Orgánico Funcional

Introduction

Limb Girdle Muscular Dystrophy (LGMD) is a heterogeneous group of muscle disorders characterized by a late onset and progressive muscle weakness and wasting [30]. Mutations in at least 21 different genes result in a similar clinical presentation, but it is largely unclear whether or not the causative gene products are mechanistically related [160]. Given the adult onset it has been suggested that LGMD is the result of impaired muscle regeneration due to exhaustion of the regenerative potential. A high level of myofiber damage would result in exhaustion of the pool of satellite cells and therefore eventually impair replacement of myofibers. Here we discuss the recent advances in our understanding of the pathophysiology of LGMD including those described in this thesis and the possibility that impaired membrane maintenance and disturbed regeneration are a common denominator of LGMD.

The most common form of LGMD in many populations is LGMD2A, and is caused by mutations in the gene encoding Calpain 3 (CAPN3) [211]. CAPN3 is a calcium-sensitive cysteine protease, and a member of the non-lysosomal non- denaturing Calpain proteases [17]. CAPN3 is often secondarily reduced in LGMD2B [7], which is caused by mutations in the gene encoding Dysferlin (DYSF) [15]. Dysferlin is a calcium-sensitive C2 domain containing single-pass transmembrane domain protein involved in membrane repair [12]. It was shown that CAPN3 can be a component of the Dysferlin protein complex [120] and both proteins were proposed to be important for repair of skeletal muscle damage [13,17]. During muscle contractions the myofibers sustain continuous damage to the internal sarcomere, the anchoring complexes that link the sarcomere to the extracellular matrix and the sarcolemma. Such damage requires rapid and efficient repair to prevent myofiber degeneration, and defects to such repair systems have been suggested as a pathogenic mechanism common to these forms of LGMD (Chapter 1). In the following sections we will discuss the myofiber’s response to damage at the level of the sarcomere, the sarcolemma and the cytoskeleton anchors, with focus on the proteins CAPN3 and Dysferlin. Finally, the remaining LGMD proteins are shortly discussed in light of the findings on CAPN3 and Dysferlin.

Sarcomere damage

During contraction the contractile apparatus sustains mechanical damage that requires efficient and rapid repair [17]. The sarcomere consists of a number of giant proteins that interact to form a dense structure that can generate mechanical force (Chapter 1, Figure 1). Apart from contraction, there is little space for movement of proteins, and the diffusion rate is estimated to be low [17]. However, when damage to the sarcomere occurs, potentially toxic protein fragments need

to be efficiently removed. Moreover, the damaged proteins must be replaced. Therefore, the muscle requires a preset system to remove and replace proteins from the dense sarcomere in response to damage. CAPN3 is one such potential mechanical sensor and actor [17].

CAPN3 is a muscle specific protease that is a member of the Calpain cysteine protease family [234]. It is highly similar to the well-studied ubiquitously expressed Calpains 1 and 2, sharing ~50% amino acid identity with these proteins [138,235]. In addition it contains three insertion sequences (IS) without any homology to other Calpains, which are used to regulate its enzymatic activity [236]. Full-length CAPN3 is proteolytically inactive. The IS1 sequence blocks the proteolytic cavity ([72], Chapter 4). Upon activation it proteolytically removes the IS1 sequence by autolysis (Figure 1). Hereupon the protease continues as a proteolytically active intramolecular heterodimer, and gains access to its potential full substrate spectrum. At the same time, CAPN3 will intermolecularly cleave itself at the remaining two insertion sequences. This feature enables the rapid deactivation of CAPN3. In vitro studies have indicated that CAPN3 is activated and deactivated within a time-span of 10 minutes [138].

In intact skeletal muscle fibers CAPN3 appears to be much more stable compared to in vitro [191]. CAPN3 contains a functional calcium sensor, consisting of five consecutive EF hands in its C-terminus, similar to Calpains 1 and 2 [138,139]. Its calcium sensitivity however, lies in the nanomolar range, thereby greatly exceeding that of Calpain 1 (micromolar) and Calpain 2 (millimolar) [191]. Moreover, in skeletal muscle most of the CAPN3 protein localizes to the sarcomere [191]. There it can directly interact with Titin [232], and it has been suggested that this interaction stabilizes CAPN3 in its inactive form [233], proving a possible explanation for the apparent discrepancy between the in vivo and in vitro measurements.

Figure 1: Schematic overview of the model for CAPN3 autolytic activation and deactivation. The IS1 sequence is highlighted in red.

Inactive Activated Proteolytic Deactivated

A recent model for CAPN3 function suggests that upon sarcomere damage CAPN3 is released by Titin, whereupon it would rapidly activate, and deactivate itself [17]. Due to this short half-life, CAPN3 activity would be spatially restricted [17]. Thereby proteolytic processing of the local sarcomere unit is achieved, thus allowing for the relaxation of the dense sarcomere structure and subsequent replacement of damaged proteins, while at the same time safeguarding a temporally restricted enzymatic activity. CAPN3 would therefore be involved in controlled degeneration of the myofiber, and indeed it was shown to function upstream of the ubiquitin-proteasome pathway [153]. This model suggests that the sarcomere has a build-in highly reactive protease to control sarcomere remodeling.

In support of this model it was shown that CAPN3 deficient mice cannot undergo extensive sarcomere remodeling in response to muscle atrophy and degeneration in hind-limb suspension experiments [153]. Transgenic mice that overexpress CAPN3 show no phenotype, but the myofibers appear immature [237]. In fact western blot analysis showed that most of the CAPN3 protein is still present in its inactive form, suggesting that the muscle contains excess docking sites to store the excess CAPN3 protein [237]. As most of the CAPN3 molecules are stored at the sarcomere the best candidate docking protein is Titin. Indeed it has been estimated that the number of CAPN3 binding sites on available Titin molecules greatly exceeds the number of expressed CAPN3 molecules [17]. Interestingly, miss-sense mutations in the CAPN3-binding site in Titin also cause LGMD [104]. Elegant proof for the stabilizing role of Titin comes from experiments with transgenic mice. Mice that carry these same Titin mutations in the CAPN3- binding site show a mild progressive myopathy reminiscent of LGMD in humans. When the a-phenotypic CAPN3 overexpressing mice were crossed onto the Titin transgenic mice, this severely aggravated that muscle phenotype, suggesting that the muscle had lost its buffering capacity for the excess CAPN3 [122]. Crossing CAPN3 null mice onto this Titin mutant background had no aggravating effect.

It is hypothesized that upon sarcomere damage, the locally stored CAPN3 is released by Titin [17]. Dependent on the free calcium concentration it rapidly activates itself through autoproteolysis [17,191]. The dense structure of the sarcomere restricts its activity to its direct vicinity [17,191]. It will rapidly break down all giant proteins in its direct neighbourhood, and through continuing autoproteolysis, quickly deactivate itself, before its activity can spread beyond where it is needed [17]. The question that remains is therefore: what are the in vivo substrates of CAPN3?

Due to its instability only few in vivo substrates of CAPN3 are known. Using a combination of bioinformatics, biochemistry and cell biology we could show that these few substrates share a common sequence motif in the vicinity of the

CAPN3 cleavage site (Chapter 4). We could subsequently show that this motif can transform non-substrates into substrates, and that this motif is shared by >300 other proteins. Based on bioinformatics analyses, the majority of the CAPN3 target proteins are involved in cytoskeleton organization. This confirms that indeed CAPN3 will target local structural proteins upon activation.

In conclusion, CAPN3 is stored locally in inactive form to enable rapid remodeling of local cytoskeleton architecture when needed. It is therefore a highly plausible candidate for a muscle repair agent.

Dysferlin and membrane maintenance

Dysferlin is critical for calcium-dependent sarcolemmal repair [13]. Dysferlin localizes to the sarcolemma and to intracellular vesicles [13]. Upon membrane damage and calcium entry it rapidly accumulates at the site of the lesion, and is thought to deliver excess membrane for patch-fusion repair of the sarcolemma [12,13]. Indeed Dysferlin deficient myofibers cannot efficiently repair laser- inflicted membrane wounds, and show sub-sarcolemmal vesicle accumulation [13]. Moreover, this repair function appears to be conserved in other Dysferlin expressing cell types, such as macrophage-like THP1 cells (Chapter 7). Dysferlin contains seven C2 domains, which are believed to function in calcium-sensitive interactions with proteins and phospholipids [244]. Such interactions would aid Dysferlin in performing its maintenance function.

Among the identified interaction partners are the membrane fusogens Annexin A1 and Annexin A2, of which the latter binds in a calcium-dependent manner [162]. These proteins are involved in membrane repair in non-muscle cells [179] and are predicted to aid in Dysferlin vesicle docking to the cell membrane. Trim72 (MG53) is a redox sensor that functions upstream of Dysferlin and is important for vesicle nucleation [34–36]. Lastly, Caveolin 3, an essential component of caveolae and mutated in LGMD1C, interacts with Dysferlin [172], and is involved in Dysferlin trafficking [110,111].

By immunoprecipitation followed by mass spectrometry analysis we have shown that Dysferlin forms a complex with many other proteins, in a context-dependent manner during myogenic differentiation (Chapter 2). Among its complex partners are vesicle-related proteins including those that are involved in endocytosis (Chapter 2). This is consistent with the observation that in adult muscle most of Dysferlin staining is found at the sarcolemma. Moreover, experiments in caveolin negative cells strongly suggested that Dysferlin needs to be anchored at the sarcolemma to prevent its rapid endocytosis [111]. Dysferlin is often secondarily reduced in LGMD1C, or caveolin 3 deficiency [172]. This suggests that Dysferlin also has a function while being localized to the sarcolemma.

Cortical cytoskeleton restructuring

For membrane repair to occur two separate processes require tight coordination [180]. First the local cortical cytoskeleton needs to be reshaped to allow for subsequent restructuring of the damaged membrane. In a second phase excess membrane from intracellular stores is recruited to enable patch fusion repair. In the absence of Dysferlin this second phase is blocked as evidenced by the accumulation of subsarcolemmal vesicles [13]. A possible role for Dysferlin in the first phase has not previously been considered.

We showed that Dysferlin forms a protein complex with focal adhesion components such as Vinculin, Talin and α-Actinin, both in cells of myogenic origin (myoblasts, myotubes, mature muscle tissue) and in differentiated macrophages (Chapters 2 and 7). Focal Complexes are based on the transmembrane Integrin receptors, which can interact with extracellular matrix molecules such as fibronectin and collagen. Upon direct interaction between Integrin heterodimers (α- and β-subunit) and the extracellular matrix the cell responds by utilizing this attachment site as a transient scaffold for the intracellular filamentous Actin network. Proteins such as Vinculin, Talin, paxilin and β-Parvin (PARVB) regulate this scaffolding function [40,41].

A possible explanation for this observation is that Dysferlin links the two steps of repair. Recent elegant biochemical experiments showed that at the site of membrane damage, focal adhesion proteins such as Talin, and Vimentin, are proteolytically modified by Calpain proteases [183]. This is rapidly followed by the arrival of new vesicles, which are enriched for Actin and Integrin β1 (ITGB1) [183]. Presumably, these proteins will allow for restoration of the resting situation. We propose a model (Figure 2) in which Dysferlin can target vesicular membrane stores, and through its interaction with focal adhesion components it can coordinate cytoskeletal remodeling. Previous work showed that Dysferlin Figure 2: A schematic model for Dysferlin function in membrane maintenance. A) In resting conditions Dysferlin is at the cell membrane, in a macromolecular complex with AHNAK, PARVB and Vinculin, and thereby attached to focal adhesions. This occurs nearby caveolea, which prevent Dysferlin endocytosis. CAPN3 is docked to AHNAK. B) Upon membrane wounding and Calcium entry, the Dysferlin/AHNAK interaction undergoes a conformational change and CAPN3 is released. S-AHNAK is released from L-AHNAK. C) CAPN3 autolytically activates and cleaves all structural proteins in its close vicinity, including AHNAK, Vinculin and Talin, to ensure focal cytoskeleton remodeling. Dysferlin is endocytosed and S-AHNAK shifts to the nucleus. D) Dysferlin containing repair vesicles nucleated by Trim72 allow for patch-fusion repair and ATP release, and restoration of the resting situation. The membrane source of these vesicles is unclear.

recruits PARVB to the sarcolemma [174]. PARVB can directly interact with Integrin linked kinase (ILK), and is important for stabilizing focal adhesions [84,173,261]. In the absence of Dysferlin PARVB does not localize to the sarcolemma [174],

Ca2+ [Ca2+] Ca2+ [Ca2+] Endosome Nucleus Ca2+

Endosome? ER/Golgi? Mitochondrion?

A

B

C

D

Cell Membrane _{Cell Membrane}

Cell Membrane Cell Membrane CAV3 CAV3 TLN TLN PARVB PARVB ITGA ITGB ITGAITGB F-ACTIN _F-ACTIN L-AHNAK S-AHNAK L-AHNAK S-AHNAK ILK ACTN ACTN Dysferlin CAPN3

but the functional consequence of this is unknown. Upon membrane damage and calcium entry, Dysferlin might change confirmation, allowing rapid disintegration of the cytoskeleton anchors and subsequent repair processes. It will be interesting to see with which other focal adhesion components Dysferlin directly interacts and if this occurs in a calcium-sensitive manner. In addition, localization of these proteins in Dysferlin deficient muscle might give clear indications for the validity of this model. It is intriguing to speculate that Dysferlin acts as a sensor to coordinate the remodeling of structural proteins in addition to aiding patch-fusion of membranes.

Calpain proteolysis and the cortical cytoskeleton

Calpain proteolysis is important for cytoskeleton remodeling in response to membrane damage. It is therefore interesting that we could identify the skeletal muscle specific Calpain family member CAPN3 as a component of the Dysferlin protein complex. As described above, the majority of CAPN3 protein localizes to the sarcomere, where it most likely functions as a build-in proteolytic switch to enable local sarcomere remodeling. It is less clear what the function of CAPN3 outside the sarcomere is. We hypothesized that CAPN3 is involved in Dysferlin dependent membrane repair, by focal remodeling of the cortical cytoskeleton, to clear the way for repair vesicles.

When ectopically expressed CAPN3 increases the turnover of focal adhesions by proteolytic cleavage of the focal adhesion components Vinculin and Talin [242], two proteins also identified in the Dysferlin protein complex (Chapters 2 and 7). Subsequently cells become round and lose their normally tight adherence [242]. This is in agreement with the model of CAPN3 proteolysis regulating the cortical Actin cytoskeleton. Moreover, in skeletal muscle cryosections CAPN3 staining was observed at costameres [118] (Chapter 3), which are important adhesion sites of the mature myofibers, and are strongly enriched for Integrin complexes. This suggests that in mature muscle CAPN3 localizes to Integrin-based adherence complexes, possibly, in analogy to sarcomere remodeling, to regulate cytoskeletal remodeling.

Recent experiments with CAPN3 deficient myoblasts showed that membrane repair processes are not severely impaired in the absence of functional CAPN3 protein [182]. Therefore, CAPN3 is dispensable for membrane repair, contrary to Calpain 2, which is essential [182,183]. However, it cannot be excluded that CAPN3 is involved in sarcolemmal repair. Possibly, there is functional redundancy between the Calpains, as their substrate targets largely overlap (including Vinculin, Talin and Vimentin). It would therefore be interesting to develop a chemical or molecular sensor to monitor subcellular CAPN3 activity, with the objective to

acquire insight into where and when this protease is active. We could show that a ten amino acid motif is sufficient for CAPN3 proteolysis, which would theoretically allow for the development of a fluorescent sensor based on this short peptide sequence, fused to a fluorescent group, similar to what has been achieved for Calpain 1 and 2.

Due to the instability of CAPN3 the question remains how the myofiber retains a local inactive pool of CAPN3 molecules at these peripheral sites. A candidate for this function might be the giant protein AHNAK, which has overlapping localization in skeletal muscle and can directly interact with recombinant proteolytically inactive CAPN3 (Chapter 3).

The AHNAK connection

AHNAK is a 700 kDa protein that is strongly expressed in cells with barrier properties and cells with a high susceptibility to mechanical stress [220]. In the cytosol two distinct major AHNAK pools exist, being at the cytoskeleton [19], and on vesicles [53]. It is found on the luminal side of enlargeosome vesicles, and can shift from the cytoplasm to the cell membrane upon calcium entry or cell contact formation [52,53]. We observed that the giant protein AHNAK is directly interacting with Dysferlin C2A through its most C-terminal 500 amino acids [119]. In addition the same AHNAK domain can directly interact with inactive CAPN3 [118].

AHNAK has been hypothesized to function as molecular scaffold to integrate multiple functions [176]. It can interact with filamentous Actin and is important for laminin-based cell-matrix adhesion in myelating Schwann cells [220]. A C-terminal 72 kDa fragment promotes Actin polymerization [101], and loss of peripheral AHNAK destabilizes the cortical Actin cytoskeleton [19,220]. In skeletal muscle cryosections we observed that AHNAK localizes to costameres and the sarcomere [118,119]. Interestingly, we observed that in skeletal muscle myoblasts recombinant miniAHNAK redistributes in response to Integrin inhibition (Chapter 6). This suggests that it is in close vicinity to and possibly part of Focal Complexes.

AHNAK is part of the Dysferlin protein complex [119]. Moreover, AHNAK is secondarily reduced in LGMD2B muscle [119], suggesting that Dysferlin is important for localizing or stabilizing AHNAK to the sarcolemma. The direct interaction between Dysferlin and AHNAK is calcium insensitive [119], but regulated by CAPN3 proteolysis ([118], Chapter 3). CAPN3 can directly interact with AHNAK and cleave at its N- and C-terminus [118]. Cleaved AHNAK fragments have lost their affinity for Dysferlin, indicating that CAPN3 mediated proteolysis regulates the interaction between Dysferlin and AHNAK [118]. This is consistent with our hypothesis that CAPN3 regulates structural components to aid Dysferlin function. From these data we hypothesized that AHNAK is important in connecting the cortical cytoskeleton

with the sarcolemma, and that its interaction with Dysferlin is important for the integration of membrane and cytoskeletal architecture.

We propose that CAPN3 is docked in inactive form on peripheral AHNAK. AHNAK is localized to the sarcolemma through its direct interaction with Dysferlin [119]. In addition it binds the cortical Actin cytoskeleton [19] and stabilizes adhesion sites. Upon damage to the sarcolemma, a conformational change in the Dysferlin- AHNAK complex results in the release and subsequent activation of CAPN3. CAPN3 subsequently proteolyzes AHNAK and other cytoskeletal components, thereby allowing for Dysferlin trafficking and the turnover of cell membrane and cytoskeletal components.

We observed that the AHNAK gene generates a small and a large protein isoform, which can directly interact (Chapter 5). The large isoform localizes to the peripheral Actin cytoskeleton. The small isoform is also seen in nuclear speckles enriched for spliceosomal proteins, and there it affects mRNA splicing. It would be interesting to speculate that the small isoform functions as a mRNA modifying factor in response to cortical cytoskeleton remodeling.