The C-terminal IgI domains of myosin-binding proteins C and H (MyBP-C and MyBP-H) are both necessary and sufficient for the intracellular crosslinking of sarcomeric myosin in transfected non-muscle cells

A hallmark of skeletal and cardiac muscle is its cross-striated appearance,conferred by the lateral alignment of sarcomeres in its cylindrical,contractile organelle, the myofibril. Each sarcomere is composed of three filament systems: the unipolar thin filaments anchored at their barbed ends to the Z-band, the bipolar thick filaments linked to one another in the M-band by a group of M-band-associated proteins, and the unipolar titin filaments that anchor the thick filaments from the M-band region all the way to the Z-band and confer elastic properties on each sarcomere(Gregorio et al., 1999). Although the sarcomere is one of the best understood organelles in eukaryotic cells in terms of composition, structure and physiology, its assembly, growth and turnover remain unresolved: to date the organelle has not been assembled in a cell-free system from its defined components. To better understand formation of this organelle, especially the A-band, we have been examining two myosin-associated proteins, myosin-binding C (MyBP-C) and myosin-binding protein H (MyBP-H), that appear to function in the assembly and registration of thick filaments during myofibrillogenesis(Winegrad, 1999).

In addition to myosin, each thick filament is composed of several major proteins including MyBP-C (Offer et al.,1973) and MyBP-H (Starr and Offer, 1982). MyBP-C is an asymmetric protein of ∼130 kDa that accounts for approximately 2% of myofibril protein mass(Offer et al., 1973). Three isoforms have been identified: cardiac, fast skeletal and slow skeletal, each encoded by separate genes (Vaughan et al.,1993a; Vaughan et al.,1993b). MyBP-H is ∼55 kDa and is mainly found in fast skeletal muscle; only one isoform and one gene have been identified. Within the sarcomere, MyBP-C and MyBP-H are distributed in highly regular patterns that are species- and fiber-type specific. MyBP-C is found in the C zone of the A-band (Squire, 1990) in association with a set of 11 transverse stripes, 43 nm apart, in each half A-band (Craig and Offer, 1976;Dennis et al., 1984). In the chicken pectoralis muscle, MyBP-H has a similar distribution to MyBP-C(Bahler et al., 1985), but different patterns have been described in mammalian muscle(Bennett et al., 1986).

Nucleotide sequencing of MyBP-C and MyBP-H mRNAs has shown that both proteins are constructed of two repeating modules: the immunoglobulin I (IgI)and fibronectin III (FnIII) motifs (Fig. 1). Skeletal MyBP-C is composed of a linear array of seven IgI and three FnIII domains, each containing ∼100 amino acids, in the order of IgI-IgI-IgI-IgI-IgI-FnIII-FnIII-IgI-FnIII-IgI, numbered C1 to C10 from the N-to C-terminus (Vaughan et al.,1993a; Weber et al.,1993). MyBP-H consists of a unique N-terminal leader sequence containing two motifs of alternating alanine and proline residues; it is this N-terminal leader domain that causes the anomalously slow mobility of chicken MyBP-H. The C-terminus of MyBP-H is composed of four repeating modules with the order FnIII-IgI-FnIII-IgI, an identical arrangement to the C-terminus of MyBP-C (Vaughan et al.,1993a). The sequence homology between the C-termini of both proteins is approximately 50%. Most studies of MyBP-C have focused on the protein's biochemical properties, of which the best characterized is its affinity for myosin (Moos et al.,1975; Offer et al.,1973). The protein binds to F-actin(Moos et al., 1978) and titin(Freiburg and Gautel, 1996). In vitro binding studies demonstrated that the LMM-binding domain of MyBP-C resides in the C-terminal IgI (C10) domain(Okagaki et al., 1993). A comparable LMM-binding domain resides in the C-terminal IgI motif (H4) of MyBP-H (Alyonycheva et al.,1997). A titin-binding site on MyBP-C has been localized within the C7-C10 module (Freiburg and Gautel,1996). Using transient transfections in cultured embryonic myocytes, it has been demonstrated that the final four domains of both MyBP-C and MyBP-H are necessary for proper localization of these proteins in the A-band (Gilbert et al., 1998;Gilbert et al., 1996). Deletion of the C-terminal IgI domain of either MyBP-H or MyBP-C prevents the proteins from incorporating into the A-band. Mutants containing the C-terminal IgI domain, but lacking the adjacent three upstream modules, do not properly localize to the A-band, even though they include the myosin rod-binding IgI domain (Gilbert et al., 1998;Gilbert et al., 1996). In a recent study designed to determine amino acids on the surface of C10 that may interact with LMM, it was found that four surface regions of C10 interact with the myosin rod. All of these contain charged residues, presumably involved in ionic interactions with complementary residues on the myosin rod. These data suggest that C10 behaves as a multivalent ligand, crosslinking three or four molecules of the myosin rod (Miyamoto et al., 1999).

In the present study we have used the COS cell transfection system(Moncman et al., 1993;Vikstrom et al., 1993) to define the regions of MyBP-C and MyBP-H that may be responsible for crosslinking myosin in thick filaments. We found that the C-terminal IgI module in both MyBP-C and MyBP-H is both necessary and sufficient to induce cable formation when these MyBPs are co-expressed with MyHC. In addition, we present evidence for previously unidentified myosin-interaction sites within domains C7-9 and H1-3 of MyBP-C and MyBP-H, respectively.

Full-length chicken skeletal MyBP-H, MyBP-C and the last four domains of MyBP-C (C7-10) were N-terminally tagged with green fluorescent protein (GFP)by cloning them into the pEGFP-C1 expression vector (Clontech, Palo Alto, CA)(Fig. 1A). GFP- tagged full-length MyBP-H (GFP/H) was constructed by excising a 1763 bp cDNA fragment from pMH (see below) using EcoRI and XbaI and cloning this fragment into the EcoRI and XbaI sites of pEGFP-C1. Similarly, a 1246 bp cDNA fragment encoding the last four domains of MyBP-C was generated by digesting pC7-10 (see below) with EcoRI and XbaI and then ligating it to EcoRI- and XbaI-digested pEGFP-C1 to make GFP/C7-10. GFP-tagged full-length MyBP-C (GFP/C) was created by digesting pMC (see below) and pGFP/C7-10 with EcoRI and NotI. The resulting 5432 and 2692 bp fragments were then ligated together. Myc-tagged MyBP-C expression plasmids pMC, pC7-10,p10 and p1-9 (Fig. 1B), and MyBP-H expression plasmids, pMH, HΔU, HΔU1 and HΔ4(Fig. 1C), were described previously (Gilbert et al.,1998; Gilbert et al.,1996). pmtα, encoding rat α MyHC, was gift a from Leslie Leinwand (University of Colorado).

COS-1 cells were propagated in Dulbecco's modified Eagle's medium (DMEM)containing 10% fetal bovine serum (FBS), 2 mM L-glutamine, 50 units/ml penicillin G and 50 μg/ml streptomycin (penstrep) in 100 mm dishes with glass coverslips. Transfections were performed with cell cultures of 40 to 60%confluence using lipofectamine PLUS reagents (Life Technologies, Rockville,MD) according to the manufacturer's instructions. COS cells were processed for analysis two days after transfection. Primary cultures of myoblasts were prepared from 11-day-old embryonic chicken pectoralis muscle as described previously (Gilbert et al.,1996). The myoblasts were cultured on collagen-coated, two-well chamber slides (NUNC, Naperville, IL) in DMEM, 10% FBS, 2 mM L-glutamine and 2% chick embryo extract. One day after plating, the cells were transfected using lipofectamine PLUS reagent. On the following day the cells were switched to DMEM containing 10% horse serum instead of FBS. The media was changed every two days and the cells allowed to differentiate for five days after transfection before being fixed and processed for immunostaining.

Transfected COS cells were prepared for western blotting by twice washing with phosphate-buffered saline (PBS), then scraping the cells into 200 μl of Laemmli sample loading buffer (Laemmli,1970). Cell suspensions were briefly sonicated and boiled for three minutes. 10 to 40 μl of each cell lysate sample, with the exception of cells transfected with pC10, were subjected to electrophoresis on a 10% SDS PAGE gel. pC10-transfected cells were analyzed on a 15% SDS-PAGE gel. After electrophoresis, the proteins were transferred electrophoretically to nitrocellulose for 1 hour at 100 V in 25 mM Tris base, 192 mM glycine, 0.1%SDS and 20% methanol. Blots were then immersed in 5% dry milk in PBS, 0.05%Tween 20 and 1% BSA (blocking solution) for a minimum of 30 minutes. The membranes were then incubated with monoclonal antibodies (mAbs) against GFP(Covance, Princeton, New Jersey) and myc (9E10) at a dilution of 1:10,000 and undiluted, respectively. Following primary antibody incubations, the COS cells were reacted with horseradish peroxidase (HRP)-conjugated goat affinity-purified anti-mouse antibody (Promega, Madison, WI). COS cells transfected with MyBP-H constructs were reacted with a 1:1000 dilution of purified anti-MyBP-H polyclonal antibody(Gilbert et al., 1998)followed by horseradish peroxidase (HRP)-conjugated affinity-purified goat anti-rabbit antibody (Promega, Madison, WI). The antibody complexes were then visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech, Inc.,Piscataway, NJ).

COS cells cultured on coverslips and myoblasts cultured on chamber slides were both processed at room temperature. COS cells were first washed twice with PBS and then fixed with 2% formaldehyde in PBS. The cells were permeabilized with 1% triton in PBS for 10 minutes and blocked with 1% BSA,0.05% Tween 20 in PBS for 30 minutes. For both primary and secondary antibody incubations, all dilutions were made in blocking buffer. Cells were first probed with a 1:10 dilution of F59, a mAb specific for sarcomeric MyHC (gift of Frank Stockdale, Stanford University) (Miller et al., 1985) for 1 hour. After the F59 incubation, the cells were incubated for 30 minutes with Alexa-594-conjugated goat affinity-purified anti-mouse antibody (Molecular Probes, Eugene, OR) at a dilution of 1:200. The cells were then reacted with biotinylated conjugated 9E10 (Covance, Princeton, NJ) at a dilution of 1:10. Myc-tagged MyBPs were then detected using Alexa 488 strepavidin (Molecular Probes, Eugene, OR) at a dilution of 1:200. Coverslips were mounted in Airvol(Air Products, Allentown PA) with 100 mg/ml 1,4-diazobicyclo (2.2.2)-octane(Sigma, St. Louis, MO) to reduce photobleaching. Cultured myoblasts were processed in a similar manner but probed only with F59 and Alexa-594-conjugated goat, affinity-purified anti-mouse antibody. The cells were examined with a Nikon epifluorescence microscope using a 100×objective. Images were digitally captured using the Sony DKC-5000 digital camera system (Sony Corporation, Japan) using P2 system Software (P2 System,Budapest, Hungary).

To test the interactions of myosin and MyBPs in the absence of other sarcomeric proteins, and to define which domains of MyBP-C are necessary for this interaction, we have employed the non-muscle COS cell transfection system(Seiler et al., 1996;Vikstrom et al., 1993). Epitope-tagged proteins have been used to define regions of skeletal MyBP-C and MyBP-H required for colocalization and the formation of co-polymer aggregates with the MyHC molecule. GFP-tagged myomesin has recently been used to define domains involved in M-band targeting of this myosin-binding protein(Auerbach et al., 1999). Our N-terminally GFP- tagged expression constructs consisted of full-length MyBP-C and MyBP-H, termed GFP/C and GFP/H respectively(Fig. 1). In addition to these,we constructed a GFP-tagged truncation mutant of MyBP-C containing only the C-terminal four domains (GFP/C7-10). To test the stability of GFP-tagged proteins, COS cells were transfected with the expression plasmids GFP/C,GFP/C7-10 and GFP/H, and whole cell lysates were prepared. Typical transfection efficiencies for GFP/C7-10 and GFP/H ranged from 10-30%, whereas GFP/C was consistently lower, ranging from 1-5%. Immunoblotting of transfected whole cell lysates with anti-GFP mAb revealed a single band for each MyBP construct (Fig. 2). As a positive control for the GFP antibody, the cell lysate from COS cells transfected with the GFP cloning plasmid, pEGFP-C1, was also examined. As expected, transfection with GFP/C and GFP/C7-10 produced immunoreactive bands of 155 and 73 kDa. GFP/H, with a predicted molecular mass of 85 kDa, ran with a slower mobility than expected. This effect is attributed to an extended motif of alternating proline and alanine residues within the N-terminal leader region of the molecule (Vaughan et al.,1993a), similar to the motif identified in MyLC-1f(Nabeshima et al., 1984). These data demonstrate that GFP-tagged full-length and truncated forms of MyBPs can be stably expressed in COS cells.

In an earlier publication, we showed that myc-tagged MyBP-C, MyBP-H and C7-10 proteins encoded by expression plasmids could incorporate into the A-bands of myofibrils in cultured, differentiating myotubes(Gilbert et al., 1998). To test whether the GFP-tagged versions of these recombinant proteins were competent to interact properly with thick filaments, transient transfections of 11-day-old embryonic chick myoblasts were performed. Expression of GFP/C,GFP/C7C10 or GFP/H resulted in the clear localization of the expressed protein to the A-band (Fig. 3B,E,F,respectively). When muscle cells transfected with GFP/C were probed with F59,a mAb specific for sarcomeric myosin, we observed that distribution of the GFP/C was similar to the distribution of myosin(Fig. 3C). This co-distribution is obvious when the overlapping staining patterns of both proteins were examined (Fig. 3D). The colocalization of GFP/C7-10 and GFP/H and myosin also was observed when these transfected muscle cells were stained with mAb F59 (data not shown). These data demonstrate that the GFP moiety did not interfere with the ability of these recombinant proteins to localize properly to the A-band.

Sarcomeric MyHCs, when expressed in COS cells, form clusters of spindle-shaped filament aggregates throughout the cytoplasm of the transfected cells (Moncman et al., 1993;Rindt et al., 1993;Straceski et al., 1994;Seiler et al., 1996;Vikstrom et al., 1992). By contrast, GFP/C, GFP/C7-10 and GFP/H, when expressed in COS cells, were distributed diffusely throughout the cytoplasm(Fig. 4A-C). We occasionally observed an apparent nuclear localization of the transiently expressed MyBPs,(Fig. 4A-C), but we did not test whether this signal was coming from within or around the nuclei. When co-expressed with MyHC, MyBP-C and MyBP-H had a striking effect on the organization of MyHC (Seiler et al.,1996). In the presence of the MyBPs, the MyHC formed long peri-nuclear cables rather than the spindle-like filament aggregates observed with MyHC alone. Double immunostaining revealed that MyBP-C or MyBP-H colocalized with the MyHC in these cables. Transfection of GFP/C with MyHC also resulted in co-polymer cable formation of GFP/C(Fig. 4D) and MyHC(Fig. 4G). The overlap of the GFP/C and MyHC was not complete (Fig. 4J); although the MyBP-C co-distributed with the MyHC, some MyHC remained in spindle form, and these spindle aggregates did not contain MyBP-C. We interpret this as reflecting a sub-stochiometric concentration of GFP/C relative to myosin. Since co-distribution of GFP/C and MyHC was observed only in the cables, we conclude that MyBP-C is required for cable formation.

The same phenomenon was observed when GFP/H was co-expressed with MyHC. Both GFP/H (Fig. 4F) and MyHC(Fig. 4I) organized into cables. The overlapping distribution of GFP/H and MyHC indicates that the both proteins can effectively colocalize. Although, the majority of GFP/H was found in the peri-nuclear cables (Fig. 4L) some myosin was detected in spindle-like aggregates lacking GFP/H. GFP/C7-10 was also an efficient promoter of cable formation(Fig. 4H). Often cells were found in which GFP/C7-10 colocalized with MyHC into co-polymer cables(Fig. 3E). However, we often observed self-aggregation of the GFP/C7-10 peptide in COS cells(Fig. 4E). This phenomenon was not observed in differentiating muscle (data not shown). In view of the fact that there were such high levels of GFP/C7C10 expression in the COS cells,some co-distribution of the two proteins may have been obscured by the intense GFP fluorescence in the transfected cells. These experiments demonstrate that GFP-tagged MyBPs are competent to induce co-polymer cable formation with MyHC. They also show that cable-forming activity is within the C-terminal four domains of MyBP-C.

The cloning of myc-tagged MyBP-C and MyBP-H mutants has been described in an earlier publication (Gilbert et al.,1998; Gilbert et al.,1996). Four of those plasmids expressing myctagged MyBP-C were reused in the present study (Fig. 1B): full-length MyBP-C; C7-10, a C-terminal fragment containing the A-band localization region; C10 (previously termed Δ1-9), encoding the C-terminal IgI-domain-containing the major myosin rod-binding domain; and C1-9 (previously termed Δ10), a MyBP-C fragment lacking the C-terminal IgI domain. Four myc-tagged MyBP-H constructs were also tested for myosin binding and cable formation in COS cells(Fig. 1C): full-length MyBP-H;HΔU, a MyBP-H peptide lacking the unique N-terminal region; HΔU1,which lacks the both the unique region and the first FnIII domain; and HΔ4, a recombinant molecule missing the C-terminal myosin rodbinding IgI domain. Our earlier studies showed that truncation mutants of the MyBPs that lack the C-terminal IgI domain, for example, C1-9 and HΔ4, lose the ability to localize in the A-bands of transfected embryonic myotubes. Before studying the cellular distribution of myc-tagged MyBPs mutants in COS cells,we tested the stability of these proteins using western blots(Fig. 5). Whole cell lysates from COS cells transfected with MyBP-C, C7-10, C10 and C1-9 were prepared two days after transfection, western blotted and probed with an anti-myc mAb. All constructs generated proteins that reacted with the myc antibody and eletrophoresed at the expected relative mobilities(Fig. 5A). Likewise,transfection of the four MyBP-H constructs produced immunoreactive products with the myc mAb. Of note, only HΔU and HΔU1 ran with motilities consistent with their molecular mass of 47 and 35 kDa. MyBP-H and HΔ4,with predicted molecular masses of 60 and 50 kDa, contain a unique proline and alanine leader region, and these ran slower than their expected masses. In general, MyBP-C constructs produced less myc-tagged protein in transfected COS cells than did the MyBP-H constructs. These studies demonstrated that myc-tagged full-length MyBPs and truncation mutants can be stably expressed in COS cells.

We next tested the antigenicity of the MyBP-H mutants with a previously described MyBP-H-specific pAb (Gilbert et al., 1998) (Fig. 5C). This pAb reacted with a single band in lysates from COS cells transfected with MyBP-H. This band had an identical relative mobility to full-length MyBP-H (Fig. 5C)and proved that recombinant myc-tagged MyBP-H, expressed in transient transfections of COS cells, was immunoreactive with a pAB specific for chicken skeletal MyBP-H. Also showing immunoreactivity were HΔU and HΔ4 lysates. COS cells transfected with HΔU1, although immunoreactive with the anti-myc pAb (Fig. 5B), did not react with the MyBP-H pAb (Fig. 5C). The positive reactivity of HΔU and the absence of reactivity with HΔU1 indicate that the epitope for this polyclonal antibody resides in domain H1.

Of the MyBP-C recombinant fragments tested, only full-length MyBP-C and C7-10 targeted to the A-bands of transfected, differentiating myotubes(Gilbert et al., 1996). C10 and C1-9 did not localize at all to the A-bands. Biochemical studies, however,have shown that the C-terminal 14 kDa fragment of MyBP-C (C10) contains a LMM-binding domain (Okagaki et al.,1993). Furthermore, a second myosin-binding domain within the C1-C2 region of cardiac MyBP-C has been identified(Gruen and Gautel, 1999). To delineate the functional domain responsible for cable formation, we transfected COS cells with full-length MyHC in combination with MyBP-C, C7-10,C10 or C1-9. The cells were then fixed and doubly immunostained with mAbs against sarcomeric myosin and myc (Fig. 6). When expressed individually, all of the MyBP-C recombinant proteins distributed diffusely throughout the cytoplasm, with occasional nuclear staining (Fig. 6A-D). When co-transfected with MyHC, myc-tagged MyBP-C and C7-10 induced co-polymer cable formation in a manner identical to their GFP-tagged counterparts(Fig. 6E,I,M and 6F,J,N). Transfection with only the last domain of MyBP-C (C10) with MyHC also promoted cable formation (Fig. 6G,K,O). These results demonstrate that although the C10 domain could not localize to the A-band of cultured myotubes, it could bundle sarcomeric myosin in this non-muscle cell system.

To determine whether C10 was necessary for cable formation, we performed a complementary experiment co-transfecting a construct lacking the C-terminal IgI domain, C1-9, with MyHC. This truncation mutant did not induce cable formation (Fig. 6H,L) but did colocalize with the spindleshaped clusters of MyHC(Fig. 6P). These data suggest the presence of other myosin-binding sites within the skeletal MyBP-C molecule besides those in C10. Our MyBP-C transfection data are presented graphically in Fig. 8A. The percentage of doubly transfected cells, demonstrating a colocalization phenotype of recombinant MyBP-C protein with MyHC exceeded 90% for each of the expression constructs. MyBP-C and C7-10 were the most effective promoters of cable formation, with an efficiency of 78 and 86%. Cables also formed in 57% of the transfected cells expressing C10 and MyHC. By contrast, only 5% of the cells co-transfected with C1-9 and MyHC formed cable-like structures. These data demonstrate that the C-terminal IgI domain of MyBP-C is both necessary and sufficient for inducing MyHC cable formation in COS cells.

The amino-acid sequences of the four C-terminal domains of MyBP-C and MyBP-H have 49% identity, with additional 23% conservative substitutions. Both proteins showed similar cable formation when co-transfected into COS cells(Fig. 7). Single transfections of MyBP-H, HΔU, HΔU1 and HΔ4 resulted in diffuse expression of the expressed proteins (Fig. 7A-D). When myc-tagged MyBP-H was co-expressed with MyHC, it behaved identically to GFP/H, forming large cables(Fig. 7E,I,M). HΔU, a MyBP-H analogous to fragment C7-10, promoted cable formation(Fig. 7F,J,N) with 80%efficiency, the same as full-length MyBP-H(Fig. 8B). These results demonstrate that the unique N-terminal region of MyBP-H does not contribute to the cable-forming properties of MyBP-H when co-expressed with MyHC. When the unique region and first FnIII domain were deleted (HΔU1), the MyBP-H truncation mutant still formed cables with MyHC(Fig. 7G,K,O) but with a 22%lower efficiency than full-length MyBP-H or HΔU(Fig. 8B). Finally, we tested the ability of construct HΔ4 to form cables. Similar to C1-9, HΔ4 lacks the C-terminal myosin rod-binding region of MyBP-H. When co-transfected with MyHC, the expressed protein colocalized with spindle structures(Fig. 7H,L,P) in 91% of the doubly transfected cells (Fig. 8B). Cable-like structures, however, were rarely observed; only 6%of the cells weakly manifested this phenotype. This demonstrated that domain H4 of MyBP-H is necessary for copolymer formation. Since construct HΔU1 co-distributed with MyHC in spindle-like structures, domains H1-3 must contain some myosin recognition sequences that are incapable of crosslinking MyHC into bundles.

In these experiments we have utilized the non-muscle, COS cell system pioneered by Leinwand and Winkelmann(Vikstrom et al., 1993;Moncman et al., 1993) to explore the mechanisms by which two myosin-binding proteins, C and H,influence the intracellular organization and polymerization of skeletal myosin in the absence of other sarcomeric proteins. This experimental model provides an intracellular assay system to identify those domains in the MyBPs that bind to and co-distribute with MyHC and those which function in cable formation, an intracellular assay of myosin crosslinking. We demonstrated that the C-terminal IgI domains of MyBP-C and MyBP-H (domains C10 and H4, respectively)are capable of binding to and bundling the MyHC when these domains are transiently co-expressed with MyHC in COS cells. Co-expression of GFP- or myc-tagged MyBP with sarcomeric MyHC results in the formation of large co-polymeric, cable-like structures that are readily detected by either fluorescence or electron microscopy. Earlier studies(Moncman et al., 1993;Straceski et al., 1994;Vikstrom et al., 1993) have shown that overexpression of sarcomeric MyHC in non-muscle cells will result in the formation of short, spindle-like, birefringent aggregates of myosin filaments. However, when MyBP-C or -H is co-expressed with MyHC, the short spindle-like structures are no longer formed; instead the cells exhibit large co-polymeric cables that encircle the nucleus(Seiler et al., 1996). In this report, we have shown that the C-terminal IgI domain, also termed the 14 kDa MyBP-C peptide (Okagaki et al.,1993), is sufficient to reorganize the MyHC into these long cable-like structures. This is the same region of MyBP-C that has been identified in vitro as its LMM-binding domain(Alyonycheva et al., 1997;Okagaki et al., 1993). MyBP-C and MyBP-H lacking this domain do not form cables. However, MyBP-C or MyBP-H constructs that have been truncated to delete this IgI domain still colocalize with the MyHC spindle-like aggregates even though they fail to induce cable formation. These data suggest the presence of at least two independent domains in the C-terminal half of the MyBP molecule that are capable of interacting with MyHC, thus explaining their co-distribution in the spindle-like structures. Since one of these binding sites, the C-terminal IgI domain,induces cable formation, we suggest that the IgI domain can bind to more than one myosin molecule; thus it serves as a multivalent crosslinker. The other binding site(s) maps outside the C-terminal IgI domain of MyBP-C and MyBP-H;its precise location requires further study, but it is not likely to be the upstream myosin-binding site identified by Gautel and colleagues(Gruen and Gautel, 1999), as MyBP-H constructs containing only H1, H2 and H3 modules colocalize with the myosin spindles. We suggest that this second myosin-binding region may reside in the H1-3 and C7-9 modules of MyBP-H and MyBP-C, respectively. With the exception of the MyBP-C motif, defined as the region between domains C1 and C2 in cardiac MyBP-C (Gruen and Gautel,1999), all of the regions in MyBP-C and MyBP-H that interact with myosin have been ascribed to the last four domains of each molecule. These different functional domains are shown schematically inFig. 9. In vitro biochemical studies have shown that a myosin-binding domain is contained in the C-terminal IgI domain (Miyamoto et al.,1999; Okagaki et al.,1993), and the titin-binding domain is within the last three modules (Freiburg and Gautel,1996). In cultured skeletal myotubes, however, these domains are not sufficient for proper insertion of the molecule into the A-band; an additional FnIII module (C7) is needed. In the present report, we have shown in a non-muscle transfection system that C10 is both necessary and sufficient for the intracellular crosslinking of MyBP-C and sarcomeric MyHC that elicits cable formation. In addition, we provide evidence for a second myosin interaction region, capable of binding to but not crosslinking MyHC. This domain is upstream of C10, probably with the three domains N-terminal to C10 or H4.

In a previous study, we proposed a space-filling model for the interaction of C10 with the myosin rod (Miyamoto et al., 1999). A hypothetical structure of C10, assumed to be aβ-barrel (Okagaki et al.,1993), was generated on the basis of the cystallographic coordinates of telokin, an homologous sequence within the C-terminal region of smooth muscle myosin light chain kinase(Holden et al., 1992). To determine which amino acids on C10 interact with LMM, we generated charge reversal mutations in positively or negatively charged residues on the surface of this domain and assayed effects on binding to or polymerization of light meromyosin (LMM). Amino-acid residues that positively or negatively affected binding, but did not alter C10 structure — as determined by circular dichroism measurements — were hypothesized to form ionic interactions with the LMM polymer at physiological ionic strength. By examining the distribution of surface charges along the coiled-coil of the myosin rod, we were able to fit C10 into a surface groove of the thick filament backbone structure put forth by Chew and Squire, in both parallel and anti-parallel orientations (Chew and Squire,1995). We concluded that three quarters of the C10 surface interacted with myosin; a single molecule of C10 could potentially crosslink three or four molecules of the myosin rod, thus explaining the well known reduction in critical concentration for myosin polymerization caused by MyBP-C(Davis, 1988). This model of C10 interacting with multiple myosin molecules is consistent with the transfection data presented in the present report. In order for myosin filaments to form higher ordered structures, for example, the cables observed in our co-transfection experiments, single molecules of C10 must be capable of interacting with at least two or more myosin rods. In other words, the C10 domain must be capable of crosslinking parallel myosin rods, in effect behaving as a multivalent ligand. This property seems to be unique to the C-terminal IgI domains of MyBP-C or MyBP-H as recombinant MyBPs lacking these domains do not form cables in COS cells co-transfected with MyHC. The precise location of MyBP-C in the vertebrate thick filament is still uncertain, but clearly reactive epitopes must reside in an accessible location on the filament since antibodies to MyBP-C strongly decorate the A-band filaments.

Although capable of interacting with myosin in the COS cell system, C10 does not insert into the A-band when overexpressed in developing muscle cells(Gilbert et al., 1996). This domain of MyBP-C appears to have a lower affinity for myosin than full-length MyBP-C does: C10 is incapable of displacing full-length MyBP-C from myosin filaments (D.A.F., unpublished). We suggest that in the absence of full-length MyBP-C, C10 is capable of binding to myosin filaments and crosslinking them into bundles, but in native myofibrils this domain cannot displace endogenous MyBP-C, thus explaining its inability to target to the A-band in developing muscle (Gilbert et al.,1996).

Under suitable salt conditions, full-length myosin polymerizes in vitro to form bipolar structures that resemble native thick filaments, but even with these impressive self-assembly characteristics, the in vitro filaments exhibit wide variations in length, rarely matching the in vivo situation(Barral and Epstein, 1999;Harrington and Joseph, 1968;Huxley, 1963;Lowey et al., 1969;Philpott and Szent-Gyorgyi,1954). LMM also aggregates in vitro at physiological ionic strength, but its polymers are neither filamentous nor typically bipolar(Atkinson and Stewart, 1991). It is likely that other components of the cell (e.g., microtubules and intermediate filaments) and of the thick filament (e.g., titin, myomesin and the MyBPs) play important but poorly understood roles in this process. In differentiating chick skeletal muscle, the spatial and temporal expression of MyBP-C coincides with the emergence of myofibrillar cross-striations(Lin et al., 1994). Furthermore, C-terminal truncations of MyBP-C that lack C10 cause a dominant interference with myofibril assembly when overexpressed in cultured myotubes(Gilbert et al., 1996). These observations are consistent with the notion that MyBP-C may function as a rate-limiting molecule in the lateral registration of thick-filaments during A-band assembly. The impact of the MyBPs on myosin organization is readily apparent in the prominent reorganization of MyHC spindles to long peri-nuclear cables, which result from the co-expression of MyBP-C or MyBP-H with MyHC. A number of publications have demonstrated that MyBP-C has a significant effect on the length of myosin polymers formed in vitro(Davis, 1988;Miyahara and Noda, 1980). In addition to its effects on lateral thick filament alignment, MyBP-C could function in both the organization and stability of already formed thick filaments.

In conclusion we have defined the essential domains of MyBP-C and MyBP-H responsible for intracellular myosin crosslinking. Our findings suggest the presence of a third myosin interaction domain shared by MyBP-C and MyBP-H. It remains to be seen how this region of the MyBPs functions within developing and mature striated muscle. Although, MyBP-H shares many structural and functional similarities with MyBP-C, especially in the C-terminal one-third of the latter molecule, MyBP-H lacks all of the putative regulatory residues found upstream in MyBP-C. Further studies on MyBP-H are clearly warranted,since its function in vivo is unknown.

We express our appreciation to Takashi Mikawa for his many helpful suggestions and criticisms during the course of this study. Additional thanks are extended to Catarina Miyamoto, Patrick Nahirney, Takashi Obinata and Naruki Sato for their valuable advice. This work was supported by grants from the National Institutes of Health (AR32157) and a postdoctoral fellowship from NHLBI (5T32 HL07423) to R.E.W.