Multiple Molecular Interactions Implicate the Connectin/Titin N2A Region as a Modulating Scaffold for p94/Calpain 3 Activity in Skeletal Muscle*

p94/calpain 3 is a skeletal muscle-specific Ca2+-regulated cysteine protease (calpain), and genetic loss of p94 protease activity causes muscular dystrophy (calpainopathy). In addition, a small in-frame deletion in the N2A region of connectin/titin that impairs p94-connectin interaction causes a severe muscular dystrophy (mdm) in mice. Since p94 via its interaction with the N2A and M-line regions of connectin becomes part of the connectin filament system that serves as a molecular scaffold for the myofibril, it has been proposed that structural and functional integrity of the p94-connectin complex is essential for health and maintenance of myocytes. In this study, we have surveyed the interactions made by p94 and connectin N2A inside COS7 cells. This revealed that p94 binds to connectin at multiple sites, including newly identified loci in the N2A and PEVK regions of connectin. Functionally, p94-N2A interactions suppress p94 autolysis and protected connectin from proteolysis. The connectin N2A region also contains a binding site for the muscle ankyrin repeat proteins (MARPs), a protein family involved in the cellular stress responses. MARP2/Ankrd2 competed with p94 for binding to connectin and was also proteolyzed by p94. Intriguingly, a connectin N2A fragment with the mdm deletion possessed enhanced resistance to proteases, including p94, and its interaction with MARPs was weakened. Our data support a model in which MARP2-p94 signaling converges within the N2A connectin segment and the mdm deletion disrupts their coordination. These results also implicate the dynamic nature of connectin molecule as a regulatory scaffold of p94 functions.

p94/calpain 3 is a skeletal muscle-specific Ca 2؉ -regulated cysteine protease (calpain), and genetic loss of p94 protease activity causes muscular dystrophy (calpainopathy). In addition, a small in-frame deletion in the N2A region of connectin/titin that impairs p94-connectin interaction causes a severe muscular dystrophy (mdm) in mice. Since p94 via its interaction with the N2A and M-line regions of connectin becomes part of the connectin filament system that serves as a molecular scaffold for the myofibril, it has been proposed that structural and functional integrity of the p94-connectin complex is essential for health and maintenance of myocytes. In this study, we have surveyed the interactions made by p94 and connectin N2A inside COS7 cells. This revealed that p94 binds to connectin at multiple sites, including newly identified loci in the N2A and PEVK regions of connectin. Functionally, p94-N2A interactions suppress p94 autolysis and protected connectin from proteolysis. The connectin N2A region also contains a binding site for the muscle ankyrin repeat proteins (MARPs), a protein family involved in the cellular stress responses. MARP2/Ankrd2 competed with p94 for binding to connectin and was also proteolyzed by p94. Intriguingly, a connectin N2A fragment with the mdm deletion possessed enhanced resistance to proteases, including p94, and its interaction with MARPs was weakened.
Our data support a model in which MARP2-p94 signaling converges within the N2A connectin segment and the mdm deletion disrupts their coordination. These results also implicate the dynamic nature of connectin molecule as a regulatory scaffold of p94 functions.
Calpain (EC 3.4.22.18, clan CA, family C2) is a family of Ca 2ϩ -requiring and papain-like proteases comprising the products of 15 different genes in humans (1)(2)(3). Calpain is considered a modulator protease, since it modulates functions of substrates. More than 10 calpain molecules are expressed in skeletal muscle, including p94/calpain 3, a skeletal muscle-specific isoform, as well as the conventionaland m-calpains, whose catalytic subunits are called calpain 1 and 2, respectively. The functions of calpain in skeletal muscle have been investigated regarding pathogenic conditions, such as atrophy and muscular dystrophy (4 -6), and the molecular mechanisms involved in myogenesis (7)(8)(9).
The physiological relevance of p94 to skeletal muscle integrity is an urgently pursued issue. Defective p94 protease activity by gene mutations causes a muscular dystrophy categorized as calpainopathy (10). The calpainopathy-type muscular dystrophy contrasts with other muscular dystrophies with regard to the mode of calpain involvement. Unlike the conventional calpains that are hyperactivated as a consequence of advanced dystrophic phenotype (i.e. an aberrant increase in intracellular [Ca 2ϩ ]) and, in turn, aggravate the symptoms (11), the dystrophic phenotypes of calpainopathy are caused by loss of p94 protease activity from skeletal muscle (12). Phenotypes of transgenic mice in which the p94 protease activity is manipulated in various ways show that the regulated proteolytic action of p94 on target proteins is critical for the maintenance of skeletal muscle functions (13)(14)(15)(16). Thus, identification of p94 substrates is a key to clarifying underlying mechanisms.
One of the interesting properties of p94 is its very rapid and exhaustive autolysis in protein expression systems examined so far, whereas native p94 protein can be detected at significant quantities in skeletal muscle, where it constitutes a complex with other myofibril components, especially connectin/titin (17)(18)(19)(20)(21). Therefore, connectin is a candidate that regulates p94 stability and activity. Two distinct regions in connectin, the N2A and C terminus regions, have been identified as p94-binding sites by yeast two-hybrid (YTH) 4 screening (19,22). These interactions are thought to impede autolytic disassembly of p94, which has been demonstrated for the N2A fragment in a proteinase-trapping assay, a phenomenon we call "p94 trapping" (23,24). Interestingly, in primary cultures of skeletal muscle cells, p94 translocates from the M-line to the N2A region as myofibrillogenesis proceeds or as the sarcomere lengthens (25). Together, these observations suggest that p94 and connectin function as a complex in skeletal muscle cells.
The importance of connectin as a scaffold for multiple molecular interactions for both the structural and force-generating elements has been recognized (26 -29). In fact, the N2A region of connectin appears to be a major site involved in signal transduction in striated muscle tissues. N-terminal to the binding site for p94 in the N2A region of connectin is a specific insertion sequence ("is"), which binds to muscle ankyrin repeat proteins (MARPs) (30) (Fig. 1A). There are three MARP paralogues: MARP1 expressed primarily in cardiac muscle (also referred to as CARP or Ankrd1), MARP2 (Ankrd2 or Arpp) expressed primarily in skeletal muscle, and MARP3 (DARP) expressed in both muscle tissues. MARPs become strongly upregulated in myocytes under a variety of acute stresses: cardiac injury and hypertrophy (MARP1) (31)(32)(33), stretch or denervation of skeletal muscle (MARP2) (34,35), and metabolic challenge (MARP3) (36). MARP1 and -2 are up-regulated in skeletal muscle in chronic pathogenic conditions, such as Duchenne muscular dystrophy (37)(38)(39). MARPs are localized both in the nucleus and the sarcomere N2A region, and MARP1 and -2 also interact with nuclear factors involved in transcriptional regulation (40 -42). Therefore, MARPs are thought to link nuclear transcriptional activity and sarcomere functions.
Another example of the physiological significance of the connectin N2A region is that the homozygosity of the in-frame deletion in the connectin N2A region, Ttn mdm/mdm , causes muscular dystrophy, mdm, in mice (43)(44)(45). The mdm deletion overlaps partially with the C terminus of the p94 binding region and compromises the interaction between connectin I81-I83 (Fig. 1A) and p94 (23,44). Intriguingly, in mdm skeletal muscle, MARP1 and -2 are up-regulated significantly, respectively, and their specific targeting to the sarcomeric I-band region were observed in microscopic analysis of myofibrils from Ttn mdm/mdm mice (44).
Most previous studies have focused on a single molecule, such as calpain, MARP, or connectin. Here, we have focused on the simultaneous relationships among these three molecules.
We examined the effect of the mdm deletion on properties of connectin fragments containing binding sites for both MARPs and p94. As a result, the sensitivity of N2A connectin to proteases, including p94, was shown to be suppressed by the mdm deletion. Identification of multiple binding sites for p94 in N2A connectin and of N2A connectin and MARP2 as substrates for p94 protease activity provided a new aspect of the relationships among these molecules. We propose that p94-connectin interaction regulates and facilitates proteolytic modulation by p94 on the signaling cascade through N2A connectin and MARP2.

EXPERIMENTAL PROCEDURES
Mouse Experiments-All procedures used for experimental animals were approved by the Experimental Animal Care and Use Committee of the Tokyo Metropolitan Institute of Medical Science.
cDNA Constructs-The cDNAs for human and mouse p94/ calpain 3 were subcloned into expression vectors pSRD and pAS2-1 or pAS2-1c for protein expression in mammalian and yeast cells, respectively, as described previously (12,46). The yeast expression vectors for the full-length and IS2 region of rat p94 were described previously (19). The mouse cDNAs encoding various regions of N2A connectin/titin and full-length MARP1, -2, and -3 were amplified by PCR from mouse skeletal muscle cDNA using Pfu DNA polymerase (Stratagene, La Jolla, CA). Connectin fragments were expressed as the N-terminally FLAG-tagged proteins using the expression vector pSRD (Table 1). MARP1 with an N-terminal hemagglutinin (HA) or MYC epitope was expressed using pSRD. MARP2 and -3 were expressed using pcDNA3.1 as the N-terminally HA-tagged protein (47). An expression vector, pACT2 (U29899; Clontech, Mountain View, CA) was used to express cloned connectin cDNAs in a YTH assay. p94:D607A was constructed by introducing an Asp 607 3 Ala mutation into the human p94 cDNA and inserting this into the XbaI-SacI sites of human p94 cDNA in pFastBac1 (Invitrogen). This mutant p94 was characterized using p94 trapping (23) and used for the biochemical analyses of p94 as described previously (24). cDNA for the FLAG-tagged I80-PEVK connectin fragment was subcloned into pFastBac-HTa. Enzymes used for manipulating recombinant DNA were purchased from Takara Bio (Shiga, Japan) or New England Biolabs (Ipswich, MA). Mutations described here were introduced by long PCR using Pfu-Turbo DNA polymerase as described previously (12). Every nucleotide of all of the constructs was verified by DNA sequencing.
the supernatant was incubated with anti-FLAG M2 affinity gel (Sigma) according to the manufacturer's instructions. For detecting the p94-connectin interaction and MARP-connectin interaction, the incubation was carried out for 2 and 4 h, respectively. Immunoprecipitates were collected by centrifugation and were rinsed twice with lysis buffer and then twice with wash buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl or CsCl), followed by an incubation with 3ϫ FLAG peptide (150 ng/l in wash buffer) for 30 min at 4°C. The supernatant was subjected to SDS-PAGE and Western blot analysis. Western Blot Analysis-Proteins were separated by SDS-PAGE and transferred onto PVDF membranes (Millipore, Billerica, MA). Membranes were probed with appropriate primary antibodies and horseradish peroxidase-coupled secondary antibodies (Nichirei, Tokyo, Japan) followed by visualization using a POD immunostaining kit (Wako, Osaka, Japan) or ECL TM Western blotting detection reagent (GE Healthcare).
Preparation of an Extract from Mouse Skeletal and Cardiac Muscle-The skeletal muscle (hamstring) and cardiac muscle tissues dissected from mice were frozen in liquid nitrogencooled isopentane. Cryosections from each muscle tissue, 20 m thick per section, were lysed in homogenizing buffer (50 mM Tris-HCl, pH 7.5, 5 mM EDTA-K, pH 8.0, and 1 mM dithiothreitol) containing protease inhibitors as follows: 1 mM PMSF, 0.1 mM pepstatin A, and 50 M calpastatin peptide 42 (aa 140 -181 of human calpastatin; NP_001035911). The homogenate (total) was fractionated into the soluble fraction (supernatant) and insoluble fraction (pellet) by centrifugation at 20,630 ϫ g for 30 min at 4°C. An equal amount of protein, 5 g for each sample, was separated by SDS-PAGE and subjected to Western blot analysis.
Expression of Recombinant Proteins in Sf-9 Cells-Recombinant baculovirus was generated according to the protocol provided by Invitrogen, and the recombinant proteins were expressed as described previously (48). Briefly, Spodoptera frugiperda (Sf-9) cells were suspended in infection medium containing one-tenth volume of each baculovirus stock solution at a concentration of 1.0 ϫ 10 7 cells/ml, left for 1 h with gentle agitation every 15 min, diluted to a concentration of 1.0 ϫ 10 6 cells/ml, and shaken at 140 rpm at 27°C for 44 -48 h. Preparation of the cell lysate and immunoprecipitation were performed as described above for analyzing proteins expressed in COS7 cells.
Bacterial Expression and Purification of Recombinant Proteins-cDNA fragments corresponding to the N2A region of human connectin, I80 -I83 (nucleotides 15307-16851 in X90569) and human MARP2 (nucleotides 294 -1289 in NM_020349) were amplified by PCR from human skeletal muscle cDNA and cloned into the pET vector. Proteins were expressed in Escherichia coli BL21(DE3) and purified as described previously (49).
Proteolytic Assay for Calpains-One g of recombinant human MARP2 was incubated with 0.25 g of recombinant human -calpain prepared as described previously (48) in 20 l of incubation buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA-K, pH 8.0, 1 mM dithiothreitol, 1 mM PMSF, 0.1 mM pepstatin A) with or without 15 mM CaCl 2 at 30°C for 60 and 90 min. To measure p94 autolysis and proteolysis of connectin fragment activity, the lysate of Sf-9 cells expressing p94 and/or the connectin fragment was incubated with 5 mM CaCl 2 at 37°C. The protein concentration of cell lysates was adjusted to 0.5 g/l. For the proteolytic assay of MARP2, recombinant His-MARP2 was added to the lysate of cells expressing p94 at a concentration of 20 ng/l. The reaction was stopped by the addition of SDS sample buffer, and the sample was subjected to SDS-PAGE followed by silver staining or Western blot analysis. Where indicated, protease inhibitors were added at the following concentrations: 1 mM PMSF, 1 mM leupeptin, 0.1 mM E64c, and 7 M calpastatin domain I (Takara Bio).
Peptide Sequencing-After SDS-PAGE, the proteins were blotted onto Pro-Blot membrane (Applied Biosystems, Foster City, CA) and then visualized by Coomassie Brilliant Blue G-250 staining. The target protein bands were excised and washed three times with an excess amount of 50% (v/v) methanol and then with absolute methanol. The N-terminal sequence was determined using a 491cLC protein sequencer (Applied Biosystems) according to the manufacturer's instructions.
YTH Assay-Saccharomyces cerevisiae strain AH109 was transformed with a series of combinations of expression vectors for various regions of connectin and p94 WT and mutants using the Fast TM -Yeast Transformation kit (G-Biosciences/Genotech, St. Louis, MO), according to the manufacturer's instructions. Cotransformants were selected on plates with SD medium that lacked Leu and Trp (SDϪLW), and the expression of reporter genes by growth on plates that also lacked His and Ade (SDϪLWHA) was measured according to the manufacturer's instructions (Clontech).
In apparent contrast to our previous conclusion (23), the mdm deletion in a connectin fragment I80-PEVK did not abolish the interaction with p94 (also confirmed when using protease-inactive mutant p94:C129S (p94:CS); data not shown). However, subsequent analysis indicated that the I80-PEVK fragment contains in addition to the previously identified p94 binding site in I82-I83 (see Fig.  1A, PpBS (primary p94 binding site)) at least one additional binding site.
Identification of Novel p94 Binding Sites in the N2A-PEVK Region of Connectin-Coimmunoprecipitation studies using p94:CS located additional binding sites for p94 within the connectin I80-is and PEVK-N regions, respectively (Fig. 1C, lanes 12 and 13), which were designated as secondary p94 binding site 1 and 2 (SpBS1 and -2), as shown in Fig. 1A. . . anti-pIS2 anti-FLAG FIGURE 1. Interaction between p94 and the connectin N2A-PEVK region. A, schematic structure of the N2A-PEVK connectin/titin region and constructs used in this study. The numbers correspond to aa in the mouse connectin, NP_035782. The horizontal bars above connectin indicate the binding regions for MARP (is) and p94 (PpBS, primary p94 binding site) identified previously by YTH. The 83-aa span deleted in mdm is indicated by a bracket. I80 -I83, is, and PEVK indicate the 80th to 83rd immunoglobulin motifs, a specific insertion sequence, and a region rich in Pro, Glu, Val, and Lys, respectively, in the connectin I-band region. Vertical arrows [1]- [3] and bidirectional arrows indicate the proteolytic sites and p94 binding sites (SpBS, secondary p94 binding site), respectively, identified in this study. The ability of each fragment to coimmunoprecipitate p94:WT or p94:CS and its susceptibility to endogenous proteolysis at the junction of I83 and PEVK are summarized on the right. NA, not applicable; ND, not determined. B, connectin constructs indicated and p94:WT were coexpressed in COS7 cells (Input) and immunoprecipitated by anti-FLAG (IP). Note that the full-length 94 kDa band for p94:WT (open arrowhead), was hardly detectable in the cell lysate before immunoprecipitation (lanes 1-6), but was efficiently coimmunoprecipitated with I80-PEVK, I80-PEVK(mdm), and I81-I83(mCN48) (lanes 8, 9, and 12). C, mapping of novel p94 binding sites using protease-inactive mutant p94:CS. p94:CS was coimmunoprecipitated with PEVK-N and I80-is (lanes 12 and 13), and these regions were designated as SpBS1 and -2 as in A. D, summary of YTH assays for binding of p94 to two different regions of connectin, I81-I83 (PpBS) and PEVK-N (SpBS2). Arrows [1]- [3] refer to p94 autolytic sites. The region containing IS2 is enlarged to show the sequences encoded by exons 14 -16. Since interaction between p94:WT and I80-I81 was not detectable (Fig. 1B, lane 11), autolyzed fragments appear to be insufficient for mediating interaction with SpBS1 only. It is also possible that SpBS1 only can interact with p94 very weakly and cannot prevent p94 from autolysis. On the other hand, I82-PEVK, in which PpBS remains intact, did not coimmunoprecipitate either p94:WT or p94:CS, regardless of whether the FLAG tag is N-terminal ( Fig. 1, B (lane 10) or C (lane 5), anti-pIS2) or C-terminal (data not shown). This suggests that the connectin local structure affects its ability to bind to p94.
p94 Interacts with PEVK Connectin through the Proximity of the IS2 Region-Because the interaction between PEVK-N and p94:CS was detectable in the YTH assay, we examined further the binding region for SpBS2 in p94 and compared this with that for PpBS (Fig. 1D). The interaction was complicated but suggested that the SpBS2 binding region overlaps with that for PpBS in the N terminus in the proximity of the IS2 region, with a slight extension toward the C terminus (Fig. 1D, p94:1-600 and p94:1-594). Interestingly, the sequence encoded by exons 15 and 16 of p94 is not necessary for its interaction with SpBS2 if the C-terminal region of p94 is intact (Fig. 1D, p94:ex15 Ϫ 16 Ϫ ). These data can be explained by assuming that SpBS2 binds to an intact structure in the proximity of the p94 IS2 region encoded by exon 14, which requires several extra amino acids at its N and C termini.
The proteolytic site in I82-PEVK was determined to be the N terminus of Ser 8934 in the mdm deletion region (Fig. 1A, arrow  1); Fig. S1B). Expression of various fragments encompassing Ser 8934 with or without the mdm deletion in COS7 cells suggested that fragments without the mdm deletion are commonly proteolyzed at the same site (Fig. S1C). These results indicate that the N2A-PEVK junction is susceptible to unidentified protease(s) in COS7 and Sf-9 cells in the context of fragment structures used in this study and that the mdm deletion confers resistance to this proteolytic attack.
A specific antibody for the N terminus of the proteolyzed fragment b (Fig. 2B), anti-TPALKK, was generated using the corresponding mouse connectin sequence. This antibody reacted with fragment b but not with a, c, or d (data not shown; see "Experimental Procedures") and was used for further analysis of connectin proteolysis.
Connectin Is Proteolyzed by p94 and an Unknown Endogenous Protease at the Same Site as -Calpain-The detected amount of I80-PEVK decreased in the presence of coexpressed p94:WT, but not p94:CS, both in Sf-9 (data not shown) and, to a lesser extent, COS7 cells (see Fig. 5A, lane 9 versus 10, Input, anti-FLAG), suggesting that I80-PEVK is a substrate for p94. Therefore, we further examined proteolytic processes of I80-PEVK using Sf-9 and p94:D607A (p94:DA) missense mutant (24).
Incubation of the lysate of Sf-9 cells expressing either p94:DA or p94:CS demonstrated that p94:DA has moderate Ca 2ϩ -dependent activity, which was inhibited by leupeptin plus E64c but not by calpastatin, representing a unique property of p94 ( Fig. S2A) (24). Moreover, as in the case of p94:WT, the fulllength p94:DA was predominantly coimmunoprecipitated with the connectin fragment (Fig. S2B, lane 3, open arrowhead).
When p94:DA was coexpressed, proteolysis of I80-PEVK was accelerated (Fig. 2C, lanes 7-12, anti-FLAG). The peak of the 43 kDa band amount shifted to 40 min from 60 min of that without p94:DA (lane 5 versus lane 10, anti-TPALKK). These data suggested the involvement of p94:DA in the proteolysis of I80-PEVK directly and/or indirectly; p94:DA proteolyzes I80-PEVK in the same manner as -calpain and/or activates unknown Sf-9 protease(s). In contrast, coexpression of p94:CS did not accelerate, but slightly slowed, the proteolysis of I80-PEVK (Fig.  2C, lanes 13-18). It is thus inferred that interaction with p94:CS protects I80-PEVK from the endogenous protease(s), and accordingly, it is likely that p94:DA binds to I80-PEVK and directly proteolyzes it when activated by Ca 2ϩ . Differential Proteolytic Susceptibility of I80-PEVK and I80-PEVK(mdm)-Judging from patterns of breakdown products, proteolysis of I80-PEVK (Fig. 2, B and C (e)) probably occurs first at the N terminus of Ser 8934 , generating the 62 kDa fragment (Fig. 2B, arrow 1 and fragment f), followed by that of Ser 8558 (arrow 2 and fragment g) and then that of Gln 8646 / Thr 8647 (arrow 3 and fragment h; ϳ10 kDa, too small to detect).
Our results suggest that the "is" region of I80-PEVK becomes further sensitized to proteases, including p94, after a rate-limiting proteolytic step occurred in the junction of I83-PEVK. This model would explain why I80-PEVK(mdm) is resistant to proteolysis. Additionally, our data suggested that p94:CS protects connectin from proteolysis at sites in the "is" region as well as in the junction of I83-PEVK. These results indicate that the region C-terminal to "is" (including those connectin sequences deleted in mdm) affects MARP-connectin interaction, possibly by modulating the structure of "is" surrounded by immunoglobulin domains. Additionally, MARP1 interferes with MARP2 in binding to connectin, both in WT and in mdm.

Effect of Connectin Structure on the MARP-Connectin Interaction-
Distinct Subcellular Distribution of MARP1 and -2 in Striated Muscle-To examine whether the in vitro MARP-connectin interaction observed above corresponds to the in vivo situation, expression of MARP1 and -2 in muscles from WT and Ttn mdm/mdm mice (n ϭ 3 for each) were compared (Fig. 3D,  WT-1-3 and mdm-1-3, respectively).
In both WT and Ttn mdm/mdm , MARP1 was enriched in cardiac muscle in the insoluble myofibrillar bound fraction, consistent with its strong binding to connectin. In skeletal muscle, MARP1 was observed in insoluble fraction only in Ttn mdm/mdm ,  The interaction between MARP2 and I80-PEVK or I80-PEVK(mdm) was strongly suppressed by coexpression with MARP1, whereas MARP1 was not affected (lanes 6 and 7, anti-HA; lanes 5 and 6, anti-MYC). D, the amount of MARP1 and -2 in whole tissue homogenates (T) and soluble (S) and insoluble fractions (P) of skeletal and cardiac muscles from three independent wild type (WT-1 to -3) or Ttn mdm/mdm mice (mdm-1 to -3) was examined using anti-MARP1 and -MARP2 antibodies. For a summary of the results, see Table 2. Open and closed arrowheads in A-C, the full-length and the N-terminal proteolyzed fragments of FLAG-connectin, respectively. where MARP1 is up-regulated as previously reported (44). In contrast, MARP2 was detected predominantly in the soluble fraction of skeletal muscle and undetectable in cardiac muscle, suggesting very weak, if not zero, interaction of MARP2 with myofibrils. In Ttn mdm/mdm skeletal muscle, where it was robustly up-regulated, MARP2 was detected slightly in the insoluble fractions as well. These data show that MARP1 and -2 differ in their cellular distribution in a manner consistent with our in vitro results (i.e. connectin binding activity is MARP1 Ͼ Ͼ MARP2). It should be noted that these molecular properties are essentially the same both in WT and Ttn mdm/mdm (Table 2), which raises a question as to a relationship between MARP1 and MARP2 under mdm conditions.
Characteristics of MARP1 and -2 in the Mode of Induction in Ttn mdm/mdm Skeletal Muscle-To characterize the expression of MARP1 and -2 in relation to the integrity and the fiber type in Ttn mdm/mdm skeletal muscle, expressions of MARP1, MARP2, dMHC, and sMHC were simultaneously analyzed in situ.
As previously reported, dMHC-positive fibers were absent in WT (data not shown), whereas degeneration/regeneration of muscle fibers was apparent in 10-week-old Ttn mdm/mdm skeletal muscle (Fig. 4, B and D, dMHC), and abnormal morphology, such as clusters of small sized regenerating fibers and a large percentage of fibers with irregular outlines and central nuclei, was apparent (Fig. 4C, HE) (45). In WT, essentially no MARP1-positive fiber was detectable (data not shown), and MARP2 was detectable only at low frequency and signal intensity (Fig. 4A), consistent with the above Western blot data (Fig. 3D). Some MARP2-positive fibers were also positive for sMHC (Fig. 4A, asterisks).
In Ttn mdm/mdm muscle, MARP1-positive fibers were sparsely identified, and significantly more fibers were positive for MARP2, sMHC, and/or dMHC in mdm than in WT (Fig. 4, B-E, and Table 3). MARP1 was not detectable in dMHC-positive fibers (Fig. 4B), suggesting that induction of MARP1 is not significant in regenerating fibers. There was a trend that MARP1-positive fibers were also MARP2-positive and had central nuclei (Fig. 4C, arrow). Expression of MARP2 seemed independent from fiber types specified by MHC (Fig.  4, C and D). Fibers with relatively intense MARP2 signal tended to have central nuclei regardless of their fiber sizes (Fig. 4C, arrowhead). Although frequency was low, MARP1

. Expression of MARP1 and MARP2 in WT and
Ttn mdm/mdm mouse skeletal muscle. MARP1 or -2 (red) and sMHC or dMHC isoform (green) were detected by double-labeled immunofluorescence on sections of posterior parts of hind limbs from WT (A) and Ttn mdm/mdm (B-E) mice using specific antibodies. Nuclei were stained with 4Ј,6-diamidino-2-phenylindole (blue). Representative images were selected. A, in WT muscle, fibers faintly positive for MARP2 were identified at a low frequency. Some of these fibers were sMHC-positive (*). B, in the corresponding area of Ttn mdm/mdm muscle, fibers expressing MARP1 were sparsely observed, where central nuclei were also identified (arrowhead). Expression of MARP1 was not detected in fibers expressing dMHC (*). C, two serial sections were stained with anti-MARP1, -MARP2, and -sMHC. Fibers expressing MARP2 and/or sMHC were observed more frequently compared with WT. Expression of MARP1 and -2 is not mutually exclusive and is independent of the muscle fiber types. Examples of combination of MARP1, MARP2, and sMHC expressed in the same fiber were shown by arrows (ϩ, ϩ, Ϫ), arrowheads (Ϫ, ϩ, Ϫ), and asterisks (Ϫ, ϩ, ϩ). D, expression of MARP2 was observed both in dMHC-positive (*) and -negative fibers (arrowheads). E, occasionally, MARP1 and MARP2 were localized in or at the periphery of central nuclei in Ttn mdm/mdm myofibers. Bars, 20 m.

Combinations of expression at a muscle fiber level (mdm)
a Expression of each protein examined by immunofluorescent study on cross-sections of a posterior segment of hind limbs was categorized based on the positive rate. ϩϩϩ, Ͼ50%; ϩϩ, 10 -50%; ϩ, 1-10%; Ϫ Ϸ 0%. b Trend in coexpression of proteins in the same muscle fiber from Ttn mdm/mdm muscle is summarized. c Mice homozygous for the wild type allele of Ttn, Ttn ϩ/ϩ . d Mice homozygous for the mdm allele of Ttn, Ttn mdm/mdm . and -2 were detected in or at the periphery of central nuclei in Ttn mdm/mdm (Fig. 4E). Because of low frequency, correlation between nuclear localization of MARP1/2 and the fiber type or size was not clear.
These observations demonstrate that MARP1 and 2 are not mutually exclusive as to their up-regulation in Ttn mdm/mdm muscle. The presence of myofibers in which only MARP2 increased (Fig. 4C, arrowhead) implies different time points of action for MARP1 and -2 during regeneration of muscle and could explain a small population of MARP2 recovered in the insoluble fraction (Fig. 3D).
Relationships among p94, MARP2, and Connectin-Because both p94 and MARP2 show predominant expression in skeletal muscle and interaction with N2A connectin, the relationship between them may be deteriorated by the mdm deletion.
Less p94:CS and MARP2 coimmunoprecipitated with I80-PEVK when both were coexpressed (Fig. 5A, IP, anti-pIS2 6), indicating that p94 and MARP2 affect their interaction with connectin and with each other. Furthermore, the lesser MARP2 coimmunoprecipitated with connectin when p94:WT was used (Fig. 5A,  IP, anti-HA, lanes 6 -8). Considering that the amounts of  4, 9, and 14). Open and closed arrowheads indicate the full-length and autolyzed fragments of p94, respectively. is p immunoprecipitated I80-PEVK were similar regardless of coexpression of p94:WT or CS (Fig. 5A, IP, anti-FLAG, lanes 6 -8) and that p94:WT protein existed much less than p94:CS, the observed decrease in coprecipitated MARP2 probably results from the proteolysis of MARP2 by p94:WT during their interactions with connectin.
Thus, proteolysis of MARP2 by p94 and the effect of connectin on it were further examined. First, purified recombinant MARP2 was proteolyzed Ca 2ϩ -dependently in the Sf-9 lysate expressing p94:DA, which was inhibited by leupeptin plus E64c, but not by calpastatin (Fig. 5B, lanes 11-15, anti-MARP2). The proteolysis was not observed in the presence of p94:CS or absence of p94 (Fig. S2C). Therefore, in addition to I80-PEVK, our studies also identified MARP2 as a substrate for p94, at least in vitro.

DISCUSSION
In this study, we characterized the relationships among p94, MARPs, and N2A connectin in the WT and mdm contexts. We found that 1) novel binding sites for p94 exist in the N-terminal (SpBS1) and C-terminal (SpBS2) regions adjacent to the previously identified site (PpBS); 2) preautolytic full-length p94 pref-erentially interacts with N2A connectin; 3) proteases, including p94 and the conventional calpains, proteolyze the N2A region of connectin at several sites (Fig. 7A, closed arrowheads), which is compromised in the fragments with the mdm deletion (Fig.  7B, open arrowheads); 4) the efficiency of binding between the connectin "is" region and MARPs is also affected by the mdm deletion.
Considering the huge molecular size of connectin, characteristics of the subfragment might provide only limited functional insights into whole connectin molecules in vivo. However, our in vitro results demonstrate for the N2A region of connectin a propensity to undergo dynamic changes imposed intramolecularly by the local molecular structures. Identification of N2A connectin and MARP2 as possible p94 substrates implicates that the N2A region of connectin serves as a versatile scaffold for p94, which stabilizes p94 and facilitates proteolysis of MARP2 by p94 to modify its functions (Fig. 7A).
Localization of p94 Activity and Targets-Identification of sarcomeric proteins, such as ␣-actinin and connectin, as p94 binding partners emphasizes the importance of these interactions to the regulation of p94, especially its stability (19,21,23,25), and that p94 is also a component for signal transduction inherent in skeletal muscle structure. So far, the physiological relevance of p94 localization to the sarcomere has been unclear. Our results present three possibilities, which may be interrelated.
First, more efficient binding of the full-length p94 than its autolyzed fragment to N2A connectin can be inferred as a molecular mechanism for condensing protease activity in the proximity of p94 substrates (i.e. connectin itself and MARP2) and for releasing a remnant of enzyme after completion of p94 functions in a given context (Fig. 7A).
Second, the protection effect of p94:CS against proteolysis of N2A connectin may result from alteration of connectin local structures by p94 binding so that connectin is not proteolyzed spontaneously and/or randomly by yet unknown proteases, one of the candidates for which is conventional calpain.
Third, the negative effect of p94:CS on the MARP2-connectin interaction could alternatively control the extent of proteolysis of MARP2 by p94 as well as its functions. One of the novel p94 binding sites of connectin (e.g. SpBS1) may directly compete with MARP binding to "is". Furthermore, the presence of several p94 -connectin interactions in I80-PEVK may trigger structural changes that affect the efficiency of MARP binding. FIGURE 7. A hypothetical scheme of the molecular complex at the N2A connectin. A, summary of the N2A region in connectin and the relationships between p94 and MARPs in WT skeletal muscle. MARP2 binds to the connectin "is" region, which is suppressed by the interaction between p94 and connectin. At least three regions in N2A connectin are involved as scaffolds in the p94-connectin complex. The closed arrowheads indicate the proteolytic sites in N2A connectin and MARP2 determined in this study. Unproteolyzed connectin orchestrates multiple signal transductions (top). Activation of proteolytic machinery (i.e. p94-connectin complex) results in proteolyzed connectin fragments whose interaction with MARP2 as well as other functions are modulated (bottom). MARP2 is also proteolyzed to generate the free N-terminal fragment, MARP2-(N), and the C-terminal connectin-associated fragment, MARP2-(C). Dissociation from connectin prompts autolysis of p94 to terminate its activity. Coordination of these reactions prevents aggravation of muscle damage triggered under various physiological conditions. B, stalled proteolysis incited by the mdm deletion in connectin. In Ttn mdm/mdm skeletal muscle, expression of MARP1 and -2 is up-regulated as a result of aberrant signal transduction by connectin with the mdm deletion. When MARP1 and -2 are coexpressed in the same muscle fiber, "is" is dominantly occupied by MARP1. On the other hand, several fibers express only MARP2, which binds to "is" without interference by MARP1 and is subjected to proteolysis. The mdm deletion impairs the nature of connectin as a substrate for proteases and distorts interaction between p94 and connectin, which compromises functions of the p94-connectin complex. These abnormalities cause deficits in cellular response to muscle damage (i.e. dystrophic symptoms observed in Ttn mdm/mdm skeletal muscle). The open arrowheads indicate proteolytic sites in connectin apparently invalidated by the mdm deletion. The wavy lines represent distorted interaction between p94 and connectin.
Activity of the p94-Connectin Complex in Ttn mdm/mdm Skeletal Muscle-The effect of the mdm deletion on p94 stabilities was apparently in conflict with previous results. Decrease of p94 in Ttn mdm/mdm skeletal muscle (43) corroborates the idea that the mdm deletion abrogates p94-connectin interaction required for p94 regulation. In contrast, I80-PEVK(mdm) was able to coprecipitate p94:WT (Fig. 1B), and autolysis of p94:DA coexpressed with I80-PEVK(mdm) was retarded (Fig. 2C). One interpretation is that I80-PEVK(mdm) can bind p94 at SpBSs but is distinct from WT not only in its resistance to proteolysis but also in that it hinders certain aspects of p94 protease activity.
In this context, we hypothesize that the p94-connectin N2A complex represents one of functional p94 protease units, p94 as a catalytic subunit and connectin as a regulatory subunit. The observation that gait deficits found in Ttn mdm/ϩ mice were restored by p94 overexpression indicates that the mdm deletion does not produce dominant phenotypes but causes a decrease in p94 activity (52).
Structure of Connectin Fragments Affecting Its Susceptibility to Proteolysis-The physiological relevance of connectin proteolysis previously reported (53) remains unknown to date. In cardiac myocytes, doxorubicin treatment causes degradation of connectin by m-calpain to produce a proteolyzed fragment designated T2 (54). The major proteolytic sites of connectin were suggested to be in the elastic I-band region, where N2A and its alternative splicing isoform, N2B, are located, and the adjacent PEVK region (55). -calpain is also reported to interact with connectin and generate proteolyzed fragments encompassing N1 (the C-terminal to Z-line) and N2A regions (56).
Proteolysis by p94 of connectin fragments corresponding to the Z-line, PEVK, and M-line regions has also been reported (16,50). In this study, p94 was shown to target the same proteolytic site in N2A connectin as -calpain. Because this proteolytic site resides in "is", the binding site for MARP2 and for p94 itself, our results provide a decisive basis for further investigation on the effect of p94 on the MARP2connectin interaction.
We have shown that the mdm deletion caused protection from proteolysis in the region of the mdm deletion and, rather unexpectedly, within "is". These results indicate that proteolysis of N2A connectin proceeds step-by-step, changing its structure accordingly (i.e. proteolytic sites within the "is" region are exposed only after the proteolysis between I83 and PEVK). One should, thus, consider that structures of connectin fragments used in experiments may affect the results as to the identification of proteolytic sites.
Relationships between MARPs and Connectin with or without the mdm Deletion-Genetic studies using mice lacking MARPs suggested that MARP paralogues have functional redundancy as both structural components and signaling molecules (57). Our study, however, showed that MARPs are different in their connectin-binding activity and, accordingly, in cellular distribution. MARP1 and -2 were shown to be differently induced under the mdm condition. This is consistent with previous studies indicating that MARP1 and -2 respond differentially depending on the quantity and/or quality of "stress" (58). Multiple cellular events should control MARP-connectin interaction in muscles (59), and whether MARPs interact with connectin as a monomer, dimer, or complex with other proteins is one of the critical future issues to be investigated (49).
Studies on fiber type specificity of MARPs showed that high level expression of MARP1 is often associated with small regenerating fibers, which are labeled by embryonic MHC (corresponding to dMHC in this study), in Duchenne muscular dystrophy but not in other muscular dystrophies (60). In Ttn mdm/mdm skeletal muscle, the expression of MARP1 was not detected in dMHC-positive fibers, whereas MARP2 was detected in fibers positive for sMHC or dMHC or negative for both. The same trend of MARP2 was reported in amyotrophic lateral sclerosis (38).
Altogether, these observations suggest that the molecular mechanisms underlying the induction of MARP1 and -2 in Ttn mdm/mdm are distinct and that these mechanisms are different from those in other muscular dystrophy conditions. In other words, different primary causes (e.g. deletion in dystrophin or connectin) of the symptom collectively described as "dystrophy" distinguish the molecular mechanism of each symptom.
Consistent with a short life span (10 -12 weeks) of Ttn mdm/mdm mice, skeletal muscle from 10-week-old mice examined in this study presented advanced dystrophic symptoms. The expression trend of MARP1 and -2 observed here may be related to a stage of disease progression. One of the directions for future study is comparative analysis of the time course for expression of MARP1 and -2 in relation to muscle fiber types and sizes.
Regulated Proteolysis as a Modulator for Muscle Functions-Proteolysis of connectin in the "is" region by calpains, including p94, is predicted to release the connectin-MARP complex from the entire connectin molecule, which may desensitize the connectin-MARP interaction to respond to a stretch signal transmitted as structural changes in connectin. Since a release of the MARP2 N-terminal part upon proteolysis by -calpain retained binding of MARP2 to connectin, the interaction of MARP2 with molecules other than connectin could be affected. In this regard, previously reported interactions of MARP2 with YB-1 or myopalladin (42,59), which is mediated by the N-terminal region of MARP2, represent good candidates.
As a condition where the proteolysis of sarcomeric components described above bears biological significance, the maintenance and remodeling of once established muscle tissues should be considered. This process must require coordinated events, including proteolytic dismissal of damaged proteins and reorganization of newly synthesized proteins (61)(62)(63). Together with reported functions of muscle proteins, our data suggest that the N2A complex comprising connectin, MARP2, and p94 is capable of playing a key role in such a scenario.
This protease machinery is allowed to be dynamic and sensitive accordingly to cellular context because of unique proper-ties of connectin, including its huge molecular size, intramolecular effects on its interaction with MARPs and on susceptibility to proteolysis, and the selective recruitment of p94 activity. Moreover, the proposed functions of MARPs in accommodating muscles to stress by regulating gene expression indicates a link between the N2A complex and the nucleus. Recent proteomic approaches also indicate that p94 is involved in the regulation of metabolism and protein synthesis (64,65). The interlocking behavior of connectin with multiple activities of skeletal muscle would assure a balanced projection of p94 activity to molecules of diverging functions. Perturbing this by, for example, the mdm deletion should lead to serious dysfunction of muscles.