A single point mutation in the LN domain of LAMA2 causes muscular dystrophy and peripheral amyelination

The surface of every muscle fiber and peripheral nerve fiber is covered by basal lamina (BL), a dense, mechanically stable sheet of extracellular matrix. The major glycoprotein in muscle and nerve BLs is laminin 2 (Lama2), also called merosin or LM-211 (Aumailley et al., 2005; Leivo and Engvall, 1988). Laminin 2 contains the α2, β1 and γ1 subunits, or chains, each encoded by a distinct gene. Laminin α2 mutations are the most common cause of early-onset muscular dystrophy (MDC1A) in humans (note: the human laminin α2 gene symbol is LAMA2), and cause a similar disorder in dogs and cats (Shelton and Engvall, 2005; Tome et al., 1994). However, disease mechanisms have been best studied in Lama2-mutant strains of mice, known as dystrophia muscularis (dy) mice (note: the mouse laminin α2 gene symbol is Lama2, but the mutation was first identified as dy) (Michelson et al., 1955; Miyagoe et al., 1997; Xu et al., 1994a; Xu et al., 1994b). Characteristically, dystrophy from Lama2 is compounded by developmental defects in myelination, and BLs on muscle fibers and Schwann cells appear thin and discontinuous (Madrid et al., 1975; Nakagawa et al., 2001; Xu et al., 1994a). Accordingly, myofiber BLs are thought to provide a stable anchor for adhesion receptors in the sarcolemma and thus to absorb cytoskeletal stresses during movement. Similarly, Schwann cell BLs have been thought to orient Schwann cell growth and differentiation during myelination.

Biochemical, immunochemical, and genetic studies buttress this model. Laminins contain specific binding domains for receptors and for polymerization, and these occur in a polarized distribution that predisposes laminins to form sheet-like networks on cell surfaces. Laminins, like collagens, are composed of three extended polypeptides (α, β and γ chains) entwined cord-like over much of their length. However, the amino-halves of laminin chains are separated, forming `short-arms' that branch from the main trunk. Globular `LN' (laminin N-terminal) domains, one at the tip of each short arm, bind in a tripartite fashion to link laminin heterotrimers into extended networks. The carboxyl end of the trunk ends in a cluster of α-chain `LG' (laminin G-like) domains, which control binding specificity for receptors such as integrins, dystroglycan and sulfatide. Thus, laminin biochemistry appears to be especially devoted to framing BLs (Colognato and Yurchenco, 2000).

In vivo and in vitro studies demonstrate that BL assembly requires at least one polymerizing laminin (Sasaki et al., 2004; Yurchenco et al., 2004a). Laminin 2 is the predominant isoform in nerve and muscle BLs, outside intercellular junctions (Patton et al., 1997). All Lama2 mutations studied in mice disrupt BLs, and all laminin α2 mutations studied in humans and mice result in upregulation of the laminin α4 chain, expressed as laminin 8 (α4β1γ1) (Patton et al., 1999; Patton et al., 1997; Ringelmann et al., 1999). However, the α4 chain lacks the LN domain, and its upregulation is insufficient to sustain BL structure or prevent disease in Lama2-mutant mice, further suggesting that loss of BL structure is pathogenic. By contrast, transgenic expression of the LN-containing α1 chain restores BL integrity, reduces myodegeneration and increases myelination in α2-mutant mice (Gawlik et al., 2004; Gawlik et al., 2006).

However, two prominent observations remain unexplained. First, the sarcolemma is reportedly stable in laminin-α2-mutant mice, in contrast with mice lacking dystrophin or the dystrophin-glycoprotein complex (DGC) (Straub and Campbell, 1997). The cytopathological mechanism is therefore unknown. Second, loss of BL structure might not account for differences in the severity of myodegeneration associated with Lama2 mutations. For example, whereas Lama2-null mice (dy^3K) have severe dystrophy and die by 8 weeks, the dy^2J mutation, which causes an internal deletion in the α2 LN domain and disrupts BL integrity, causes moderate dystrophy and no mortality. Rescue studies confirm that mortality is due to dystrophy (Kuang et al., 1998). It is possible that the severity of dystrophy reflects the overall loss of laminin 2 protein affected by specific mutations, with myodegeneration varying primarily from progressive loss of BL stability. Alternatively, laminin 2 might act through multiple mechanisms that are selectively inactivated by specific mutations. These questions are directly relevant to human MDC1A, in which complete LAMA2 deficiency is associated with perinatal onset and death in adolescence, whereas mutations that leave C-terminal portions of α2 intact are associated with distinctly milder dystrophy (Cohn et al., 1998; Sewry et al., 1997).

In contrast to muscle, all mouse Lama2 mutations similarly disrupt peripheral myelination (Biscoe et al., 1975; Bradley and Jenkison, 1973; Nakagawa et al., 2001; Weinberg et al., 1975). Immature Schwann cells lacking laminin 2 fail to proliferate and differentiate properly during the perinatal process of axonal (radial) sorting. As a result, nerves in dy-strain adults contain large bundles of `amyelinated' axons. Myelination in spinal roots is almost entirely dependent on laminin 2, whereas myelination in distal nerves is more equally dependent on laminin 2 and laminin 8, making the ventral roots a sensitive indicator of laminin 2 function in myelination (Yang et al., 2005). Loss of BL structure on Schwann cells lacking α2 was initially thought to be responsible for amyelination (Bunge et al., 1986; Madrid et al., 1975). However, considerable evidence now argues that BL formation per se is not germane (Nakagawa et al., 2001; Yang et al., 2005). Nevertheless, the dy^2J deletion of the α2 LN domain almost completely inactivates the ability of laminin 2 to promote myelination. We have suggested that laminin 2 polymerization mediated by the LN domain is required to foster receptor aggregation over short distances, rather than BL formation (Yang et al., 2005). Alternatively, the α2 LN domain might have a specific receptor-mediated function in promoting myelination, or amyelination in dy^2J mice might result from reducing laminin 2 levels below a critical threshold (Sunada et al., 1995; Xu et al., 1994b).

Here, we identify a Lama2 point mutation in nmf417 mice that causes muscular dystrophy and amyelination similar to dy and dy^2J, but without loss of laminin 2 expression or BL stability previously associated with the disorder. The mutation changed Cys79 to Arg, thereby disrupting a paired cysteine motif (CxxC) conserved in all known laminin LN domains, both vertebrate and invertebrate. These findings suggest that the widely held hypothesis that laminin 2 promotes myelination and prevents dystrophy primarily through its role in maintaining BL stability is overly simplistic. Observations in nmf417 nerves demonstrate a role for the LN domain of α2 that is separate from overall laminin 2 expression, and suggest that additional factors play major roles in modulating BL stability.

A program of chemical mutagenesis at The Jackson Laboratory, designed to identify genetic causes of neurological disease in mice, isolated a heritable recessive mutation, nmf417, resulting in overt neuromuscular dysfunction. In initial litters from unaffected parents, four of 21 offspring developed general skeletal muscle wasting at juvenile ages and bilateral contractures of the hind limbs as young adults. Preliminary histological examination of muscles at 8 weeks found hallmarks of muscular dystrophy, including widespread endomysial fibrosis, focal necrosis and central nuclei typical of regenerated fibers. In addition, sections through the spine revealed severe hypomyelination in nerve roots, but not in spinal cord. Combined skeletal muscular dystrophy and peripheral hypomyelination is consistent with the phenotype of dy, which results from loss-of-function mutations in Lama2 (Sunada et al., 1995; Xu et al., 1994b). We tested for genetic non-complementation by mating mice known to be heterozygous for the nmf417 and Lama2^dy (dy) mutations. Three of ten offspring were dystrophic, indicating that nmf417 is a Lama2 allele.

To confirm and identify a Lama2 mutation, the entire open reading frame of cDNA generated from an affected nmf417 mouse was sequenced and compared with isogenic C57BL/6J sequence. Consistent with ethylnitrosourea (ENU)-induced mutations, a single base change from T to C was found at the first position of codon 79, which converts Cys79 to Arg (Fig. 1A-C). Primer-mismatch PCR based on this mutation gave specific but distinct bands in nmf417 and wild-type mice, and was used for subsequent genotyping (see Materials and Methods and supplementary material Fig. S1). There were no other coding-sequence differences between isogenic control C57BL/6J and nmf417. Using cDNA prepared from heterozygous mice, chromatogram peaks of nearly equal intensity were observed for both T and C, suggesting that wild-type and nmf417 Lama2 mRNA transcripts are similarly stable in vivo (Fig. 1B).

Cys79 (C79) is conserved in all known LAMA2 sequences, including mammals, birds, and the Drosophila and Caenorhabditis elegans orthologs (Fig. 1D,E). Within the α2 chain, Cys79 is centered in the LN domain at the free end of the short arm (Fig. 1F), which is implicated in mediating laminin-laminin interactions and BL assembly (Cheng et al., 1997; Yurchenco, 1990). The previously described dy^2J allele of mouse Lama2 generates an internal deletion within the LN domain that includes C79 and disrupts laminin 2 polymerization and BL formation (Sunada et al., 1995; Xu et al., 1994b; Colognato and Yurchenco, 1999). Therefore, the C79R mutation in nmf417 might identify a key residue for promoting BL formation. Alternatively, the C79R mutation could decrease overall laminin 2 expression or stability, or act by disrupting LN domain activities unrelated to BL formation. To distinguish these possibilities, we assessed the severity of the nmf417 allele, determined levels of laminin 2 protein expression in nmf417, and assessed whether myodegeneration and dysmyelination had a similar relationship to BL structure as that observed in other Lama2 alleles (Table 1).

We assessed general disease progression in nmf417 homozygous mice. The body weight of nmf417 homozygotes did not significantly diverge from control values until the fifth week (supplementary material Fig. S2, P<0.01 at 5 weeks). This is markedly later than dy^w or dy^3K homozygous mice, which show reduced size by 2 weeks of age, but is earlier than dy^2J homozygotes, in which growth is not strongly affected (Kuang et al., 1998; Miyagoe et al., 1997; Moll et al., 2001). The nmf417 homozygotes showed no mortality by 6 months, and bred and reared litters efficiently. However, their movements became visibly impaired by weaning, and hindlimbs became partially paralyzed by 6 weeks of age, although some hind-limb function was retained at 8-10 months. Mice homozygous for severe Lama2 alleles such as dy^3K or dy^W become cachexic and die by 8-12 weeks, whereas mice homozygous for the original dy allele rarely breed, develop permanent hindlimb contractures at 10-12 weeks of age and most die by 6 months. Mice homozygous for the dy^2J allele have permanent contractures by 12 weeks, but are fertile and do not have a shortened lifespan. Thus, the overt phenotype of nmf417 homozygotes is most similar to mice homozygous for the dy^2J allele, although the nmf417 growth rate was more impacted. Interestingly, however, nmf417 hindlimb paralysis was also less severe than that of dy^2J homozygous mice, suggesting partially divergent phenotypes for the nmf417 mutation and dy^2J internal deletion in nerve and muscle.

We next assessed laminin 2 production in muscle and nerve using antibodies specific to the α2 chain. Consistent with the milder phenotype of the homozygous mice, the intensity of immunoreactivity for α2 in cryostat sections of adult nmf417/nmf417 muscles was similar to age-matched littermate controls, and α2 was co-concentrated with β1 and γ1 chains along myofiber surfaces, and was generally absent inside myofibers and between fibers (Fig. 2A-F). Similar results were obtained with antibodies directed against the short arms and LG domains of α2, and in fixed and unfixed tissues (not shown). The results suggested that the C79R mutant α2 chain is correctly synthesized and assembled into laminin 2 heterotrimers, secreted and, stably incorporated into myofiber surface matrix. To verify these conclusions, we prepared sarcolemma fractions from leg muscles of nmf417/nmf417 and control mice, separated insoluble matrix proteins from other membrane-associated proteins by detergent extraction, and immunoblotted under non-reducing conditions to preserve the chain linkage within laminins. Blots revealed stable matrix incorporation and co-migration of C79R α2 chain with β and γ chains at normal heterotrimer mobilities (Fig. 2G). The abundance of α2 in mutant muscle was determined by western blotting directly from tissue homogenates using denaturing and reducing conditions (Fig. 2H). The primary band detected in both nmf417/nmf417 and control samples migrated at the expected size of 320 kDa for the major α2 fragment. Lanes were loaded with equal amounts of sample protein (20 μg) and signal intensities were standardized for loading variation by re-probing blots with an antibody against myosin heavy chain. Using this standard, the mutant samples showed an increase in laminin α2 abundance (142±22% of control, mean ± s.e. in analysis of eight mutant and seven control lanes prepared from four nmf417/nmf417 and four control samples, 8-9 weeks of age). This apparent increase might partly reflect a loss of myosin in the dystrophic muscles, but indicates that there is no decrease in α2 levels below a critical threshold in the nmf417/nmf417 muscle. Thus, the C79R mutation does not significantly affect α2-chain size, stability, heterotrimer assembly, basement membrane incorporation or abundance. The normal mobility of nmf417 laminin 2 also suggests that glycosylation is not affected by the C79R mutation. The nmf417 allele, therefore, contrasts with the dy and dy^W mutations, which substantially reduce laminin 2 levels in myofiber BLs, and dy^3K, which eliminates it.

At neuromuscular junctions (NMJs), identified by α-bungarotoxin staining for acetylcholine receptors, the α2 chain is assembled in trimers with the β2 chain as laminin 4 (α2β2γ1) (Patton et al., 1997; Sanes et al., 1990). We found that synaptic BLs at nmf417/nmf417 NMJs contained normal levels of the α2 and β2 chains, indicating normal assembly and targeting of laminin 4, as well as of other synaptic extracellular matrix components (Fig. 2I,J and supplementary material Fig. S3). Laminin 4 also replaced laminin 2 at nmf417 myotendinous junctions, as in normal muscle (not shown). These results are in contrast to those from dy muscles, in which the α2 chain is selectively absent from synaptic BLs, while low levels of laminin 2 persist in extrasynaptic BLs (Patton et al., 1997; Sewry et al., 1998).

Previous α2 mutations diminish laminin 2 levels in nerves more than in muscles. Laminin 2 is undetectable on Schwann cells in dy, dy^W or dy^3K homozygous mice (Gawlik et al., 2006; Nakagawa et al., 2001; Patton et al., 1997; Sewry et al., 1998; Sunada et al., 1994). In dy^2J homozygotes, laminin 2 levels on Schwann cells are decreased and what remains is distributed in a granular pattern, consistent with disruption of endoneurial BL structure observed by transmission electron microscopy (TEM) (Yang et al., 2005). By contrast, we found no quantitative decrease or qualitative disruption in the distribution of the mutant α2 chain in nmf417/nmf417 sciatic nerves compared to littermate controls (Fig. 2K,L). Qualitatively similar results were observed in nmf417/nmf417 spinal roots. Immature Schwann cells on the surface of axon bundles expressed laminin α2 with other BL components, including laminin β1, γ1 and nidogen (Fig. 2M,N). Thus, in nerve, as in muscle, the α2 C79R mutation had little or no effect on the secretion or incorporation of laminin 2 into the cell surface matrix. The C79R mutation is unique in separating loss of a specific α2 function from more general effects on laminin 2 expression or cell-surface-receptor interactions manifested in other dy-mutant strains.

Muscle and nerve structure were assessed at young adult ages (7-8 weeks). In normal muscles, myofibers generally had similar diameters, were closely associated with neighboring fibers and had myonuclei located immediately beneath the sarcolemma (Fig. 3A). In nmf417/nmf417 muscles, the myofiber population included small as well as large caliber fibers separated by monocyte-rich endomysial connective tissue, and many contained centrally located nuclei (Fig. 3B), which mark fibers that have regenerated postnatally in place of original embryo-derived fibers. Qualitatively similar defects were present in all homozygous nmf417 mice examined, in axial as well as limb muscles, and in muscles composed of predominantly fast fibers (plantaris, Fig. 3) and slow fibers (soleus, supplementary material Fig. S4). In direct comparisons, dystrophic changes in nmf417 mice were comparable to dy^2J mice and less severe than dy (Fig. 3C,D) [n=5 nmf417, 4 dy^2J, 3 dy homozygotes, and >10 littermate control mice examined; all at postnatal day (P)59]. The regenerative capacity of nmf417/nmf417 plantaris muscle was also assessed in these samples by counting overall fiber number and the percentage of central nuclei (supplementary material Fig. S5). In both nmf417 and dy^2J homozygotes, muscle-fiber number was not significantly decreased and a similar percentage of fibers showed central nuclei, an indicator of regeneration. By contrast, dy homozygotes showed reduced fiber number and a trend towards a lower percentage of fibers with central nuclei, indicating more-severely impaired regenerative capacity in the dy allele than in nmf417 or dy^2J. Thus, in all regards, the level of myodegeneration in nmf417 homozygotes was consistent with the less severe overt phenotype of this allele.

We assessed myelination in the spinal roots, in which axonal sorting is almost completely dependent on laminin 2. Defects in axonal sorting preserve embryonic organization, with immature Schwann cells bordering large `amyelinated' axon fascicles, and this defect is thus readily distinguished from demyelinating neuropathy. All previously reported Lama2 mutations cause severe amyelination in spinal roots. Similarly, we found that most axons in nmf417/nmf417 roots lacked ensheathing processes of Schwann-cells and were grouped in large bundles (Fig. 4A-D). Electron microscopy confirmed that axons in bundles were generally devoid of Schwann cell processes (Fig. 4F-G), as described in dy and dy^2J mice (Bradley and Jenkison, 1973; Bradley and Jenkison, 1975; Weinberg et al., 1975). The number and distribution of myelinated axons in nmf417 roots was similar to dy and dy^2J roots (n=4 nmf417, 4 dy^2J, 6 dy homozygous mice; age 59 days for the nmf417 and dy^2J, 5-9 months for dy; >10 littermate controls were examined). Thus, the dy, dy^2J and nmf417 mutations result in similarly severe forms of amyelination. Previous work indicates the internally deleted laminin 2 of dy^2J is essentially inactive for myelination, because residual myelination is completely dependent on laminin 8 (Yang et al., 2005). The phenotypic similarity suggests that the nmf417 α2-C79R mutation also inactivates laminin 2 for myelination.

We noted two differences in the axon–Schwann-cell relationship between nmf417, dy and dy^2J mutants. First, the extent of amyelination phenotype was consistent in nmf417/nmf417 mice but varied in dy^2J homozygotes (supplementary material Fig. S6). Second, the immature Schwann cells on the surface of bundles in nmf417 homozygotes typically adhered loosely to axons, and did not extend processes into bundles (Fig. 4E), a phenotype reminiscent of mice lacking Rac1 (Benninger et al., 2007; Nodari et al., 2007). By contrast, immature Schwann cells along dy and dy^2J bundles were tightly associated with axons, and often extended invasive processes into bundles, albeit without completing myelination (Fig. 4F,G) (n=4 nmf417/nmf417 mice, 4 dy^2J/dy^2J mice and 3 dy/dy mice; four littermate controls of the nmf417 mice were examined in nerve TEM studies; all mice were P59 except dy/dy which were 218±64 days old).

These data extend observations in dy^2J/dy^2J sciatic nerves (Yang et al., 2005) to suggest that Schwann cells in dy^2J/dy^2J roots retain a limited capacity to initiate radial sorting, but that progress towards myelination is highly susceptible to interruption. The dy^2J mutation causes a splicing defect that results in exon 2 being skipped. Variability in dy^2J/dy^2J roots could derive in part from variable efficiency of splicing between cells. Regardless, Schwann cells in nmf417/nmf417 roots appeared comparatively inactive even at the earliest steps of radial sorting.

Laminin-α2-related dystrophy and amyelination are associated with defects in BL formation (Madrid et al., 1975; Moll et al., 2001; Xu et al., 1994a). Using TEM, BLs are seen to consist of two main layers: the relatively unstained lamina rara immediately adjacent to the plasma membrane, and the heavily stained lamina densa (Fig. 5A). Consistent with previous reports, the lamina densa on muscle fibers in dy mutants was typically discontinuous, sparse and enmeshed with collagen fibrils from the collapse of the outer reticular lamina onto the plasma membrane (Fig. 5B). Unexpectedly, myofibers in nmf417 muscles possessed a continuous lamina densa that lacked collagen fibrils and was typically thicker than controls (Fig. 5C). Essentially identical results were observed in all muscle examined in each cohort (n=5 nmf417 mice, 3 dy mice, and 7 littermate controls, all P59).

We next investigated whether nmf417 Schwann cells also assembled BLs. In mammals, Schwann cells in contact with axons form BLs containing a continuous lamina densa that separates the abaxonal plasma membrane from the collagen-rich interstitial matrix (Fig. 6A). In direct comparisons, we found that Schwann cell BLs in dy^2J mice had a discontinuous lamina densa, and dy Schwann cells had very sparse BLs (Fig. 6B,C), consistent with levels of laminin 2 protein in these nerves. Interestingly, in nmf417 homozygotes, all myelinating Schwann cells possessed intact BLs that were indistinguishable from wild type (Fig. 6D). However, all non-myelinating Schwann cells in nmf417 mice (including premyelinating cells bordering amyelinated axons) lacked BLs entirely, similar to their counterparts in dy and dy^2J mice (Fig. 6E-G). Laminin 2 was expressed on premyelinating Schwann cells in nmf417 mice (Fig. 2M,N) but, despite its presence, these cells do not form a compact BL that is visible by TEM. Therefore, the C79R mutation does compromise the ability of laminin 2 to organize BLs on a subset of cell types. Importantly, the inability of the nmf417 mutant form of laminin 2 to establish stable interactions on immature Schwann cells brings the C79R mutation into line with previous interpretations regarding the mechanisms of laminin 2 in myelination (Yang et al., 2005). However, it also raises the issue of why the C79R mutation does not disrupt BL formation on myelinating Schwann cells, or on myofibers. BL stability is evidently modulated by additional factors that are differentially expressed by myelinating and non-myelinating Schwann cells.

Primary BL components other than laminins, including type IV collagens, nidogen and perlecan, were present at normal levels in nmf417 nerve and muscle as were the primary laminin receptors, the DGC and β1-integrins (Figs 7, 8). Combined with TEM (Figs 5, 6), these results indicate that the extracellular matrix of nmf417 muscle fiber contains an intact basement membrane of normal composition. Although alternative laminin α-chains containing an LN domain were primary candidates to maintain BL stability in nmf417 mice, we found no compensatory upregulation of α1, α3 or α5 (n=5 mutant and littermate controls each, at age 59 days). Interestingly, however, only a minority of myofibers in nmf417 mice expressed elevated levels of laminin α4. This result is unique among Lama2 alleles. Two manipulations reliably increase α4 expression: muscle injury leads to re-expression of α4 by immature regenerating myofibers (Patton et al., 1999), and all previously studied α2 mutations in mice and human MDC1A have been accompanied by marked and uniform increases in α4 laminin on both mature myofibers and myelinating Schwann cells (Nakagawa et al., 2001; Patton et al., 1999; Patton et al., 1997; Ringelmann et al., 1999). The ability of the α4 protein to compensate for the nmf417 mutation is doubtful, because α4 lacks the N-terminal LN domain. In normal development, α4 laminins are highly expressed by immature Schwann cells and myofibers (Patton et al., 1997), and are then downregulated perinatally (Ringelmann et al., 1999). In adults, α4 is normally undetectable in extrasynaptic myofiber BLs and low on myelinating Schwann cells, but remains high on non-myelinating Schwann cells (Patton et al., 1997; Previtali et al., 2003; Yang et al., 2005). The limited upregulation of α4 in nmf417 muscle is most consistent with transient expression by regenerating fibers. Thus, nmf417 is unique for not expressing α4 laminins on Schwann cells and myofibers in response to a loss-of-function α2 mutation. Of particular interest, we noted that high laminin α4 expression therefore correlates directly with loss of BL stability in nmf417, and also dy^2J, mice.

The results of the study of the Lama2^nmf417 allele allow us to draw several mechanistic conclusions. First, C79 in the CxxC motif of the laminin α2 chain is very specifically affected in the nmf417 allele, demonstrating it to be crucial for the onset of myelination by Schwann cells in spinal roots and for the stability of mature skeletal muscle fibers. The CxxC motif is invariant among laminin LN domains, and is shared by other LN-domain proteins, such as netrin, but its function has not previously been tested. Second, the resulting dystrophy and amyelination are not caused by decreased laminin α2 abundance, either through inefficient translation or instability; nor are they caused by protein mislocalization, a failure to assemble laminin trimers or by detectable changes in the composition of the extracellular matrix. Third, the C79R mutation does not disrupt the ultrastructural integrity of the BL on myelinating Schwann cells and muscle fibers. Each of these conclusions is distinct from all previously described loss-of-function mutations in laminin α2. Together, these new results argue strongly that the muscular dystrophy and amyelination resulting from laminin α2 mutations are not a simple derivative of the extent to which the BL is compromised or levels of laminin 2 are reduced. This conclusion is fully consistent with a previous study of laminins in myelination that found that BL integrity, per se, has little or no role (Yang et al., 2005). However, it differs from previous views regarding laminin-deficient muscular dystrophy.

The mechanisms leading to dystrophy in nmf417 homozygotes probably encompass much of the pathogenic mechanism in dy^2J homozygotes. Dystrophy in nmf417 and dy^2J homozygotes are remarkably similar in onset, progression and severity. C79, which is disrupted in nmf417, is included in the 57-residue deletion caused by the dy^2J mutation (Sunada et al., 1995; Xu et al., 1994b). However, the dy^2J deletion generates an array of molecular abnormalities, including decreased α2 protein levels and disrupted basal laminae, which obscures the relative contribution of each to the phenotype. Nmf417 is therefore remarkable for its phenotypic severity despite its molecular normality.

Because BL stability and levels of laminin 2 are not decreased in nmf417-mutant muscle, we infer that the highly conserved CxxC motif is involved in a more specific molecular interaction that is crucial for proper laminin α2 function. Candidate interactions include binding of heparan sulfate or integrins on the cell surface, activities that have been mapped to the LN domain in previous studies (e.g. Ettner et al., 1998). Alternatively, the CxxC motif might be required for proper higher-level organization of the laminin meshwork within the mature BL (Yurchenco, 1990; Yurchenco et al., 2004a). LN-domains are thought to scaffold BL assembly by forming tripartite interactions between the α, β and γ chains of neighboring laminin heterotrimers. The CxxC motif could selectively disrupt pairings within the tripartite complex, permitting the assembly of a BL with an abnormal lattice. This interpretation is consistent with the increased BL thickness we often observed on nmf417 mutant myofibers. The dy^2J mutation, which decreases overall polymer stability and BL structure, might therefore reflect a more complete loss of tripartite-complex formation.

Distinguishing these possible mechanisms will require additional experiments to study laminin polymerization, and to identify LN-domain binding partners and how these are altered by the C79 disruption. Ideally, these studies could be combined with structural analyses to determine whether C79 is involved in disulfide crosslinking, in which case the dysfunction could arise from gross alteration in N-terminal conformation. Future studies in which the nmf417 allele is used to investigate laminin 2 polymerization are also appealing for two reasons. First, we have shown that other polymerizing α chains (α1, α5) are not upregulated in nmf417 homozygous mice to compensate for the mutation in α2. Indeed, α1 can substitute for α2 to prevent dystrophy and improve peripheral myelination, suggesting that the phenotype results from general laminin interactions or activities such as polymerization in the BL rather than activities that are specific to α2 (Gawlik et al., 2004; Gawlik et al., 2006). Second, the non-polymerizing α4 chain, which does not contain an N-terminal extension and LN domain, is also not upregulated. This aspect of nmf417 is unique among mouse and human Lama2 mutations (Patton et al., 1999). The lack of α4 upregulation is significant for interpreting polymerization studies because laminin 8 (α4β1γ1) might actually inhibit laminin polymerization in places where it is colocalized with laminin 2, when laminin 2 function is compromised. This competitive relationship is suggested by previous studies of laminins in myelination (Yang et al., 2005), and by the present study, in which we showed an absence of BL surrounding immature non-myelinating Schwann cells that express α4. At NMJs, another site of α4 localization, a BL still forms in mice homozygous for other Lama2 alleles examined (Gilbert et al., 1973; Law et al., 1983; Patton et al., 2001), but this is presumably owing to the presence in the synaptic BL of α5, which serves as an alternative polymerizing α chain (Miner et al., 1997).

Mechanical properties of the BL, such as torsional flexibility or elasticity, might also be lost in nmf417. Such properties might relate to either laminin polymerization or other molecular interactions between BL components or between the BL and cell surface (Yurchenco et al., 2004b). A function for laminin 2 in maintaining sarcolemmal integrity during mechanical stress is attractive, and is consistent with mutations in other DGC genes that also cause dystrophy. However, Lama2 mutations reportedly do not cause consistent sarcolemmal-integrity defects, as assayed by uptake of the azo-dye Evan's Blue, although scattered positive fibers are present in muscles and vary with age and severity of the allele studies (Straub et al., 1997) (B.L.P. and R.W.B., unpublished observations). Furthermore, a function in mechanical stabilization is more plausible in a contracting muscle than in the myelination of nerves during development. The failure of myelination is a consistent feature in human MDC1A and Lama2-mutant mice, but is not seen in most other dystrophy-causing mutations. Laminin 2 might be playing distinct roles in nerve and muscle BLs, but it is interesting that both of these functions would still depend on C79 and the LN domain of α2.

Nmf417 also offers several points of relevance for human MDC1A. First, loss-of-function LAMA2 mutation can occur without loss of α2 protein. Some forms of MDC1A might therefore be misdiagnosed by immunocytochemistry, and ultimately require analysis of the entire LAMA2 coding sequence for their correct diagnosis. Second, nmf417 represents a reference for point mutations that perturb specific domain functions, without overall loss of expression or BL integrity. Several human LAMA2 point mutations cause MDC1A, including L2564P, C862R and C527Y (He et al., 2001; Tezak et al., 2003), although none are in the LN domain. Third, nmf417 might prove particularly useful in defining the unique contribution of laminin 2 to myelination. It will also be of great interest to determine whether defects in CNS myelination in human MDC1A and in dy^2J mice are replicated in nmf417. Finally, nmf417 offers advantages over dy^2J as an experimental model. Mutant mice are readily genotyped as pups and can be bred as homozygotes, produce a defined mutant protein and present with consistent phenotypic severity, all of which favor a reliable platform for interpreting therapeutic strategies.

Therefore, although many questions remain unanswered, we suggest that the phenotype of the nmf417 allele indicates a crucial function for the LN domain, and more specifically the conserved C79, in laminin biology. Furthermore, nmf417 mice provide an in vivo model for these studies without confounding issues such as decreased protein levels or the upregulation of other laminin α chains. Thus, the nmf417 allele is a useful and convenient experimental model of MDC1A, which suggests new mechanistic relationships between BL formation, muscle fiber stability and Schwann cell differentiation.

The mice referred to in this study have the following official nomenclature: nmf417 is Lama2^dy-7J; dy^2J is Lama2^dy-2J; dy is Lama2^dy; dy^W is Lama2^tm1Eeng; and dy^3K is Lama2^tm1Stk. All mice used were in a C57BL/6J genetic background, except dy, which were on a mix of C57BL/6 and 129P1/Re.

The nmf417 allele was generated in a mutagenesis screen for recessive neurological mutations. Male C57BL/6J (B6) mice were mutagenized using ENU, allowed to recover fertility and bred to B6 females. Male offspring in generation 1 (G1), each carrying a unique assortment of mutations, were mated to B6 females to generate G2 females, which were bred back to the G1 male ancestor, producing homozygous recessive mutations in G3. G3 offspring were screened for neurological phenotypes using a battery of behavioral, clinical and histological analyses. The nmf417 strain was initially identified based on overt hind-limb wasting and dysfunction, followed by histological characterization revealing skeletal muscular dystrophy and peripheral hypomyelination, and complementation with a known Lama2^dy allele heterozygote.

All mice were housed in PIV caging on a 12/12-hour light/dark cycle, and provided with food and water ad libidum. Protocols and procedures were designed in accordance with the Guide for the Care and Use of Laboratory Animals, and approved by the Animal Care and Use Committees of The Jackson Laboratory and Oregon Health Science University. Numbers and ages of mice used for each experiment are stated throughout the text.

Trizol (Invitrogen) was used according to the manufacturer's instructions to prepare total RNA from the brains of a B6 control, a known nmf417 heterozygote and an affected (presumed homozygote) nmf417 mouse. 5 μg total RNA was reverse transcribed to first-strand cDNA using Superscript III reverse transcriptase (Invitrogen) and a mix of oligo-dT and random primers. The Lama2 coding sequence was amplified by PCR in 12 overlapping reactions of approximately 800 bp each, and the products sequenced directly on both strands and compared to sequence from the mouse genome project (www.Ensembl.org) and from isogenic control mice using the Sequencher analysis program.

The nmf417 allele was genotyped using a PCR primer-mismatch assay in which a common reverse primer to Lama2 intron2 is paired with wild-type and mutant forward primers in which the single mutated base is the second-to-last (penultimate) nucleotide. Primer sequences: nmf417 F Wild-Type, 5′-CTGTGAGGAACCCTCAGTG-3′; nmf417 F Mutant, 5′-CTGTGAGGAACCCTCAGCG-3′; nmf417 R Intron2, 5′-CATTTCAGGACCTGTGTTGA-3′. Reaction conditions, using HotMaster Taq Polymerase (Eppendorf): wild-type specific reaction, 95°C for 5 minutes; 95°C for 25 seconds, 62°C for 35 seconds, 72°C for 45 seconds, for 30 cycles, and 72°C for 5 minutes; mutant specific reaction, 95°C for 5 minutes; 95°C for 25 seconds, 64°C for 35 seconds, 72°C for 45 seconds, for 38 cycles, and 72°C for 5 minutes.

Sources for primary antibodies were: laminin α1 mAb-198, provided by L. Sorokin (University of Münster, Germany); laminin α2 mAb 4H8-2 from Alexis Pharmaceuticals (San Diego, CA); affinity purified rabbit polyclonal Abs to the α2 G-domains provided by Peter Yurchenco (Robert Wood Johnson Medical School, Piscataway, NJ) and Takako Sasaki (#1078+E; Shriners Hospital, Portland, OR); laminin β1 (MAB1928), laminin γ1 (MAB1914), nidogen-1 (MAB1946), perlecan (MAB1948) and collagen IV (AB756P) all from Chemicon (Temecula, CA); rabbit antibodies to laminin α5 (1113+) and γ3 (1138+) from T. Sasaki (OHSU, Portland, OR); rabbit laminin 5 antisera (including the α3, β3 and γ2 chains, from M. P. Marinkovich (Stanford University, Palo Alto, CA); affinity-purified rabbit antibodies to agrin from M. Ferns (McGill University, Montreal, Quebec) and D. Glass (Regeneron, Tarrytown, NJ). Additional antisera to laminin β2, α4 and α5 are described (Miner et al., 1997). Alexa-fluorochrome-conjugated second antibodies were from Molecular Probes (Eugene, OR).

For histology, muscles were dissected free, immersed using Bouin's fixative, embedded in paraffin, microtome sectioned at 5 μm and stained using Hemotoxylin and Eosin (H&E) by standard protocols. Nerve tissue was dissected free; fixed in 2% glutaraldehyde, 2% paraformaldehyde in 0.1 M cacodylate; post-fixed in 1% OsO₄; embedded in plastic; sectioned at 0.5 μm; and stained with Toluidine Blue for light microscopy. For TEM, 50- to 100-nm plastic sections of nerve and muscle were examined on a Jeol 1230 80 kV electron microscope and imaged with a Hamamatsu digital camera. For immunohistochemistry, unfixed tissues were flash frozen in liquid-N₂ cooled freezing compound and cryosectioned at 8 μm. Prior to freezing, normal and mutant nerves were sandwiched in parallel between pieces of diaphragm, to ensure identical processing.

Muscle extracellular matrix fractions were prepared similar to methods previously described for kidney (Miner et al., 1997). Tissue was homogenized at 0-4°C in five volumes of low-ionic-strength buffer containing protease inhibitors. A crude membrane fraction (30 minutes at 30,000 g) was washed once, resuspended in three volumes of buffer, supplemented with 1.0 M NaCl and 1% Triton X-100 and sonicated 90 seconds, and centrifuged for 30 minutes at 100,000 g. Protein content of matrix-rich insoluble pellets was determined by a detergent-compatible Lowry assay (Peterson, 1977) with BSA as standard. Samples of control and nmf417 matrix fractions containing equal protein content were heated in SDS sample buffer without reducing agents, loaded into neighboring slot wells (5 μg/mm) of Laemmli polyacrylamide gel, and separated as described (Miner et al., 1997). Nitrocellulose transfers were blocked with Tween-20 and parallel strips were cut from each sample area, numbered and incubated overnight in a single set of diluted antibodies. Reducing westerns for α2 quantification were performed as above, except that tissue homogenates were used to avoid potential differences in matrix stability; 100 mM DTT was added to the gel sample buffer. Equal amounts of total protein were loaded and blots were probed with antibodies against α2 (pAb 1078+E), and the same blots were reprobed with an antibody against myosin heavy chain, which served as a normalization control for sample loading. Bound antibodies were detected with HRP-conjugated second antibodies, chemiluminescent substrate (Supersignal WestFemto, Pierce, Rockford, IL) and Kodak X-OMAT film.

All statistical comparisons were done using two-tailed t-tests and a significance threshold of P<0.05 was used.

We thank Takako Sasaki, Lydia Sorokin and Peter Yurchenco for generously providing antibodies; Kate Miers, Steve Rauch, Joel Gay and Janis McFerrin for technical assistance; Gregory Cox for comments on the manuscript; and Pete Finger and the Histology and Microscopy service at The Jackson Laboratory, which is supported by the NCI Cancer Center (CA034196). Work was supported by grants from the Muscular Dystrophy Association and NIH (NS040759) to B.L.P. and by NIH (UO1-NS41215) for the mutagenesis program at The Jackson Laboratory.