the minority of the fibers that appear brightly stained for E-Tmod are not stained for fast C-protein (Fig. 9, A and B, long arrow), indicating that E-Tmod is enriched in slow muscle fiber types. Conversely, most fibers that do not stain for E-Tmod stain brightly for fast C-protein (compare Fig. 9, A and B, double  arrowhead). On the other hand, fibers stained for Sk-Tmod also contain fast C-protein (Fig. 9, B and C, short arrow, double arrowhead) and fibers with hardly any detectable Sk-Tmod are brightly stained for E-Tmod but not fast C-protein (Fig. 9, A-C, long arrow). Additionally, we found that by Western blot of two-dimensional gels, Sk-Tmod and E-Tmod are both present in BF, whereas Sk-Tmod is the only tropomodulin isoform detected in posterior latissimus dorsi muscle, which consists of fast muscle fiber types (data not shown).
Together, these results suggest that Sk-Tmod predominates in fast twitch fiber types. However, fibers that are moderately stained for fast C-protein are also stained for E-and Sk-Tmod (Fig. 9, A-C, short arrow) indicating that some fibers contain both tropomodulin isoforms. Hybrid fibers containing both fast and slow isoforms of different myofibrillar protein isoforms are reported to be quite frequent and may indeed represent the rule rather than the exception (13).
E-Tmod Colocalizes with ␣-Spectrin in Distinct Subsarcolemmal Domains-E-Tmod was first identified as a component of the spectrin-actin membrane skeleton in human erythrocytes (15,39). Interestingly, we noticed E-Tmod staining associated with the sarcolemma of many muscle fibers in BF muscle (Fig.  9A, arrowhead). This staining is specific for E-Tmod because it is eliminated by pre-incubating mAb 95 with a 100-fold molar excess of purified E-Tmod, but not Sk-Tmod, prior to incubation with the muscle sections (data not shown). Furthermore, we obtained the same sarcolemmal staining using another E-Tmod specific monoclonal antibody (mAb 9) (Fig. 10A), which recognizes an epitope of E-Tmod located in a different region of the molecule than the epitope recognized by mAb 95 (data not shown). These observations suggested that E-Tmod might be a component of the subsarcolemmal spectrin-actin membrane skeleton in skeletal muscle.
␣-Spectrin has been described to be concentrated at subdomains along the sarcolemma that overlay the I bands in skeletal muscle and are termed costameres (41). To explore the association of E-Tmod with the subsarcolemmal spectrin lattice, we double-stained cross-sections of BF muscle for E-Tmod and ␣-spectrin. Fig. 10 shows that E-Tmod as well as ␣-spectrin are associated with the sarcolemma of all muscle fiber types contained in this muscle (Fig. 10, A and B, arrows). A higher magnification view of an oblique BF section demonstrates that E-Tmod staining is not uniformly distributed along the sarcolemma, and instead appears to be concentrated in distinct and periodic puncta ϳ2 m apart, which colocalize with the ␣-spectrin puncta (Fig. 10, compare C with D, arrowheads).
This result suggests that E-Tmod is associated with the FIG. 5. Sk-Tmod is located at the pointed ends of skeletal muscle thin filaments. Myofibrils were isolated from relaxed chicken breast muscle and double-stained for F-actin (A) and Sk-Tmod (B) or F-actin (D) and ␣-actinin (E) as described under "Experimental Procedures." The merged images for the pairs are shown in C-F. Sk-Tmod is localized at the pointed ends of thin filaments in a doublet flanking the H zone (A-C, arrowheads). In contrast, ␣-actinin is localized at the Z line of the sarcomere (D-F, arrows). Bar, 5 m.
␣-spectrin-containing costameres overlying the I bands in chicken skeletal muscles. DISCUSSION Tropomodulins are a family of actin filament pointed end capping proteins. The first tropomodulin isoform identified in vertebrates was E-Tmod, a ϳ40-kDa protein whose expression in mammals is restricted to terminally differentiated cells including erythrocytes, lens fibers, neurons and striated muscle (9,28,42). Subsequently, another ϳ40-kDa tropomodulin isoform, N-Tmod, was identified in embryonic and adult neurons (27). ϳ40-kDa tropomodulin homologs have also been identified in flies (Sanpodo) (31) and in C. elegans (28,31). Larger, more distantly related proteins with limited regions of similarity to portions of the ϳ40-kDa tropomodulins have also been identified, including a ϳ64-kDa protein in smooth muscle and extraocular muscle (43,44) as well as a longer ϳ60-kDa (predicted) protein from C. elegans (28,31). In this study, we report the identification of a third ϳ40-kDa tropomodulin isoform from vertebrates, which we have named Sk-Tmod, based on its high level of expression in skeletal muscle.
The three known ϳ40-kDa vertebrate tropomodulin isoforms are products of different genes and are all about 60 -65% identical (75-90% similar) to each other at the amino acid level. The high sequence homology between tropomodulin isoforms implies that they most likely share functional properties. Indeed, here we show that Sk-Tmod shares all the known functional properties described for E-Tmod (9); it is associated with the pointed ends of muscle thin filaments in vivo and completely blocks actin elongation from the pointed ends of tropomyosinactin filaments in vitro (K d Ͻ 7 nM). Like E-Tmod, Sk-Tmod also inhibits elongation from the pointed ends of pure actin filaments but with considerably lower affinity (K d ϳ 0.2 M). Thus, we propose that a defining feature of this family of proteins is the ability to block the elongation of tropomyosinactin filaments from their pointed ends, mediated by their binding to both actin and tropomyosin (9). N-Tmod has been shown previously to interact with tropomyosin similarly to E-Tmod, but its actin capping activity has not yet been investigated (27). Thus, verification of this hypothesis will require determination of the tropomyosin-actin filament capping activities for N-Tmod and the other recently identified tropomodulin homologs from invertebrates.
On the other hand, the sequences of tropomodulins in each independent group in the tropomodulin family (E-, N-, and Sk-Tmod groups) are highly conserved across vertebrates (28 and Fig. 4 of this study). This might indicate that different tropomodulin isoforms have acquired minor but functionally significant differences in their sequences that have been preserved through evolution. This study and those of others have described several observations supporting this idea. First, tropomodulin isoforms exhibit strikingly different tissue expression patterns (28,42). For example, in adult chickens, Sk-Tmod is found in fast skeletal muscle, erythrocytes, and lens. In contrast, E-Tmod is found in heart and slow skeletal muscle while N-Tmod is expressed in brain. 4 A second observation, and probably the one that most strongly supports the idea of functional specificity for each tropomodulin isoform, is the sorting of Sk-and E-Tmod isoforms to different cytoskeletal structures within the same skeletal muscle cell (e.g. myofibrils and costameres, respectively). This may indicate that other regions of the molecule, not involved in tropomyosin or actin binding, might be responsible for interactions with other sarcomeric or cytoskeletal components and determine unique functions, which are as yet undescribed.
On the other hand, the tropomodulin isoform composition in the same tissue appears to differ between species. For instance, chicken lens and erythrocytes contain Sk-Tmod (Fig. 7) while mouse lens and erythrocytes contain E-Tmod. 4 This may indicate that tropomodulins could have redundant functions and that their differential expression in tissues and between species may reflect the regulatory constraints of different developmental programs. Future studies on the genomic organiza- tion and the intron/exon structure of tropomodulins may also provide an understanding of the role that regulatory or structural regions play in generating and maintaining this tropomodulin isoform sequence diversity.
Expression of Different Tropomodulin Isoforms in Muscle Correlates with Changes in Thin Filament Lengths-Our results show an interesting correlation between the tropomodulin isoform expressed and the changes in thin filament lengths that take place during the development of chicken breast muscle. Decoration of thin filaments from adult breast muscle with antibodies to troponin invariably revealed 24 striations at regular intervals of 38 nm, whereas in embryonic muscle the number of striations varied between 25 and 29 (45). This sug-gests that the thin filaments are all ϳ0.9 m long in adult breast muscle, whereas they vary from ϳ0.95 to 1.1 m in embryonic breast muscle. Our results show that E-Tmod is present in embryonic breast muscle and thus appears to be associated with sarcomeres in which the thin filaments are longer and have a wider range of lengths. In contrast, Sk-Tmod is present in adult breast muscle and thus appears to be associated with sarcomeres in which the thin filaments are shorter and have a more precise length distribution. To further explore this idea, it will be necessary to compare directly the expression of tropomodulin isoforms in other muscles in which thin filament lengths have been measured.
How might different tropomodulin isoforms influence the relative precision or variability in thin filament length regulation? One possibility is that tropomodulin isoforms could have different effects on the dynamics of actin monomer exchange at the thin filament pointed ends. For instance, in vivo, Sk-Tmod might provide, directly or indirectly, a tighter capping activity that would more precisely stabilize thin filament lengths in fast skeletal muscles. In contrast, E-Tmod might cap the thin filaments less tightly in embryonic breast muscle, heart, or adult slow skeletal muscle leading to a more extensive exchange of monomeric actin at their pointed ends, and consequently a wider variation in lengths.
Although Sk-Tmod has a similar affinity as E-Tmod for tropomyosin-actin filament pointed ends in vitro, differences in the affinity of tropomodulin isoforms for thin filament pointed ends in vivo could be influenced by other sarcomeric components. Potential candidates are tropomyosin isoforms and nebulin. Skeletal muscle cells have two tropomyosin isoforms, ␣ and ␤, that are developmentally regulated at the transcriptional level such that the ␤-tropomyosin gene is repressed at hatching (46). Thus, in adult breast muscle, ␣/␣ is the principal isoform found (47)(48)(49)(50). Furthermore, the ratio of ␣/␣, ␣/␤, and ␤/␤ dimers differs between mature muscle fiber types (51,52).
Nebulin is a long actin-binding protein that extends from the Z line to the thin filament pointed end and has been proposed to function as a molecular template to specify thin filament lengths in skeletal muscle (53)(54)(55). The existence of different nebulin isoforms which differ in their molecular size has been correlated with the length of thin filaments in different skeletal muscle types and during differentiation of chicken breast muscle (56 -58). One can speculate that distinct nebulin isoforms might interact with specific tropomodulin isoforms to regulate tropomodulin affinity for the pointed ends and thereby produce thin filaments of different lengths for physiologically distinct types of muscle (5). However, cardiac muscle does not contain nebulin (55), yet it contains E-Tmod, which might suggest that nebulin is not necessarily required by E-Tmod to regulate the length of thin filaments.
Sorting of Tropomodulin Isoforms to Different Cytoskeletal Structures-Interestingly, in fast skeletal muscle fibers, which coexpress both E and Sk-Tmod, E-Tmod appears to be associated with the ␣-spectrin-containing costameric network overlying the I bands at the sarcolemma while Sk-Tmod is associated with the myofibrils within the same cell. What can account for restriction of E-Tmod to the sarcolemma in hybrid muscle fibers which express both E and Sk-Tmod? One possibility is that Sk-Tmod binds thin filament pointed ends more tightly and thus competes for E-Tmod binding to this site. Similarly, E-Tmod may bind more tightly to costameric actin filaments than Sk-Tmod. In this case, we might expect the existence of specific sequences in tropomodulin isoforms to target the tropomodulins to their appropriate binding partners in each structure. However, it is important to point out that the basis for sorting of different tropomodulin isoforms to different cytoskeletal structures is likely to be complex since in some muscle fibers (i.e. slow and mixed) E-Tmod is detected associated with myofibrils and at the sarcolemma.
Costameres were first described as vinculin-rich areas forming rib-like elements flanking the Z lines of underlying myofibrils in skeletal muscle (59,60). Subsequently, spectrin was found to be enriched in costameres overlying the I bands, in transverse elements over the M line and in fine longitudinal strands connecting the costameres (41,(61)(62)(63)(64)(65). To date, a variety of other proteins including desmin, integrins, dystrophin, and components of the spectrin-based membrane skeleton have also been localized to elements of the costameric network (64, 66), but very little is known about their molecular organization. Our observation that E-Tmod is a component of the costameric FIG. 9. Sk-Tmod predominates in skeletal fast muscle fiber types while E-Tmod predominates in slow muscle fiber types. Serial cross-sections of frozen adult chicken BF muscle were stained for E-Tmod (A), fast twitch C-protein (B), or Sk-Tmod (C). E-Tmod is most abundant in slow muscle fiber types (A, long arrow), where fast twitch C-proteins (B, long arrow) and Sk-Tmod (C, long arrow) are hardly detected. Some fibers contain a moderate amount of E-Tmod (A, short arrow), fast twitch C-protein (B, short arrow), and Sk-Tmod (C, short arrow). Sk-Tmod is most abundant in the fast muscle fiber types that are brightly stained for fast C-protein (B and C, double arrowheads) and do not contain E-Tmod (A, double arrowhead). E-Tmod staining is also detected at the sarcolemma of muscle fibers (A, arrowhead). The brighter staining obtained for fast C-protein on the right edge of the tissue section is most likely due to a fold or a difference in thickness of the section. Bar, 50 m.
FIG. 10. E-Tmod colocalizes with ␣-spectrin at discrete subsarcolemmal domains in skeletal muscle fiber cells. Serial crosssections of frozen adult chicken BF muscle were double stained for E-Tmod (A), or ␣-spectrin (B). E-Tmod together with ␣-spectrin localizes at the sarcolemma of all the muscle fiber types (A and B, arrows). Higher magnification images show that E-Tmod is localized in distinct puncta along the sarcolemma (C, arrowhead), which colocalize with ␣-spectrin (E, arrowhead). The merged image is shown in D (merged). The asterisk in D and E indicates a capillary brightly stained for ␣-spectrin. Bar in A and B, 25 m. Bar in C-E, 2 m. network suggests that similar to erythrocytes, the spectrinbased membrane skeleton underlying the costameric network is built by short actin filaments capped by E-Tmod at their pointed ends and cross-linked by spectrin (9).
It has been proposed that costameres could play a mechanical role in anchoring the sarcomeric cytoskeleton of the most peripheral myofibrils to the membrane via intermediate filaments and also in linking the sarcolemma to the extracellular matrix via integral membrane proteins (60,63,64,67,68). In addition to a mechanical role, costameres have also been proposed to be involved in signal transduction across the membrane through focal adhesion proteins such as integrins, talin, and vinculin, or through the dystrophin-based membrane skeleton, which appears to interact with enzymatic or regulatory proteins (64). Thus, E-Tmod in costameres may function in a structural capacity to stabilize and limit the lengths of actin filaments in a spectrin-actin network, thereby influencing the clustering of focal adhesion components and/or dystrophin-associated proteins at costameres. This could regulate, in turn, the signaling pathways controlling skeletal muscle fiber growth or atrophy.
Prospects-While studies of isolated proteins have been useful in identifying protein function, they may be less helpful in distinguishing properties of closely related isoforms. The correlation of tropomodulin isoform expression with differences in the contractile properties of different muscle types suggests the possibility that tropomodulins might also be involved in thin filament linked regulation of contraction together with tropomyosins and troponins (for a discussion, see Ref. 5). In vivo, it is likely that the functional advantages produced by each molecular phenotype have contributed to preserve certain combinations of sarcomeric isoforms in each muscle. Recently, several functional studies have been published reporting that closely related sarcomeric isoforms of tropomyosin and troponin are not functionally redundant in vivo (Refs. 69 -71; for a review, see Ref. 13). Thus, genetic approaches to gene inactivation and gene replacement of tropomodulin in animal models may provide new insights into the physiological consequences of altering tropomodulin isoform expression during development and differentiation.