Laminin-deficient Muscular Dystrophy: Molecular Pathogenesis and Structural Repair Strategies

Abstract

Laminins are large heterotrimers composed of the α, β and γ subunits with distinct tissue-specific and developmentally regulated expression patterns. The laminin-α2 subunit, encoded by the LAMA2 gene, is mainly expressed in skeletal muscle, Schwann cells of the peripheral nerve and astrocytes and pericytes of the capillaries in the brain. Mutations in LAMA2 cause the most common type of congenital muscular dystrophies, called LAMA2 MD or MDC1A. The disorder manifests mostly as a muscular dystrophy but slowing of nerve conduction contributes to the disease. There are severe, non-ambulatory or milder, ambulatory variants, the latter resulting from reduced laminin-α2 expression and/or deficient laminin-α2 function. Lm-211 (α2β1γ1) is responsible for initiating basement membrane assembly. This is primarily accomplished by anchorage of Lm-211 to dystroglycan and α7β1 integrin receptors, polymerization, and binding to nidogen and other structural components. In LAMA2 MD, Lm-411 replaces Lm-211; however, Lm-411 lacks the ability to polymerize and bind to receptors. This results in a weakened basement membrane leading to the disease. The possibility of introducing structural repair proteins that correct the underlying abnormality is an attractive therapeutic goal. Recent studies in mouse models for LAMA2 MD reveal that introduction of laminin-binding linker proteins that restore lost functional activities can substantially ameliorate the disease. This review discusses the underlying mechanism of this repair and compares this approach to other developing therapies employing pharmacological treatments.

Introduction

Basement membranes (BMs) are cell-adherent extracellular matrices (ECMs) that are essential for the structural support, maintenance and differentiation of animal tissues [1–3]. Their assembly, structure and functions depend upon laminin heterotrimeric glycoproteins consisting of α-, β- and γ subunits. In mammals there are a total of five different α subunits (α1 – α5, encoded by LAMA1 to LAMA5), four different β chains (β1 – β4, encoded by LAMB1 – LAMB4) and three γ chains (γ1 – γ3, encoded by LAMC1 – LAMC3). During protein synthesis and secretion α, β and γ chains assemble via specific interactions of their coiled-regions. So far, 15 different laminins derived from the many possible combinations have been described, which differ in function and expression pattern [4, 5]. The essential role of the laminins for the tissue integrity and function is best supported by mouse studies showing that deletion of the gene encoding a particular chain results either in embryonic lethality or severe disorders after birth. For example, mouse embryos deficient for Lamc1 (gene coding for the γ1 chain), which is the γ chain of ten different laminins, die at pre-implantation stage [6].

The fundamental role of laminins is based on their creating of a primary scaffold that (1) attaches the extracellular matrix to the cell surface and (2) serves as a platform to which other extracellular matrix components, such as the nidogens and collagens, become stably attached. The main laminin receptors expressed on cell surfaces are α-dystroglycan, specific integrins and sulfated glycolipids (reviewed in [1, 2]). Binding of laminins to their receptors affects assembly of the BM, mechanical linkage of the cell to BM and signaling. We would like to emphasize that the receptor binding of the laminins in the context of laminin-deficient muscular dystrophy is particularly important for the assembly of the sarcolemmal and endoneurial BM and that the different receptors can, at least partially, compensate for each other.

LAMA2 mutations cause congenital muscular dystrophy

ECMs stabilize cells and tissues by providing solid-phase cell adhesion substrates, linking ECM to cytoskeleton [7]. The consequences of ECM linkage loss is particularly evident in the heritable muscular dystrophies and skin blistering diseases. In both cases, mutations in distinct laminins and collagens cause severe diseases [8, 9]. The reason for the severe phenotype is the disruption of the ECM-to-receptor-to-cytoskeletal linkage. In skin, the main receptors of the ECM components are particular integrins, which then connect to several cytoskeletal components [9]. In skeletal muscle, the main ECM receptors are α-dystroglycan (αDG) and α7β1 integrin and dystrophin is an important linker to the cytoskeleton. Mutation in the genes coding for those proteins cause muscular dystrophies supporting the concept that a disruption of transverse linkages from stroma to cytoskeleton results in a failure of sarcolemmal stability and muscle function [8, 10].

Laminin-211 (α2β1γ1 chain composition; abbreviated Lm-211) is expressed mainly in adult muscle and peripheral nerve. Mutations within the LAMA2 gene cause a particular subtype of congenital muscular dystrophies (CMDs), which are characterized as early onset, often very severe muscular dystrophies [8] (Bönnemann, this issue). LAMA2 MD (also called MDC1A) is the largest subgroup of the CMDs. A recent study of 249 CMD patients in United Kingdom [13] revealed that LAMA2 mutations were the most common (37.4%) followed by dystroglycanopathies (26.5%) and Ullrich-CMD (15.7%). Mutations within the LAMA2 gene are spread throughout the gene (Fig. 1A). The majority of LAMA2 mutations listed in the Leiden Data Bank are nonsense mutations that result in a premature chain truncation to cause a severe, non-ambulatory dystrophy (Fig. 1A). The patients with such mutations lack any laminin-α2 staining in muscle biopsies but express similar amounts of the β1 and the γ1 laminin chains as detected in healthy controls (Fig. 1B). As a consequence of laminin-α2 deficiency, myofibers become damaged under the load of myofiber contraction. Muscle fibers undergo apoptotic degeneration followed by regeneration, chronic inflammation, and fibrosis (Fig. 1C). Analysis of laminin-α2 deficient mice has shown that regeneration fails and causes regenerating cells to undergo apoptosis [14, 15]. Patients with null-expression mutations are floppy at birth, often fail to thrive, never ambulate, and often die of muscle wasting and respiratory failure by the second decade [8, 13].

Figure 1. Mutations and muscle phenotype in LAMA2 MD.

A. Domain structure of laminin-α2 (top), location of mutations that cause the severe, non-ambulatory form (middle) or the ambulatory form of LAMA2 MD (bottom). The mutations of the severe form span the entire length of the gene and its 65 exons and intervening introns. They completely lack expression of the laminin-α2 resulting from premature chain-termination mutations. The ambulatory dystrophy is typically seen with missense mutations causing reduced and sometimes even normal expression of laminin-α2, resulting from amino acid substitutions or small deletions. Note that 17 of these mutations are clustered in the N-terminal LN domain that enables the laminin to polymerize. B. Cross-sections of muscle biopsies from normal and LAMA2 MD patients stained with antibodies directed to laminin-α2 or laminin-β1 and –γ1. Note that LAMA2 patient biopsies are negative for laminin-α2 but express similar amount of the laminin-β1 and -γ1. C. Haematoxylineosin stained muscle cross-sections from healthy controls and LAMA2 MD patients. Note the loss of muscle fibers and replacement with non-muscle tissue and the heterogeneous size of muscle fibers in LAMA2 MD patients compared to healthy controls.

While the absence of laminin-α2 results primarily in skeletal muscle damage, it also affects the peripheral nerve and the brain. The peripheral nerve defect alters nerve conduction [16, 17] and there is a report that a LAMA2 mutation can also result in limb-girdle muscular dystrophy with involvement of the peripheral nerve [18]. In the brain, LAMA2 MD patients often show changes in white matter density detected by T2-weighted magnetic resonance deduced to reflect increased water density rather than changes in myelination [19–22]. Additional brain abnormalities are those of cerebellar hypoplasia and, rarely, occipital lobe neuronal migration defects [8]. A subset of patients exhibits seizures or moderate mental retardation.

Other mutations in LAMA2 cause milder forms of the disease [23]. They are often missense mutations localized to the 5’ region of LAMA2 (red in Fig. 1A) that result in only reduced or near-normal expression of laminin-α2. Based on the recent solution of the laminin LN domain by X-ray crystallography [24, 25], several of the dystrophic mutations (Q167P, Y138H, G284R) map to a polymerization patch on one face of the LN domain. Others (C86Y, W152G, S157F, S277L, S204F, L243P) reside in the interior of the LN domain and are likely to disrupt the protein fold. An additional eight mutations are located downstream of the LN domain on the short arm, one is an in-frame deletion within the L4b domain [26]. The molecular basis for the alteration of laminin function by these mutations is unknown given no activities other than LN domain spacing has been postulated to date. In summary, most of the nonsense mutations cause absence of laminin-α2 and thus result in severe, non-ambulatory forms of LAMA2 MD while some of the missense mutations are predicted to affect laminin polymerization because they reside in the LN domain that has been demonstrated to affect laminin self-assembly in vitro and on cultured cells (see below).

Animal models of LAMA2 MD

There are several autosomal recessive mouse models of Lama2 mutations that cover the range of clinical severity seen in humans (Table 1). The first of those mutants was identified more than 60 years ago as a spontaneous mutant, called dystrophia muscularis (dy/dy) [27]. While dy/dy mice represent an intermediate to severe form of LAMA2 MD, the other spontaneous mutants, called dy^2J/dy^2J [28] and dy^nmf417/dy^nmf417 [29] display milder forms. Besides the spontaneous mutants, dy^W/dy^W [14, 30] and dy^3K /dy^3K mice [31] were generated by homologous recombination to represent the severe forms of LAMA2 MD. Because the mutation in the dy/dy mice has not been mapped, dy^W/dy^W and dy^3K /dy^3K mice are the two models that are used as models for the severe form. The dy^3K/dy^3K mice are complete Lama2 knockout mice [31]. They show a very severe phenotype that becomes evident soon after birth; they do not thrive and develop severe muscular dystrophy evidenced by histology and functional weakness. They develop hindlimb paralysis but the mice die at an age of a few weeks before lameness becomes the predominant phenotype. A very similar phenotype is seen in dy^W/dy^W mice, which are the best characterized and most often used mouse model for LAMA2 MD. The median survival of dy^W/dy^W mice is between 5 to 16 weeks. The big variation seems to largely depend on housing conditions and the genetic background [32]. The overall phenotype of dy^W/dy^W mice is slightly less severe compared to dy^3K /dy^3K mice, which might be based on the presence of low levels of a truncated laminin-α2 fragment [33]. Because of the longer survival of dy^W/dy^W mice compared to dy^3K /dy^3K mice, the hindlimb paralysis becomes more dominant at later time points.

Table 1.

Mouse models of laminin-deficient muscular dystrophy

mouse	mutation	laminin- α2 pro- tein	median survival (weeks)	body weight	muscular dystrophy	Ref.
dynmf417/dynmf417(dy7J/dy7J)	ENU-induced missense mutation	C79R mutation; normal level	no reduction reported	normal	mild	[29]
dy2J/dy2J	Spontaneous splice-site mutation	truncated and reduced level	no reduction reported	slight reduction	mild	[28] [34]
dy/dy	Unknown, spontaneous mutation	very low level to absent	20–35	reduction	moderate	[27, 107]; [108] [109]
dyW/dyW	targeted knockout	truncated; very low level to absent	5–16	strong reduction	severe	[76] [33] [32]
dy3K/dy3K	targeted knockout	absent	3–5	strong reduction	severe	[31]

Mice representing the mild, ambulatory form of LAMA2 MD are dy^2J/dy^2J and dy^nmf417/dy^nmf417 mice. The Lama2 mutation in dy^2J/dy^2J mice is a splice donor defect that causes an in-frame 57 amino acid deletion (residues 34–90) of the α2LN domain [34, 35]. Muscle histology shows a reduction in cross-sectional area with rounded contours, fibrosis, chronic inflammation, and regeneration (measured by the presence of central nuclei). This is accompanied by prominent amyelination in the sciatic nerve and roots [36] and there is evidence of leakage of water from the blood-brain barrier capillaries [37, 38]. Starting from about 3 weeks of age, the mice exhibit a stereotyped behavior of retracting their hindlimbs when suspended by the tail. The mice display weakness of all limbs and several weeks after weaning develop permanent hindlimb extensor contractions. Weights are near-normal and survival is typically greater than one year. A very similar phenotype to the dy^2J/dy² mice has been described for dy^nmf417/dyn^mf417 mice [29]. A missense mutation in Lama2 converts the cysteine residue at position 79 to an arginine (C79R). This residue is located in the α2LN domain suggesting that impaired laminin polymerization (see below) contributes to the disease phenotype. Of note, levels of laminin-α2 in muscle and nerve are not reduced by immunostaining and the BM ultrastructure appears unchanged in dynm^f417/dynmf4¹⁷ mice [29] Thus, dy^2J/dy^2J and dyn^mf417/dy^nmf417 mice are good models of the LAMA2 MD patients who suffer from the less severe, ambulatory form of the disease. There are also zebrafish, dog, cat models for LAMA2 MD [39–41].

Pathogenic mechanisms of LAMA2 MD

Laminins are essential for BM assembly, initiating the process by binding to surface receptors and receptor-like molecules (Fig. 2A). In muscle this initiating adhesion event is mediated by the principal laminin, Lm-211. The main laminin-binding loci responsible for initial anchorage to the cells are located in the LG domains of laminin-α2 chain [42, 43]. These domains bind to αDG, integrin α7β1, and very possibly to sulfated glycolipids (SGLs), similar to what was found in Schwann cells [44]. Although SGLs may substantially contribute to BM assembly by providing additional cell surface anchorage, it is αDG (via β-dystroglycan) and α7β1 integrin that establish critical BM anchorage to the underlying cytoskeleton [45, 46]. Of note, α7β1 integrin-binding is restricted in its binding repertoire, interacting only with α2, α1, and α5 laminins [47]. Integrin α7β1 binds to laminin domains LG1–3 and requires the presence of the adjacent coiled-coil domain [48, 49]. Binding of Lm-211 to αDG maps to the C-terminal LG1–3 and LG4–5 domains of laminin-α2 [2] and requires the presence of mannosyl O-linked carbohydrates [50]. A unique repeating disaccharide consisting of xylose and glucoronate, coded by the LARGE glycosyltransferase and located on the neck-like region of αDG, is required for binding to the LG domains of laminins, agrins, and perlecan [50–56]. A comparison of Dag1 (encoding α- and β-dystroglycan) and Itga7- (encoding α7 integrin) deficient muscles in mice indicates that the interaction of the BM with αDG is more critical for muscle maintenance than the α7β1 integrin, as Dag1-deficient but not Itga7-deficient muscle fibers become detached from the BM and renders fibers prone to contraction-induced injury [57]. As Lm-411 has barely detectable binding to αDG and Lm-511 binds poorly [58], the increase of these laminins is inadequate to compensate for the loss of Lm-211.

Figure 2. Normal and dystrophic basement membrane assembly and composition.

A. In skeletal muscle, Lm-211 forms the initial polymer nascent scaffolding by binding to integrin α7β1, αDG, and possibly also to sulfated glycolipids (SGL) via the LG domains. The α7β1 integrin and αDG form linkages through cellular adaptor proteins to actin and other cytoskeletal cables. Lm-21 1 also polymerizes through the three different LN domains that interact to form a ternary node. Nidogen-1 (and −2) binds to the laminin-γ1subunit (domain LEb3) and to collagen-IV, acting as a bridge, with the collagen polymerizing to form a second network. Agrin binds to the coiled-coil of the Lm-211 via its NtA domain and to αDG with the C-terminal LG domains, possibly acting as a collateral linker. All structural components become directly or indirectly tethered to cell receptors through the polymerizing laminin. Lm-411 is a component of the peripheral nerve Schwann cell basement membrane but absent in normal sarcolemmal basement membrane. B. Immuno-stained cross-section of skeletal muscle from a healthy control and a LAMA2 MD patient. In healthy controls, laminin-α4 is largely restricted to blood vessel and the neuromuscular junction. In LAMA2 MD patients, laminin-α4 is found in muscle basement membrane hereby substituting for laminin-α2.

Concomitant with receptor-mediated anchorage of Lm-211, laminin self-polymerization contributes to the formation of a provisional extracellular matrix. One effect of this self-polymerization is to confer an approximately three-fold increase of the amount of laminin bound to the surface of cultured myotubes and an even higher increase in Schwann cells [12, 43]. Polymerization occurs through ternary binding of the LN domains of the α-, β- and γ chains, each laminin forming three ternary nodes with adjacent laminins [24, 59–63]. Absence of any of the three different LN domains prevents polymerization [43, 62]. As in muscle all the expressed β- and the γ-laminins possess LN domains, the N-terminal end of the long α-laminins decides on the polymerization properties of the native heterotrimer. Thus, laminins that contain a short α-laminin, i.e. laminin-α3A or laminin-α4, are predicted not to polymerize. Laminin-α4 is normally found in the adult Schwann cell BM of peripheral nerve, in blood vessels of nerve and muscle, and at the neuromuscular junction, but it is not found in non-synaptic regions of the muscle BM [36, 64, 65]. Anchorage in nerve and vessels may depend upon interactions with sulfated glyco-lipids and CD-146 given that αDG and integrin interactions appear to be quite weak [47, 66–69].

The laminin accumulation by cell surface binding and self-polymerization provides the scaffold for nidogens (mainly nidogen-1) to form a critical high-affinity bridge between laminins and to type IV collagen [43]. This is achieved by the binding of nidogen-1 to the short arm of laminin-γ1 [70, 71] and its binding to collagen type IV [60, 70]. The resulting co-polymer of the laminin and the collagen type IV networks provides structural support that resists damage through muscle contractions and also acts as a signaling platform.

Loss of laminin-α2 in LAMA2 MD patients does not result in the loss of staining with antibodies directed against the β1- and the γ1 chain (Fig. 1B), suggesting compensatory expression of genes coding for other α chains. Indeed, muscle biopsies of LAMA2 MD patients are strongly positive for laminin-α4 [58] (illustrated in Fig. 2B). The same compensatory expression of laminin-α4 has been documented in dy/dy and dy^W/dy^W mice [72–74]. Laminin-α4 is expressed in skeletal muscle during embryonic development. It disappears after birth from the muscle BM except at the neuromuscular junction and it also outlines blood vessels in muscle [58, 64]. Besides laminin-α4, there is also some increase in the levels of laminin-α5 in dy^W/dy^W mice and in LAMA2 MD patients [58, 73].

In muscle, the loss of laminin-α2 and the compensatory presence of laminin-α4 increases the fraction of laminin that can be extracted with isotonic buffer, reflecting reduction of the tight attachment of Lm-411 to the muscle BM [12, 58]. The loose attachment of Lm-411 to muscle BM is paralleled by a complete lack of accumulation of recombinant Lm-411 to the surface of cultured C2C12 myotubes [58]. This loss of surface binding by Lm-411, which contrasts the still significant attachment of Lm-411 to BM in vivo, is thought to be largely due to low concentration of components assembling in an open culture system compared to the high concentration for assembly in the diffusion-limited potential space that becomes the BM in a tissue. Effectively, the readout of laminin accumulation in tissue culture system is very sensitive to differences that are obscured in vivo, a valuable characteristic for discerning among different repair treatments.

The reasons for the lack of Lm-411 accumulation in tissue culture and its poor incorporation into muscle BM are the weak binding of the laminin to αDG and to integrins [47, 69], absence of alternative receptors, and inability to self-polymerize. Similarly, the other compensatory laminin, Lm-511, although it binds well to several integrins (including α7β1) and polymerizes, binds only weakly to αDG compared to Lm-211 [47, 58, 75]. Thus, LAMA2 MD may well be a disease that is largely caused by the insufficient assembly of the laminin network and the poor connection to the plasma membrane of the underlying muscle fiber. Indeed, recent experiments that aimed at restoring those two functions have provided strong support for this hypothesis (see below).

Treatment strategies

Replacing lost laminin

The most obvious strategy to treat LAMA2 MD would be to express laminin-α2 using gene therapeutic approaches (Table 2). In a proof-of-principle experiment, Engvall and colleagues have shown that muscle-specific, transgenic expression of a cDNA encoding the human version of laminin-α2 is capable of restoring muscle histology and allowing dy^W/dy^W mice to survive for more than one year (there were no attempts made to measure lifespan). Because the transgene was not expressed in the Schwann cells of the peripheral nerve, hindlimb paralysis was still present [76]. In an analogous approach, the Durbeej laboratory expressed cDNA encoding laminin-α1 transgenically using the chicken β-actin (CAG) promoter [77, 78]. As this transgene is expressed ubiquitously (including Schwann cells in the peripheral nerve), laminin-α1-expressing dy^3K /dy^3K mice did not suffer from hindlimb paralysis and the mice survived for up to two years [78]. This work therefore shows that laminin-α1 can fully substitute for the function of laminin-α2. As cDNAs encoding laminin-α2 or laminin-α1 by far exceed the size limit of adeno-associated viral (AAV) vectors, this approach cannot be translated into a treatment of LAMA2 MD patients. An alternative to gene therapy could be the injection of recombinant Lm-111. Indeed, injection of EHS-derived Lm-111 into dy^W/dy^W mice has been shown to be efficacious [79]. However, translation of such a protein-based therapy remains a challenge because it may be difficult to efficiently incorporate a protein of the size of 1,000 kDa into the muscle BM after injection. Moreover, production of sufficient amounts of fully active, recombinant human Lm-111 has not yet been achieved. Another recent approach to increase LAMA1 expression has used electroporation of skeletal muscle to deliver a guide RNA (gRNA) that targets the Lama1 promoter and a “dead” version of Cas9 coupled with the VP160 transcription activation domain. Applied to mdx mice, a mouse model for Duchenne Muscular Dystrophy (DMD), this approach led to increased expression of Lama1 and detectable levels of Lm-111 [80]. While there is some controversy as to whether Lm-111 in DMD patients would result in a beneficial effect [81], such an approach, when applied to LAMA2 MD patients is likely to improve the dystrophy.

Table 2.

Basement membrane structural repair strategies evaluated in mouse models.

Type of Repair	Mouse	Approach	Observed benefits	Comments	Ref.
LAMA1 gene replacement	dy3K/dy3K	Transgenic expression of laminin-α1 using general CAG promoter	Substantial repair of muscle and peripheral nerve with restoration of weight, strength and behavior.	This approach is not amenable to gene therapy as cDNA too large (9.3 kb) for somatic viral-mediated delivery.	[78]
Parenteral delivery of Lm-111	dyW/dyW	EHS Lm-111 injected i.v., i.p. or i.m.	Lm-111 detected in sarcolemmal pattern. Median survival increased from ~83 to 292 days; slight weight and muscle force increases; increased activity; decreased inflammation.	EHS Lm-111 protein therapy also reported to improved mdx mice (DMD model); however, findings not replicated by laminin-α1 transgene in mdx.	[79, 110]; [81]
Exon skipping	dy3K/dy3K	PMO-mediated skipping of exon 4 of mouse premature mRNA.	Trend to improve survival.	Exon-skipping, under intense investigation for DMD, seems unlikely to be of general value for the LAMA2 gene because the exon borders do not match protein domain borders.	[83]
Laminin receptor anchorage	dyW/dyW	Transgenic muscle-expression of miniaturized agrin (mag)	Substantial increase in survival, fiber size, BM ultrastructure, weights, and walking time with reduced muscle creatine kinase activity.	Mag, by binding to the compensatory Lm-411, increases anchorage to αDG. AAV-mediated expression improved dyW/dyW dystrophy.	[73, 98]
Laminin polymerization	dy2J/dy2J	Transgenic muscle-expression of a laminin-nidogen fusion protein (αLNNd)	Restoration of fore-limb grip strength to normal. Forelimb muscle histology fiber sizes/appearance restored to normal with elimination of fibrosis. Regeneration greatly reduced.	Lm-211 is unable to polymerize. αLNNd binds to laminin-γ1 near the intersection of the laminin cross. It acts primarily by enabling polymerization of truncated Lm-211 expressed in dy2J/dy2J mice	[12]
Combined laminin anchorage + polymerization	dyW/dyW	Transgenic muscle expression of mag and αLNNd	Survival increased greater than five-fold to near-normal levels. Histology (fiber size/fibrosis) improved. Increased weights and muscle strength.	Improvement is mediated by enabling Lm-411 to polymerize and bind to αDG. Benefit was significantly greater than that seen with either transgene alone. Sets stage for double AAV somatic gene delivery.	[58]
Non-homologous end-joining correction of splicing defect	dy2J/dy2J	AAV delivery of CRISPR-Cas9 gene-editing components	Restored expression of full-length laminin-α2. Substantial improvement of muscle histology with reduced fibrosis. Increased mouse activity and muscle force of contraction.	General applicability for splice donor mutations. Would not correct nonsense and missense mutations.	[84]

Exon-skipping to correct out-of-frame mutations has been used successfully to treat dystrophin-deficiency in mice and human. A first exon-skipping oligonucleotide has recently been conditionally approved by Food and Drug Administration (FDA) for the treatment of DMD [82]. Exon-skipping has also been discussed for LAMA2 MD but appears most likely to be problematic in that skipping of nearly all LAMA2 exons will result in either cysteine mispairing with improper domain folding and disulfide-bond formation and/or major loss of functional domains. In a recent trial to restore laminin-α2 in dy^3K /dy^3K mice, exon-skipping had some benefit but did not significantly increase lifespan [83].

AAV-mediated CRISPR-Cas9 genome editing, targeted to both the muscle and the nerve, was recently used to substantially ameliorate the muscle and peripheral nerve phenotype of the dy^2J/dy^2J mice by correcting the splicing defect [84]. The basic therapeutic design should have wide application for splice donor defects. In a recent study of congenital muscular dystrophies in U.K. it was reported that 19/89 LAMA2 MD arose from splice-site mutations.

Restoration of muscle BM by linker proteins

Another approach to restore the linkage of the muscle BM to the plasma membrane in LAMA2 MD patients uses the upregulated Lm-411 (and Lm-511) as scaffold for the attachment of small, bi-functional linker proteins. The first modification to make Lm-411 behave more like Lm-211 was made many years ago by using a specifically designed, miniaturized version of the protein agrin, called mini-agrin or briefly mag (Fig. 3) [73]. Agrin is concentrated at the neuromuscular junction where its motor neuron-derived splice isoforms are responsible for the formation of postsynaptic specializations [85]. In non-synaptic regions of the muscle fibers, expression of agrin is low [54] and thus is unlikely to effectively compensate for a loss of laminin-α2 in dy^W/dy^W mice [86]. Design of mag was based on work revealing the binding activities of the muscle (non-neural) agrin isoform. Mag consists of the high-affinity, laminin coiled-coil domain-binding NtA domain [87, 88] and one follistatin repeat fused to the LG/LE C-terminal domains responsible for high-affinity binding to αDG [89]. Thus, mag links Lm-411 with its extremely low affinity for αDG and α7β1, to αDG. Transgenic, muscle-specific expression of mag in dy^W/dy^W strongly improves muscle histology and survival [73]. While survival was not measured, mag greatly improved muscle histology and regeneration also in dy^3K /dy^3K mice [90]. In dy^W/dy^W mice, similar efficacy like mag was also achieved with a fusion protein between the agrin NtA domain and the αDG-binding domain V of perlecan [91], corroborating the idea that this disease-ameliorating effect is based on the linkage of Lm-411 to αDG. Finally, substantial repair was even achieved when expression of mag in dy^W /dy^W mice was initiated after birth [91].

Figure 3. αLNNd and mag repair of laminin function.

A. Domain structure and functional activities of αLNNd and mag. Regions derived from laminin-α1 are in green; regions derived from nidogen-1 are in orange. Mag is a miniaturized version of agrin with N-terminal regions (blue) and C-terminal parts (red). B. In the ambulatory form of LAMA2 MD and its dy^2J/dy^2J mouse model, a truncated version of Lm-211 (“dy2J–Lm-211”) is expressed. αLNNd binds to the nidogen-binding site and creates an artificial short arm with a functional LN domain. This enables polymerization and promotes assembly of a stable basement membrane. C. In the absence of laminin-α2, which causes the severe non-ambulatory form of LAMA2 MD, the amount of Lm-41 1 is increased. Lm-411 is unable to polymerize and binds poorly to integrin α7β1 and to αDG. Co-expression of αLNNd and mag provide the necessary domains for polymerization and αDG anchorage. Together the two linker proteins restore laminin assembly and muscle binding, which results in strong amelioration of the severe muscular dystrophy in dy^W/dy^W mice.

While mag improves Lm-411 binding to αDG, this approach is still unable to cause self-polymerization. To enable Lm-411 self-polymerization, another linker protein was generated that consists of the laminin-α1 LN-LEa domains fused to a nidogen-1 fragment extending from the G2 to the G3 domain [92]. This fusion protein, called αLNNd, was shown to bind to Lm-111 at the nidogen-binding locus (Lmγ1 domain LEb3) and to collagen-IV. Lm-111 binding is mediated by the C-terminal G3 domain of αLNNd while collagen-binding is mediated by the G2 domain (Fig. 3A,C). The presence of the α1LN domain enables polymerization of laminins that lack either the αLN domain or the entire α-short arm [92]. On cultured C2C12 myotubes, co-incubation of αLNNd with α-short arm-truncated Lm-111 restores binding to levels indistinguishable to full-length Lm-111 [12]. The idea that restoration of laminin self-polymerization may provide benefit to the mild forms of LAMA2 MD, was tested by the muscle-specific transgenic expression of αLNNd in dy^2J/dy^2J mice, which bears an internal in-frame α2LN deletion and is defective of self-polymerization [11]. Indeed, expression of αLNNd resulted in a strong amelioration of the dystrophic phenotype. The effect was manifested as a prevention of fibrosis and restoration of fore-limb grip strength, both to normal levels. αLNNd also improved myofiber shape, size, and numbers to control levels. It led to the restoration of near-normal levels of Lm-211 in the muscle BM, suggesting that this reflects improved assembly of the truncated Lm-211 expressed in dy^2J/dy^2J into the BM [12].

While the success with the αLNNd transgene was an exciting advance that also helped confirm the model of BM assembly, only a small fraction of patients would be candidates for repair with αLNNd alone. For most patients there is a complete or near-complete loss of Lm-211. As highlighted above and in Fig. 2C, there is substantial compensation by Lm-411 in these severe forms of the LAMA2 MD. If indeed self-polymerization and cell anchoring are the two key functions of Lm-211, restoration of self-polymerization should provide additional benefit to mag-expressing dy^W /dy^W mice, representing the severe form of LAMA2 MD. This hypothesis was well supported by the observation that combination of αLNNd and mag restores binding of Lm-411 to myotubes that reaches levels that are only observed with Lm-211, while either of the linker proteins led to a moderate increase in Lm-411 binding [58]. Most importantly, transgenic co-expression of αLNNd and mag in dy^W /dy^W mice completely restored BM stability with substantial recovery of muscle force and size, increased body weight, and the extension of survival from a few weeks to more than 1.5 years [58]. About one third of the double transgenic dy^W/dy^W mice lived more than two years, comparable to wild-type mice. These results are thus clear proof for the model that laminin self-polymerization and its anchorage to the muscle fiber surface are the two key activities that are needed to stabilize muscle BM.

The cDNAs for αLNNd and mag are in the size range that can be packaged into adeno-associated virus (AAV) with an appropriate promoter. AAV is one of the most promising of the somatic gene delivery systems in which high expression can be achieved in muscle and other tissues. The experience of human clinical trials with AAV to correct muscular dystrophies has been encouraging. While protein can be lost due to host cellular immune responses to transgene products and AAV capsid [93], this problem has been reduced by avoiding the creation of transgene neoantigens (note that the domains of αLNNd and mag are expressed in LAMA2 MD patients), by optimizing serotype, and by adding immunosuppressive therapy [94]. Promising success has been recently achieved in the treatment of limb-girdle dystrophy type 2D (α-sarcoglycan deficiency) with recombinant AAV with persistence of protein expression, restoration of the sarcoglycan-complex, and increased muscle fiber size [95]. Furthermore, simultaneous infection with two AAV vectors has been successfully accomplished for the delivery of dysferlin, dystrophin or CRISPR-Cas9 genome-editing components to muscle [84, 96, 97]. By the same token, it should be possible to separately deliver the two linker proteins αLNNd and mag to LAMA2 MD patients with paired AAV somatic gene delivery in muscle and peripheral nerve. It should further be noted that expression of mag using systemic infection with AAV has been shown to provide similar benefit to dy^W/dy^W mice as the transgenic expression [98]. Thus, while the combined repair strategy is more challenging than one using a single AAV, such strategy is compatible with existing technologies and limitations.

Interference with secondary events of disease

Treatment strategies using small molecules attempt to halt disease progression by interfering with downstream consequences of Lm-211 deficiency (Table 3). Such strategies are aimed to inhibit the apoptosis, inflammation and fibrosis that all result from the primary injury. Drug therapies to treat the sequelae of the dystrophy show improvements, but are limited in that they do not correct the underlying structural defect. The most advanced of those drugs is omigapil. It is an orally available deprenyl-analog that binds to GAPDH. The drug inhibits apoptosis and was originally developed by Novartis and is now further developed by Santhera Pharmaceuticals Ltd. Treatment of dy^W/dy^W and dy^2J/dy^2J mice was shown to ameliorate disease progression by inhibition of apoptosis, positively affecting fibrosis and improving respiration [99, 100]. Omigapil is currently being evaluated in a clinical phase 1 trial for congenital muscular dystrophies (CALLISTO, NCT01805024, Carsten Bönnemann and Reghan Foley, NINDS, NIH). Losartan is an antagonist of angiotensin-II type I receptor that is approved for the treatment of high blood pressure. The drug was found to reduce fibrosis by acting on TGF-β1 activity. In dy^W/dy^W and dy^2J/dy^2J mice, it inhibits TGF-β signaling, lowers TGF-β1 levels in serum, reduces fibrosis, and increases the number of myofibers [101–103]. Inhibition of inflammation in dy/dy mice with prednisolone, a potent steroid, improves survival and forelimb strength [104]. Bortezomib, a proteasome inhibitor, was evaluated in dy^3K /dy^3K mice [105] and was found to increase survival from a mean of 23.5 days to 32 days, to slightly increase weights and the number of quadriceps myofibers, and to decrease the proportion of caspase-3 positive fibers. However, the same treatment regimen in dy^2J/dy^2J mice did not show much benefit [105, 106]. All those treatments are potentially of value for the LAMA2 MD patients, it remains to be seen, however, whether they are efficacious enough to provide benefit in a clinical trial.

Table 3.

Drug treatments directed at sequelae of muscle damage evaluated in mouse models

Inhibition target	Mouse	Treatment	Observed Bene- fits	Comments	Ref.
Inflammation	dyW/dyW	Prednisolone (10 µg/gm/wk p.o.)	Survival increase from 66 to 90% at 24 weeks; no change in weight; 16% increase in grip strength.	Long-term steroid use many side-effects that may make this drug problematic in LAMA2-MD.	[104, 111]
Protein degradation (proteosome genes)	dy3K/dy3K & dy2J/dy2J	Bortezomib (0.4 mg/kg i.v.)	Median survival, weight and fiber size in dy3K/dy3K mildly increased with decreased regeneration.	No beneficial effect in dy2J/dy2J mouse.	[105, 106]
Apoptosis	dyW/dyW & dy2J/dy2J	Omigapil (0.1–1 mg/kg oral)	Improves respiration, locomotion and weight and protects from early mortality Modest reduction of fibrosis.	Under phase 1 clinical trial for LAMA2 and collagen-VI deficiencies.	[99, 100]
Fibrosis	dyW/dyW & dy2J/dy2J	Losartan (0.6 mg/ml drinking water)	Decreases TGF-β signaling. Reduced fibrosis and inflammation. Improved locomotion, strength and weight.	Losartan is an angiotensin II receptor type 1 antagonist shown to block pro-fibrotic action of TGF-β.	[101–103]
Other	dyW/dyW	Doxycycline (6 mg/ml drinking water)	Median survival increased from 32 to 70 days; small weight increase; improved behavior and histology.	Tetracycline derivatives may benefit the dystrophy by apoptosis inhibition.	[112]

Concluding Remarks

Considerable progress has been made in developing treatments for LAMA2 deficiency. The greatest degree of phenotypic improvement has been observed in mice in which the underlying structural defect was corrected. Nonetheless, significant improvements were also obtained with agents that reduce the untoward sequelae. We cannot be confident that a single therapy will suffice to effectively treat dystrophic patients. The different therapies are not mutually exclusive and it stands to reason that the most successful protocols may prove to be combinatorial ones that repair the structural defect but also separately reduce the apoptotic, inflammatory and fibrotic sequelae.

Highlights.

• Basement membranes have reached center stage due to their involvement in many physiological and pathological conditions.

• Genetic diseases of basement membranes are debilitating and affect many organs resulting in diverse phenotypes.

• We provide an overview of the field and discuss developmental, structural and biochemical aspects of basement membranes.

• We introduce the special issue and outline key components of basement membrane biology.