The Hydrophilic Domain of Small Ankyrin-1 Interacts with the Two N-terminal Immunoglobulin Domains of Titin*

Little is known about the mechanisms that organize the internal membrane systems in eukaryotic cells. We are addressing this question in striated muscle, which contains two novel systems of internal membranes, the transverse tubules and the sarcoplasmic reticulum (SR). Small ankyrin-1 (sAnk1) is an ∼17-kDa transmembrane protein of the SR that concentrates around the Z-disks and M-lines of each sarcomere. We used the yeast two-hybrid assay to determine whether sAnk1 interacts with titin, a giant myofibrillar protein that organizes the sarcomere. We found that the hydrophilic cytoplasmic domain of sAnk1 interacted with the two most N-terminal Ig domains of titin, ZIg1 and ZIg2, which are present at the Z-line in situ. Both ZIg1 and ZIg2 were required for binding activity. sAnk1 did not interact with other sequences of titin that span the Z-disk or with Ig domains of titin near the M-line. Titin ZIg1/2 also bound T-cap/telethonin, a 19-kDa protein of the Z-line. We show that titin ZIg1/2 could form a three-way complex with sAnk1 and T-cap. Our results indicate that titin ZIg1/2 can bind sAnk1 in muscle homogenates and suggest a role for these proteins in organizing the SR around the contractile apparatus at the Z-line.

Little is known about the mechanisms that organize the internal membrane systems in eukaryotic cells. We are addressing this question in striated muscle, which contains two novel systems of internal membranes, the transverse tubules and the sarcoplasmic reticulum (SR). Small ankyrin-1 (sAnk1) is an ϳ17-kDa transmembrane protein of the SR that concentrates around the Z-disks and M-lines of each sarcomere. We used the yeast twohybrid assay to determine whether sAnk1 interacts with titin, a giant myofibrillar protein that organizes the sarcomere. We found that the hydrophilic cytoplasmic domain of sAnk1 interacted with the two most N-terminal Ig domains of titin, ZIg1 and ZIg2, which are present at the Z-line in situ. Both ZIg1 and ZIg2 were required for binding activity. sAnk1 did not interact with other sequences of titin that span the Z-disk or with Ig domains of titin near the M-line. Titin ZIg1/2 also bound T-cap/ telethonin, a 19-kDa protein of the Z-line. We show that titin ZIg1/2 could form a three-way complex with sAnk1 and T-cap. Our results indicate that titin ZIg1/2 can bind sAnk1 in muscle homogenates and suggest a role for these proteins in organizing the SR around the contractile apparatus at the Z-line.
As striated muscle develops, the basic contractile unit, the sarcomere, is assembled before the transverse tubules (T-tubules) 1 and the sarcoplasmic reticulum (SR) mature (1,2). The contractile cycle in striated muscle normally requires the spread of the action potential along the T-tubules into the interior of the muscle fiber, where depolarization induces the release of Ca 2ϩ from the terminal cisternae of the SR, causing contraction (2,3). Relaxation follows the re-uptake of Ca 2ϩ into a distinct domain of the SR, which has been referred to as the longitudinal or network SR (4). Typically, the network SR is positioned around the Z-disks and M-lines of each sarcomere, but the structural elements that determine its location have not been determined. Early ultrastructural studies demonstrated the presence of numerous filaments joining the periphery of sarcomeric Z-disks to adjacent SR membranes (5), but the molecular identity of these structures remains elusive.
We have searched for protein partners of small ankyrin-1 (sAnk1), a muscle-specific isoform of the erythroid ankyrin-1 gene that is concentrated in the network SR of striated muscle fibers, surrounding the Z-disks and M-lines (6). Ankyrins are a family of proteins that possess binding sites for diverse integral membrane proteins as well as cytoskeletal components (7)(8)(9).
To date, molecular cloning has identified three distinct ankyrin genes in mammals (Ank1, Ank2, and Ank3) that are expressed as tissue-specific, alternatively spliced isoforms (10 -12). Ank1 is expressed predominantly in erythroid cells, striated muscle, and brain (13)(14)(15); Ank2 in brain and cardiac muscle (16 -19); and Ank3 in cells of epithelial origin and striated muscle as well as in lysosomes and Golgi membranes in a wide variety of cells (20 -23). The large canonical ankyrins share a similar structure, consisting of an N-terminal ϳ89-kDa membranebinding domain, a central ϳ62-kDa spectrin-binding domain, and a C-terminal ϳ55-kDa regulatory domain (10,11). In striated muscle, the products of the Ank1 gene include the large (ϳ210 kDa) and small (ϳ17-19 kDa) ankyrin isoforms (6,15). sAnk1 lacks both the membrane-and spectrin-binding regions of the larger form and has a C-terminal domain that is significantly shortened (24,25). The N-terminal portion of sAnk1 contains a unique 73-amino acid segment, whereas the C-terminal 82 residues are identical to the C-terminal sequence of the large ϳ210-kDa ankyrin-1. The first 29 residues of sAnk1 are highly hydrophobic and target the molecule to the SR membrane, whereas the remaining 126 amino acids extend into the myoplasm. 2 Thus, the hydrophilic tail of sAnk1 is appropriately oriented in the cytoplasm of striated muscle fibers to serve as a binding site for sarcomeric proteins.
Here, we describe a direct and specific association between sAnk1 and titin, also known as connectin (25)(26)(27)(28)(29). Titin is a giant (ϳ2.7-4 MDa) protein that extends from the Z-disk to the M-line within the sarcomere, which it helps to organize. It is highly modular: ϳ90% of its mass is composed of repeating Ig-C2 and fibronectin-3-like domains that provide binding sites for myofibrillar proteins (31,32). The remaining ϳ10% consists of unique non-repetitive sequence motifs, including phosphorylation sites, binding sites for muscle-specific calpain proteases, and C-terminal Ser/Thr kinase domains (30,(33)(34)(35).
The C-terminal 2 MDa of titin are located within the A-band, where titin tightly associates with the myosin thick filaments and several A-band proteins such as C-protein, M-protein, and myomesin (36 -38). The most C-terminal end of the molecule (ϳ200 kDa), which is embedded in the M-line, contains a Ser/ Thr kinase domain, which implicates titin in myofibrillar signal transduction pathways (37)(38)(39)(40). In the I-band, titin (ϳ800 kDa to 1.5 MDa) carries proline/glutamate/valine/lysine-rich sequences, which confer great extensibility to the titin filaments (35,(41)(42)(43)(44), in addition to numerous Ig domains. At the junction of the I-band with the Z-disk, titin interacts with the actin thin filaments, although it is still unclear which titin motifs mediate this interaction (45,46). The N-terminal 80-kDa region of titin spans the entire Z-disk (47). Several copies of a 45-residue repeat, called the Z-repeat, bind ␣-actinin within the Z-disk (47)(48)(49). The two most N-terminal Ig domains of titin, which are constitutively expressed in all titin isoforms and reside in the periphery of the Z-disk, bind a recently identified, 19-kDa protein of striated muscle, referred to as T-cap or telethonin (47,50).
Titin has two functions in striated muscle: as a "molecular blueprint" for sarcomeric protein assembly during myofibrillogenesis and as a "molecular spring" that maintains myofibrillar integrity during contraction, relaxation, and stretch (27,30,32). Our results show that, in addition to binding T-cap, the two N-terminal Ig domains of titin interact specifically with sAnk1, suggesting that titin also coordinates the assembly of the contractile apparatus with the network SR that surrounds the Z-disk.
For generation of MBP-T-cap fusion protein, the full-length pET9D-T-cap plasmid (a generous gift from Drs. S. Labeit (European Molecular Biology Laboratory, Heidelberg, Germany) and C. C. Gregorio (University of Arizona, Tucson, AZ)) was used as template to obtain a PCR fragment that contained amino acids 1-140 (47). Primers 7 (5Ј-ACGT-GAATTCATGGCTACCTCAGAGCTG-3Ј, sense) and 8 (5Ј-ACGTGTC-GACTCATGTCTCCAGCGCCAG-3Ј, antisense), carrying EcoRI and SalI sites, respectively, were used for insertion into the pMAL-c2X vector. T-cap-(1-140) was also introduced into the pGEX4T-1 vector at EcoRI/XhoI sites (XhoI and SalI have compatible ends) to produce a GST fusion protein. The authenticity of the obtained constructs was verified by sequencing analysis. GST and MBP recombinant polypeptides were expressed by induction with 0.5 mM isopropyl-␤-D-thiogalactopyranoside for 3 h and purified by affinity chromatography on glutathione-agarose (for GST fusion proteins) (Amersham Biosciences) or amylose resin (for MBP fusion proteins) (New England Biolabs, Inc.) columns following the manufacturers' instructions.
Liquid ␤-galactosidase assays were performed as described in the Clontech Yeast Protocols Handbook using chlorophenol red ␤-D-galactopyranoside as substrate. For each interaction tested, four independent colonies were assayed, and each experiment was repeated twice. Results represent average values.
Blot Overlay-The blot overlay assays were performed as previously described with minor modifications (51). Briefly, ϳ2.5 g of bacterially expressed, affinity-purified GST and GST-ZIg1/2 proteins were separated by 12% SDS-PAGE and transferred to nitrocellulose. Nonspecific sites on the nitrocellulose membranes were blocked in buffer A (50 mM Tris (pH 7.2), 120 mM NaCl, 3% bovine serum albumin, 2 mM dithiothreitol, 0.5% Nonidet P-40, and 0.1% Tween 20) plus protease inhibitors for 3 h at 25°C and then incubated with 3 g/ml MBP-sAnk1 fusion protein in the same buffer for 16 h at 4°C. Blots were washed five times (15 min each) with buffer A and once with buffer B (1ϫ phosphatebuffered saline (pH 7.2), 10 mM NaN 3 , and 0.1% Tween 20). Subsequently, they were incubated in buffer C (1ϫ phosphate-buffered saline (pH 7.2), 10 mM NaN 3, 0.1% Tween 20, and 3% dry milk) and probed with antibodies to sAnk1, diluted in buffer C. In a set of parallel assays, increasing concentrations of affinity-purified GST-ZIg1/2 fusion protein (i.e. 5 and 10 g) were added to buffer A along with MBP-sAnk1, and blots were processed as just described.
In Vitro "Competition" Assay-Equivalent amounts of GST protein, GST-sAnk1, and GST-T-cap bound to glutathione matrices were allowed to interact with 5 g of recombinant MBP-ZIg1/2, MBP-sAnk1, or MBP-T-cap in 250 l of binding buffer (50 mM Tris (pH 7.2), 120 mM NaCl, 10 mM NaN 3 , 2 mM dithiothreitol, 0.1% Tween 20, and 10 mM maltose plus protease inhibitors) for 12 h at 4°C. Subsequently, the supernatants were removed, and the glutathione beads were extensively washed with a solution containing 50 mM Tris (pH 7.2), 120 mM NaCl, 10 mM NaN 3 , and 0.1% Tween 20. In a parallel set of experiments, GST-sAnk1 and GST-T-cap attached to glutathione matrices were initially allowed to interact with 5 g of bacterially expressed MBP-ZIg1/2 for 12 h at 4°C. Following removal of the supernatants, 5 g of affinity-purified MBP-T-cap or MBP-sAnk1 were added to the GST-sAnk1⅐MBP-ZIg1/2 or GST-T-cap⅐MBP-ZIg1/2 complexes, respectively, and allowed to interact for another 12 h in the cold. At the end of the incubation period, the glutathione beads were washed four times (15 min each) and dissolved in 2ϫ SDS Laemmli sample buffer. The soluble proteins were fractionated on 12% SDS-PAGE, transferred to nitrocellulose, and probed with the appropriate antibodies.
Materials-Unless otherwise noted, all reagents were from Sigma and were the highest grade available.

Binding of sAnk1 to the Two Most N-terminal Ig Domains of
Titin Shown with the Yeast Two-hybrid Assay-sAnk1 is an ϳ17-19-kDa integral membrane protein of the network SR that has a 126-amino acid sequence extending into the sarcoplasm surrounding Z-disks and M-lines (6). 2 We used the yeast two-hybrid assay to test the idea that the hydrophilic sequence of sAnk1 (sAnk1-(29 -155)) interacts with the giant myofibrillar protein titin, which spans each half-sarcomere from the Z-disk to the M-line. We inserted cDNA encoding the hydrophilic cytoplasmic domain of sAnk1 (sAnk1-(29 -155)) (25) into the yeast two-hybrid pLexA bait vector (Fig. 1A) and assayed its ability to interact with the N-terminal ϳ80-kDa portion of titin that resides in the Z-line (47), expressed by a series of constructs inserted into the yeast two-hybrid pB42AD prey vector (Fig. 1B). Specifically, the PCR products of titin we assayed were ZIg1/2 (amino acids 1-200), ZIg3 (amino acids 201-557), the Z-repeats or Zr domain (amino acids 558 -910), and ZIg4/5 (amino acids 911-1118) (see "Experimental Procedures"). Yeast two-hybrid analysis followed by qualitative liquid ␤-galactosidase assays (Fig. 1C) indicated that sAnk1-(29 -155) specifically interacted with the two most N-terminal Ig domains of titin, ZIg1/2 (amino acids , which reside at the edge of the Z-disk (28,47). No specific association between sAnk1-(29 -155) and the remaining ϳ60-kDa portion of Z-disk titin could be detected (Fig. 1C).
In additional tests of the specificity of the interaction with ZIg1/2, we generated two additional "prey" constructs encoding tandem Ig domains that reside in the M-line region of titin (37)(38)(39). These included MIg1/2 (amino acids 25250 -25422) and MIg5/6 (amino acids 26281-26478). When their ability to interact with sAnk1-(29 -155) was tested in a yeast two-hybrid assay, no specific interaction was observed (Fig. 1C). Although we were unable to test the ability of other M-line domains of titin to interact with sAnk1 in this assay, our results suggest that sAnk1 interacts preferentially with Ig domains at the N-terminal region of titin, located at the Z-disk.

FIG. 1. Yeast two-hybrid analysis identifies titin as a cytoplasmic ligand for sAnk1.
A, sAnk1 consists of an hydrophobic N-terminal sequence that anchors the molecule to the SR membrane, followed by a hydrophilic sequence that is exposed on the cytoplasmic face of the membrane. We introduced the C-terminal hydrophilic "tail" of sAnk1 (amino acids 29 -155) into the pLexA bait vector of the Matchmaker yeast two-hybrid system. B, consecutive PCR fragments spanning the entire length of the Z-band region of titin (i.e. ZIg1/2, ZIg3, Zr, and ZIg4/5) or parts of the M-line region of titin (e.g. MIg1/2 and MIg5/6) were inserted into the pB42AD prey vector. C, yeast two-hybrid analysis followed by liquid ␤-galactosidase (␤-gal) assay indicated that sAnk1 specifically interacted with the two most N-terminal Ig domains of titin, ZIg1/2. Other regions of titin from the N-terminal region, associated with Z-disks, or the C-terminal region, proximal to M-lines, failed to interact with sAnk1.

FIG. 2. Identification of the binding domains of sAnk1 and titin.
To determine the regions of sAnk1 and titin ZIg1/2 required for their association, we generated a series of deletion constructs and introduced them into the bait and prey vectors, respectively. A, yeast two-hybrid analysis followed by ␤-galactosidase (␤-gal) assay showed that amino acid residues 61-89 of sAnk1 mediate the binding of sAnk1 to titin. B, both N-terminal titin Ig domains are required for binding to sAnk1. tosidase assays with different combinations of truncated sAnk1 and titin ZIg1/2 constructs indicated that sAnk1-A (amino acids 29 -89) and sAnk1-C (amino acids 61-130) elicited similar results, suggesting that the amino acids shared by the two subfragments (i.e. amino acids 61-89) contain the binding site for titin ZIg1/2 ( Fig. 2A). Furthermore, both ZIg1 and ZIg2 were required for titin to bind to sAnk1 because sAnk1 failed to interact with the individual Ig domains (Fig. 2B).
Binding of sAnk1 in Muscle Homogenates to GST-ZIg1/2 in a Pull-down Assay-To confirm the specificity of the interaction between sAnk1 and titin ZIg1/2, we performed a GST pulldown assay using homogenates of skeletal muscle from adult rats. We expressed ZIg1/2, ZIg3, Zr, ZIg4/5, MIg1/2, and MIg5/6 as GST fusion proteins (Fig. 3). The calculated molecular masses of the GST fusion proteins are ϳ47, ϳ65, ϳ64, ϳ48, ϳ44, and ϳ47 kDa, respectively (Fig. 3A). GST-ZIg3 and GST-Zr showed some degradation, probably due to endogenous bacterial proteases, whereas GST-ZIg4/5 migrated with an apparent molecular mass of ϳ60 kDa instead of the calculated ϳ48 kDa. Equivalent amounts of these proteins and control GST (25 kDa) were bound to glutathione matrices and incubated with homogenates of quadriceps muscle. The matrix-  bound GST fusion proteins were examined by Western blotting for their ability to adsorb native sAnk1. Only GST-ZIg1/2 specifically retained native sAnk1 (Fig. 3B). None of the other Z-disk or M-line fragments of titin that we examined adsorbed sAnk1 from muscle homogenates, consistent with the yeast two-hybrid data (Fig. 1C). These results indicate that the two most N-terminal Ig domains of titin can bind to sAnk1 in muscle homogenates, consistent with the association of sAnk1 with this region of titin in vivo.
To determine whether sAnk1 binds titin ZIg1/2 directly, we performed an overlay assay with bacterially expressed MBP-sAnk1-(29 -155) (ϳ56 kDa) and GST-ZIg1/2 (ϳ47 kDa) fusion proteins (Fig. 4A). Equivalent amounts of GST-ZIg1/2 and control GST were subjected to SDS-PAGE, transferred to nitrocellulose membranes, and overlaid with affinity-purified MBP-sAnk1- (29 -155). Recombinant sAnk1 specifically bound to GST-ZIg1/2 fusion protein, but not to control GST protein, as shown by immunoblotting with antibodies to sAnk1 (Fig. 4B). The ability of recombinant sAnk1 to bind GST-ZIg1/2 directly was inhibited by including soluble GST-ZIg1/2 (5 and 10 g) in the overlay buffer along with MBP-sAnk1-(29 -155) (Fig. 4B), confirming the specificity of the direct interaction between sAnk1 and titin ZIg1/2. sAnk1 and T-cap Bind to Titin ZIg1/2 Simultaneously-The two most N-terminal Ig domains of titin were shown previously to bind to a striated muscle-specific, 19-kDa Z-disk protein named T-cap or telethonin (47,50). Our yeast two-hybrid and in vitro binding studies indicated that titin ZIg1/2 also bound to a 29-amino acid segment (i.e. amino acids 61-89) within the hydrophilic segment of sAnk1. To determine whether titin ZIg1/2 can bind simultaneously to sAnk1 and T-cap or whether binding to one abolishes binding to the other, we performed an in vitro competition assay using bacterially expressed sAnk1, T-cap, and titin ZIg1/2 proteins.
Subcellular Distribution of sAnk1, Titin ZIg1/2, and T-cap in Adult Skeletal Muscle Fibers-It has been well documented that sAnk1, titin ZIg1/2, and T-cap are present at the Z-lines of sarcomeres (6,47,54). To examine their topography with re- Titin is a giant myofibrillar protein that spans the half-sarcomere, with its N terminus near the Z-disk and its C terminus embedded in the M-line. Residues 1-200 of titin contain two Ig domains (ZIg1 and ZIg2) that reside at the periphery of the Z-disk and that are constitutively expressed in all known isoforms of titin. Both ZIg1 and ZIg2 are required for a specific, direct, and presumably physiologic association with the cytoplasmic sequence of sAnk1 as well as with the Z-disk protein T-cap. sAnk1 does not compete with T-cap for binding to titin ZIg1/2, suggesting that a complex of these three proteins can form in situ. However, their relative locations suggest that a three-way complex can form only at the periphery of the Z-disk. The binding of sAnk1 to titin ZIg1/2, whether the latter is also bound to T-cap or not, implies a role for these proteins in coordinating the organization of the network SR with that of nearby contractile structures. spect to the Z-disk, we used antibodies against sAnk1, titin ZIg1/2, and T-cap to label longitudinal sections and cross-sections of muscle fibers by immunofluorescence, followed by confocal microscopy. As expected, each of the antibodies labeled the Z-lines in longitudinal sections of adult rat quadriceps muscle (Fig. 6, A-C); in addition, anti-sAnk1 antibody also labeled M-lines (Fig. 6A) (6). Cross-sections of muscle labeled with the same panel of antibodies showed sAnk1 in a reticular pattern, consistent with its distribution in the network SR (Fig.  6D) (6). By contrast, both titin ZIg1/2 and T-cap concentrated at the Z-disks (Fig. 6, E and F). No labeling was detected when primary antibodies were replaced with nonimmune rabbit or goat IgG (Fig. 6, G and H). Thus, titin ZIg1/2 and T-cap are present at the Z-disk, whereas sAnk1 is concentrated in the SR surrounding the Z-disk. Thus, if a complex between sAnk1 and titin forms in skeletal myofibers, with or without T-cap/telethonin, it would be limited to the periphery of the Z-disk. DISCUSSION A key question in the biology of striated muscle is how the internal membranes of the SR and the T-tubules become precisely aligned with the contractile apparatus. We have begun to address this question by identifying ligands of sAnk1, a structural protein of the network SR that, we hypothesize, helps to coordinate the alignment of the SR with nearby M-lines and Z-disks. Because titin serves as a molecular blueprint for the assembly of other myofibrillar elements, we postulated that it might also provide a site for anchoring the SR at the level of the Z-disk. We found that the hydrophilic sequence of sAnk1, which extends from the SR membrane into the sarcoplasm (6), specifically and directly interacted with the two most N-terminal Ig domains of the giant myofibrillar protein titin. These two domains, ZIg1 and ZIg2, are present in all titin muscle isoforms identified to date and are localized at the periphery of the Z-disk lattice (47). These domains of titin are therefore appropriately positioned to anchor sAnk1 in the network SR to the Z-disk.
sAnk1 carries a C-terminal hydrophilic sequence that differs significantly from the C-terminal portion of the large canonical form of ankyrin-1. It is considerably shortened (ϳ14 versus ϳ55 kDa) and contains a unique peptide sequence (amino acids 29 -73); the remaining 82 residues (amino acids 74 -155) are shared by both small and large splice forms of ankyrin-1, followed by a common translation stop codon (24,25). The results of our yeast two-hybrid experiments suggest that a peptide 29 amino acids long (residues 61-89) contains the minimal sequence in sAnk1 with binding activity for titin ZIg1/2. Thus, the titin-binding site on sAnk1 may be comprised of residues unique to sAnk1 (i.e. amino acids 61-73) and amino acids shared by both small and large forms of Ank1 (i.e. amino acids 74 -89). It is of course possible that the binding of sAnk1 to titin ZIg1/2 requires only some of these 29 residues. Future studies will delineate more precisely whether amino acids that are unique to sAnk1 or shared with larger forms of Ank1 are needed to form the titin-binding site on sAnk1.
Both the ZIg1 and ZIg2 domains of titin were required for binding to sAnk1. This is not surprising, as the binding of many titin ligands, such as myomesin, M-protein, myosin-binding protein C, obscurin, and T-cap/telethonin, requires the presence of pairs of titin Ig domains (37, 38, 41, 47, 50, 54 -56). Indeed, the Ig domains of the defining members of the Ig superfamily, the immunoglobulins, act in tandem to form the binding sites of antibodies (57). The requirement of both ZIg1 and ZIg2 implies that the binding site of titin for sAnk1 includes residues from both Ig domains. Preliminary observations from our laboratory indicated that sAnk1 may homodimerize or multimerize in vivo. Thus, a sAnk1 dimer or multimer may be the active ligand for the ZIg1/2 region of titin.
Interestingly, ZIg1 and ZIg2 share a highly conserved peptide (i.e. SGXYS⌽XATN, where X is a nonconserved amino acid and ⌽ is a nonpolar hydrophobic residue) that could serve as the binding site of a potential sAnk1 dimer. Although further experimentation will be required to address this issue, the results of our in vitro binding studies suggest that dimerization of sAnk1 may not be required for it to bind to titin ZIg1/2.
The two N-terminal Ig domains of titin were previously shown to contain the binding site of a Z-disk protein referred to as T-cap or telethonin (47,50). Similar to sAnk1, the binding of T-cap to titin requires the presence of both ZIg1 and ZIg2. Titin ZIg1/2 can bind simultaneously to both sAnk1 and T-cap, indicating that these proteins can form a three-way complex.
The subcellular location of these proteins places limits on where this complex could form. T-cap and titin ZIg1/2 co-localize at the edge of the Z-disk lattice, where their binding is believed to anchor the N-terminal portion of titin to the Z-disk (47). By contrast, sAnk1 is limited to the SR at the periphery of the Z-disk, where its C-terminal hydrophilic sequence extends from the SR membrane (6). 2 sAnk1 is therefore likely to interact with titin and possibly form a three-way complex with T-cap only at the periphery of the Z-disk (Fig. 7). As postulated, this interaction would link the network SR, where sAnk1 is concentrated, to the Z-disk.
T-cap was recently shown to bind to MinK, the ␤-subunit of the potassium channel of the transverse tubular membranes (58). This suggests that, just as sAnk1 links titin at the periphery of the Z-disk to the SR, T-cap may link the same population of titin molecules to the T-tubules. These interactions can occur in mammalian cardiac muscle and in avian and amphibian striated muscle, where both the SR and T-tubules are located around the Z-disk. In mammalian skeletal muscle, however, this function is likely to be limited to sAnk1, as only the SR is concentrated around the Z-disk. The T-tubules in skeletal muscle are present at the junction of the A-and I-bands and so are unlikely to be anchored via T-cap to titin at the Z-disk. Thus, through its ability to bind simultaneously to sAnk1 and T-cap, titin may simultaneously serve as a scaffold for assembling not only the contractile apparatus, but also the network SR and, in many (but not all) striated muscles, the T-tubules.
Titin is the second myofibrillar protein that we have identified as a major cytoplasmic ligand for sAnk1: obscurin also binds to sAnk1 (53). Obscurin is a giant (ϳ800 kDa) sarcomeric Rho guanine nucleotide exchange factor protein with homology to titin (55). Immunolocalization studies have shown that obscurin closely surrounds the myofibrils at the Z-disks and Mlines of each sarcomere (53). Thus, it appears that, whereas titin associates with sAnk1 at the level of the Z-disk, obscurin interacts with sAnk1 around both Z-disks and M-lines.
We have previously shown that amino acids 61-130 of the C-terminal hydrophilic sequence of sAnk1 contain the binding site for obscurin (53). In the present study, we have shown that residues 61-89 of sAnk1 are likely to contain the binding site for titin ZIg1/2. These findings suggest that the sAnk1 binding sites for titin and obscurin may overlap. Future studies will examine whether titin and obscurin compete with each other for binding to sAnk1 around the Z-disk in developing or mature striated muscle and whether one or both of these giant sarcomeric proteins play an important role in aligning the SR with the contractile apparatus.