Membrane Repair: Mechanisms and Pathophysiology

Abstract

Eukaryotic cells have been confronted throughout their evolution with potentially lethal plasma membrane injuries, including those caused by osmotic stress, by infection from bacterial toxins and parasites, and by mechanical and ischemic stress. The wounded cell can survive if a rapid repair response is mounted that restores boundary integrity. Calcium has been identified as the key trigger to activate an effective membrane repair response that utilizes exocytosis and endocytosis to repair a membrane tear, or remove a membrane pore. We here review what is known about the cellular and molecular mechanisms of membrane repair, with particular emphasis on the relevance of repair as it relates to disease pathologies. Collective evidence reveals membrane repair employs primitive yet robust molecular machinery, such as vesicle fusion and contractile rings, processes evolutionarily honed for simplicity and success. Yet to be fully understood is whether core membrane repair machinery exists in all cells, or whether evolutionary adaptation has resulted in multiple compensatory repair pathways that specialize in different tissues and cells within our body.

I. INTRODUCTION

A. The Vulnerability of a Single Membrane Bilayer

Unlike bacterial cells, eukaryotic cells are not protected by a hardened and impermeant cell wall. The “naked” membrane bilayer covering early eukaryotes permitted the evolution of phagocytic vesicles for the uptake of nutrients, and secretory vesicles for the extrusion of waste products, enzymes, and signaling factors. The loss of a cell wall also led to the development of a new internal protective skeleton, the cytoskeleton. Together, cytoskeletal networks working in concert with internal membranes led to the development of the eukaryotic endomembrane system.

However, an unprotected bilayer member renders eukaryotic cells more vulnerable to mechanical and chemical stressors. Consequently, plasma membrane disruption is a common type of cellular injury in eukaryotic cells, and effective membrane repair mechanisms have evolved to rapidly reseal a membrane breach to ensure cell survival. These repair mechanisms utilized the newly evolved endomembrane and cytoskeletal systems. Within this review we outline the subcellular and molecular events that restore bilayer integrity after a membrane disruption injury, highlighting the protein families implicated in membrane repair, and the ancient biology that underpins membrane resealing and cell survival from a membrane breach.

B. Membrane Injury Underlies Many Human Pathologies

Many human pathologies are characterized by membrane injury, and modulation of membrane repair pathways holds tremendous therapeutic potential. Plasma membrane disruptions have been documented under physiological conditions in many mechanically active tissues, such as in the stratified epithelium that covers our body, the endothelia that line our blood vessels, and the epithelial barrier of our gastrointestinal tract (178). Disruptions are especially frequent in skeletal muscle, especially when it undergoes high-force, eccentric contractions (91, 180, 199). In certain forms of muscular dystrophy, the frequency of disruption initiated by physiological contractions is far higher than in normal muscle (54, 180).

Membrane disruptions are also caused by bacterial pore-forming toxins (PFTs) that are potent virulence factors secreted by most pathogenic bacteria (120). As the name suggests, PFTs form stable membrane pores that perforate the plasma membrane of host cells. Pore formation by bacterial pathogens is thought to serve many purposes, the most obvious being lysis and induction of cell death programs in immune cells, to mute immune cell activity and thus facilitate bacterial infection. Pores may also serve as channels for the bacteria to deliver other virulence factors and to access cellular nutrients from infected cells for their own metabolic growth, such as amino acids, ions, and ATP (165). Large pores formed by the cholesterol-dependent cytolysins can span 40 nM (257) and are also permeable to cellular proteins. However, in moderate doses, cells and organisms survive the onslaught of PFT perforation, and we will discuss recent developments regarding membrane repair mechanisms mobilized for survival from bacterial pores.

Cells within our vital organs also suffer membrane damage with ischemia-reperfusion injury, as occurs following heart attack and stroke. Ischemic membrane injury represents a complex cascade of events that results from an interruption to the circulation that feeds an organ oxygen and nutrients. A lack of oxygen causes depletion of ATP. ATP-dependent pumps begin to fail, resulting in disequilibrium in the potassium-sodium gradient, acidosis, and an inability to extrude or sequester calcium. Sodium influx causes cell swelling, and calcium influx induces proteolysis and triggers mitochondrial dysfunction, production of free radicals, and apoptosis. Cell swelling, acidosis, and oxidation compromises the plasma membrane. Membranes become leaky, with breaches sufficiently large to allow the release of cellular enzymes (125). Particularly in the case of contractile cells of the heart, contraction with reperfusion exacerbates membrane injury, and a cascade of necrosis follows. Indeed, traumatic brain injury is also characterized by widespread disruption of neuronal plasma membranes. It has been proposed that these membrane disruption events initiate a “death cascade” that is a major contributor to patient morbidity (39).

C. The Influx of Calcium Through a Membrane Breach Is the Key Trigger for Membrane Repair

Universally accepted within the membrane repair field is the critical role of calcium as an activating trigger for the rapid membrane repair of large lesions. Indeed, initiation of membrane repair may represent one of the most primitive forms of calcium signaling. Eukaryotic cells possess an innate ability to repair very large wounds. Microinjection of human oocytes for in vitro fertilization creates an enormous lesion encompassing hundreds of square micrometers, but is readily survivable. Indeed, sea urchin oocytes can repeatedly reseal sequential wounds encompassing a thousand square micrometers of surface, though they cannot survive a single insult if extracellular calcium is removed (181). Cells can also repair abundant smaller wounds (nanometer diameter), such as those induced by bacterial PFTs, using both calcium-dependent (123, 129) and calcium-independent processes (164).

D. Size Matters With Membrane Injury

1. Spontaneous resealing

Tiny membrane injuries (less than a nanometer), such as those created by electroporation or proteins that induce lipid disorder, may repair spontaneously. Exposure of hydrophobic domains of lipids rapidly results in diffusion of lipids around the break site to form a curved edge. If the membrane consists only of a phospholipid bilayer, lipid disorder present on the curved edges of a disruption provides a driving force for resealing, and is a function of disruption diameter squared (Figure 1, A1–4). However, when the injured membrane belongs to a eukaryotic cell and the phospholipid bilayer is tethered to an underlying cytoskeleton (Figure 1, B1–4), the opposing force of membrane tension, a function of disruption diameter cubed, prevents spontaneous repair of biological membranes when the disruption exceeds a certain diameter. Based on experimental and theoretical data, that diameter is in the nanometer range (106).

Figure 1.

Spontaneous resealing of plasma membrane injuries in the nanometer range is opposed by the forces of the underlying membrane cytoskeleton. For an injury to a phospholipid bilayer alone (A1-4), the lipid disorder present on the curved edges of the disruption provides the driving force to spontaneously reseal the injury and is a function of disruption diameter squared. However, if the injured phospholipid bilayer is tethered to underlying cytoskeleton (B1-4), the membrane tension from adhesion to the cytoskeleton confers an opposing force for resealing, a function of disruption diameter cubed, and prevents spontaneous repair of membrane disruptions that exceed diameters in the nanometer range (106).

2. Active resealing pathways

Injuries larger than a few nanometers in diameter require the help of an active membrane repair mechanism. For these larger membrane disruptions, the opposing forces of membrane tension preclude spontaneous membrane resealing, and in the case of bacterial pore-forming toxins, the stable proteinaceous structure of the pores cannot be simply resealed.

Repair of very large disruptions, hundreds to thousands of nanometers in diameter, requires calcium-dependent exocytosis (27, 267), involving both vesicle-vesicle and vesicle-plasma membrane fusion to crudely “patch” the compromised plasma membrane (285). The sea urchin egg can replace over 2,000 square microns of surface membrane ripped from its surface in <5 s (285), and neurons and muscle cells can survive complete transections (45, 101, 146). In the sea urchin egg, large secretory granules form a membrane patch at large injuries. However, which vesicle population(s) are utilized for membrane patching in different mammalian cells and tissues is not yet clear (see sect. IIIA). Calcium-triggered exocytosis is also thought to reduce membrane tension, which facilitates resealing driven by lipid disorder (Figures 2 AND 3) (292). How exactly the exocytic addition reduces membrane tension is an important unanswered question. The answers likely relate both to the delivery of membrane lipids to reduce lipid packing, as well as the associated remodeling of the submembraneous cytoskeletal that is required for exocytic delivery of vesicles.

Figure 2.

Calcium-activated exocytosis reduces membrane tension and promotes spontaneous repair driven by lipid disorder for injuries hundreds of nanometers in diameter. With larger injuries, the opposing force of membrane tension exceeds the resealing forces of lipid disorder at the edges of the disruption, negating the driving forces of spontaneous membrane resealing. Eukaryotic cells have been shown to utilize calcium-activated exocytosis to reduce membrane tension and promote repair via lipid-disorder driven attractions. The reduction in membrane tension is likely due directly to the addition of phospholipids to reduced lipid packing, as well as due in part to the cytoskeletal remodeling associated with vesicular transport at the plasma membrane.

Figure 3.

Very large plasma membrane disruptions (micron diameter) require membrane patching. The calcium influx of a membrane injury activates vesicular exocytosis and homo- and heterotypic fusion of cytoplasmic vesicles. Exocytic fusion reduces membrane tension, and vesicle-vesicle fusion events provide a patch as a replacement for the membrane barrier missing at the disruption site. The membrane patch may serve only temporarily as a surface barrier replacement that is subsequently remodeled and removed via exocytic and/or endocytic machinery.

Smaller injuries such as those caused by bacterial PFTs (nanometer range) do not appear to utilize “membrane patching” and in fact cannot be repaired but must be removed. Thus exosomal secretion and endosomal uptake are employed for removal of bacterial-lined pores (14, 58, 123, 129, 280) (see sects. IIA and III, D and E). These membrane repair mechanisms are likely also important to remove and remodel hastily repaired larger lesions as part of the membrane repair process (Figure 3).

II. PHYSIOLOGIES OF MEMBRANE INJURY

A. Pore-Forming Toxins

We begin with what was likely an early ancestral challenge requiring membrane repair, eukaryotic cell survival from bacterial PFTs.

1. Evolution and structure

The plasma membrane of a cell represents a remarkable landscape of ordered and partitioned proteins and lipids, working together to maximize cellular signaling and communication. Pathogens have evolved that exploit the predictable clustering of receptors within cholesterol-rich plasma membrane microdomains. These pathogens release pore-forming proteins, named for their capacity to perforate the plasma membrane of their target cell. Pore-forming proteins are best characterized in bacteria (for detailed reviews, see Refs. 30, 105, 120, 213), but are also produced by many higher organisms such as sea anemones and jellyfish (8, 318), earthworms (260), and plants (70). Interestingly, the mammalian immune system has reciprocally adopted a pore-forming strategy to lyse invading pathogens. Several members of complement membrane attack complex function as pore-forming proteins to lyse bacteria and other pathogens (167), and natural killer cells and cytotoxic T-cells expressing the pore-forming protein perforin to provide passage for granzymes into target pathogens to initiate a cell death cascade (19, 149, 239). Mammalian cells also use a pore-forming strategy to activate apoptosis via the Bcl2 protein Bax (145, 217, 251).

PFTs effectively induce lytic death of many cell types in vitro, although the role of pore formation in vivo during infection is more complex (for a comprehensive recent review, see Ref. 165). PFTs specifically target an organism's immune defense by inducing lytic death of immune cells (via pores) as well as inducing cell death programs triggered by potassium efflux, calcium influx, ATP depletion, mitochondrial damage, disrupted ion homeostasis, and swelling. Pores are thought to also provide an entry point for other bacterial virulence factors that aide their infectivity, replication, and ability to escape immune detection. One of the main causes of patient morbidity relates to the deleterious overstimulation of inflammatory pathways that eventually compromises the integrity of epithelial and endothelial barriers, allowing the infection to spread, and disrupting the fidelity of the vasculature (165).

a) pfts are the largest class of virulence-related bacterial toxins. PFTs represent the largest class of bacterial toxins and play a major role in their virulence. Pathogenic bacteria such as Escherichia coli, Pneumococcus, Streptococcus, Staphylococcus, Vibrio cholera, Clostridia, Listeria, and Diphtheria, to name but a few, each produce pore-forming proteins that contribute worldwide to infection-related morbidity and mortality. Pore-forming proteins from different species typically do not share high sequence homology, but are known to assume similar tertiary structures (93, 111, 301, 312) and are thought to function using the same mode of action. There are two main classes of PFTs: the α-PFTs that adopt an α-helical fold for membrane insertion (for example, colicins from Escherichia coli) and β-PFTs that adopt a β-barrel conformation for membrane insertion and pore formation (the most common form of PFTs) (213). PFTs are released as soluble proteins by bacteria and bind to target membranes, recognizing specific GPI-anchored proteins (1, 60, 198) or lipid compartments, such as the cholesterol-dependent cytolysins (240, 257, 296). Binding to specific receptors or lipid raft regions on the plasma membrane increases their local concentration, facilitating their oligomerization, which is an essential step in pore formation. The capacity of PFTs to metamorphosize from their soluble form into oligomers, capable of transmembrane insertion and subsequent formation of a pore, is a truly remarkable feature.

2. The membrane repair response to PFT infection

The response of an intoxicated cell to PFTs depends on the type of toxin, the levels of toxin, and the period of exposure to toxin (for reviews, see Refs. 30, 46). Interestingly, cells require much longer to recover from infection by small-diameter pores (∼2 nm) such as Staphylococcus α-toxin (6 h or more) (118), compared with large-diameter pores (40 nm) such as Streptococcus streptolysin-O (SLO), that are removed by microvesicular shedding and/or endocytosis within minutes (14, 123, 129, 219). The differences in membrane repair response between small and large pores are most likely due to the fact that small pores are not permeable to calcium (305, 321), and thus do not activate rapid calcium-activated exocytic and endocytic responses. Recovery of plasma membrane integrity following infection by bacterial streptolysin-O (SLO) is perhaps one of the most well-characterized examples of how cells are able to utilize both exocytic and endocytic responses to survive a membrane injury.

a) survival from pft utilizes both exocytic and endocytic responses. SLO treatment can induce cell blebbing and release of microvesicles containing pores and SLO protein (14, 118, 136) (see Figure 4). However, perforin (287, 288) and SLO (122, 280) have also been shown to induce endocytosis into an abnormally enlarged endosomal compartment. Following pore insertion into the plasma membrane, a local elevation in intracellular calcium induces lysosome fusion with the plasma membrane (123, 229) and release of acid sphingomyelinase (280). It is proposed that hydrolysis of sphingomyelin head groups leads to the formation of ceramide-enriched plasma membrane micro-domains that activate endocytosis, and rapid removal of the SLO pores into the degradative endosomal pathway (280) (Figure 4). Endocytic removal of toxin pores presents its own set of problems, as these pores may remain conductive in the early endosomal pathway and release acid hydrolases into the cellular cytoplasm. However, several PFTs induce endocytic removal from the plasma membrane as part of their pathogenic entry into the cell (96). Thus how endosomal permeability contributes to PFT toxicity is only beginning to be teased out.

Figure 4.

Survival from bacterial pore-forming toxins utilizes both exocytic and endocytic responses. Bacterial pore-forming toxins oligomerize and insert in the plasma membrane of target cells forming a diffusible pore. Evidence suggests these pores are removed both by endosomal degradative pathways (123, 164, 280) and exosomal shedding (14, 118, 136). Shed microvesicles containing streptolysin-O have been shown to also contain annexins A1 and A6 (219). In C. elegans, Rab 5 endocytic and Rab 11 recycling pathways are implicated in pore removal (164). ESCRT machinery (pink helix) has been implicated in exosomal shedding (129), although other exosomal machinery may also be involved (indicated by a question mark).

In vivo studies in C. elegans have addressed the question of how cells escape PFT attack, and also emphasize the unique interplay between exocytic and endocytic aspects of the repair response (118, 164) (Figure 4). PFT infection of C. elegans gut epithelial cells by S. aureus α-toxin (118), B. thuringiensis Cry5B, and V. cholerae cytolysin (VCC) (164) induces rapid endocytosis and also exosomal shedding of pore-containing vesicles. Studies by Los et al. (164) demonstrated these responses were dependent on Rab-5 and Rab-11 (164), master regulators of, respectively, endosome formation and exocytic recycling. RNAi knockdown of Rab-5 or Rab-11 resulted in decreased endocytosis and worm hypersensitivity to PFTs. Importantly, Rab-11 depletion specifically prevented microvillar shedding. The authors propose that the worm's epithelial cells use both routes of PFT elimination. Toxins are directed via Rab-5-based endocytosis into the lysosome and are also shed via Rab-11-based exosomal shedding of microvillar membrane vesicles into the gut lumen (Figure 4).

Data suggest that endocytic and exosomal pathways to remove toxin-lined pores can occur via both calcium-dependent and calcium-independent pathways. α-Toxin, Cry5B, and VCC form small-diameter pores (1-2 nm) (213, 320) and are thought not to be permeable to calcium (305, 321). Moreover, endocytosis and exosomal shedding of Cry5B and VCC following in vivo infection of C. elegans were shown to occur independently of extracellular calcium in the medium (164). In contrast, removal of large calcium-conductive SLO pores in mammalian cells occurs by rapid calcium-dependent endocytosis (123) and calcium-dependent exosomal shedding (14). Additionally, the ESCRT machinery has been recently implicated in the calcium-dependent exosomal shedding of bacterial pore-forming toxins (129), and this is discussed in greater detail in section IIIF. Therefore, whether rapid calcium-dependent or slower calcium-independent pathways are utilized, likely relates to the size of the pores and whether they are conductive to calcium.

B. Muscle Injury and Muscular Dystrophy

1. Eccentric injuries and t-tubules

Muscle fibers are particularly prone to injury when subjected to lengthening contractions, referred to as eccentric damage (180). Eccentric injury occurs because whilst lengthening a muscle, for example, your quadriceps as you stride down hill, you simultaneously ask the muscle to contract against a lengthening stretch, when the membrane tension is significantly increased. The muscle plasma membrane transverse tubule (t-tubule) network is particularly sensitive to eccentric stretch, as this network of small-diameter tubules runs perpendicular to the long axis of contraction. T-tubules are invaginations of the muscle plasma membrane, ∼20–40 nM in diameter (size differs slightly amongst species) (90), that penetrate deep into the interior of the myofiber. T-tubules are anchored at precise intervals along the sarcomere (the contractile unit of muscle) and contain the voltage-gated channels responsible for initiating the wave of calcium release that activates muscle contraction. T-tubules are essential to rapidly conduct the electric impulse from the nerve into the myofiber interior, such that all of the voltage-gated channels open in unison, and contraction occurs simultaneously among all of the bundles of myofibers within a muscle group. However, eccentric stretch can result in disruption of the sarcomeric apparatus (5), and because the t-tubule network is firmly anchored to these contractile units, it too suffers a lateral stretch and gets pulled out of position (221). The fine longitudinal tubules that connect adjacent transverse tubules (80) rapidly and reversibly swell with eccentric stretch, reducing the efficiency of electrical conductivity and contraction.

a) intense eccentric exercise in untrained subjects can kill muscle fibers. When untrained normal healthy controls are subject to shorts bouts of repeated eccentric stretch, for example, 20 min of stepping up and down a stair (stepping down the stair is the eccentric stretch), the muscle membrane is injured and allows the release of the muscle enzyme creatine kinase into the serum over the following hours and days (199). Eccentric muscle damage is characterized by a feeling of weakness and wobbliness immediately after the exercise, with muscles becoming tender, sore, and stiff 1 or 2 days after the injurious event (254). Ultrastructural analysis of muscle biopsies taken from control subjects who have undergone a protocol of eccentric stretch reveals marked disruption in the organization of the skeletal muscle contractile apparatus immediately after the exercise (91, 221). Indeed, the eccentric stretch injury induces such significant damage that over the following days and weeks, the pool of muscle stem cells, called satellite cells, are activated and proliferate to repair and rebuild lethally injured myofibers (91, 131).

2. Persistent membrane injury in muscular dystrophy

In patients with muscular dystrophy, creatine kinase is persistently elevated (regularly hundreds of times higher than control levels) and muscle biopsy samples show histopathological signs of ongoing degeneration and regeneration of muscle fibers. The most common form of muscular dystrophy is Duchenne's muscular dystrophy (DMD), an X-linked inherited disorder affecting ∼1:3,000 boys, due to mutations in the cytoskeletal protein dystrophin (113).

Dystrophin is part of the spectrin superfamily of cytoskeletal proteins and is thus proposed to lend elasticity to the muscle plasma membrane, as well as provide the structural cornerstone of the muscle costamere. The costamere is a focal adhesion-like complex assembled at regular intervals along the sarcomere, providing a stabilizing connection between the cytoskeletal apparatus via the intermediate filament network, through the plasma membrane, and out to the extracellular matrix (85). At costameres, dystrophin forms part of a large transmembrane glycoprotein complex (86, 121). The lack of dystrophin, or other components of the dystrophin-glycoprotein complex, causes different forms of inherited muscular dystrophies (55) and renders muscle much more susceptible to injury with eccentric stretch (216) (for a review, see Ref. 56).

In 1999, a new gene was identified as the cause of a form of inherited muscular dystrophy, dysferlin (22, 161). Dysferlin did not sediment with the dystrophin-glycoprotein complex and instead was shown to be required for acute resealing of laser-injured myofibers (18). Thus, rather than playing a structural role in the sarcolemmal stability, dysferlin was proposed as a key mediator of calcium-dependent muscle membrane repair and is discussed within section IIIB. A primary defect in skeletal muscle membrane repair characterizes dysferlinopathy, but is also a feature of muscle fibers from diabetic mice (116) and cells from patients with lysosomal storage disease (49, 119, 280) (discussed further in sect. IIIF). Furthermore, evidence for t-tubule injury and repair is also a feature in statin myopathy (77, 302). Thus an imbalance in susceptibility to membrane injury, and capacity for membrane repair, may be common in many forms of myopathy.

a) what does muscle membrane injury look like? We do not really know the precise nature of membrane disruptions caused by repeated eccentric stretch, by lengthening strain with a sporting injury, or suffered with more routine physical activity in a patient with muscular dystrophy. What we do know is that the egress of creatine kinase (199) and the uptake of albumin into eccentrically injured muscle fibers (180) indicates sites of membrane permeability are sufficiently large to allow the transfer of large macromolecules. The dimensions of serum albumin are ∼80 Å × 80 Å × 30 Å (272) (roughly 8 nM × 8 nM × 3 nM). Therefore, for significant levels of albumin to enter an injured muscle fiber, there must either be multiple large lesions encompassing hundreds to thousands of square nanometers, a multitude of smaller lesions hundreds of square nanometers, or perhaps a mixture of both. All possibilities are consistent with the notion that mechanical stretch leads to physical disruption of the plasma membrane and t-tubule network (180).

Studies of dystrophin-deficient fibers isolated from the mdx murine model of DMD and loaded with various sodium- and calcium-sensitive dyes could find no evidence for overt membrane tears following eccentric stretch (316). Rather, Yeung et al. (316) provide evidence that dystrophin deficiency alters the properties of stretch-activated calcium channels that leech calcium into the cell after a bout of eccentric injury. The dysregulation in calcium homeostasis leads to calpain overactivity, mitochondrial dysfunction, and oxidative damage (for review, see Ref. 6). Oxidation of membrane lipids and proteins is thought to render muscle fibers more susceptible to injury with subsequent challenges.

However, these were single fibers studied in isolation. Studies of whole muscles derived from mdx and control mice concurred that calcium appears to “leak” into dystrophin-deficient fibers after bouts of contraction, inducing local hypercontraction. This region of local hypercontraction exerts a lateral strain on adjacent lengthened regions (see Figure 5). These regions then incurred a more significant injury that caused global hypercontraction and “concertina-like” retraction of fibers within the muscle bundle (53). This sort of fiber retraction, ripping away from adjacent muscle fibers, would reasonably cause large membrane injuries, sufficient to explain the large egress of creatine kinase in patients with muscular dystrophy (Figure 5). Even in healthy muscles, an in vivo model of a large strain injury caused membrane injuries of sufficient magnitude to allow uptake of large dextran macromolecules, which are excluded at later time points, but does not activate satellite cell repair (235, 236). This provides direct evidence that an injurious strain to healthy muscle causes membrane injuries that are repaired in a regeneration-independent manner (see sect. IIIB).

Figure 5.

What might membrane injury to muscle fibers look like? Muscle fibers have a complex plasma membrane network with a repeating register of deep plasma membrane invaginations called the t-tubule network. Muscle fibers are subject to huge variations in membrane tension, due to their contractile activity. Repeated eccentric exercise in healthy subjects (i.e., stepping down for 20 min) is known to induce damage so severe that muscle fibers degenerate over the following days and weeks (91, 131, 199). Patients with muscular dystrophy are more susceptible to injury from eccentric stretch (216), with studies in mouse models suggesting susceptibility to injury can escalate with multiple insults (53). One model explaining membrane injury in dystrophin-deficient muscle fibers proposes that an initial injury causes a local influx of calcium and a local region of hypercontraction. These shortened sarcomeres induce a concomitant lengthening of adjacent sarcomeres and increased lateral strain to the plasma membrane. Subsequent insult(s) of eccentric stretch result in a more severe wound and global hypercontraction, producing fiber retraction within the muscle bundle (53). As muscle fibers have strong interfiber connections, muscle injuries may manifest both as shearing of the membrane from increased membrane tension and strain, as well as ripping of plasma membrane regions from fiber retraction or hypercontraction.

C. Ischemic Membrane Injury

1. Ischemia-reperfusion injury is a leading killer in the Western world

Disorders associated with ischemia-reperfusion injury, such as heart attack, stroke, and vascular disease, encompass the largest causes of mortality and morbidity in the Western world. Disruption in the blood flow to a tissue or organ can result in large regions of cell death (infarct), and often occur as a consequence of a clot within a major artery delivering circulation. The size of the infarct is primarily determined by the length of time blood flow is obstructed and the nature of the blood vessel affected. Blood delivers oxygen and glucose to tissues, both of which are required for the production of ATP by the mitochondrial respiratory chain. Thus ischemia induces a rapid loss of cellular ATP, and a complex cascade of events ensues from ATP depletion. For comprehensive reviews in the area of ischemia-reperfusion injury, please refer to References 132, 193, and 247 and references therein.

2. Cells need ATP to maintain ion homeostasis

ATP drives the sodium and potassium exchange pumps that maintain plasma membrane potential, and vitally, the F₁/F₀-ATPase used to generate mitochondrial membrane potential (Δψ). ATP also drives the pumps that extrude calcium from the cell (sodium-calcium exchanger) and the pumps that drive cytoplasmic calcium back into the endoplasmic or sarcoplasmic reticulum after each contraction in heart and muscle (via sarcoplasmic endoplasmic reticulum calcium ATPase, SERCA). During ischemia, cells are obliged to utilize anaerobic metabolism to produce ATP. This results in a rapid increase in lactate and protons (acidosis), and from here, there is a downhill spiral of compensatory mechanisms that conspire to result in calcium overload and toxicity.

To try and reneutralize normal cytosolic pH, cells furiously pump out the protons in exchange for sodium using the Na⁺/H⁺ exchanger (191). These pumps cannot keep up with demand, particularly with low cellular ATP, and the massive influx of sodium ions results in cell swelling. Attempts are also made to extrude the sodium in exchange for calcium using the sodium/calcium exchanger that begins to operate in reverse, allowing more lethal calcium into the cell. Unfortunately, with ATP in short supply, a large problem for contractile heart cells is reuptake of calcium following contractions, whereby ATP-dependent SERCA is compromised, and calcium release by the ryanodine receptor is enhanced (279).

3. Calcium toxicity, cell swelling, and oxidative damage conspire to injure membranes

Reperfusion adds fuel to the fire. Although reestablishing the oxygen supply is essential for cell survival, restoration of extracellular pH to neutrality increases the proton gradient for cells with an acidic intracellular pH; the Na⁺/H⁺ exchanger works busily and sodium levels increase, followed by an even larger influx of calcium via the sodium/calcium exchanger (192). Elevated levels of submembraneous calcium activate calpains that cleave focal adhesion complexes and cytoskeletal elements (see sect. IIIC), compromising the fidelity of the plasma membrane. Also with this flush of oxygen, reactive oxygen species (ROS) are produced by the mitochondria, due to damage to the electron transport chain incurred during ischemia, defective transport of electrons, and production of superoxide (10, 94, 322). ROS cause widespread oxidation of both protein and lipids, another insult for the plasma membrane.

Increased levels of cytosolic calcium and ROS can activate a large mitochondrial conductance channel called the mitochondrial permeability transition pore (MPT) (16). The MPT pore allows protons into the mitochondrial matrix, ablating mitochondrial membrane potential, uncoupling the electron transport chain, and inhibiting ATP production. Water also flows through the MPT pore, causing mitochondria to swell and rupture. If a large number of mitochondria within a cell activate the MPT pore, the cell will not be able to synthesize sufficient ATP, ion homeostasis will be lost, the cell will swell, the plasma membrane will rupture, and in many cases the cell will initiate cell death cascades through either necrotic, apoptotic, or autophagic pathways (103, 104).

4. Brain is particularly sensitive to ischemia

Different tissues differ in their sensitivity to ischemic injury, with brain being the most sensitive, then heart, kidney, liver, and lastly skeletal muscle, which is remarkably resistant to ischemia-reperfusion injury. Sensitivity to ischemic injury seems to relate to the capacity of a tissue to transition successfully to anaerobic metabolism, the capacity of its fuel stores, and its intrinsic resistance to oxidative stress (132). Brain is utterly dependent on oxidative metabolism and has very low levels of stored glycogen, in contrast to liver and skeletal muscle, for example. Brain also has lower levels of antioxidants such as superoxide dismutase, glutathione peroxidase (4), and heme oxygenase (68). Collectively, these attributes render the brain particularly sensitive to ischemic injury, with stroke being a leading cause of worldwide morbidity and mortality.

5. What does ischemic membrane injury look like?

The precise structural form of ischemic membrane injury is not known. Membrane permeability most likely occurs as a combination of effects related to calcium toxicity, calpain activation, cell swelling due to altered sodium/potassium homeostasis, oxidative damage, and in the case of cardiac myocytes, contraction-induced mechanical injury of a damaged plasma membrane. The heart cannot stop beating. Like skeletal muscle injury, ischemic membrane permeability is inferred by the egress and uptake of large macromolecules such as lactate dehydrogenase and albumin that are unable to cross an intact lipid bilayer. Thus we know the membrane is permeable, and permeable enough to allow high levels of macromolecular passage, but we do not know exactly what is the size, number, or disposition of the membrane breaches.

III. THE PROTEIN MACHINERY OF ACUTE MEMBRANE REPAIR

A. SNAREs

1. Evolution and structure

The SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) comprise three families of proteins that interact to form the core machinery of vesicle fusion: syntaxins, SNAPs (soluble N-ethylmaleamide attachment proteins), and VAMPs (vesicle associated membrane proteins). SNARE proteins have ancient eukaryotic origins and are present in protozoans, early unicellular eukaryotes, fungi, yeast, and animals, and are central players in the evolution of the eukaryotic endomembrane system (65–67).

a) snare complex and how vesicle fusion happens. Three proteins form a SNARE complex: two proteins from the target membrane, syntaxin (26) and SNAP (207), and one from the vesicle membrane, VAMP (295). Most SNARE proteins are tail-anchored transmembrane proteins, with the “business side” of the protein facing the cytoplasm, and only a short luminal or extracellular tail (see Figure 6). A common feature of all SNAREs is the presence of a cytoplasmic α-helical heptad repeat domain, consisting of ∼67 amino acids, often positioned just before the transmembrane domain. This α-helical sequence is referred to as the SNARE motif and facilitates interaction of the three SNARE proteins into a parallel coil-coiled structure (275). The quaternary α-helical complex formed by assembly of syntaxin, SNAP, and VAMP is extremely stable and resistant to heating (but not boiling) in sodium dodecyl sulphate (SDS) lysis buffer. These ternary complexes self-assemble to form a ring (50), zippering from the NH₂ terminus to the COOH terminus to draw the opposing membranes of the vesicle and target membranes together for fusion (for recent reviews, see Refs. 210, 234).

Figure 6.

Schematic representation of the structural features of the protein families implicated in membrane repair.

The surface of exocytic vesicles is negatively charged due to polar phosphate headgroups, providing a natural vesicle-vesicle and vesicle-plasma membrane repulsion that must be overcome for membrane fusion. Thus membrane fusion is thought to be driven by the strong association of the SNARE components, the binding of positively charged calcium to neutralize the negative repulsion and repel water between the fusing membranes (128), and local disruption of the membrane curvature of the opposing bilayers by the transmembrane domains of the SNARE proteins (266; for recent reviews, see Refs. 127, 133).

b) synaptotagmins are late evolutionary arrivals to the snare complex. At neural synapses, synaptotagmins 1 and 2 trigger the calcium-activated fusion of neurotransmitter-containing vesicles (35, 38, 81) through interaction with SNARE vesicle fusion machinery (253). Synaptotagmins 1 and 2 localize to secretory vesicles and are type I transmembrane proteins with two, tandem cytoplasmic C2 domains and show calcium-regulated interaction with members of the SNARE complex. However, the precise mechanism by which synaptotagmins facilitate or accelerate calcium-regulated exocytic fusion remains the topic of intense debate. It is thought that calcium-binding to synaptotagmin triggers phospholipid binding of its C2 domains to the vesicle and plasma membrane, as well as binding to the SNARE complex, releasing complexins, which negatively regulate SNARE assembly (282).

Interestingly, phylogenetic analyses reveal that synaptotagmins originated in multicellular eukaryotes (61) evolutionarily postdating syntaxins (65, 66), suggesting core SNARE machinery could once function independently of synaptotagmin in primitive cells, or alternately worked in concert with a more evolutionary ancient predecessor. The synaptotagmin gene family rapidly expanded in metazoans, with correlation between multicellular complexity and the number of synaptotagmin paralogs particularly apparent in the evolution of green plants (248). In humans, there are 16 synaptotagmins (61) and 15 syntaxins (284), each displaying a distinct pattern of subcellular localization to mediate the trafficking of vesicular cargo between different intracellular destinations. Synaptotagmins-1 and -2 are essential for rapid, synchronous neurotransmission at vertebrate synapses and induce neonatal lethality in knockout mice (97, 209).

2. Role in membrane repair

a) parallels between membrane repair and synaptic exocytosis. Steinhardt et al. (267) significantly advanced our understanding of how membrane repair works, when they showed that membrane resealing utilized a process with parallels to synaptic neurotransmission. Confocal microscopy of damaged sea urchin embryos or unfertilized eggs showed that a membrane puncture induced the exocytic fusion of secretory yolk granules (27, 267). This injury-induced exocytic fusion was wholly dependent on calcium, not activated by other cations, and antagonized by magnesium, with each property also being a feature of synaptic exocytosis. Larger membrane disruptions required vesicle-vesicle fusion to form a patch, whereby steps of “vesicle-to-vesicle” fusion, and integration of fused “patches” into the plasma membrane, were both strictly dependent on extracellular calcium (179, 285). These studies also showed that successful membrane repair correlated with the degree of exocytic fusion and that membrane repair was blocked or severely inhibited by clostridial neurotoxins in sea urchin eggs (docked cortical granules), embryos (undocked cortical granules), and mammalian Swiss 3T3 cells (nonsecretory cells) (27, 267). The clostridial neurotoxins are proteases that specifically cleave SNARE proteins and induce paralysis in infected victims (29, 252, 314). Further experiments using recombinant fragments of synaptotagmins or SNAREs, or inhibitory antibodies recognizing synaptotagmins or SNAREs, collectively implicated SNARE proteins in the vesicle fusion of membrane repair (72, 291).

b) a role for synaptotagmin vii and lysosomal exocytosis for membrane repair? In 2001, Reddy et al. (229) reported that membrane repair in nonsecretory skin fibroblasts was mediated by synaptotagmin VII (SytVII) via calcium-activated lysosomal exocytosis (229). Membrane injury induced the calcium-dependent appearance of the luminal epitope of lysosomal associated membrane protein (LAMP-1) on the cell surface of wounded skin fibroblasts, and treatment with anti-SytVII antibodies, recombinant SytVII C2A, or anti-LAMP-1 antibodies impaired lysosomal exocytosis and membrane repair (229). A SytVII knockout mouse was then generated and showed normal growth and development, but developed an inflammatory myopathy with elevated creatine kinase and muscle weakness with other autoimmune symptoms, such as dermatomyositis and an antinuclear antibody response (47). Embryonic fibroblasts from SytVII knockout mice were shown to be defective in lysosomal exocytosis and membrane resealing after wounding (47). The authors later showed that loss of SytVII resulted in fewer lysosomal fusion events overall and the properties of the fusion events were different (126). In wild-type cells, calcium influx triggers lysosomal fusion events characterized by a small fusion pore and minimal diffusion of lysosomal transmembrane proteins into the plasma membrane. In contrast, SytVII knockout fibroblasts showed more complete lysosomal fusion events with merging of lysosome membrane contents with the plasma membrane, suggesting SytVII regulates and restricts fusion pore formation (126). Further studies are required to reconcile how these aberrations in lysosomal exocytosis relate to failed repair in SytVII null cells.

More detailed studies of the effect of recombinant domains of synaptotagmin I and VII on membrane repair showed that the C2A domain of SytVII had no effect on membrane resealing, but did impair the facilitated response to a second injury at the same site (258). Interestingly, treatment with recombinant C2B of SytVII inhibited both initial membrane resealing and the capacity of an injured cell to more readily repair a second injury. Similar inhibition of membrane resealing was observed with recombinant C2B of synaptotagmin 1. Inhibitory effects were dependent on the calcium sensitivity of the C2 domains, with no effects elicited by calcium-binding mutants of each domain. However, the authors stress that treatment with recombinant C2 domains of synaptotagmins likely indirectly impacts membrane repair by promiscuously binding SNARE machinery and target phospholipids (258).

3. Perspectives

Evidence suggests that lysosomes are not membrane repair organelles and do not directly contribute to the exocytic formation of “membrane patches” for lesion repair. Although lysosomal exocytosis was shown to occur in response to ballistics injuries in skeletal muscle cells, these fusion events occurred distal to the injury site, and lysosomal markers did not demark vesicles recruited and enriched at injury sites labeled by dysferlin (150). Similarly, zebrafish lysosomal membrane proteins did not accumulate at injury sites in zebrafish larval muscle cells, although lysosomes were observed to fuse with the plasma membrane when near an injury site (238). However, there is an expanding body of evidence that lysosomal exocytosis occurs in response to the acute elevation in intracellular calcium caused by membrane injury and that machinery of the late endosomal pathway play key roles in membrane repair. Specific roles in membrane repair for late endosomal machinery ESCRT (129, 250) and mucolipin-1 (49) are discussed in section IIIF. Exocytic fusion of late endosomes/lysosomes may function to reduce membrane tension, to deliver protein machinery required for vesicle formation and exosomal shedding (129, 250), as well as to release enzymes such as acid sphingomyelinase to activate endocytosis (280).

Collectively, there is a strong body of evidence showing that membrane repair is inhibited by clostridial neurotoxins, recombinant protein fragments of SNARE proteins and synaptotagmin C2 domains, and by antibodies that functionally inhibit SNAREs and synaptotagmins. Thus, although SNARE machinery is squarely implicated in the vesicle fusion of membrane repair, it is yet to be determined whether there is an ancient and preserved set of core SNARE machinery for membrane repair, or whether there have been evolutionary adaptations in the repair mechanism to best suit the types of injury encountered, and the available endogenous vesicle fusion machinery, in different cell lineages.

While the extremely high concentration of local calcium around a wound site is not altogether dissimilar from an active synaptic zone, or directly adjacent to the calcium-release channels of the triad junction in skeletal muscle, this level of intracellular calcium would be an unusual event in a nonsecretory cell and would certainly create a unique and very active local environment. The flood of extracellular calcium thought the breached plasma membrane will promiscuously initiate calpain-mediated proteolysis, calcium second messenger signaling, and calcium-regulated fusion of any nearby primed or “primable” vesicles. In the setting of a membrane injury where it is repair or die, the evolutionary pressure on the effectiveness of membrane repair means that each of these cascades must play an important and interrelated role. It is difficult to reason whether it is more likely that a core, evolutionary ancient membrane repair machine exists in all cells, or whether promiscuity in the fusion machinery and vesicle populations utilized for membrane repair is central to the survival response itself.

B. Dysferlin

1. Evolution and structure

Dysferlin is a large (∼240 kDa) tail-anchored transmembrane protein that bears the unique feature of seven tandem C2 domains within its large cytoplasmic domain, the most of any protein family. C2 domains are independently folding protein motifs comprised of ∼110–130 amino acids, originally identified in protein kinase C (202). C2 domains mediate lipid and protein binding, often regulated by coordination of calcium ions within a negatively charged binding pocket comprised of highly conserved acidic residues (usually aspartate) (51). The crystal structures of many C2 domains have been solved and feature a folded sandwich of two β-sheets, each containing four anti-parallel β-stands (87, 274). Clustered at the end of the β-sheet sandwich reside three variable connecting loops that, in the case of calcium-sensitive C2 domains, contain the highly conserved acidic residues that form the binding pocket for multiple calcium ions. The binding of calcium within this pocket directly facilitates membrane interaction (196), with the amino acid composition of the loop region shown to influence phospholipid selectivity for targeting specific membrane compartments (51).

a) dysferlin belongs to an ancient family of vesicle fusion proteins. There are six mammalian ferlin proteins: dysferlin (Fer1L1), otoferlin (Fer1L2), myoferlin (Fer1L3), Fer1L4, Fer1L5, and Fer1L6, each characterized by a cytoplasmic domain bearing between five and seven tandem C2 domains anchored by an extreme COOH-terminal transmembrane domain (for a recent review, see Ref. 151). Our phylogenetic studies have revealed that ferlins have ancient origins in eukaryotic biology (152). Ferlins are detected in all eukaryotic kingdoms, including unicellular phytoplankton and in protozoans, indicating origins predating evolutionary branching. Ferlins have not yet been identified in fungi or plants, suggesting they may have been lost from these evolutionary lineages. Invertebrate and vertebrate animal models of ferlin deficiency are united by pathologies linked to defective calcium-activated vesicle fusion (3, 18, 59, 205, 242). Thus it is proposed that ferlins are a family of calcium-binding vesicle fusion proteins for regulated exocytosis, with evidence suggesting more ancient evolutionary origins than the classical mediators of vesicle fusion, the synaptotagmins (see section IIIA) (151).

2. Localization

Dysferlin is expressed ubiquitously in mammalian tissues, with high levels in skeletal muscle and heart (9). Dysferlin localizes to the skeletal muscle plasma membrane, also called the sarcolemma (9, 218), to the invaginating t-tubule network (7, 135, 138, 163). Dysferlin has also been shown to localize to the apical plasma membrane of syncytiotrophoblasts (297), suggesting a role in polarized membrane trafficking in the placenta. Studies of dysferlin trafficking in transfected C2C12 mouse myoblasts reveal that dysferlin shuttles between the plasma membrane and the endo-lysosomal pathway (88).

3. Role in human disease

Dysferlin mutations underlie a form of autosomal recessive inherited muscular dystrophy, called limb girdle muscular dystrophy type 2B (LGMD2B) (22, 161). Dysferlinopathy is a late-onset form of muscular dystrophy, manifesting in older teenagers or adults, and is characterized by absence or marked reduction of dysferlin protein in the skeletal muscle of affected patients. Curiously, prior to their presentation, dysferlinopathy patients show no evidence for subclinical muscle weakness as children, with many patients reporting sporting distinction in their youth. This differs from other later onset muscular dystrophies and myopathies, in which there is often a long history of poor sporting performance and avoidance of strenuous physical activities. In dysferlinopathy, physically able teenagers often suffer an injury that is difficult to recover from, and begin to experience unexplained fatigue and muscle pain, followed by progressive muscle weakness. Unfortunately, once dysferlin muscular dystrophy manifests, the physical decline can be rapid, from being able-bodied to nonambulant (unable to walk) in 4–8 yr (200). It is not understood why dysferlin deficiency does not clinically affect young muscles of affected children yet results in severe weakness and muscle degeneration in adulthood, with some patients also presenting with mild cardiac involvement (143, 310).

4. Role in membrane repair

a) dysferlin plays a key role in calcium-dependent membrane repair. Dysferlin-deficient mice also exhibit a late-onset progressive muscular dystrophy (18, 32, 112). A seminal study by Bansal et al. in 2003 (18) revealed that muscle fibers lacking dysferlin did not show the same membrane fragility as dystrophin-deficient fibers (the basis of Duchenne muscular dystrophy, the most common human muscular dystrophy), and instead demonstrated defects in calcium-activated membrane resealing following laser injury. Control muscle fibers were shown to effectively reseal a laser-induced plasma membrane injury and exclude the styryl dye FM1-43 within 30 s, whereas dysferlin-null mouse muscle fibers showed increased and prolonged entry of FM1-43 up to 2 min following the laser injury. Indeed, the kinetics and magnitude of FM1-43 dye entry in injured dysferlin-null fibers resembled results from wild-type fibers damaged in the absence of calcium. Given the homology of dysferlin to the C. elegans protein Fer-1, previously shown to regulate calcium-activated vesicle fusion (3, 307), dysferlin was therefore proposed to be a key mediator of calcium-activated vesicle fusion for muscle membrane repair.

A role for mammalian ferlins in calcium-activated vesicle fusion was further strengthened by the discovery that otoferlin, the genetic basis of a form of inherited nonsyndromic deafness in humans (315), was due to defective calcium-activated auditory neurotransmission at the cochlear inner hair cell synapse (242). It remains debated whether the role of otoferlin in the cochlear relates solely to functions as a calcium-activated trigger for synaptic vesicle exocytosis, or whether its role may instead/also relate to endocytosis and recycling of synaptic vesicles to restock the ready releasable pool of neurotransmitter containing vesicles required for sustained, high-frequency firing of the highly demanding inner hair cell synapse (211).

Although dysferlin is expressed ubiquitously, the skeletal muscles are particularly affected by loss of dysferlin. This may relate to the specialized architecture of the skeletal muscle t-tubule membrane network and its vulnerability to eccentric stretch. Dysferlin is abundantly expressed within the t-tubule network, and dysferlin-deficient muscle fibers show major t-tubule abnormalities after a bout of in vivo lengthening strain injuries (135). Indeed, dysferlin-deficient fibers suffer significant damage after long strain injury, requiring satellite cell-mediated repair pathways to rebuild necrosing muscle fibers following a stretch protocol (236). In contrast, wild-type fibers readily survive a lengthening strain injury without evidence for necrosis, and without requiring satellite-cell mediated repair (236). These data suggest dysferlin is particularly important for repair and remodeling of t-tubule membranes.

b) dysferlin translocation to injury sites. Bansal et al. (18) studied the localization of dysferlin in myofibers injured via aspiration through an 18-gauge needle, using fluorescent dextran to demark injured fibers. Confocal microscopy revealed intensely labeled “patches” of dysferlin that appeared to correlate with potential sites of plasma membrane disruption, inferred by brightfield images showing small regions of thickened and nonuniform sarcolemma, visually discontinuous with an otherwise uniform section of myofiber sarcolemmal membrane, and consistent with a small repair patch. These regions showed reduced or absent labeling for constitutively expressed sarcolemmal proteins caveolin-3 and δ-sarcoglycan, consistent with recently repaired injury “patches” derived from nonsarcolemmal membrane sources. Dysferlin is also observed to intensely label the fine, meshlike network of longitudinal tubules of the t-tubule network that become vacuolized and injured with over-stretch (304).

Many groups have attempted to study the recruitment of heterologously expressed dysferlin to sites of membrane injury in cultured myotubes. Klinge et al. (138) suggested that an NH₂-terminal EGFP-dysferlin fusion protein showed calcium-dependent and injury-dependent plasma membrane enrichment in C2C12 myotubes injured by rolling glass beads. Dysferlin enrichment appeared broad and generalized over large areas of a myotube, in stark contrast to the tightly refined patches of endogenous dysferlin recruited to proposed injury sites in mature skeletal myofibers reported by Bansal et al. (18). This may reflect immaturity of the myogenic model employed by Klinge et al. (138), comparatively larger areas of membrane injury induced by the rolling beads, or perhaps disruption of normal dysferlin behavior through epitope-tagging. Arguably however, the precise site of membrane injury could not accurately be determined in either case, and therefore, it is difficult to qualify whether areas of dysferlin enrichment represent injury sites, or not.

Live cell imaging experiments in transfected murine C2C12 myoblasts indicated that EGFP-dysferlin is recruited to sites of laser injury or needle microinjection only when coexpressed with the muscle-specific membrane repair cofactor mitsigumin-53 (MG53, discussed in detail in sect. IIIE) (42). However, zebrafish dysferlin expressed as a COOH-terminal fusion with monomeric teal fluorescent protein showed rapid recruitment to injury sites during in vivo imaging of laser-injured zebrafish myofibers, and zebrafish do not express a MG53 paralog (238). Thus whether MG53 is required for injury recruitment of dysferlin in muscle cells remains unclear. The sophisticated in vivo imaging experiments described by Roostalu and Strahle (238) revealed that a short dysferlin COOH-terminal fragment (the extracellular domain of 22 amino acids, transmembrane domain, and only 29 of the ∼2,000 amino acids of the cytoplasmic domain) was sufficient to confer targeting to recruited repair membranes. In contrast, truncated dysferlin NH₂-terminal domains were unable to be effectively recruited, suggesting cellular targeting of dysferlin by its transmembrane domain to appropriate membrane compartments is vital for its mobilization and recruitment to injury sites.

Lek et al. (150) recently developed a ballistics model of membrane injury in cultured human myotubes, producing readily identifiable enface injuries, suitable for high-resolution imaging. Three-dimensional structured illumination microscopy (3D-SIM) was used to reconstruct the recruitment of dysferlin to injury sites (150). Super-resolution imaging of injury sites resolved the rapid recruitment of dysferlin-containing cytoplasmic vesicles to sites of membrane injury within 10 s, undergoing calcium-dependent integration into plasma membrane compartments decorated by MG53 (Figure 7). This process is surprisingly consistent with the mechanism proposed by Steinhardt et al. (267) nearly 20 years ago entitled: “membrane repair occurs via a process analogous to synaptic exocytosis.”

Figure 7.

A cartoon depicting the potential role of dysferlin-mediated vesicle fusion in membrane repair. Membrane injury causes a local influx of calcium and activation of calpains. MG53 (40) shows diffuse enrichment at injury sites within 2 s of membrane injury in a calcium-independent manner (150). The signal to activate recruitment of MG53 to injury sites is not clear, but may relate to its role as a ubiquitin ligase to target substrate(s) damaged as a consequence of the membrane injury. Dysferlin is not detected at injury sites until 10 s postinjury, a delay we attribute to an intermediary step involving calpain cleavage. Activated calpains cleave dysferlin within a motif specifically encoded by alternately spliced exon 40a (230). As dysferlin may only be detected at injury sites with antibodies recognizing COOH-terminal epitopes, and not several antibodies to NH₂-terminal or central domains (150), data suggest the COOH-terminal cleaved fragment termed mini-dysferlin_C72 is the form specifically recruited to injury sites. It remains uncertain whether full-length dysferlin is also present at injury sites but is structurally precluded from antibody labeling and whether the dysferlin recruited for membrane repair is derived from plasma membrane or intracellular membrane compartments. Super-resolution 3-dimensional structured illumination microscopy (3D-SIM) reveals dysferlin-laden vesicles undergo calcium-dependent integration at injury sites (150). At 10 s postinjury, dysferlin and MG53 form a lattice that intensely labels the edges of the lesions. Lesions expand and are filled in by a dysferlin and MG53 lattice. At 90 s postinjury, filled lesions are characterized by a dominant arc of dysferlin and MG53 labeling we believe represents the original edges of the lesion that have been drawn together by cytoskeletal motors for resealing. [3D-SIM images from Lek et al. (150), with permission from The Journal of Neuroscience.]

Immediately following ballistics injury, dysferlin specifically and intensely labeled the exposed phospholipids encircling the ballistics injuries (10 s post injury) (150). At later time points, dysferlin formed an intricate lattice (with MG53) labeling the broader surrounds of the membrane lesions (20–60 s post injury). One to two minutes after injury, dysferlin labeled a bright arc of two closely opposed membranes lying in a bed of dysferlin and MG53 lattice, positioned parallel to the long axis of the cultured myotubes (Figure 7). These observations were consistent with a model whereby dysferlin initially labels the periphery of the lesion, forms a repair lattice that infiltrates the lesion surrounds as new membrane is delivered, and cytoskeletal motors “zipper” the ballistics lesions together by drawing opposing membranes together lengthwise along the long axis of the myotube.

c) injury-activated calpain cleavage of dysferlin: emergency production of a synaptotagmin-like vesicle fusion module for membrane repair? Perplexingly, dysferlin could only be detected at injury sites using an antibody recognizing the extreme COOH terminus of the dysferlin cytoplasmic domain, Hamlet-1, and not with three other anti-dysferlin antibodies recognizing more NH₂-terminal epitopes (150). Biochemical analyses subsequently revealed that dysferlin was cleaved by activated calpains with membrane injury, releasing a COOH-terminal fragment of 72 kDa, termed mini-dysferlin_C72. Interestingly, mini-dysferlin_C72 bears the last two most ancestrally conserved C2 domains and transmembrane domain (152), with structural parallels to the classical vesicle fusion proteins, the synaptotagmins. Results therefore suggest that it may not be full-length dysferlin that is recruited to injury sites in cultured human myotubes, but a COOH-terminal fragment of dysferlin, mini-dysferlin_C72.

We have subsequently shown that calcium-dependent cleavage of dysferlin is mediated by the ubiquitous calpains (calpains-1 and -2) via a cleavage motif encoded by an alternately spliced exon, exon 40a (230). Thus not all dysferlin isoforms may be cleaved by calpains in response to injury. Interestingly, other members of the ferlin family are also cleaved by calpains to release similar COOH-terminal modules (230). Evolutionary conservation of this feature implies that calpain cleavage of ferlins bestows an important functional modification in settings of intense calcium signaling.

Interestingly, while dysferlin transcripts bearing exon 40a are abundantly expressed in many human tissues (40–60% of all transcripts in kidney, lung, liver, placenta, and pancreas) (230), somewhat counterintuitively, only ∼15% of dysferlin transcripts in skeletal muscle contain exon 40a (220, 230). Thus not all dysferlin isoforms in skeletal muscle can be cleaved by calpains. Further studies are needed to clarify the respective roles of full-length dysferlin (without exon 40a) and cleaved mini-dysferlin_C72 for membrane repair of skeletal muscle and other tissues.

d) a patient mini-dysferlin restores membrane repair, but not dystrophic pathology. Further evidence that cleaved mini-dysferlin_C72 plays a specialized role in membrane repair has been provided serendipitously, through studies of a naturally occurring truncated dysferlin based on a patient with a genomic deletion within the dysferlin gene, that is very similar to calpain-cleaved mini-dysferlin_C72 (140). In this patient, a genomic region encompassing exons 2–40 of dysferlin are deleted, and the patient employs a cryptic splice site to express a truncated dysferlin with 13 amino acids at the NH₂ terminus derived from intronic sequences, followed by exons 41–55. Calpain-cleaved mini-dysferlin_C72 bears approximately seven residues of exon 40a (although calpains do not strictly cleave at one site, and may cleave either side of their preferred site), followed by residues encoded by exons 41–55. The patient presents with a mild-moderate dysferlinopathy and transgenic expression of the patient mini-dysferlin in dysferlin-null muscle fibers restored normal membrane repair (140). These data support our proposal that calpain-cleaved mini-dysferlin_C72 is an important mediator of membrane repair. However, transgenic expression of the patient mini-dysferlin did not prevent development of a dystrophic pathology in dysferlin-null mice (140, 166). Thus defective membrane repair does not appear the sole factor underlying the pathology of dysferlinopathy.

e) which pool of dysferlin is cleaved? A recent study created a transgenic mouse model expressing dysferlin (without exon 40a) with an extracellular pHluorin tag (172). pHluorin is a genetically modified form of GFP that is pH sensitive (184), in this case showing reduced fluorescence in acidic compartments. McDade et al. (172) showed that immediately following membrane injury, dysferlin-pHluorin fluorescence diminished in regions surrounding, and distal to, the injury site, and cytoplasmic accumulations of dysferlin formed in regions distal to the injury site. These results suggest dysferlin may initially be endocytosed into acidic compartments in response to membrane injury. Is it endocytosed dysferlin that is cleaved by calpains, then vesicles laden with cleaved mini-dysferlin_C72 exocytosed at injury sites (Figure 7)?

5. Perspectives

It is clear that dysferlin plays a key role in membrane repair. However, exactly what dysferlin does to mediate membrane repair is only beginning to be teased out. From the perspective of the pathogenesis of dysferlinopathy, collective evidence suggests defective membrane repair may be only one contributing factor to disease pathology. Dysferlin deficiency also affects the trafficking and signaling of growth factor receptors (71) and adhesion molecules (256). Muscle injury may therefore present a “perfect storm” for dysferlinopathy patients, where a defective response to an acute membrane injury collides with a static defect in day-to-day trafficking of dysferlin-specific cargo, affecting muscle, vascular, and immune cells and producing a poor regenerative environment. Given the late presentation of dysferlin disease, one can only assume there are intrinsic differences in requirements for dysferlin-dependent trafficking and membrane repair pathways between growing muscle fibers of children versus fully mature adult myofibers.

Our recent discovery revealing activated calpains specifically cleave dysferlin in response to membrane injury provides the first step linking membrane repair roles separately established for dysferlin and calpains (discussed in detail in sect. IIIC). It remains to be experimentally determined whether mini-dysferlin_C72 possesses specialized vesicle fusion activity that is important for membrane repair, and whether dysferlin-laden cytoplasmic vesicles interact with classical SNARE fusion machinery. Also yet to be determined is the role of the cleaved dysferlin NH₂-terminal domain. Modal functions of different dysferlin C2 domains makes sense in light of the evolutionary preservation of each of the ferlin C2 domains, that are highly divergent from one another, but very similar to the analogous C2 domain in other ferlins. This tells us that each C2 domain is functionally specialized.

Endocytosis, calpain cleavage, then reexocytosis seems a labored path for a rapid emergency response. However, evidence that dysferlin may participate in both endocytic and exocytic pathways during the membrane repair response is consistent with the dual exocytic and endocytic recycling roles proposed for otoferlin in auditory neurotransmission (211, 242). Determination of the nature and the cargo contained within dysferlin-laden vesicles recruited to injury sites may provide valuable clues: are these purely a source of membrane lipids to patch a hole, or are soluble and transmembrane proteins codelivered that play an integral role in the repair and remodeling process?

C. Calpains

1. Evolution and structure

Calpains are an ancient family of thiol-proteases, present in protozoa, plantae, and eukaryota kingdoms (62, 84). The ubiquitous calpains, calpain-1 and calpain-2, typically exist as heterodimers, with a large catalytic subunit of ∼80 kDa and a smaller regulatory subunit of 28 kDa (for comprehensive review, see Ref. 102). The large catalytic subunit of calpain consists of four different domains (115, 270). Domain I is an α-helical domain comprised of 10–20 amino acids and is divergent among calpain orthologs and paralogs; with some isoforms bearing Zn-finger or transmembrane helices within domain I (263). Proteolytic removal of this NH₂-terminal helix plays a role in the activation of some calpains (137). Domain II forms the catalytic core, rich in essential catalytic residues cysteine, histidine, and asparagine. Domain III consists of an eight β-strand sandwich with structural resemblance to a C2 domain, and thought to confer calcium and phospholipid binding. Domain IV bears five tandem calcium-binding EF hand domains (33, 159). The presence of a calmodulin-like penta-EF-hand module characterizes the “typical calpains” such as ubiquitous calpain-1 and -2 and occurred late in calpain phylogeny, present only in metazoans (animal lineage). Protozoan, plant, and fungal calpains do not possess an EF-hand containing domain IV; moreover, several “atypical” mammalian calpains also lack this domain (62, 102).

a) calpains selectively modify target substrates. Calpains target selective substrates and use complex substrate recognition motifs dictated both by primary and secondary protein structure (162, 294). Calpains are not terminal degradative enzymes but, rather, selectively modify their target substrate. More than 100 proteins have been identified as calpain substrates; many are cytoskeletal proteins, such as vimentin, talin, desmin, troponin, dystrophin, and spectrin, but substrates also include receptors, ion channels, transcription factors, signaling proteins, and enzymes (102).

2. Expression and localization

There are 14 genes encoding large catalytic calpain subunits in humans. Calpain-1 and -2 are ubiquitously expressed and are also referred to as micro (μ)- and milli (m)-calpain, respectively, thus named according to their activating calcium concentration for proteolytic activity: 10–50 μM for μ-calpain and 0.25–0.35 mM for m-calpain. Many calpains show tissue-specific expression, for example, calpain-3a in skeletal muscle, calpain-6 in placenta, calpain-8 and -9 in the gastrointestinal tract, and calpain-11 in testis (see reviews in Refs. 102, 276).

The ubiquitously expressed calpain-1 and -2 are the best studied of the calpains. In cultured cells, calpains show cytosolic and membrane localization and are widely described to translocate to the plasma membrane in response to cellular signaling by calcium (see, for example, Refs. 99, 100) and with growth factor receptor activation (153, 255). The localization of calpains has been widely studied in skeletal muscle fibers, with most labeling detected within the myofibrillar apparatus (142). More recent studies in skinned skeletal muscle fibers show that calpain-1 can freely diffuse out of the myofibrillar compartment at resting calcium levels, but becomes tightly bound within skinned fibers with increased cytosolic calcium (20 μM) (195). Muscle cells present a complex environment when considering calpain activity; the calcium transients of muscle contraction would theoretically provide sufficient cytosolic calcium to activate proteolytic activity of μ-calpain but, fortunately for muscle fibers, does not. It is therefore proposed that calpain activity in skeletal muscle is tightly regulated by calpastatin (an abundantly expressed natural and specific inhibitor of calpain-1 and -2), their localization, and accessibility to their discreet repertoire of protein substrates.

3. Role in human disease

Calpains play a key role in development, with targeted knockout of calpain-2 resulting in embryonic lethality due to an implantation defect (11, 79). Although calpain-1 and -2 cleave many of the same substrates in vitro, knockout of calpain-1 produces viable mice that are morphologically normal but show defects in platelet function (13). This highlights different functional roles for calpain-1 and calpain-2, whereby calpain-1 is unable to compensate for calpain-2 deficiency during early embryonic development.

Disturbances in calpain behavior are implicated in numerous human pathologies (see review and articles within Ref. 317). The crux of the problem in most cases stems from calpain “overactivity” due to aberrant calcium handling and is often associated with pathologies of membrane injury, such as muscular dystrophy (262, 289), cardiac ischemia-reperfusion injury (124), traumatic brain injury (243) and stroke (17, 64), Alzheimer's disease (273), multiple sclerosis (259, 271), and cataract formation (31). Moreover, calpains are becoming increasingly implicated in cancer biology, with roles in pro-survival or apoptotic decision-making, and cytoskeletal remodeling and migration, highly relevant to tumorigenesis and metastasis (268).

A polymorphism within intron 3 of calpain-10 has been implicated as a susceptibility factor for type 2 diabetes (114). However, a direct role for calpains in human disease is best described by a monogenic form of autosomal recessive limb girdle muscular dystrophy (LGMD2A) due to mutations in the gene encoding calpain 3a (CAPN3) (233). The mechanism underpinning LGMD2A remains relatively poorly understood. Calpain-3, although abundant, is extremely labile and partitions to the myofibrillar compartment of skeletal muscle (194). The current paradigm for the pathogenesis of LGMD2A centers upon a role for calpain-3 as a “sarcomeric remodeler” (24, 283), although different studies have separately implicated roles in muscle maturation (264), myonuclear apoptosis (15), and maintenance of the costameric dystrophin-associated complex via targeted cleavage of filamen-C and δ- and γ-sarcoglycan (107).

4. Role in membrane repair

a) activation of calpain is vital for the acute membrane repair response. Interestingly, modulating calpain activity bestows both a blessing and a curse for cellular survival following membrane injury. Although treatment with calpain inhibitors improves physiological outcomes and the degree of cell and tissue death following pathologies of membrane injury such as ischemia-reperfusion injury in the heart, stroke, and traumatic brain injury (317), activation of calpain is vital for the survival of an acute membrane injury (183).

In 1991, Xie and Barrett (313) presented evidence showing that calcium-activated proteases facilitated membrane resealing of transected mammalian neurites; calpain inhibitors strongly inhibited resealing of severed axons, suggesting a requirement of calpain activity for neurite sealing. A role for calpains in neurite sealing was confirmed by Godell et al. (101), who demonstrated that application of exogenous calpain restored the resealing capacity of crayfish medial giant axons (MGAs) in calcium-free media, and could also induce sealing in transected squid giant axons (GAs) that otherwise do not seal (with or without calcium). Godell et al. (101) showed that both crayfish MGAs and squid GAs recruited vesicles to the transected end of the axon. In the presence of calcium, vesicles recruited to the end of transected crayfish MGAs fuse to form a dye-impermeable barrier, whereas vesicles in squid GAs do not. Treatment of transected MGA neurites with calpain inhibitors did not attenuate vesicle formation or recruitment, but prevented their fusion, in some cases inducing the formation of abnormally large vesicles that failed to fuse with the cut axonal end. The authors therefore proposed that calpains enhanced the fusion of recruited vesicles to reseal transected neurites and suggested squid GAs possess all of the machinery for effective resealing, although must lack sufficient active calpain to facilitate vesicle fusion.

Calpains have also been shown to be vital for membrane resealing of mammalian somatic cells. In the acute setting of membrane scrape-injury, treatment with calpain inhibitors markedly impairs cell survival of immortalized mouse embryonic fibroblasts (MEFs) and primary human skin fibroblasts and neonatal rat cardiomyocytes (182, 183). Using mouse embryonic fibroblasts with targeted knockout of the CAPNS1 regulatory subunit of calpain-1 and -2 (11), Mellgren et al. (183) were able to show that calpain-1 and/or calpain-2 were required for acute membrane repair (183). In contrast, calpain-3a did not contribute to survival after scrape damage of skeletal myoblasts or to acute repair of a laser damaged membrane in skeletal myotubes derived from CAPN3 knockout mice (182).

b) the puzzle: how do calpains underpin the calcium-dependence of membrane repair? Significantly, these studies suggested that calpains were primary mediators of the calcium dependence of membrane repair. CAPNS1-deficient MEFs did not show improved membrane repair outcomes in the presence of calcium, in contrast to wild-type MEFs, or virally rescued CAPNS1 knockout MEFs. The calcium dependence of membrane repair is a central dogma within the membrane repair field, historically attributed to a requirement for calcium-dependent exocytosis (27, 267) and vesicle-vesicle fusion to facilitate membrane “patching” of plasma membrane disruptions (181, 285). Thus how do calpains regulate calcium-dependent exocytic fusion underpinning the membrane repair paradigm?

A role for calpain-1 and -2 in membrane repair of adult cardiomyocytes was recently confirmed using a cardiac-specific knockout of CAPNS1 (281). Cardiac CAPNS1 null mice showed normal heart histology and function at baseline, although developed symptoms of congestive heart failure following a surgical protocol to constrict the aorta and induce pressure overload to the heart. Cardiac CAPNS1-null mice showed increased levels of fibrosis, poorer cardiac contractility, and greater permeability to Evan's Blue dye uptake following the hemodynamic stress protocol. Isolated cardiomyocytes subjected to the two-photon laser damage assay (18) showed a significant defect in plasma membrane repair, although whether CAPNS1-null cardiomyocytes still possessed calcium-dependent membrane repair was not examined (281).

Although calpain is known to cleave numerous substrates, the functional specialization that calpain cleavage imparts upon target substrates is largely unknown, with few exceptions. Recent results from our own research illuminate dysferlin as a specific substrate of injury-induced calpain activation, with a clear link to membrane repair. Since dysferlin is proposed to be the “calcium-sensing vesicle fusion protein” for exocytic membrane repair (18, 181), could calpain cleavage of dysferlin to release a synaptotagmin-like module, mini-dysferlin_C72, explain how calpains may regulate the exocytic fusion of membrane repair?

5. Perspectives

Calpain inhibition is being explored as a therapeutic strategy in numerous pathologies (see Ref. 44). The challenge with emerging calpain-inhibitor-based therapies is to specifically target the appropriate calpain and to define the therapeutic window that permits the acute resealing response, but protects against the uncontrolled calpain activity that follows, leading to a negative spiral of deleterious calpain-mediated degradation of cellular proteins that results in cytotoxicity.

The superphysiological activating calcium concentration for m-calpain has puzzled biochemists for the last two decades, and much effort has been devoted to deciphering molecular modifications that occur in vivo to lower its required calcium concentration for activation. Membrane injury seems tailor-made for m-calpain; full activating calcium concentrations, discrete substrate specificity, and a repertoire of protein substrates that one can easily fit into a model for membrane repair: 1) dysferlin and voltage-gated calcium channels for vesicle fusion; 2) matrix proteins and submembraneous cytoskeleton to allow delivery of exocytic vesicles for patch-repair as well as internalization of endocytic vesicles to remove oxidized lipids and damaged receptors; 3) dystrophin, talin, and spectrin to severe adhesion anchors and improve membrane flexibility to allow remodeling and resealing of large injuries; and 4) annexins to compartmentalize and order the resulting mish-mash of lipid membranes delivered in the emergency response to restore membrane integrity.

On balance, several lines of evidence are converging to triangulate calpains as central regulators of membrane repair. One begins to question whether dysferlin is indeed a master regulator of membrane repair, or one of several effectors activated by calpain during its boisterous response to the unregulated influx of calcium that uniquely characterizes the local environs of a membrane injury.

D. Annexins

1. Evolution and structure

The annexins are an ancient family of calcium-sensitive phospholipid binding proteins (98). There are 12 annexins in chordates (A1-A11, A13), with more than 500 members found across all major eukaryotic phyla (190). Like dysferlin (152), annexins are also present in primitive unicellular protists and are absent from prokaryotes and yeast (190).

There are three major classes of structurally diverse calcium-binding domains found in eukaryotic proteins: the β-sheet sandwich of the C2 domain, the helix-loop-helix module of EF hands, and the superhelical annexin calcium binding domain. Annexins bear a conserved domain structure of a variable NH₂-terminal domain and a COOH-terminal core domain of four, tandem, calcium-binding repeats (eight in annexin VI) (98). Each repeat is ∼70 amino acids in length, forming a superhelix of 5 inter-wound α-helices (158). The four superhelical domains of annexin pack into a compact disc, slightly convex on one side and slightly concave on the other side. The calcium-coordinating residues lie on the convex face of the disc and are highly evolutionarily conserved (190). Thus the annexin core functions as a four-domain module that binds negatively charged phospholipids via its calcium-bound convex surface, with its concave surface facing the cytoplasm.

The annexin NH₂-terminal domains vary in length and are thought to lie on the concave side of the core module where they confer specificity for different binding partners. In the case of annexin A1, the NH₂-terminal domain was shown to be buried within the core of the molecule and released in a conformational shift with calcium binding (241). Thus calcium coordination by annexin A1 simultaneously mediates phospholipid binding of the convex face, and, effector interaction by the NH₂ terminus on the cytoplasmic concave face. For annexins A1 and A2, the unique NH₂-terminal domains also convey specificity for binding the EF-hand accessory proteins S100A10 (annexin II) and S100A11 (annexin I) (156).

2. Expression and localization

Many annexins are ubiquitously expressed (A1, 2, 4, 5, 6, 7, 11), highly abundant proteins, although several annexins show tissue-specific expression; for example, A3 in neutrophils, A8 in placenta and skin, A9 in the tongue, and A10 and A13 in the gastrointestinal tract (98). The term annexin is derived from annexare (Latin) and annexer (French) meaning to join, to connect, to attach; the primordial role of all annexins is thought to relate to their capacity to aggregate and organize membranes in response to calcium.

Annexins generally show a cytoplasmic distribution, but partition to discrete membrane compartments upon calcium signaling. Studies of GFP-annexin fusion proteins have revealed that different annexins target different subcellular membrane compartments with calcium signaling, and indeed, respond to different activating calcium concentrations (78, 188). Annexins A1 and A2 shuffle between plasma membrane and endosomal compartments (170, 231), with emerging evidence for extracellular activities in anti-inflammatory and anticoagulation pathways (108, 214). Annexin 5 localizes to several different compartments of the biosynthetic pathway, including the endoplasmic reticulum, Golgi, late endosomes, and nuclear envelope (21, 73, 225). Annexin A5 is also widely used as a marker of apoptotic cells via its specificity for calcium-dependent binding to exposed phosphatidlyserine (139).

3. Role in membrane repair

Annexins are becoming widely implicated in different aspects of membrane compartmentalization, stabilization, and remodeling associated with membrane repair (76). Annexin A1 interacts with dysferlin (154) and was the first of the annexins to be implicated in membrane repair (176). Annexin A1 translocates to the plasma membrane with injury (14, 176), and an anti-annexin A1 blocking antibody, a small peptide inhibitor, and a calcium-binding annexin A1 mutant were each shown to inhibit resealing in HeLa cells subjected to mechanical scrape injury or laser irradiation (176). Levels of annexin A1 are upregulated in patients with muscular dystrophy (304), and annexin A1 labels vacuolized longitudinal membranes of the t-tubule network (a response to stretch injury) in patients with muscular dystrophy (304) and statin myopathy (302).

a) annexins incrementally respond to different degrees of calcium influx. Subsequent studies of SLO perforated HEK293 cells have revealed unique interplay between annexins 6 and 1 following membrane injury (14, 219). Annexin 6 shows higher sensitivity to elevations in intracellular calcium, and rapidly and reversibly translocates to the plasma membrane in cells suffering minor SLO perforation, inducing local hot spots of elevated calcium in the range of 5–10 μM. Annexin 6 specifically targets these injured regions of the plasma membrane and is encapsulated together with SLO perforated membranes into microvesicles that are then shed by the cell. With more severe SLO perforation, cytosolic [Ca]_i exceeds 10 μM and activates the less sensitive annexin A1, which translocates to the nuclear envelope and plasma membrane and is associated both with microvesicle shedding of perforated membranes as well as internalization of ceramide platforms (14). The careful work of this research group has revealed cooperative roles of annexins A1 and A6 that can incrementally respond to different degrees of membrane damage according to local elevations in [Ca]_i, and either package small lesions into shed microvesicles, or alternately internalize larger areas of membrane damage demarked by ceramide plaques (76).

b) annexin a5: a shield for injury sites? A recent study has shown that calcium-dependent binding of annexin A5 to exposed phosphatidylserine plays a biological role in the membrane repair response of damaged perivascular cells (37). Annexin A5 is known to self-assemble into trimers that interconnect to form two-dimensional structural arrays once bound to negatively charged phospholipids (206). This property plays an important role in its anti-thrombolytic and anticoagulant activity (237, 278, 286). Maternal anti-annexin A5 autoantibodies that disrupt this crystalline annexin A5 anticoagulant shield that covers the placenta is thought to be the basis for thrombosis and pregnancy loss in human anti-phospholipid syndrome (226).

Annexin A5 displays rapid recruitment to injury sites in damaged C2C12 myoblasts (40) and perivascular cells (37). Moreover, exogenous application of recombinant annexin A5 improves membrane repair in wild-type perivascular cells and rescues the membrane repair defect in annexin A5 null perivascular cells (37). Through design of an annexin A5 mutant unable to form two-dimensional arrays from assembled trimers, Bouter et al. (37) were able to demonstrate that formation of crystalline arrays is crucial for the role of annexin A5 in membrane repair.

c) annexin a6 participates in skeletal muscle membrane repair. Intravital imaging of annexins in the zebrafish model of sarcolemmal membrane repair also identified a role for annexin A6 (238). Annexin A6 showed immediate (10–20 s), dysferlin-independent translocation to injury sites, temporally preceding recruitment of annexin 11a (40 s), annexin 2a (60–80 s), and annexin 1a (200–240 s). This is consistent with the temporal progression of annexin recruitment to injury sites in SLO-treated HEK293 cells (219).

Morpholino knockdown of annexin A6 resulted in similar, though more mild, symptoms to dysferlin morphants, with a curved trunk, myofibrillar misalignment, and a reduction in birefringence (238). Double knockdown of both dysferlin and annexin A6 caused a severe phenotype, with all animals presenting with a curved trunk, severe cardiac edema, myofibrillar disorganization, and almost absent birefringence (238). Knockdown of annexin A6, dysferlin, or both prevented injury-activated accumulation of annexin A1, and slowed recruitment of annexin A2, with more profound effects observed with the double knockout. Collectively, results suggest that dysferlin and annexin A6 are independently recruited to injury sites, and work cooperatively to facilitate the subsequent accumulation of slower responding annexins A1 and A2. These data are supported by studies in human myotubes, where dysferlin rapidly recruits to injury sites and specifically labels the edges of the lesion, whereas annexins A1 and A2 do not show enrichment until 30 s postinjury, and do not specifically label injury sites but instead show more generalized enrichment in a large zone around the ballistics lesion (150).

Recently, annexin A6 has been identified as a potential disease modifier of muscular dystrophy, which is characterized by a high degree of pathological variability in human patients. A truncated version of A6 was found in a genome-wide scan of mice, and this abnormal A6 was shown to interfere with membrane repair by disrupting the recruitment of normal A6 to disruption sites (277).

4. Perspectives

The ancient phylogeny of the annexins, their high abundance, and their ability to stabilize and order phospholipid membranes makes them good candidates as primordial mediators of a membrane repair response. It makes sense that cells will activate and employ different coping mechanisms, depending on the location and extent of the membrane injury, whether that be rapid shedding of microvesicles or internalization of larger sections of the bilayer with more substantial damage. Ceramide accumulation at annexin A1-labeled injury sites, or crystalline annexin A5 arrays may function threefold: to provide a mechanism for the cell to recognize a damage site and target endocytic machinery, to provide a scaffold to aide in the recruitment and assembly of other repair and remodeling factors, and, more intriguingly, perhaps to provide a temporary shield to delay an immune or apoptotic response to allow time for repair.

Although calcium signaling may provide the ultimate “membrane injury” signal to activate a tailored response by different cytoplasmic annexins in many cells and tissues throughout the body, it is difficult to reconcile how such a mechanism works effectively in the muscle myoplasm. The cytoplasm of contractile muscle cells is routinely flooded with extraordinarily high levels of calcium during contraction; ∼1–3 μM cytoplasmic calcium with ∼100 μM locally at the triad junction (23, 148). These cytoplasmic calcium concentrations would rarely be encountered in other cell types, except in the local environs of specific calcium signaling or a small membrane rupture. Perhaps it is this specific property of muscle that confounds an ancestral membrane repair response, which perhaps requires further refinement by muscle-specific effectors such as MG53.

E. TRIM72/MG53

1. Evolution, structure, and localization

Mitsigumin-53 (MG53) was first identified as part of an immunoproteomic study to derive monoclonal antibodies against skeletal muscle triad proteins, mitsigumin being the Japanese word for triad. MG53 is highly expressed in skeletal and cardiac muscle and immunolocalizes to the t-tubule/sarcoplasmic reticulum triad junction in murine muscle (308), with more variable cytosolic, sarcolemmal, and t-tubule labeling in human skeletal muscle fibers (304). MG53 is a member of the TRIM family of E3 ubiquitin ligases and is also known as TRIM72.

There are more than 70 TRIM family proteins in mammals, thus named due to their characteristic NH₂-terminal tripartite motif, a RING finger domain conferring ubiquitination activity, one or two zinc-binding B-box domains; TRIM-family specific sequences thought to mediate protein-protein interactions, and a coiled-coiled domain that mediates hetero- and homo-oligomerization (130, 232). TRIM proteins play diverse roles in development, differentiation, oncogenesis, and immunity (175, 197, 208). Phylogenetic studies of TRIM-family specific B-box sequences reveal the presence of this domain in protozoa and unicellular eukaryotes, but the tripartite motif that characterizes the TRIM family E3 ligases is exclusive to multicellular metazoans (249).

The NH₂-terminal tripartite motif is generally thought to collectively mediate E3 ligase activity of TRIM proteins, whereas the COOH-terminal motif is though to confer substrate specificity and varies throughout the TRIM protein family. MG53 (TRIM72) is a group 2 TRIM protein bearing a COOH-terminal PRY-SPRY domain (249). This class of TRIM proteins are evolutionary newcomers, present only in chordates and showing rapid gene expansion in humans. The PRY-SPRY domain of MG53 has been crystallized, revealing a compact globular structure characterized by a convex surface on one side and a concave surface on the other that includes a large binding pocket (212).

2. Role in human disease

a) mg53. MG53 has not been implicated in human muscle disease but shows upregulation in patients with muscular dystrophy (304). Recently, MG53 has been implicated in the metabolic syndrome that results in type II diabetes and insulin resistance (261). Levels of MG53 were elevated in animal models with insulin resistance and metabolic syndrome, as well as in obese humans. Upregulation of MG53 in skeletal muscle via transgenic overexpression caused whole body insulin resistance and metabolic syndrome, whereas mice deficient in MG53 were protected from the metabolic effects of high fat feeding. Biochemical studies showed that MG53 overexpression induced ubiquitination and degradation of the insulin receptor and the insulin receptor substrate IRS1, therefore functioning as a muscle-specific negative regulator of insulin signaling (261).

b) other trim family proteins. Mutations in several other TRIM-family ubiquitin ligases cause different forms of inherited disorders in humans. MID1 (midline-1) causes Opitz Syndrome (223), and TRIM37 causes Mulibrey (muscle-liver-brain-eye) nanism (12); both are developmental disorders affecting several organs and tissues. Interestingly, another TRIM protein is implicated in the pathogenesis of muscular dystrophy in humans. Mutations in TRIM32 cause an autosomal recessive muscular dystrophy (LGMD2H) (92), although the disease mechanism is unknown. TRIM32 belongs to a more ancient subgroup of TRIM proteins present in invertebrates, but shows only weak amino acid identity to MG53 and does not contain the COOH-terminal PRY-SPRY domain. It remains to be determined whether TRIM32 mutations impact skeletal muscle membrane repair in affected patients.

3. Role in membrane repair

In 2009, a MG53 knockout mouse was derived and found to show exercise intolerance with downhill running and increased muscle damage detected by Evans Blue dye uptake (40). Closer examination revealed histopathological signs of a mild, progressive muscular dystrophy. Membrane damage assays using laser irradiation and microinjection revealed that MG53 was rapidly recruited to sites of membrane injury in a calcium-independent manner and that MG53^−/− null myofibers and myotubes possessed a primary defect in membrane resealing. MG53 was shown to oligomerize via cysteine 242, with mutation of C242 to alanine ablating MG53 injury recruitment and membrane repair activity (40). MG53 was thus proposed to be an oxidative sensor of membrane injury, forming an oligomerized scaffold for assembly of the membrane repair complex (for review, see Ref. 177).

MG53 was shown to coimmunoprecipitate dysferlin and the muscle-specific caveolin-3 (42). Caveolins are the major protein constituents of membrane caveolae, and mutations in caveolin-3 cause a form of autosomal dominant muscular dystrophy (LGMD1C) (186). A Golgi-retained muscular dystrophy mutant of caveolin-3 (P104L) caused coaccumulation of dysferlin and MG53 in the Golgi apparatus of skeletal myofibers, resulting in defective membrane repair (42). Overexpression of MG53 in C2C12 mouse myoblasts was shown to activate membrane exocytosis, inducing masses of filopodic extensions (41), a process reversible by coexpression of caveolin-3. Collectively, these findings implicated functional interplay between MG53, caveolin-3, and dysferlin in membrane trafficking events required for membrane repair. MG53 recruitment and anchorage at sites of membrane injury have since been to shown to require the cytoskeletal motor nonmuscle myosin IIA (160) and caveolae regulatory protein PTRF (polymerase I and transcript release factor, also known as cavin-1) (319).

Application of recombinant MG53 can improve resealing outcomes when applied to the extracellular media bathing dysferlin-deficient myofibers and can improve dystrophic pathologies when systemically delivered to the mdx mouse, a model of Duchenne muscular dystrophy (309). MG53 has also been implicated in the response to membrane injury in the heart (43, 306).

4. Perspectives

MG53 is expressed only in chordates, with no ortholog identified in chicken or zebrafish (249). As membrane repair is a feature of both vertebrate and invertebrate cells, one must consider that MG53 is not an ancestral membrane repair component, although it may have assumed the function of an evolutionarily more ancient predecessor or perhaps lent a tissue-specific specialization to the membrane repair response. In the setting of a membrane injury with unregulated calcium influx, many plasma membrane proteins and receptors will be damaged by oxidation and calpain cleavage. Therefore, it seems likely that the trigger for MG53 recruitment relates to its role as a ubiquitin ligase, utilizing existing trafficking machinery that ubiquitin ligases transit to rapidly label target substrate(s) for endocytosis and degradation.

The recent discovery of the insulin receptor as a specific substrate of MG53 ubiquitin ligase activity (261) is not easily reconciled with the role of MG53 in membrane repair (40). It is possible that MG53 has several ubiquitination targets and that one of these additional targets plays an important role in the response to, and the signaling of, a membrane injury. Given the ancient eukaryotic origins of other proteins implicated in membrane repair (SNAREs, ferlins, annexins, and calpains), one might suppose that the ubiquitination target of MG53 belongs to an ancient receptor or channel family, which has perhaps evolved a chordate-specific or tissue-specific isoform that MG53 has recently evolved to specifically target.

F. Late Endosomal Pathway And Membrane Repair

Recent research is expanding the body of evidence implicating machinery of the late endosomal pathway in membrane repair, including components of the endosomal sorting complex ESCRT-III (129, 250) and an endolysosomal transient receptor potential (TRP) cation channel, mucolipin-1 (TRPML-1) (49).

1. ESCRT

a) evolution, structure, and localization. ESCRT (endosomal sorting complexes required for transport) plays a key role in the endocytic trafficking and lysosomal targeting of ubiquitinylated transmembrane receptors and proteins (for recent reviews, see Refs. 110, 224). ESCRT proteins were first identified in yeast, as class E vacuolar protein sorting (Vps) genes, required for sorting of transmembrane proteins into intraluminal endosomal vesicles (228, 298). The ESCRT complex is comprised of five subcomplexes: ESCRT-0, ESCRT-I, ESCRT-II, ESCRT-III, and the VPS4 ATPase, and it works in concert with a wide range of adaptor proteins. Components of ESCRT complexes-I, -II, -III, and the VPS4 ATPase are conserved throughout eukaryotic evolution, present in protozoan parasites through to metazoa (155). Furthermore, there is evidence for ancestral ESCRT-III proteins and the VPS4 ATPase in archae-bacteria (89, 203). However, ESCRT-0 is found only in Opisthokonts (animal and fungi kingdoms) and is not present in primitive eukaryotes (155).

The ESCRT complexes are best known for their role in endosomal sorting, whereby ubiquitinylated receptors are endocytosed into clathrin-coated endosomes, clustered within the endosomal membrane, then partitioned into membrane compartments that invaginate and bud-off into the endosomal lumen to form multivesicular bodies (MVBs) (204). MVBs then fuse with lysosomes, resulting in the eventual degradation of the ubiquitinylated cargo (117). ESCRT-0 bears a ubiquitin-interacting domain and a FYVE domain that specifically recognizes phosphatidylinositol 3-phosphate lipids highly enriched within the endosomal membrane (95, 144). ESCRT-0 is thought to recognize the ubiquitinylated cargo within the endosomal membrane and facilitate recruitment of ESCRT-I, which in turn mediates interaction with ESCRT-II. The serial assembly of ESCRT-0, -I and -II complexes provides the platform for assembly of ESCRT-III, the core workhorse of the ESCRT machinery that forms filamentous circular arrays to drive deformation of the endosomal membrane and formation of intraluminal endosomal vesicles (109, 224). The VPS4 ATPase disassembles the ESCRT-III machinery and facilitates recycling of its protein constituents (244).

In addition to a role in endosomal sorting, components of ESCRT play key roles in viral budding (269, 303) and the abscission step of cytokinesis (189, 246, 265) (for recent reviews, see Refs. 171, 174, 245).

b) role in membrane repair. Recently, membrane fission roles of the ESCRT complex have been implicated in the closure of small wounds to the plasma membrane (129) (Figure 4). Using several modes of membrane injury (microinjection, laser ablation injury, pore formation), Jimenez et al. (129) show recruitment to injury sites of several members of the ESCRT-III complex (CHMP4, 3, 2A, and 2B), the Vsp4 ATPase, and the adaptor protein ALIX known to facilitate assembly of ESCRT-III in the absence of ESCRT-I and II complexes. Interestingly, recruitment of ESCRT proteins was calcium-dependent, but not accompanied by vesicular recruitment of late endosomes/lysosomes, and was not blocked by microtubule deploymerization. These data suggest ESCRT proteins are not delivered via vesicular transport but are recruited as soluble, cytoplasmic proteins.

Using quantitative analyses of live imaging data combined with scanning electron microscopy, the authors revealed the ESCRT proteins play a role in repair of small lesions <100 nM, including small lesions made by bacterial pore-forming toxins. Repair and removal of the bacterial pores or small microinjuries are related to outward budding of the decorated wounds and pinching-off via exocytic shedding (129) (see Figure 4). The capacity to survive the membrane injury and shed ESCRT-positive exosomes were both ATP-dependent, suggesting the Vsp4 ATPase may be required to disassemble the ESCRT-III complexes so the protein constituents can be recycled, and/or play a direct role in the pinching off the exocytic buds (129).

A subsequent study used a proteomics approach to identify proteins exocytosed to the plasma membrane after an acute elevation in submembraneous Ca²⁺ (using the Ca²⁺ ionophore ionomycin), and highlighted several components of the ESCRT-III machinery (250). Importantly, Scheffer et al. (250) established that the ALIX-binding protein ALG-2 (apoptosis-linked gene-2), a penta-EF hand protein, provides the Ca²⁺-dependent component for sequential recruitment and assembly of the ESCRT-III complex (250). Using siRNA knockdown of ALG-2, ALIX, and Vsp4 ATPase, they demonstrated each of these components was required for repair, closure, and exosomal shedding of plasma membrane injuries. By studying the capacity of each knockdown cell line to recruit the different components of the ESCRT-III machinery, a model emerged whereby ALG-2 responds to the Ca²⁺ influx and facilitates recruitment of its binding partner ALIX. In turn, ALIX mediates the sequential assembly of CHMP proteins, then the Vsp4 ATPase (250). The ESCRT-III machinery deforms the plasma membrane into outwardly forming buds that are pinched off to close and/or remodel the wound, with abscission steps involving the ESCRT Vsp4-ATPase (129, 250).

c) perspectives. Interestingly, although data presented by Jimenez et al. (129) suggest ESCRT-III machinery does not influence the closure of wounds larger than 100 nM, Scheffer et al. (250) found cells with >90% knockdown of ALG-2, ALIX, and Vsp4 ATPase were more vulnerable to larger laser injuries (∼1 μM) than control cells. Thus whether ESCRT-III plays a greater role in repair of smaller versus larger injuries requires further clarification. Importantly, Scheffer et al. (250) showed ESCRT-III plays a role in membrane repair in C2C12 myoblasts and primary myofibers, and thus is a ubiquitous response, and not a peculiarity of the immortal HeLa cell line. Notably, both studies carefully demonstrated that recruitment of ESCRT-III machinery was not accompanied by lysosomal exocytosis at the site of injury, although lysosomal exocytosis was observed to occur distally in response to the membrane injury (129, 250).

Although one begins to sniff a tantalizing potential link between the E3 ubiquitin-ligase MG53 (TRIM72) and ESCRT machinery in membrane repair, immunolabeling for polyubiquitination of plasma membrane proteins following injury suggested assembly of ESCRT-III preceded polyubiquitination events (129). Again, Ca²⁺ provides a key trigger for ESCRT-III exosomal shedding for membrane repair, via ALG-2. On this basis, it will be important to determine whether other settings of ESCRT-mediated fission are also Ca²⁺-dependent and, furthermore, whether the concentration of calcium required to activate recruitment of ESCRT-III machinery to injury sites is consistent with concentrations required to activate calcium-dependent membrane repair in mammalian cells [∼150 μM in human muscle cells (230) and ∼300 μM in 3T3 fibroblasts (267)].

Yet to be determined and vital for this field are the connections and interplay between the rapid Ca²⁺-dependent exocytosis triggered in the immediate surrounds of a membrane injury (<10 s; Ref. 150), the rapid compensatory endocytosis (57, 123, 173), the slower phase of lysosomal exocytosis at regions distal to the injury site (126, 229), and the vesicle-independent accumulation of ESCRT-III machinery for exosomal shedding of the injured membrane (49, 129).

2. Mucolipin-1

a) evolution, structure, and localization. Mucolipin-1 (also known as TRPML-1) is one of three members of the vertebrate mucolipin-subfamily of the TRP (transient receptor potential) cation channels, a superfamily of nonselective cation channels bearing six transmembrane domains (S1–S6) (299). Functional TRP channels are homo- or hetero-oligomers of four subunits, with a pore domain located between transmembrane domains S5 and S6 consisting of negatively charged glutamate and aspartate residues to confer cation selectivity (157). TRP channels are best known for their roles as sensory channels for photoreception, proprioception, mechanosensation, nociception, and thermosensation (299). Members of the TRP superfamily are present in yeast through to metazoa, and although TRP-like genes are identified in unicellular phytoplankton, TRP-like genes are absent in plants (201). The mucolipin subfamily of TRP channels emerged as a single gene in invertebrates, with gene duplication events producing three mucolipin genes in vertebrates (201). The mucolipin-1 channel localizes to Rab-7-positive late endosomal and lysosomal compartments and is permeable to Ca²⁺ and Fe²⁺ (75).

b) role in disease. Mutations in the mucolipin-1 gene (MCOLN-1) cause an autosomal recessive lysosomal storage disease, mucolipidosis type IV (20). Mucolipidosis type IV is a devastating neurodegenerative disorder characterized by progressive intellectual disability and motor dysfunction, muscle weakness, retinal degeneration and clouding of the cornea, and anemia due to deficient acid secretion in the gut (acid is required for iron absorption), likely confounded at the cellular level by defects in iron-conducting properties of the mucolipin-1 channel itself. Abnormally swollen and enlarged lysosomal compartments filled with membrane and protein aggregates are a feature of patients with mucolipidosis type IV, suggesting abnormal maturation and fusion of endolysosomal organelles (48, 147).

c) role in membrane repair. Somewhat unexpectedly, the major presentation of a mouse knockout model of mucolipin-1 is an early-onset progressive muscular dystrophy, characterized by abundant internal nuclei, fibrotic and inflammatory infiltrate, and elevated serum creatine kinase (49). Dystrophic pathology is apparent as early as 1 mo of age, preceding overt histopathological symptoms of lysosomal storage disease or neuronal degeneration (apparent >5 mo of age) (49). Skeletal muscle fibers isolated from mucolipin-1^−/− mice show a major defect in Ca²⁺-dependent membrane repair, characterized by defective lysosomal exocytosis. Similarly, when mucolipin-1 channels are blocked in C2C12 myoblasts or MEFs using synthetic inhibitors, membrane repair is inhibited and lysosomal exocytosis is reduced. Thus lysosomal Ca²⁺ flux via mucolipin-1 is an important event for both membrane repair and for lysosomal exocytosis. The importance of lysosomal Ca²⁺ stores was supported through treatment with glycyl-l-phenylalanine 2-naphthylamide (a substrate hydrolysed by cathepsin C leading to lysosomal osmotic injury and depletion of Ca²⁺), and BAPTA-AM (a cell-permeable, rapid, and highly effective Ca²⁺ chelator). Both treatments inhibited membrane repair, suggesting intracellular Ca²⁺, and lysosomal Ca²⁺ in particular, is important for membrane repair. Lastly, electrophysiological analyses of wild-type cells reveals an increase in whole cell mucolipin-1 currents following membrane perforation with SLO, implying mucolipin-1 channels become incorporated into the plasma membrane following lysosomal exocytosis in injured cells.

c) perspectives. This study provides additional evidence that lysosomal exocytosis plays an important role in membrane repair. Mucolipin-1 mediates lysosomal Ca²⁺ release in response to injury, and this Ca²⁺ release is required for normal lysosomal exocytosis and membrane repair. In the context of membrane repair, it is not yet clear whether lysosomal Ca²⁺ release mediated by mucolipin-1 is required for hetero- or homotypic fusion of endolysosomal compartments (215, 222) that precede lysosomal exocytosis, or whether it provides a specific Ca²⁺ flux important for docking and fusion of lysosomes with the plasma membrane. Notably, mucolipin-1 is a binding partner of ALG-2 (300), potentially providing a key link to the ESCRT-III membrane repair pathway.

In addition to the endoplasmic reticulum, mitochondria and lysosomes are major reservoirs of Ca²⁺, with intraluminal lysosomal Ca²⁺ around 0.4–0.6 μM (52). One provocative possibility is whether the dystrophic pathology of mucolipin-1^−/− mice may relate to defects in Ca²⁺ sequestration, or more intriguingly, Ca²⁺ extrusion. Persistent elevation of intracellular Ca²⁺ in skeletal muscle, induced by transgenic overexpression of the stretch-activated TRPC3 channel, induces a severe muscular dystrophy phenotype (185). This study shows that development of a muscular dystrophy phenotype is strongly related to unregulated Ca²⁺ influx. If, like mitochondria, lysosomes play a role in Ca²⁺ sequestration following membrane injury, then levels of lysosomal Ca²⁺ will increase in the seconds and minutes following membrane injury. Despite the vast Ca²⁺-handling capabilities of the sarcoplasmic reticulum, it is conceivable that lysosomal exocytosis triggered by membrane injury may also play some role in Ca²⁺ extrusion, to help purge the cell of unwanted Ca²⁺, protect mitochondria from Ca²⁺ overload, and protect a cell from deleterious calpain overactivity.

G. Cytoskeletal Networks

1. Structure and evolution

Eukaryotic cells are differentiated from their prokaryotic and archaeal ancestors not only by their endomembrane system, but also by their complex cytoskeleton. Prokaryotic cells do not possess direct homologs of the major cytoskeletal proteins or tubulin or actin, although they express divergent filamentous proteins with remarkable structural similarity to eukaryotic tubulin (bacterial FtsZ; Refs. 69, 227) and actin (bacterial MreB; Ref. 36). A dynamic cytoskeleton underpins eukaryotic cell division, cell shape, motility, as well as the capacity to phagocytose nutrients for growth and secrete signals for cellular communication.

Isoforms of tubulin and actin are present in every phyla of the eukaryotic kingdom, together with their molecular motors kinesin/dynein and myosin (for reviews, see Refs. 82, 311). Tubulin microfilaments possess an inherent instability intrinsic to their dynamic function. Mature microtubules typically consist of 13 protofilaments that interact to form a hollow cylinder. Each protofilament results from the polymerization of GTP-bound α- and β-tubulin heterodimers; free GTP at the growing (plus) end of the filament promotes further polymerization, whereas hydrolysis of GTP to GDP-tubulin exerts a structural and energetic change that renders the microtubule prone to deploymerization (83). Dynamic microtubules form mitotic spindles for cell division and cytoskeletal highways for vesicular transport via their molecular motors dynein and kinesin.

Actin is one of the most highly conserved eukaryotic proteins, showing ∼85% identity between human, yeast, and plant actin isoforms. Actin filaments consist of two strands of polymerized actin, interwound into a tight right-handed helix (74). Actin filaments are not as intrinsically unstable as microtubules, but nevertheless are capable of rapidly polymerizing and depolymerizing through an active treadmilling process where GTP-bound actin monomers are added to the plus end of the growing filament and removed from the minus end of the filament, regulated by conserved families of monomer-binding factors, capping proteins, filament-stabilizing proteins, and actin-severing proteins. Actin-myosin contractile cycling is best characterized in skeletal and cardiac muscle, with nonmuscle isoforms of actin and myosin playing vital roles in cytokinesis, cell motility, and endomembrane transport (34).

Myosins are also an evolutionary ancient gene family, characterized by the myosin head domain that encompasses a powerful ATPase that utilizes ATP hydrolysis to power movement along an actin filament. Myosin head proteins are present in virtually all eukaryotes, as are microtubule-dependent motors, the kinesins. Indeed, structural similarity between that catalytic core of myosin and kinesin suggests a common evolutionary ancestor of both ATPase motor domains (141).

2. Role in human disease

Mutations in tubulin, actin, and myosin underpin a broad spectrum of inherited disorders in humans. Tubulin mutations have recently been implicated in inherited neurodegenerative disorders, consistent with the vital role of the microtubule cytoskeleton for neuronal function (134, 290). Mutations in members of the actin and myosin gene families cause a wide range of clinical defects, typically affecting tissues and organs where they are dominantly expressed (63). For example, mutations in skeletal actin (ACTA1) and myosin (MYH7 and MYH2) cause a skeletal myopathy, mutations in cardiac actin (ACTC1) and myosin (MYH7) cause cardiomyopathy, and mutations in smooth muscle actin (ACTA2) and myosin (MYH11) cause thoracic aortic aneurysms and dissections. Mutations within nonmuscle members of the actin and myosin gene families can induce deafness, blindness, and other neurological and immune cell dysfunction, depending on the isoform involved. Currently, no pathology caused by mutations in cytoskeletal proteins has been linked directly to defects in membrane repair.

3. Role in membrane repair

A role for cytoskeletal motors in membrane repair was proposed by Xie and Barrett (313) in the early 1990s, who revealed that microtubule destabilizing agents (colchicine) promoted resealing of transected neurons, whereas microtubule stabilizing agents (taxol) inhibited resealing. Discreet roles for microtubule/kinesin and actin/myosin motors for membrane repair was elaborated upon in greater detail by Bi et al. (28), who presented evidence for three phases of vesicle fusion associated with membrane repair of embryonic sea urchin cells: immediate, fast, and slow phases. Immediate exocytosis (<5 s) that occurred in response to membrane wounding was insensitive to both kinesin and myosin inhibitors, and presumed to involve predocked vesicles that did not require kinesin or myosin motors for delivery or fusion. The next “fast” phase (5–15 s) of vesicle fusion was sensitive to myosin inhibitors, but not kinesin inhibitors, suggesting a local pool of vesicles that can quickly respond to the calcium flux of membrane injury. The third “slow” phase (20–60 s post injury) of vesicle fusion was sensitive to kinesin inhibition, and also to myosin inhibition, suggestive of longer distance recruitment of vesicles trafficked along microtubules toward the site of membrane injury.

Inhibition of kinesin motor activity was similarly shown to impair membrane resealing of mammalian Swiss 3T3 fibroblasts (267). Treatment with the actin depolymerizing agent DNAse I improved resealing of gastric epithelial cells, whereas actin filament stabilizing agents phalloidin and jasplakinolide strongly inhibited resealing (187). There is some discrepancy regarding the effects of cytochalasin D, an actin severing agent, shown to exert inhibitory (urchin embryos, Ref. 28), facilitative (3T3 fibroblasts, Refs. 123, 292), or no effect (gastric epithelia, Ref. 187) on membrane resealing from a single wound. However, collectively, evidence using different actin stabilizing and destabilizing agents in different models suggests local deploymerization of subcortical actin facilitates the vesicle fusion of membrane repair.

Further research later refined a specific role for nonmuscle myosin IIB in wound-activated exocytosis and membrane repair of injured 3T3 fibroblasts (293). Knockdown of nonmuscle myosin IIA did not affect exocytosis or membrane repair, but rather inhibited facilitated resealing, a more rapid resealing response to a second injury at the same site due to vesicle formation and recruitment from endomembrane organelles such as the trans-Golgi network (TGN).

a) an acto-myosin ring for membrane repair. A series of elegant imaging experiments in Xenopus oocytes revealed evidence for complex interplay between acto-myosin and microtubule cytoskeletal motors for wound repair (25, 168, 169). Xenopus oocyte injury sites were shown to be encircled by a ring of nonmuscle myosin II positioned at the inside edge of a wider band of actin. This acto-myosin ring was surrounded by a radial array of microtubules that were integrally associated with actin filaments and indeed appeared to be pulled into the active zone of actin polymerization by the acto-myosin motors themselves (168). In turn, the microtubule networks controlled the breadth of the zone of actin assembly, facilitating formation of the actin-myosin II contractile ring that gradually contracts to seal the wound. Inhibition of either acto-myosin or microtubule motor systems impaired wound healing and cell survival. Importantly, wounds made in calcium-free medium failed to reseal, with injected fluorophores leaking out and failing to successfully label cytoskeletal compartments. This suggests that formation of the acto-myosin contractile ring occurs after formation of a membrane barrier that is dependent on extracellular calcium.

These studies were mirrored in wounded early Drosophila embryos that do not undergo cytokinesis for the first few nuclear divisions and exist as large syncytial cells. Both actin and myosin showed rapid accumulation at injury sites (from ∼30 s after injury), aligning in concentric circles that gradually contract to reseal the injury (2). Resealing was shown to occur via discrete stages of expansion, contraction and closure. Immediately after wounding, the lesions expand until actin is observed to accumulate at the periphery. Myosin II accumulated at the wound edge forms a contractile actomyosin ring mediating wound closure. Interestingly, this study revealed evidence for periodic connections between the actomyosin ring and the circumference of the wound sites through interactions mediated by DE-cadherin, suggesting that final stages of wound closure and remodeling following injury occurs via tethering of the contractile machinery to stable adhesive connections with the plasma membrane.

4. Perspectives

The importance of both the actin cytoskeleton (172) and microtubule/kinesin motors (173) has been confirmed for membrane resealing of murine skeletal myofibers, and it is very likely that acto-myosin contractile rings and microtubule transport networks underpin plasma membrane repair and remodeling in all of eukaryota. Whereas endocytosis and exocytic shedding provide an effective means for removal and repair of small injuries (14, 129, 164), larger injuries present a more complicated problem for resealing. As one reflects upon the central themes of membrane repair–calcium, vesicle fusion, calpains, cytoskeletal remodeling, contractile rings–there is remarkable complementarity and consistency among major findings by different research groups, despite huge variance in experimental models and approaches. Acknowledging the oversimplification of what is a complex and incompletely understood pathway, as well as the caveats of comparison between different model systems, the timelines separately proposed by individual research groups fit surprisingly well into a collective model for membrane repair and makes for interesting contemplation.

IV. LITERATURE OVERVIEW: A COLLECTIVE MODEL FOR REPAIR OF LARGE MEMBRANE INJURIES

A. 0–10 s

Calcium floods down a steep concentration to create a microenvironment of high local intracellular calcium, activating calcium-activated signaling molecules and calpains, which cleave dysferlin and other cytoskeletal and plasma membrane substrates. In skeletal muscle cells, MG53 responds to an unknown signal and is rapidly recruited to injury sites. The immediate and fast phases of vesicle fusion occur, utilizing predocked and local vesicle pools, including nearby lysosomes. Dysferlin is endocytosed from regions distal to the injury site, and dysferlin is delivered to the plasma membrane in cytoplasmic vesicles, perhaps triggering their fusion, and upon integration into the plasma membrane intensely labels the circumference of injury sites. There is reduced labeling for filamentous cortical actin surrounding the membrane injury, due to proteolytic cleavage of membrane tethers by calpain and actin deploymerization. Lesions expand.

B. 10–30 s

The slow phase of vesicle fusion begins, requiring cytoskeletal transport of vesicles upon microtubule and actin-myosin motors. ALG-2 and annexins A6 and A11 are delivered to injury sites. Calcium influx continues to trigger exocytic fusion of lysosomes in zones surrounding the lesion site. Calpains cleave cytoskeletal networks and adhesion proteins. Lesions continue to expand.

C. 30–60 s

Actin accumulates at the wound periphery, forming a dynamic zone of rapidly polymerizing and depolymerizing actin. Actin filaments interconnect with local polymerizing microtubule filaments, pulling them toward the wound site and anchoring these microtubule highways in position for rapid transport of vesicles, signaling proteins and cytoskeletal remodeling proteins. Microtubules in turn stabilize the active zone of actin polymerization to facilitate formation of an acto-myosin contractile ring. ALIX, CHMP, and Vsp4 ATPase ESCRT-III machinery sequentially accumulate at injury sites. Acid sphingomyelinase released by lysosomal exocytosis hydrolyzes sphingomyelin to form ceramide-rich platforms to activate endocytosis. Damaged plasma membrane is removed by endocytosis and exosomal shedding. Lesions contract.

D. 60–240 s

The acto-myosin ring gradually contracts and closes, reestablishing the network of cytoskeletal connections beneath the wounded plasma membrane. Annexins A1 and A2 accumulate at injury sites. Phases of exocytosis, endocytosis and exosomal shedding remodel the hastily repaired plasma membrane barrier, replacing damaged proteins and receptors, and reconstituting the normal repertoire of plasma membrane lipids and microdomains.

V. CONCLUDING REMARKS

It seems no accident that membrane repair uses ancient and robust molecular machinery, such as contractile rings and vesicle fusion; these are fundamental biological processes vital for early eukaryotic life. Along these lines, one suspects the vesicle fusion of membrane repair may be more rudimentary than that of synaptic neurotransmission, evolving 500 million years later. Importantly for the membrane repair field, the biology of vesicle fusion and cytokinesis are well defined, each with established pharmacological modifiers. The key now is to connect each research field implicated in membrane repair and better elucidate the molecular signaling governing the temporal progression of each stage in the membrane repair process. It is only then we will be able to strategize novel interventions so desperately needed to treat the huge spectrum of human disorders characterized by membrane injury.

GRANTS

S. T. Cooper is supported by an Australian National Health and Medical Research Council (NHMRC) Career Development Fellowship APP1048816, NHMRC Project Grant APP1048814, and funding from the Jain Foundation. P. L. McNeil was supported by National Institutes of Health Grants AR-060565 and DK-090703.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.