Laminins promote early stages of peripheral nerve myelination by assembling basement membranes (BMs) on Schwann cell surfaces, leading to activation of β1 integrins and other receptors. The BM composition, structural bonds and ligands needed to mediate this process, however, are not well understood. Mice hypomorphic for laminin γ1-subunit expression that assembled endoneurial BMs with reduced component density exhibited an axonal sorting defect with amyelination but normal Schwann cell proliferation, the latter unlike the null. To identify the basis for this, and to dissect participating laminin interactions, LAMC1 gene-inactivated dorsal root ganglia were treated with recombinant laminin-211 and -111 lacking different architecture-forming and receptor-binding activities, to induce myelination. Myelin-wrapping of axons by Schwann cells was found to require higher laminin concentrations than either proliferation or axonal ensheathment. Laminins that were unable to polymerize through deletions that removed critical N-terminal (LN) domains, or that lacked cell-adhesive globular (LG) domains, caused reduced BMs and almost no myelination. Laminins engineered to bind weakly to α6β1 and/or α7β1 integrins through their LG domains, even though they could effectively assemble BMs, decreased myelination. Proliferation depended upon both integrin binding to LG domains and polymerization. Collectively these findings reveal that laminins integrate scaffold-forming and cell-adhesion activities to assemble an endoneurial BM, with myelination and proliferation requiring additional α6β1/α7β1-laminin LG domain interactions, and that a high BM ligand/structural density is needed for efficient myelination.
In peripheral nerve, basement membranes (BMs) play a critical role in the maturation of Schwann cells by promoting radial axonal sorting leading to myelination (Bunge et al., 1986; Webster, 1971; Webster et al., 1973). The process is dependent upon β1 integrins that bind to BM ligands to mediate sorting and Schwann cell proliferation (Benninger et al., 2007; Berti et al., 2011; Nodari et al., 2007; Pereira et al., 2009). BMs also bind to α-dystroglycan and α6β4 integrin; however, these mediate later stages of myelination, affecting nodal development and myelin-sheath stabilization (Berti et al., 2011; Nodari et al., 2008; Saito et al., 2003). Finally BMs may tether heparin-binding growth factors such as soluble neuregulin (Ma et al., 2011). Despite this information, little is actually known about the critical receptor ligands and structural contributions of BMs that are required for the progression of stages leading to myelination.
Laminins are a family of glycoproteins that along with nidogens, type IV collagen and the proteoglycans perlecan and agrin constitute the major components of BMs. Laminins are required to initiate BM assembly (Smyth et al., 1999). Assembly depends upon laminin adhesion to cell surfaces, polymerization, and binding to nidogens and agrin (Li et al., 2005; McKee et al., 2007; Yurchenco, 2011). Laminins also mediate BM-dependent functions by interacting with integrins and dystroglycan (reviewed by Li et al., 2002). The laminin requirement poses a challenge for dissecting BM contributions given the complexity of ligands within BMs and the relative inability to distinguish laminin from non-laminin contributions.
Schwann cell laminins share the γ1 subunit in the principal endoneurial laminin-211 (Lm211, α2β1γ1 subunit composition) and the minor laminins 411 and 511. Inactivation of LAMC1 gene expression or combined inactivation of the LAMA2 and LAMA4 genes each caused a severe peripheral neuropathy characterized by amyelination and reduced proliferation (Chen and Strickland, 2003; Yu et al., 2005; Yang et al., 2005). Inactivation of the β1 integrin gene in Schwann cells caused substantial peripheral nerve disease, particularly severe in a C57Bl6 background, similar to the LAMC1 Schwann cell knockout phenotype (Berti et al., 2011).
We now describe a mouse with intermediate-level laminin γ1 subunit expression with an axonal sorting defect not accompanied by reduced Schwann cell proliferation, the latter attribute distinct from the LAMC1 laminin knockout and other laminin-deficiency states. This defect was characterized by hypomyelination, alterations of Schwann cell myelination factors, and accompanied by muscle atrophy and dystrophy. Investigation of the role of laminins and their functional domains in cultured dorsal root ganglia (DRG) rendered incompetent to secrete γ1 laminins revealed that laminin cell surface anchorage and polymerization are important for BM assembly and that laminin concentration, polymerization, and β1-integrin interactions differentially contribute to promote proliferation, ensheathment and myelination.
Laminin hypomorphic mice
LAMC1flneo/− (abbreviated flneo/–) mice were indistinguishable from their LAMC1flneo/+ (abbreviated flneo/+) littermates until 3 weeks of age. When lifted by the tail, the flneo/+ mice extended their hindlimbs outwards. In contrast, hindlimbs of flneo/− mice were slightly tremulous and retracted inwards. By 8-10 weeks of age, the affected mice exhibited sustained hindlimb contractures and reductions of muscle mass in the lower back and hindlimbs with difficulty in ambulation. The possibility that the gait abnormality might be due to the presence of the neo selection cassette was supported by the finding fl/− offspring lacking the cassette did not develop a gait abnormality and appeared normal in their behavior throughout life.
Morphology of adult peripheral nerve
Sciatic nerves and other tissues were dissected from flneo/− mice and their flneo/+ littermates (10-24 weeks) and examined by microscopy (Fig. 1; supplementary material Fig. S1). Methylene blue-stained semi-thin sections of sciatic nerve and nerve roots of flneo/− mice exhibited multiple pale-staining areas corresponding to amyelinated axons compared to littermate controls (Fig. 1A-D). Larger amyelinated areas were found in the flneo/− nerve roots (Fig. 1C,F). The ultrastructure of the sciatic nerves was examined in adults (Fig. 1G,H). Compared to littermate controls, flneo/− nerves had a twofold reduction of myelinated axons and near-complete absence of enveloped axons (Remak bundles) with many naked axons (Fig. 1; supplementary material Fig. S2A). This was accompanied by a sevenfold increase in Schwann cells that do not myelinate, mostly located adjacent to naked axons. LAMC1 flneo/− nerves were hypomyelinated with increased axon/myelin+axon ratios. The thickness of adult Schwann cell BMs was found to be the same.
Sciatic nerve ultrastructure was also examined at post-natal day 5 (P5) during the stage of active axonal-sorting (supplementary material Fig. S3). Schwann cell lamellipodial processes (LP) were noted to extend into and divide bundles of axons, and to envelop axons and axon groups, in both LAMC1flneo/+ and flneo/− sciatic nerve. Longer processes extending from Schwann cell bodies were often seen to encompass groups of axons in which the BM was always located on the opposite side as the axons (lower panel), reflecting a polarity of structure. Scattered myelinated axons were present as well. However, the naked axon bundles were larger in mutant nerve with less extensive penetration and separation by Schwann cell processes. Schwann cell BMs possessed frequent discontinuities along the LP surfaces and portions of Schwann cell bodies in the flneo/− nerves in contrast to flneo/+ nerves. The BMs of myelinated axons (MA), on the other hand, were essentially continuous in control and mutant nerves. Length measurements of LP BMs relative to the underlying plasma membranes revealed a two-fold decrease of BM.
The nerve and muscle phenotype suggested a laminin deficiency state. RNA extracts from the nerve, muscle, and other organs were measured by quantitative real-time RT-PCR (Fig. 2A; supplementary material Fig. S1A,B). There was a reduction of laminin-γ1 mRNA isolated from flneo/− mice compared to that of wild-type (+/+) mice (36% average relative to wild type), implicating the floxed-allele paired with a null-allele as the basis of reduced laminin expression. Since a lesser degree of reduction of mRNA was observed for mRNA isolated from LAMC1+/− mice (average of 67%) that have no evidence of disease, it suggests that the threshold for development of a phenotype from a LAMC1 mRNA exists between 36 and 67% of wild type. Somewhat surprisingly, the decrease in mRNA from flneo/flneo relative to wild-type mice was almost negligible, suggesting an enhancement of mRNA expression may result from an interaction between the two identical alleles not present in flneo/−. Removal of the neo cassette from the floxed allele alone was sufficient to restore mRNA expression to levels approaching those of the wild-type state (supplementary material Fig. S1A), implicating the presence of the neo cassette cis to the active allele as the cause of reduced expression. This is in keeping with published reports that the neo cassette (including that of the LAMA5 gene) can be a cause of reduced neighboring gene expression (Meyers et al., 1998; Shannon et al., 2006). The mRNA level for laminin β1, γ3, α2 and α4 (nerve, muscle), and α5 (kidney and lung) was found not to be different for flneo/+ and flneo/− littermates (supplementary material Fig. S1Bc).
Basement membrane protein in peripheral nerve and other tissues
Immunostaining of adult sciatic nerve revealed similar BM-pattern reductions of laminin subunits γ1, β1, α2, α4, γ3 (slight), collagen-IV and perlecan (Fig. 2B). Integrin subunits α6 and β1 and dystroglycan changed little if at all. Since the adult BM structural components in nerve were less bright by immunostaining without recognizable changes in ultrastructure, it was concluded that the endoneurial BMs possessed a lower density of these components. The reduction in these laminin-associated components is predicted given formation of a BM is dependent upon laminin expression (Li et al., 2002; Li et al., 2005; McKee et al., 2007). Reductions of BM components, but not integrin or dystroglycan, were noted for hindlimb muscle (quadriceps); however, these reductions were not as great as those noted in peripheral nerve (supplementary material Fig. S1C). Slight reductions of Lmγ1 were noted in lung and kidney glomeruli with no reductions appreciated in colon (supplementary material Fig. S1D,E). Hindlimb nerve at E18.5, P1 and P5 revealed differences similar to those in adults (supplementary material Fig. S4). The relative lack of laminin reductions in non-neuromuscular tissues, despite comparable reduced Lmγ1 mRNA, might be due either to a higher transcriptional efficiency or to high levels of γ1 subunit production relative to the other subunits. The latter possibility is suggested by the finding that secretion of laminins is limited by the subunit synthesized in least amount (Yurchenco et al., 1997).
Schwann cell proliferation and myelination factor expression
While there is diminished Schwann cell proliferation in the conditional Schwann cell LAMC1 gene knockout (Yu et al., 2005), reduced Schwann cell proliferation was not detected in hypomorphic nerve, even at the time of peak proliferation that occurs just before and at the time of birth (Fig. 3A,B with plots in Fig. 3G,I). The mRNA levels for Schwann cell-factors involved in myelination were determined by quantitative RT-PCR in adult and P5 sciatic nerve (Fig. 3C,D). Elevations of the transcription factors cJun, Oct6 and Sox2 and reductions of Krox20, the p75 neuregulin receptor, and myelin-protein-zero (MPZ) were detected in mutant nerve.
Muscular atrophy and dystrophy
Hindlimb muscle mass was observed to gradually become reduced in LAMC1flneo/− mice. The quadriceps were abnormal in that they were populated by scattered small myotubes and central nuclei in the absence of obvious endomysial fibrosis at the light microscopic level, although a slight but distinct increase in interstitial collagen fibrils was appreciated by ultrastructure (supplementary material Fig. S5). The sarcolemmal BMs appeared normal. Quantitation of the muscle diameter revealed significantly smaller muscle diameter (atrophy) in flneo/− mice.
Induction of myelination in dorsal root ganglion cultures with recombinant laminins
In order to better understand mechanisms underlying the nerve phenotype resulting from reduced and absent laminin expression, an analysis using laminin-deficient cultured dorsal root ganglia cultures treated with wild-type and modified recombinant laminins was conducted.
Laminin sulfated glycolipid-binding (e.g. sulfatides), integrin-binding, dystroglycan-binding, polymerization, and nidogen-binding activities map to different structural domains and have been selectively inactivated through engineered deletions or point mutations (Fig. 4) allowing for loss-of-function analysis (McKee et al., 2009; McKee et al., 2007; Smirnov et al., 2002). In addition, laminin-binding linker proteins that fuse laminin domains to nidogen domains and modified agrin allow for gain-of-function analysis (McKee et al., 2009). The C-terminal FLAG tag, present in two of the laminins, consists of eight nearly all charged amino acid residues that are located four residues downstream of a glutamic acid required for integrin interactions with laminin LG domains and that similarly impede integrin binding (Ido et al., 2007).
Previous studies of Schwann cells cultured with sensory neurons of rat dorsal root ganglia (DRG) have shown a role of BMs in myelination (Ard et al., 1987; Clark and Bunge, 1989; Eldridge et al., 1989). The presence of endogenous laminin production and baseline myelination by Schwann cells limits this approach for further dissection of laminin contributions to myelination, particularly when BM assembly is induced with ascorbate (Eldridge et al., 1989; Eldridge et al., 1987). It was found that this limitation can be overcome with DRGs isolated from conditional knockout mice treated with adeno-cre-recombinase virus to prevent expression of the laminin γ1 subunit common to Lm211, Lm411 and Lm511 (Päiväläinen et al., 2008; Yu et al., 2009), and in which BM assembly, ensheathment and myelination are then initiated by the addition of exogenous laminin to the culture medium. To analyze the contributions of laminin concentration and laminin-domain activities to these processes, the laminin-deficient DRGs were treated with recombinant laminins that are wild type or that possess structural modifications and that have been characterized by rotary shadowing, self-assembly, binding interactions and BM assembly on isolated Schwann cells (McKee et al., 2009; McKee et al., 2007; Smirnov et al., 2002; and data not shown). While Lm211 is the principal laminin of peripheral nerve, Lm111 has been found to restore myelination in nerve following transgenic expression of the Lmα1 subunit in Lmα2-null (dy3K) mice, indicating the laminin possesses the necessary domains for proper function (Gawlik et al., 2006). The evaluation was largely conducted with Lm111 and supported with Lm211 comparisons.
The cultured DRG explants consist of neuronal cell bodies concentrated in the central region that extend out axonal-like processes with Schwann cells resting on and between these processes, and with only a few scattered fibroblasts detected. In the absence of added laminin, very little if any laminin was detected by immunostaining with antibody to Lmα2 (the principal α-subunit) from the time of virus infection to six days after addition of ascorbate (supplementary material Fig. S6A). Pilot experiments with added Lm111 revealed that peak proliferation occurred about the time of addition of ascorbate and that myelination developed after ascorbate addition (data not shown). The ascorbate induction is a consequence of increased secretion of stable type IV collagen that presumably also induces laminin (Eldridge et al., 1987). Various laminins along with recombinant nidogen-1 (28 nM, to promote binding to type IV collagen) were added to the medium of the DRGs starting at the time of viral addition and extending through the switch to ascorbate-containing medium to induce myelination. Proliferation [estimated by the segmented ratio of 5-ethynyl-2′-deoxyuridine (EdU) incorporated into nuclei undergoing DNA synthesis to DAPI-stained nuclei] occurred at low levels in the absence of laminin and increased with added rLm111; however, a plateau was approached above 3.5 nM (Fig. 5A-E,K,L). In contrast, myelination (measured by the segmented ratio of MBP/NFL+gfp, or by total myelinated segment lengths/DRG area; NFL, neurofilament-200) was found to increase with increasing concentrations of exogenous rLm111 over the range examined (Fig. 5F-J,K,M; supplementary material Fig. S6B).
Schwann cell adhesion and integrin binding
The cell adhesion attributes of DRG-derived Schwann cells with respect to laminins and other BM components were examined with Schwann cells isolated from DRG cultures (Fig. 6). Schwann cells adhered to, and spread on rLm111 and collagen-IV. Additional weaker adhesion and spreading were observed for miniagrin (which contains the cell-adhesive LG domains), and only poor adhesion was observed with nidogen-1. rLm111 lacking the first three LG domains (rLmΔα1LG1-3) required for integrin binding (Nishiuchi et al., 2006) did not support adhesion at all. rLm211 also supported Schwann cell adhesion and spreading; however, a higher coat concentration was required compared to rLm111 (Fig. 6B). rLm111 or rLm211 bearing a Lmγ1 FLAG tag exhibited reduced cell adhesion compared to their normal counterparts. Binding of soluble integrins to rLm111 and rLm211 revealed that α7X2β1 was substantially reduced when either laminin possesses a C-terminal FLAG tag and that, in the case of rLm111, the FLAG tag conferred weaker binding to integrin α6β1. Furthermore, α3β1 was found to be the integrin that binds miniagrin.
The ability of these recombinant laminins to induce myelination and form a BM was examined in the virus-induced laminin-deficient DRGs (Fig. 7). Myelination was found to be extensive following treatment with 28 nM wild-type rLm111 and, to a somewhat lesser degree, with rLm211. However, myelination was considerably reduced when DRGs were incubated with FLAG-tagged rLm111F and rLm211F that were found to bind poorly to β1 integrins that normally bind to laminin LG domains. Deletion of the proximal three LG domains of rLm111 (rLmΔα1LG1-3), essential for integrin binding to laminin LG domains and also containing an LG2 sulfated-glucuronate-binding activity, prevented myelination. Incubation of DRGs with rLmΔα1LG1-3 coupled to miniagrin (binds strongly to the laminin coiled coil and interacts with cell surfaces through sulfatides, dystroglycan and α3β1 integrin) generated a slight degree of myelination. Myelination was nearly completely absent when DRGs were treated with rLm111 bearing a deletion of the distal LG4-5 domains that mediates dystroglycan, sulfatide and heparin binding (Harrison et al., 2007).
Myelination was nearly absent following treatment of DRGs with rLmΔα1LN-L4b, a laminin unable to polymerize and that lacks a secondary potential locus of cell interaction [α1LN domain binds to sulfatides, α1β1/α2β1, and heparin (Colognato-Pyke et al., 1995; McKee et al., 2009; McKee et al., 2007)]. Given that earlier studies with Schwann cells revealed a sulfatide, but not a heparan sulfate dependency of BM assembly (Li et al., 2005), and our inability to detect a contribution to Schwann cell adhesion in the absence of LG domains, it was presumed that the critical lost activities in rLmΔαLN-L4b were polymerization and possibly sulfatide binding. However, a comparable reduction of myelination was observed with rLmΔβ1LN-LEa, whose deleted domains are associated only with polymerization (McKee et al., 2009; McKee et al., 2007), confirming that ablation of polymerization alone is sufficient on its own to prevent myelination. Furthermore, evidence for a key polymerization role was provided by the observation that myelination could be significantly restored by αLNNd, a fusion protein that binds to the nidogen-binding locus and provides the critically missing α1LN domain to rLmΔα1LN-L4b (but not to rLmΔβ1LN-LEa that lacks the β1LN rather than the α1LN domain) (McKee et al., 2009). Finally, it was found that DRG myelination was reduced, albeit slightly, with a laminin (rLm111NS) unable to bind to nidogen.
Addition of rLm111 to the DRGs led to increased accumulation of laminin subunits (exogenous) as well as endogenously derived Schwann cell type IV collagen and perlecan (Fig. 7C). Thus it appears that multi-component BMs, and not just laminin ECMs, had assembled. Finally, comparisons of rLm211F with rLm211F and rLmΔαLN-L4b with rLm111 revealed that presence of the integrin-reducing modification results in a decrease in proliferation while absence of polymerization reduces laminin induction of proliferation to baseline (laminin-free) levels (Fig. 7D).
DRGs treated with laminins was examined by electron microscopy (Fig. 8; supplementary material Fig. S6E-I). In the absence of added laminin, the Schwann cells lacked BMs on their surfaces. Axons were adjacent to Schwann cells but were naked in their appearance, either in contact with Schwann cell plasma membranes or near them. At the lowest concentration of added rLm111 (3.5 nM), substantial but incomplete BM coverage on Schwann cells was noted characterized by BM discontinuities of variable lengths. Near-complete BM coverage were noted for the two wild-type laminins at 28 nM. The extent of BMs coverage of Schwann cells was found to increase as a function of rLm111 concentration. A plateau was reached by 7 nM and corresponds to the peak of proliferation. Treatment with rLm111F or rLm211F did not reduce BM. In contrast, deletion of the distal LG1-3 domains, and especially deletion of the LG1-3 domains, resulted in decreased BM. These decreases might be attributed to reduction of cell-surface sulfatide binding (LG4), dystroglycan binding (LG4), integrin binding (LG1-3) and/or sulphated glucuronate (LG2) (Hall et al., 1995; Harrison et al., 2007; Nishiuchi et al., 2006; Sung et al., 1993). However, since reduction of LG–integrin binding or dystroglycan has not been found to affect BM assembly in other studies, reduction of sulfatide-binding activity in LG4 and loss of sulfated-glucuronyl-binding activity in LG2 may account for BM reductions for LG4-5 and LG1-3 deletions (Feltri et al., 2002; Li et al., 2002; Saito et al., 2003). BM was restored with miniagrin with only a partial restoration of myelination.
Laminins lacking short-arm segments (LmΔα1LN-L4b, LmΔβ1LN-LEa) with α1LN and β1LN polymerization domains reduced BM cell coverage. BM was increased (but not to normal levels) when αLNNd was added to LmΔα1LN-L4b, but not to LmΔβ1LN-LEa. Previous analysis with modified rLm111 has shown that an absence of LN domains causes a failure of polymerization and BM assembly and that the polymer requires an αLN, a βLN and a γLN domain (McKee et al., 2009; McKee et al., 2007). Thus αLNNd provides the missing αLN domain to LmΔα1LN-L4b to enable polymerization, but fails to provide the β1LN domain missing in LmΔβ1LN-LEa. Finally, laminin bearing a point mutation (Lmγ1N802S) that ablates nidogen-1 binding resulted in BMs with only slight reduction of coverage. This suggests that nidogen mediation of high affinity bridging of laminin to type IV collagen is less critical than polymerization for maintaining a functional endoneurial BM (Fox et al., 1991).
Axonal ensheathment was also examined (supplementary material Fig. S6E-I) with respect to the degree of bundling and segmentation within Schwann cells to gauge the degree of axonal sorting (see Webster, 1993) for absent laminin, low and high rLm111, rLm211F, laminin-111 without LG4-5, and non-polymerizing laminin. The axons of DRGs not treated with laminin were nearly all naked (some in contact with Schwann cell surfaces) and often in large bundles. The axons of DRGs treated with 28 nM wild-type rLm111 were mostly segmented and in 1∶1 complexes. All other conditions examined lay between the two extremes. rLm211F had a considerably higher fraction of naked axons compared to the wild type with a distribution similar to non-polymerizing laminin and laminin lacking LG domains.
The study reveals that a partial reduction of laminin-γ1-subunit expression results in BMs with reduced levels of structural components (hence less dense) and a Schwann cell axonal sorting defect, but without reduction of Schwann cell proliferation. In addition, increases of transcription factors (cJun, Oct6) that can contribute to hypomyelination were noted (Svaren and Meijer, 2008). The hypomorphic state was unusual in that phenotypic changes were restricted to nerve and muscle, the latter mild, despite a general reduction of laminin subunit transcription. Nerve and muscle γ1-subunit deficiency lacked compensatory increases of other laminin subunits, in particular the Lmγ3 subunit that is capable of joining with the same subunits as Lmγ1. The phenotype in nerve was particularly severe, resulting in amyelination at levels comparable to the dy2J mouse. The selective defect resulting from a general transcriptional reduction suggests that a regulatory decrease in either LAMC1 or LAMB1 gene transcription could be a cause of selective peripheral neuropathy and myopathy in humans.
During peripheral nerve development, axonal sorting is normally accompanied by Schwann cell proliferation, the latter needed to insure availability of sufficient numbers of Schwann cells to wrap individual axons and with both dependent upon the activity of small GTPases (Benninger et al., 2007). It is unusual that in the laminin hypomorph there was a preservation of normal Schwann cell proliferation in the face of defective radial sorting, breaking the coupling, and leaving what may be inactive Schwann cells of indeterminate differentiation state scattered adjacent to and near naked axon bundles. Furthermore, the persistent high expression of Oct6, cJun, and Sox2 and low expression of Krox20 and MPZ as seen in the laminin-deficient nerves may reflect a deficit in the maturation of the Schwann cell population to myelinating cells (Svaren and Meijer, 2008). The high expression level of Oct6 detected was similar to that reported in transgenic mice overexpressing Oct6, with hypomyelination present in both conditions (Ryu et al., 2007).
The normal Schwann cell proliferation of the hypomorph differs from either the LAMC1 null or single and combined α-subunit deficient states in peripheral nerve (Yang et al., 2005; Yu et al., 2005). It was thought that the reduced BM density of the hypomorphic mouse (continuous BMs with reduced immunostaining intensity for all structural components) might represent an intermediate state in which there was sufficient BM for proliferation but not for myelination. This possibility and a question of what laminin-domain-specific activities are required for these functions led to an examination of myelination and proliferation in laminin-deficient DRGs. The comparison revealed that the extent of BM accumulation on Schwann cell surfaces, Schwann cell proliferation, axonal ensheathment and axonal myelination was dependent on the concentration of laminin in which high laminin concentration, corresponding to high laminin expression in vivo, is more important for achieving efficient myelination than Schwann cell proliferation. The laminin-induced increase in proliferation was found to partially depend upon a contribution for the β1-integrin-binding site (especially α7β1 and α6β1) in the LG domains and substantially depend upon the polymerization activity of laminins. The latter effect may be an indirect one due in part to reduced laminins and other BM component receptor–ligands and possibly to an alteration of BM rigidity, a property recently found to be involved in integrin activity and turnover (Du et al., 2011).
BMs have been proposed to act as signaling platforms in which multiple ligands are presented to cell integrin, dystroglycan and other receptors to induce changes in cell function (Li et al., 2005). It has been further proposed that functional activation can depend upon contributions arising from ligand composition and as well as the structural properties of the scaffold itself (Engler et al., 2004; Engler et al., 2006). Laminins, however, are only among several BM components capable of providing Schwann cell integrin-binding ligands. While integrin binding to laminin LG1-3 domains can occur with α7β1 (Lm111, Lm211, Lm511), α6β1 (Lm511, Lm111, and weakly with Lm411), and α3β1 (Lm511), type IV collagen (α1β1, α2β1), perlecan (α2β1) and agrin (α3β1) can provide relevant integrin-binding ligands as well (Yurchenco, 2011). Non-laminin components, expressed endogenously, are recruited into the endoneurial BM upon exposure of Schwann cells to laminins. Employing recombinant laminin that bind poorly to integrins, myelination was substantially reduced without apparent change in BM ultrastructure, or as seen in a previous study, without loss of the efficiency of accumulation of BM components on isolated Schwann cells (McKee et al., 2007). The finding strongly suggests that direct integrin β1 binding to laminin LG domains is a major contributor to myelination. It also adds further support to the hypothesis that integrins serve primarily as mediators of signaling and not of BM assembly in vitro (Li et al., 2002), and is consistent with the presence of BM in endoneurium in the absence of the integrin in vivo (Feltri et al., 2002).
The myelination reduction resulting from the deletion of LG4-5 is in agreement with myelination-deficient phenotype of Lmα2-null (dy3K) mouse following transgenic expression of the laminin α1 subunit bearing a deletion of LG4-5 (Gawlik et al., 2010). Deletion of laminin LG1-3, with a predicted loss of integrin binding, was found to reduce BM and myelination. This contribution to BM assembly was not appreciated in earlier studies (Li et al., 2002; Li et al., 2005). One possibility is that reduction of BM assembly arises from a loss of laminin binding to sulfated glucuronate (the HNK1 epitope that binds to Lmα1LG2) and that is expressed by Schwann cells (Hall et al., 1997; Hall et al., 1995; Saito et al., 2005). BM coverage was restored by binding the laminin to non-neural miniagrin, a protein that attaches firmly to α-dystroglycan and sulfatides (Gesemann et al., 1998; McKee et al., 2009; Moll et al., 2001). The partial increase in myelination that accompanied this might reflect the lower affinity of miniagrin for integrin α3β1 (21 nM Kd, this study) relative to Lm111 for α6β1/α7β1 [(9.5 and 1 nM dissociation constants respectively (Nishiuchi et al., 2006)] and/or a lower abundance of α3β1 compared to α6β1/α7β1 in Schwann cells.
Laminin polymerization was found to contribute to both proliferation and myelination. Treatment of laminin-deficient DRGs with non-polymerizing laminins reduced BM on Schwann cell surfaces to about one-half compared to 28 nM wild-type laminin (slightly below 3.5 nM wild-type levels) and failed to support myelination. In contrast, non-polymerizing LmΔα1LN-L4b did not support proliferation above no-treatment levels. The finding of a polymerization role for myelination may help explain the severity of the myelination defect of the dy2J mouse (Montgomery and Swenarchuk, 1978). The Lmα2 subunit of this dystrophic mouse arises from an in-frame deletion within the Lmα2 LN domain, destabilizing it and reducing polymerization of Lm211 (Colognato and Yurchenco, 1999; Sunada et al., 1995). The role of polymerization in myelination is primarily a structural one. Loss of the activity has been associated with reduction of laminin density in isolated Schwann cells (McKee et al., 2007) and, here, in BM. This reduction causes a decrease in the availability of integrin-binding LG ligands and the ligands of laminin-binding proteins that may be required for myelination. Furthermore, the finding that a laminin with a selective loss of polymerization activity (i.e. rLmΔβ1LN-LEa) demonstrates for the first time that loss of polymerization, rather than loss of LN-dependent integrin, sulfatide, and heparin binding, is likely to be responsible for the neuropathy.
Nidogen-1 is a structure-forming rather than receptor-binding glycoprotein with that binds to the laminin coiled coil through its C-terminal G3 domain and to collagen IV through its G2 and G3 domains (Fox et al., 1991; Ries et al., 2001; Salmivirta et al., 2002). Knock-in of a LAMC1 gene coding for a laminin γ1 subunit lacking the Lmγ1LEb3 nidogen-binding domain resulted in embryonic and early neonatal death prior to peak myelination (Willem et al., 2002). Analysis in this study of DRGs treated with laminin bearing the critical nidogen-binding Lmγ1N802 mutation within domain LEb3 resulted in a slight reduction of myelination.
In conclusion, the study, through comparison of mouse and in vitro models, reveals that myelination more than Schwann cell proliferation depends upon the degree of γ1-laminin expression/concentration within endoneurial BMs. It also demonstrated that scaffold-forming and integrin-receptor binding of laminins must act in an integrated fashion to promote axonal sorting, myelin wrapping and proliferation in which both laminin polymerization and integrin binding through laminin LG domain substantially contribution to myelination and proliferation.
Materials and Methods
Genetically modified mice
Generation of conditional knockout LAMC1 mice by flanking of exon-2 with P-lox (‘fl’) insertions and with a retained upstream PGKneo selection cassette has been described previously (Chen and Strickland, 2003). These mice were maintained in a homozygous state (designated LAMC1flneo/flneo) in a C57Bl/6 background and also bred two generations into 129SvEvTac. LAMC1flneo/flneo males were mated with female Sox2-cre mice to create LAMC1+/− mice. LAMC1+/− Sox2-cre-negative mice were then mated with LAMC1flneo/flneo mice to produce LAMC1flneo/+ and flneo/− offspring. No difference in nerve or muscle phenotype was observed for the two backgrounds. LAMC1flneo/flneo mice were also bred with mice expressing FLP recombinase (Jackson Laboratories) in order to delete the PGKneo cassette flanked by FRT sequences and located upstream of the LoxP inserts. Homozygous mice were mated with the LAMC1+/− mice to generate LAMC1 fl/+ and fl/− offspring that now lacked a neo cassette in the floxed gene (the neo cassette remained in the null allele). Genotyping was performed by PCR using DNA extracted from tails. Primer set 5′-CCTGGCGATCCCTGAACATGTCC-3′ and 5′-CTCTAGAGCCTCTGCTAACC-3′ was used to amplify a 300 bps product of Sox2 cre. Primer set 5′-GATTTTCAAAGAAGCAGAGTGTG-3′ and 5′-CCTACATTTTGAATGCAAGGATTGG-3′ was used to visualize a LAMC1 floxed allele at 1.6 kb and a recombined allele at 0.6 kb. The floxed allele lacking neo was detected with primers P1-12c (5′-CTCAGAGCTGGCTTCTCCATG-3′) and P1-6 (5′-CATTTCCCCACAAGTGGTTCTT-3′) yielding a 400 bp wild-type product. Animal experiments were conducted according to institutional and national guidelines under protocols approved by the institutional IACUC.
Tissues, antibodies and microscopy
Tissues were embedded in OCT (Tissue-Tek, Elkhart, IN), frozen in liquid nitrogen, and stored at −80°C. Tissue blocks were sectioned (5 µm thick) with a cryostat (Leica CM 1850) at −20°C onto charged glass slides and fixed in 3.5% paraformaldehyde for 15 min at RT or 30 min on ice.
Following overnight blocking (5% goat serum, 0.5% BSA, PBS) slides were treated with one or more of these antibodies: Lmγ1-subunit mAb (mouse, 1:400 dilution, Millipore), Lmβ1LN-LEa rabbit polyclonal [‘anti-E4’, 1-2 µg/ml (Yurchenco et al., 1997), Lmα1LG4-5 polyclonal, 5 µg/ml (Yurchenco et al., 1993)]. Lmα2LG1-5 rabbit polyclonal, 10 µg/ml (Cheng et al., 1997), Lmα4, 1:1000 (provided by Takako Sasaki, Oregon Health Sciences University), Lmα5, (Jeffrey Miner, Washington University School of Medicine), Lmγ3, 1:5000 (William Brunken, Downstate Medical School), nidogen-1 rabbit polyclonal, 10 µg/ml (Li et al., 2002), perlecan rabbit polyclonal, 2 µg/ml (Handler et al., 1997), type-IV collagen rabbit polyclonal, 0.5 µg/ml (Millipore), α-dystroglycan, mouse IgM mAb, 1:100 (Millipore/Upstate), integrin-α6 mouse mAb, 5 µg/ml (GoH3, BD Pharmingen 555734), integrin-β1 rat mAb, 10 µg/ml (MAB1997; Millipore/Upstate), neurofilament-200, 1:1000 (NFL, chicken antibody, Abcam), and myelin basic protein (MBP), 10 µg/ml (Millipore MAB386). DRG-derived fibroblasts were detected with anti-Thy1.2 rat IgG monoclonal antibody, 1:1000 (Thermo Scientific, MA1-90188). Bound antibodies were detected with Alexa Fluor 647 goat anti-rabbit IgG, Alexa Fluor 488 anti-mouse IgG and Alexa Fluor 594 anti-rat IgG secondary antibodies (1∶100, Molecular Probes). MBP was detected anti-rat 647 and NFL detected with anti-chicken 488 (Molecular Probes a11039). For detection of nuclei, slides were stained with 2 µg/ml of DAPI (4′-6-diamidino-2-phenylindole). Coverslips were mounted with SlowFade (Invitrogen/Molecular Probes).
Tissues were fixed overnight in 10% neutral-buffered formalin (NBF) and embedded in paraffin. Sections were stained with Hematoxylin and Eosin or Masson's trichrome (Tissue Analytical Service, Cancer Institute of New Jersey).
Indirect immunofluorescence and electron microscopy were performed as described previously (Yang et al., 2011).
Schwann cell proliferation in nerve was measured by injecting pups or pregnant mice i.p. with 10 µl/gram of 20 mM 5-ethynyl-2′-deoxyuridine (EdU) followed by harvest or sciatic nerve three hours later (Zeng et al., 2010). EdU incorporation was detected in frozen sections with the Click-iT EdU Alexa Fluor 594 imaging kit (Invitrogen, Carlsbad, CA) and counterstained with DAPI to detect nuclei. The fraction of cells undergoing DNA synthesis was determined by count of EdU/DAPI-staining nuclei (b). To measure DRG Schwann cell DNA synthesis, cultures were incubated with 1 µM EdU for 3 hrs, washed, fixed in 3.7% paraformaldehyde, permeabilized with 0.1% Triton X-100, reacted with Click-iT Alexa-594 dye-conjugate, and counterstained with DAPI. EdU:DAPI ratios were estimated by segmentation analysis.
Quantitative real-time PCR analysis
RNA was isolated and stabilized from tissues with reagents provide by an RNeasy Plus Mini Kit (Qiagen). Real-time reverse transcriptase polymerase chain reaction (qRT-PCR) was carried out to quantify laminin subunits, integrin β1, perlecan, c-Jun, Oct6/Scip, Krox-20, myelin protein zero (MPZ), p75-NTR, neuregulin-1, Erbb2 and FGF2 by using an ABI 7900 HT Sequence Detection System as described previously (Yang et al., 2011).
Recombinant and other glycoproteins
Type IV collagen (Col-IV) was purified from the Engelbreth-Holm-Swarm (EHS) tumor as described previously (Yurchenco and Furthmayr, 1984).
Recombinant glycoproteins: rLm111, rLm111 bearing a γ1-subunt C-terminal FLAG tag (rLm111F), rLm111 with a deletion of LG domains 1-3 and containing FLAG (rLmΔα1LG1-3), rLm111 with a deletion of LG domains 4-5 without FLAG (rLmΔα1LG4-5), rLm111 with deletion of α1-subunit short arm domains LN through L4b (rLmΔαLN-L4b) without FLAG, rLm111 with deletion of β1-subunit short arm domains LN-LEa (rLmΔβ1LN-LEa) without FLAG, rLm111 with a γ1N802S mutation without FLAG, nidogen-1 (rNd1) and non-neural miniagrin (mA) were prepared in HEK293 cells and purified as described previously (McKee et al., 2009; McKee et al., 2007). To generate deletion mutants of the Lmα1 chain for domains LG1-3 and LG4-5, an XhoI digest of the mouse α1 plasmid removed a 4.8 kb fragment leaving a 9.4 kb fragment. The LG1-3 deletion construct required overlapping PCR utilizing primer pairs ‘P1F’ and ‘P2R’ (5′-GAAAGGTGCTGTGCAGGCTATCACG-3′ and 5′-GAGTTCCCCGTGTTCTCTGTGCAGTGTTTCCGGGCCCGGCTGATCAGCAG-3′) and ‘P2F’ and ‘P1R’ (5′-CTGCTGATCAGCCGGGCCCGGAAACACTGCACAGAGAACACGGGGAACTC-3′ and 5′-GCTGATCAGCGGGTTTAAACGGGC-3′). The generated 1.8 kb and 1.2 kb products were sewn together with primers P1F and P1R, digested with XhoI and ligated to the 9.4 kb fragment. The LG4-5 deletion was generated with a single PCR product using P1F and P3R (5′-GATGCGGCTCGAGTTATCAACTCTGCATGGGCTCTTCTGCCAG-3′) and includes 2 stop codons and an XhoI site downstream of LG3. The 3.6 kb PCR product was digested with XhoI and ligated into the 9.4 kb fragment. To generate cells expressing Lm211 without a FLAG epitope tag, stable Lmγ1/Lmβ1-expressing HEK293 cells were transfected with pCis-human Lmα2 DNA and pcDNA3.1-puro as described previously (McKee et al., 2007; Smirnov et al., 2002).
Protein was purified with HA affinity gel (Sigma E677) as per the manufacture conditions. Laminin-211 bearing a γ1-subunit C-terminal FLAG epitope tag (rLm211F) was prepared as described previously (Smirnov et al., 2002). Laminin concentrations were determined from intensity of Coomassie-Blue-stained aliquots whose bands were separated by SDS-PAGE (reducing conditions) against EHS-Lm111 standards and using the following peptide molecular masses to calculate molar concentrations (McKee et al., 2007; Yurchenco and Cheng, 1993).
Dorsal root ganglia and Schwann cell cultures
Dorsal root ganglia (DRGs) were isolated by dissection from LAMC1flneo/flneo embryos at E13.5 based on the method of Päiväläinen et al. (Päiväläinen et al., 2008). One to two DRGs were placed into each collagen-treated tissue culture well (24-well dishes, treated with 2 mg/ml rat tail collagen in 0.1 M acetic acid and neutralized with ammonium hydroxide vapors) along with 20 µl of growth medium/well, and allowed to adhere followed by addition of 0.3 ml medium. The DRGs were treated 3 days later with 2×107 PFU of cre-recombinase-GFP adenovirus [DNA kindly provided by Tse-Hua Tan, Baylor College (Grewal et al., 2001) with virus generated in HEK293 cells and purified by CsCl density-gradient centrifugation]. GFP fluorescence was used to determine optimal viral titer and to confirm expression of recombinase. Recombinant nidogen-1 (28 nM), with or without recombinant laminins were added at the same time as virus. After 3 additional days, the initial medium was replaced with myelination medium containing ascorbic acid, and including different recombinant proteins. Medium was changed every 3 days for a total of 6 days. DRGs were then fixed for 15 minutes with 3.5% paraformaldehyde (RT), treated for 6 min with ice-cold methanol, followed by blocking buffer containing 0.1% Triton X-100 overnight at 4°C. DRGs were then stained with antibodies.
For isolation of Schwann cells for adhesion assays, multiple wild-type DRGs were placed in a single collagen-coated dish and maintained in growth medium for 7-10 days. Schwann cells were passaged in Schwann cell medium at a density of 1.5×106/100 mm dish and used at passage 5-6 for adhesion assays. Fibroblast contamination levels were determined by Thy1.2 antibody and DAPI staining. The antibody-stained cells were broad and flat unlike the fusiform Schwann cells. The cells used to evaluate cell adhesion were found to contain 0.7±0.3% fibroblasts (average ± s.d. of five microscope fields, total of 1762 cells).
Overlapping images (4× objective) of immunostained DRGs were recorded in IPLab and joined to create a composite image. The degree of DRG myelination was estimated from these images by several methods. (1) The first was by determining the ratio of myelinated axons stained red with MBP to NFL-stained axons and gfp-fluorescing Schwann cells (green) by segmentation analysis of photographed images in IPLab. A common baseline level to subtract non-specific fluorescence was established for each fluor within the group of DRGs to be analyzed. The segmentation parameters for myelin and axons were set to identify a maximum number of myelinated or axonal forms respectively while minimizing background. Segmented images were then used to determine values for the sum of pixel intensities. An average of three or more DRGs per condition were used to calculate an average and standard deviation. (2) The second method was by direct measurement of the lengths of myelinated forms contained within a DRG. This was accomplished by tracing the linear myelinated (MBP-stained) structures and recording the accumulation of linear pixels with the IPLab application for this purpose. The summed lengths were divided by the area of the DRG delineated by the NFL/gfp stained structures and expressed as an average and standard deviation for the DRGs for each condition evaluated. (3) The distribution of myelinated segment lengths was determined by sampling MBP-stained segments whose ends could be clearly distinguished.
Proliferation was estimated in a similar manner in DRGs by determining the ratio of EdU stained nuclei (red) to total DAPI-stained nuclei (blue).
Electron micrograph measurements were performed by tracing structures of interest in either IPLab (v.3.7, Scanalytics) in Windows or iVision (v.4.0.15, BioVision Technologies) with calculations performed in Excel. The ratio of axonal and axonal and myelin areas and distribution of axonal areas, were calculated from multiple (2300-3000× magnification) cross-section images. The ratio of BM/plasma membrane was determined after tracing contour lengths of each along multiple Schwann cell surfaces containing BMs.
Cell adhesion and integrin-binding assays
Costar 96-well polystyrene dishes (no. 3631) were coated with BM proteins in 0.15 M sodium bicarbonate (pH 9.6) overnight at 4°C. Plates were washed and blocked for 1 hr at RT with 0.3% heat-inactivated BSA in PBS. Schwann cells (30,000 cells/100 µl/well) were allowed to adhere for 80 min. Unbound cells were removed with three PBS washes and WST-1 (Clontech) was added at 1:10 into Schwann cell medium. Adherent cells were measured by absorption (492 nm) in a Tecan Spectra Fluor.
Integrin binding to recombinant laminins and miniagrin immobilized in 96-well dishes was determined with recombinant extracellular domains of integrins basically as described previously (Nishiuchi et al., 2006). In brief, integrins α6β1, α7β1, α3β1 and αvβ1 were fused to ACID/BASE α-helical peptides with FLAG/6His epitope tags and containing cysteines for disulfide bonding of subunit fragments. Plastic wells were coated with laminins (20 µg/ml, 2-3 hrs) in TBS, blocked with 3% BSA containing 0.1% Tween-20 (1 hr), and incubated with the soluble integrins at different concentrations for 2 hrs at RT in Tris-buffered saline (TBS: 50 mM Tris-HCl, 90 mM NaCl, pH 7.4) containing 1 mM manganese sulfate, 0.1% Tween-20. After washing, the wells were incubated with biotinylated rabbit anti-Velcro antibody (ACID/BASE coiled coil, 1∶1000 dilution), followed by HRP-conjugated streptavidin (1 µg/ml). Binding was detected by incubating with o-phenylenediamine with absorbance measured at 492 nm.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.107995/-/DC1
We thank Bruce Patton (Oregon Health Sciences University) for help in examining nerve and muscle pathology and useful discussions and David Harrison and Sergei Smirnov (Robert Wood Johnson Medical School) for construct design and integrin-binding assays.
This study was supported by the National Institutes of Health [grant numbers R37-DK36425 and R01-NS38469 to P.D.Y., NS038472 to S.S.]; and the Muscular Dystrophy Association [grant number MDA4066 to Z.L.C.]. Deposited in PMC for release after 12 months.