The muscle dystrophy-causing ΔK32 lamin A/C mutant does not impair the functions of the nucleoplasmic lamin-A/C–LAP2α complex in mice

Lamins are intermediate filament proteins in metazoan cells that form the lamina, a scaffold structure tightly associated with the inner nuclear membrane (Dechat et al., 2008). They provide mechanical stability for the nuclear envelope and the nucleus and help to organize higher-order chromatin structure. Lamins are grouped into A- and B-type, based on biochemical, structural and dynamic properties, sequence homologies and expression patterns. B-type lamins are ubiquitously expressed throughout development, but the major A-type lamins (lamin A and C) encoded by LMNA are expressed at later stages during development (Röber et al., 1989; Stewart and Burke, 1987). A- and B-type lamins undergo post-translational processing at their C-terminal CAAX motif, including farnesylation and carboxy-methylation (Rusiñol and Sinensky, 2006). The hydrophobic farnesyl group at the C-terminal cysteine facilitates tight interaction with the inner nuclear membrane. Mature B-type lamins remain farnesylated, but lamin A undergoes an additional endoproteolytic processing step that removes 15 residues from the C-terminus including the farnesyl group (Pendás et al., 2002). Thus mature lamin A and also lamin C, which is not processed post-translationally, lack a farnesyl group and are less tightly bound to membranes than B-type lamins. Consequently, a fraction of A-type lamins is also found in the nucleoplasm (Dechat et al., 2010).

The physiological relevance and functions of the nucleoplasmic lamin A/C pool are poorly understood, but they are probably involved in many of the reported functions of lamins in cell signaling and gene expression (Andrés and González, 2009; Dechat et al., 2010; Heessen and Fornerod, 2007; Prokocimer et al., 2009). Our recent studies showed that lamina-associated polypeptide 2α (LAP2α) regulates the localization and functions of nucleoplasmic A-type lamins (Naetar et al., 2008). LAP2α is a unique member of the LAP2 protein family (Wilson and Foisner, 2010). Although most LAP2 proteins are integral membrane proteins of the inner nuclear membrane and associate with lamins in the peripheral lamina, LAP2α localizes to the interior of the nucleus and interacts with nucleoplasmic lamins A and C (Dechat et al., 2000). Deletion of LAP2α in mice causes loss of lamins A and C in the nucleoplasm of dermal fibroblasts and proliferating tissue-progenitor cells, while lamins at the peripheral lamina are unaffected (Naetar et al., 2008). Similarly, human fibroblasts lose nucleoplasmic lamins following RNA-interference-mediated knockdown of LAP2α (Pekovic et al., 2007). Moreover, during myoblast differentiation, LAP2α expression is downregulated and nucleoplasmic lamins are lost (Markiewicz et al., 2005). Nucleoplasmic lamins and LAP2α were shown to bind directly to the tumor-suppressor retinoblastoma protein (pRb) in its active, hypo-phosphorylated form (Markiewicz et al., 2002) and to promote pRb repressor activity on pRb/E2F target gene promoters, mediating efficient cell cycle exit of proliferating cells (Dorner et al., 2006). Accordingly, LAP2α deletion in mice accompanied by loss of nucleoplasmic lamins results in hyperproliferation of tissue progenitor cells and tissue hyperplasia (Naetar et al., 2008).

Mutations in the LMNA gene and in several genes encoding lamin-associated proteins have been linked to phenotypically heterogeneous diseases generally termed laminopathies. The disease variants range from muscular dystrophies over cardiomyopathies to lipodystrophies and systemic involvement of many tissues, such as in the premature ageing disease Hutchinson–Gilford progeria syndrome (HGPS) (Worman and Bonne, 2007). The molecular disease mechanisms underlying the laminopathies are still poorly understood. One disease model proposes defects in mechanical properties of the lamina in laminopathic cells, leading to increased fragility of nuclei, whereas other models have proposed impaired functions of mutated lamins in chromatin regulation and gene expression (Gotzmann and Foisner, 2006).

In a recent study we described a novel mouse model for a severe, striated muscle-affecting laminopathy (Bertrand et al., 2012): Lmna^ΔK32/ΔK32 knock-in mice harbor a Lmna mutation that results in the loss of lysine 32 in the N-terminal domain of lamins A and C, and causes a severe form of congenital muscular dystrophy (CMD) in humans (Quijano-Roy et al., 2008). Homozygous Lmna^ΔK32/ΔK32 mice were indistinguishable from their wild-type littermates at birth but soon exhibited striated muscle maturation delay and metabolic defects and died within 2–3 weeks (Bertrand et al., 2012). Interestingly, the ΔK32 mutation was previously proposed to impair the lateral assembly of lamin A/C head to tail polymers (Bank et al., 2011). In line with this observation, mutant lamins failed to assemble at the lamina and localized predominantly in the nucleoplasm in Lmna^ΔK32/ΔK32 mice. Taking into account our previous results on the important role of nucleoplasmic lamins in tissue progenitor cells (Dechat et al., 2010), we hypothesized that a deregulated nucleoplasmic lamin pool in Lmna^ΔK32/ΔK32 mice might contribute to the pathologies. Therefore, we set out to test whether LAP2α, a major regulator of nucleoplasmic lamins in Lmna^+/+ mice (Naetar et al., 2008), is also involved in regulating nucleoplasmic ΔK32 lamin A/C, and whether and how loss of LAP2α in Lmna^ΔK32/ΔK32 mice might affect mutant lamin A/C localization and function. Using LAP2α^−/− and Lmna^ΔK32/ΔK32 double mutant mice, we show in this manuscript that loss of LAP2α does not change the localization of mutant lamins, as shown for wild-type lamins. LAP2α and ΔK32 lamin A/C, however, formed ‘functional’ complexes in the nucleoplasm, as loss of LAP2α in Lmna^ΔK32/ΔK32 mice caused hyperproliferation of epidermal progenitor cells and hyperplasia of the epidermis, a similar phenotype to Lmna^+/+ mice lacking LAP2α. These data indicate that a LAP2α-unrelated assembly defect of ΔK32 lamin may be the predominant molecular defect in Lmna^ΔK32/ΔK32 mice, while its LAP2α-dependent functions in the nucleoplasm are unaffected.

Mutant ΔK32 lamin A/C in Lmna^ΔK32/ΔK32 knock-in mice fails to assemble at the nuclear lamina and mislocalizes to the nucleoplasm (Bertrand et al., 2012). Because LAP2α has previously been found to regulate the nucleoplasmic pool of wild-type lamins A and C (Naetar et al., 2008), we wanted to examine the influence of LAP2α loss on ΔK32 lamin A/C expression and cellular distribution. We generated double mutant mice by crossing Lmna^+/ΔK32 mice with heterozygous Lap2α-deficient mice and isolated fibroblasts and myoblasts from newborn littermates. Lamin A/C protein levels were massively reduced to between 60 and 90% of wild-type lamin A/C levels in all Lmna^ΔK32/ΔK32 cells independent of the presence or absence of LAP2α (Fig. 1; supplementary material Fig. S1A). LAP2α and other nuclear envelope and/or lamina proteins, such as lamin B1 and emerin, were not affected in Lmna^ΔK32/ΔK32 cells. Similar results were obtained in lysates of liver and diaphragm derived from mice of the four genotypes (supplementary material Fig. S1B). Thus, LAP2α loss did not affect the expression level of mutant ΔK32 lamin A/C protein. Downregulation of mutant lamin A/C probably occurs post-transcriptionally, because lamin A mRNA levels in fibroblasts were similar in all genotypes (Fig. 1) in accordance with previous reports (Bertrand et al., 2012).

Immunofluorescence analyses of primary mouse fibroblasts confirmed the reduction of lamin A/C protein levels in single and double mutant Lmna^ΔK32/ΔK32 mice. Co-cultures of LAP2α-expressing and LAP2α-deficient Lmna^+/+ and Lmna^ΔK32/ΔK32 cells made it possible to identify Lmna^ΔK32/ΔK32 versus Lmna^+/+ cells in the co-culture (by the lack of LAP2α staining) and made it possible to compare lamin A/C levels and localization in the different genotypes under identical experimental conditions (Fig. 2A). Although wild-type lamin A/C was predominantly found peripherally in the nucleus with an additional weaker nucleoplasmic staining, the lowly expressed ΔK32 lamin A/C mutant was equally distributed throughout the nucleoplasm without any clear nuclear rim staining. Quantitative analyses of lamin A/C localization by plotting lamin A/C staining intensity profiles across the nuclear diameter in confocal immunofluorescence images confirmed the uniform distribution of the ΔK32 lamin A/C staining (Fig. 2B). LAP2α localization was unaffected in ΔK32 lamin A-expressing cells.

Although loss of LAP2α in Lmna^+/+ cells decreases lamin A/C levels in the interior of the nucleus (Naetar et al., 2008), the exclusive nucleoplasmic localization of ΔK32 lamin A/C mutant was unaffected by the absence of LAP2α, as revealed in the staining intensity profiles (Fig. 2B). In addition, we calculated the ratios of nucleoplasmic over peripheral mean lamin A/C fluorescence intensities for 25–30 fibroblasts for each genotype (Fig. 2C). In Lmna^ΔK32/ΔK32 fibroblasts, the ratios were significantly increased compared to Lmna^+/+ cells (n = 25, P<0.05), reflecting the lack of accumulation of mutant lamin A/C at the periphery and its even distribution throughout the nucleus. Loss of LAP2α had no effect on the distribution of mutant lamin A/C (Lmna^ΔK32/ΔK32 background), whereas in the lamin A/C wild-type background loss of LAP2α caused a significant reduction of nucleoplasmic lamins (n = 30, P<0.05).

Our data suggest that ΔK32 lamin A/C has a LAP2α-independent assembly defect, preventing it from assembling at the nuclear lamina, which is in line with previous reports in C. elegans (Bank et al., 2011). To test this hypothesis, we performed cell lysis in physiological buffer plus 0.5% Triton X-100/0.1% SDS and determined the soluble, unassembled pool of lamin A/C by immunoblotting and densitometric analyses. Although only approximately 10% of total wild-type lamin A/C was soluble, all of the ΔK32 lamin A/C was solubilized under these conditions (Fig. 2D).

Next we tested whether the mutated ΔK32 lamin A is able to directly interact with LAP2α. Bacterially expressed ΔK32 and wild-type pre-lamin A were transferred to nitrocellulose and probed with in vitro translated ³⁵S-labeled LAP2α. Autoradiography of the blot revealed binding of LAP2α to both wild-type and ΔK32 lamin A, whereas binding to a related, cytoplasmic intermediate filament protein, vimentin was not detectable (Fig. 3). To test whether ΔK32 lamin-A–LAP2α complexes also exist in vivo we lysed ΔK32 lamin-A-expressing cells in physiological buffer containing 0.5% Triton X-100/0.1% SDS, and immunoprecipitated ΔK32 lamin A using specific antibodies. Unlike in mock precipitations using empty beads, LAP2α was co-precipitated with ΔK32 lamin A (Fig. 3C).

In order to test the effect of loss of LAP2α on the organismal and tissue phenotype of Lmna^ΔK32/ΔK32 mice, we analyzed litters of mice heterozygous for both LAP2α and ΔK32 lamin A/C (Lap2α^+/−/Lmna^+/ΔK32). The breeding produced genotypes according to Mendelian ratios (n = 384; Lmna^+/+/Lap2α^+/+ = 4.7%; Lmna^+/+/Lap2α^−/− = 7.5%; Lmna^ΔK32/ΔK32/Lap2α^+/+ = 4.9%; Lmna^ΔK32/ΔK32/LAP2α^−/− = 8.4%). All genotypes were indistinguishable from wild-type littermates at birth. From postnatal day 6 on, all mutant mice homozygous for Lmna^ΔK32/ΔK32, independent of the Lap2α genotype, were smaller with a kinked tail, and showed progressive growth retardation and stagnation in weight gain, as well as atrophied muscles and reduced mobility. By post-natal day 15 for single Lmna^ΔK32/ΔK32 mutants, and day 17 for double Lmna^ΔK32/ΔK32/Lap2α^−/− mutants, only 50% of mice were alive, and none survived longer than post-natal day 21 (supplementary material Fig. S2). Thus, double mutants for Lmna^ΔK32/ΔK32 and Lap2α^−/− showed a slightly prolonged, though statistically insignificant survival in comparison to the single Lmna^ΔK32/ΔK32 littermates, indicating that the Lmna^ΔK32/ΔK32-linked phenotype was prominent.

Having shown that ΔK32 lamin A/C fails to form a lamina at the periphery of the nucleus, but was still able to interact with LAP2α in the nucleoplasm, we sought to test whether ΔK32 lamin-A–LAP2α complexes can still function in tissue progenitor cell regulation. Loss of nucleoplasmic lamin-A/C–LAP2α complexes by either deletion of Lap2α or deletion of Lmna (causing loss of nucleoplasmic and peripheral lamina) was shown to cause hyperproliferation of epidermal progenitor cells and progressive hyperplasia of the paw epidermis during post-natal life (Naetar et al., 2008). If the nucleoplasmic ΔK32 lamin-A/C–LAP2α complex is still functional in this pathway, LAP2α knockdown in Lmna^ΔK32/ΔK32 mice is expected to have a similar effect on epidermal progenitor cells and epidermal thickness as in wild-type mice. Despite the young age of Lmna^ΔK32/ΔK32 single and double mutant mice, the paw epidermis was ∼20% thicker in Lmna^ΔK32/ΔK32/Lap2α^−/− versus Lmna^ΔK32/ΔK32/Lap2α^+/+ littermates (Fig. 4A). Moreover, and in line with our previous findings, a higher number of proliferating (KI67-positive) cells was detected in double mutant Lmna^ΔK32/ΔK32/Lap2α^−/− versus Lmna^ΔK32/ΔK32/Lap2α^+/+ mice (Fig. 4B), pointing to an increased proliferation of epidermal progenitor cells. Thus, we concluded that the ΔK32 lamin A/C mutant is still active in regulating progenitor cells in conjunction with LAP2α.

LAP2α loss was previously also shown to increase the number of skeletal muscle progenitor cells (Gotic et al., 2010). To test the number of satellite cells in Lmna^ΔK32/ΔK32 mice, we enriched skeletal muscle progenitor cells (SMPC) from an isolated pool of muscle-fiber-associated cells by flow cytometry, based on the expression of CXCR4 and β1-integrin and lack of expression of CD45, Sca1 or Mac1. Immunofluorescence microscopy of these cells confirmed mislocalization of ΔK32 lamin A/C and unaltered lamin B distribution in lamin mutant versus wild-type cells (Fig. 5). Importantly, in both Lmna^+/+ and Lmna^ΔK32/ΔK32 mice the SMPC population was increased in Lap2α^−/− versus Lap2α^+/+ background (Fig. 5), supporting the hypothesis that ΔK32 lamin A/C, in conjunction with LAP2α, is equally capable of regulating muscle progenitor cells as the wild-type lamin A/C. The increase in SMCP cells in LAP2α-deficient versus LAP2α-expressing Lmna^ΔK32/ΔK32 mice might slightly improve growth capability of muscle, which may also contribute to the subtle increase in survival of Lmna^ΔK32/ΔK32/Lap2α^−/− versus Lmna^ΔK32/ΔK32/Lap2α^+/+ mice (see supplementary material Fig. S2).

Based on the reported muscle phenotype of Lmna^ΔK32/ΔK32 mice (Bertrand et al., 2012) we also compared SMPC cell numbers in Lmna^ΔK32/ΔK32 versus Lmna^+/+ littermates. The number of muscle-fiber-associated SMPC cells was indistinguishable in these genotypes (Fig. 5), indicating that an exhaustion of the SMPC pool is unlikely to contribute to muscle disease in Lmna^ΔK32/ΔK32 mice.

Histological Hematoxylin and Eosin staining of gastrocnemius and soleus muscle sections of 16-day-old Lmna^ΔK32/ΔK32 mice revealed a generally atrophied muscle, decreased fiber cross-sectional area, variability in fiber size and a significantly increased proportion of muscle fibers with centrally located or internalized nuclei (Fig. 6). Other dystrophic changes, such as replacement of functional muscle fibers with connective tissue or fat, cellular infiltrates, increased endomysial space, ruptured fibers or an increase in serum levels of creatine kinase (CK) levels as reported by Costanza and Moggio were not observed (Costanza and Moggio, 2010). This phenotype was predominantly caused by the lamin A/C mutant, as the presence or absence of LAP2α did not grossly affect this phenotype. However, one striking phenotype upon loss of LAP2α in Lmna^ΔK32/ΔK32 mice was the further increase in the number of fibers with centrally located nuclei (from 14.2% to 18.4%, n = 7, P = 0.019; Fig. 6C). This observation is consistent with a subtle increase in muscle growth in the LAP2α-deficient background probably due to the higher number of SMPCs (Fig. 5). The presence of a central nucleus within muscle fibers may also be an indicator of a muscle maturation defect as reported (Bertrand et al., 2012). Indeed, the proportion of muscle fibers expressing the embryonic form of myosin heavy chain was significantly increased in Lmna^ΔK32/ΔK32 versus Lmna^+/+ mice (Fig. 6D; n = 4, P = 0.04), but was independent of LAP2α expression.

When Lmna^ΔK32/ΔK32 myoblasts were cultivated in vitro and induced to differentiate by withdrawal of serum, we observed a delayed onset of differentiation, an reduced formation of myotubes and failure to upregulate myosin heavy chain compared to the wild-type littermates, pointing to an impairment of the differentiation potential of Lmna^ΔK32/ΔK32 myoblasts. Additional loss of LAP2α did not noticeably aggravate or ameliorate this phenotype. Fig. 7A shows a lag of Lmna^ΔK32/ΔK32 myoblast differentiation at day 3, irrespective of LAP2α expression and a massive reduction of myotube formation at day 6 of in vitro muscle differentiation, as confirmed also by immunofluorescence microscopy of myotubes at day 4 of differentiation (Fig. 7C). In line with this, qRT-PCR analyses revealed a failure of myosin heavy chain upregulation during differentiation (Fig. 7B).

In this manuscript we confirm and extend our previous findings showing that CMD-linked ΔK32 lamin A/C mutants fail to accumulate at the nuclear lamina in primary fibroblasts of Lmna^ΔK32/ΔK32 knock-in mice. Mutant lamin A/C protein is significantly reduced and localizes uniformly throughout the entire nucleus. Also in wild-type cells and tissues, a highly dynamic and cell-cycle-dependent pool of lamin A/C has been described in the interior of the nucleus (Moir et al., 2000; Naetar et al., 2008). This nucleoplasmic pool of lamin A/C has been found to associate with LAP2α and has been implicated in the regulation of proliferation and differentiation of tissue progenitor cells during tissue homeostasis (Naetar et al., 2008). It was thus tempting to speculate that an abnormally regulated pool of nucloplasmic lamins or a misbalance between lamina-associated and lamina-independent lamins were responsible for some of the pathologies described in the Lmna^ΔK32/ΔK32 mice (Bertrand et al., 2012). To test this hypothesis we investigated whether any of the previously described functions and regulation mechanisms of nucleoplasmic lamin–LAP2α complexes are impaired in ΔK32 lamin A/C knock-in mice.

Although the regulation of the intranuclear, nucleoplasmic lamin A/C pool is not completely understood yet, our previous studies revealed one direct regulator of nucleoplasmic lamins A and C, a nucleoplasmic isoform of the lamina-associated polypeptide 2 family, termed LAP2α. Although most other LAP2 isoforms are transmembrane proteins of the inner nuclear membrane and bind lamins in the nuclear lamina (Foisner and Gerace, 1993), LAP2α lacks a transmembrane domain and localizes to the interior of the nucleus (Dechat et al., 1998) and specifically binds A-type lamins (Dechat et al., 2000). We showed that loss of LAP2α in LAP2α knockout mice reduced the levels of lamins A and C in the interior of the nucleus in proliferating epidermal progenitor cells and primary fibroblasts, and that re-expression of LAP2α in LAP2α-deficient fibroblasts rescued the nucleoplasmic lamin A/C pool (Naetar et al., 2008). These data suggested that LAP2α is essential and sufficient for targeting and/or stabilizing nucleoplasmic lamins A and C. Since this activity of LAP2α required its C-terminal lamin-A/C-interaction domain, it was assumed that LAP2α regulates intranuclear lamin A/C by direct binding. Because we found here that LAP2α also interacted with mutant ΔK32 lamin A/C, we reasoned that loss of LAP2α in Lmna^ΔK32/ΔK32 mice might reduce the potentially abnormal nucleoplasmic ΔK32 lamin A/C pool in mutant mice and might allow mutant lamin to associate with the peripheral lamina. However, loss of LAP2α did not affect either the levels or localization of ΔK32 lamin A/C. Therefore, we concluded that mutant lamin A/C is incapable of assembling at the nuclear lamina even in the absence of LAP2α, supporting previous studies that suggested that the ΔK32 mutation in lamin A/C impairs the lateral association of head-to-tail dimer protofilaments into anti-parallel tetrameric filaments (Bank et al., 2011). These studies, which were performed in C. elegans did not, however, reveal a uniform nucleoplasmic distribution of mutant lamin but rather nucleoplasmic aggregates. This difference may be due to the fact that C. elegans has only one lamin gene, which encodes a farnesylated B-type lamin. In any case, our studies reveal that mouse ΔK32 lamin A shows a LAP2α-independent assembly defect.

Our binding analyses show that LAP2α and ΔK32 lamin A/C can form complexes in the nucleus. Are these complexes functional? Lamins A and C at the nuclear lamina have been implicated in a number of functions, including nuclear architecture (Sullivan et al., 1999), (hetero-)chromatin organization (Guelen et al., 2008) and signaling (reviewed by Andrés and Gonzalez, 2009; Heessen and Fornerod, 2007). We have previously shown that LAP2α and nucleoplasmic lamin A/C also function in the regulation of the pRb pathway (Dorner et al., 2006; Naetar et al., 2008). This function, which is likely to be independent of the peripheral lamina, has been proposed to regulate the proliferation and differentiation of tissue progenitor cells during tissue homeostasis. Since classical gene knockout or knock-in approaches in mouse by targeting the Lmna gene affect both the peripheral lamina and the nucleoplasmic lamin A/C pool, it has been difficult to distinguish and specifically test the activities of peripheral versus nucleoplasmic lamins. The LAP2α knockout mouse is currently the only model that selectively affects the nucleoplasmic pool of lamin A/C, while the peripheral lamina remains unaffected (Naetar et al., 2008). The fact that both the selective loss of nucleoplasmic lamins (by LAP2α deletion) and the complete loss of lamin A/C in Lmna^−/− mice (affecting nucleoplasmic and peripheral lamins) show a similar hyperproliferation of progenitor cells in the paw epidermis and epidermal hyperplasia (Naetar et al., 2008) indicates that the phenotype is directly linked to the loss of nucleoplasmic lamin–LAP2α complexes rather than the loss of lamin-independent functions of LAP2α. If the nucleoplasmic ΔK32 lamin-A/C–LAP2α complexes were still functional, we would expect similar consequences on tissue progenitor cells upon LAP2α loss in wild-type and Lmna^ΔK32/ΔK32 mice. Indeed, we observed an increase in proliferating cells in paw epidermis and a thickening of the epidermis upon knocking out LAP2α in Lmna^ΔK32/ΔK32 mice. These data show, (1) that nucleoplasmic ΔK32 lamin A/C still functions in the regulation of tissue progenitor cells, and (2) that the nucleoplasmic lamins require LAP2α for this activity. Forcing lamin A/C into the interior by, for instance interfering with their assembly at the lamina (as occurs in Lmna^ΔK32/ΔK32 mice) is insufficient to generate ‘active’ nucleoplasmic lamin complexes.

Based on our results, it is likely that the pathologies described in the Lmna^ΔK32/ΔK32 mice are primarily caused by loss of peripheral lamins and/or the downregulation of lamin protein levels. Can loss of LAP2α affect the mutant lamin A/C-linked phenotype? One of the most prominent phenotype described in the Lmna^ΔK32/ΔK32 mice is an impaired peri- and postnatal muscle maturation, reflected in an increased number of muscle fibers with centrally located nuclei and increased embryonic myosin heavy chain expression. Since LAP2α loss has been shown to increase the number of fiber-associated progenitor cells, we speculated that the larger pool of skeletal muscle progenitor cells might partially rescue the mutant lamin-A/C-mediated muscle defect. Although we saw a LAP2α loss-mediated increase in muscle progenitor cells in Lmna^ΔK32/ΔK32 mice, which might theoretically contribute to regeneration of defective muscle, we did not see any significant rescue of muscle morphology and maturation in double mutant Lmna^ΔK32/ΔK32/Lap2α^−/− versus Lmna^ΔK32/ΔK32 mice. The slight increase in muscle fibers with centrally located nuclei in muscle of Lmna^ΔK32/ΔK32/Lap2α^−/− versus Lmna^ΔK32/ΔK32 mice would be consistent with an increased regeneration/growth activity. However, the prominent lamin mutant-linked defect in myoblast differentiation in vitro may preclude any further improvement of the double mutant phenotype.

Overall, our studies show that a LAP2α-independent defect of the assembly and stability of ΔK32 lamin A/C and the accompanying loss of lamin A/C at the peripheral lamina are prominent in Lmna^ΔK32/ΔK32 mice and responsible for the pathologies. In contrast, the nucleoplasmic ΔK32 lamin A/C is still able to bind LAP2α and function in the regulation of tissue progenitor cells.

Mice were maintained in accordance with the procedures outlined in the Guide for the Care and Use of Laboratory Animals. Animal experiments were performed according to permissions from Austrian authorities. Data was obtained by observers who did not know the genotype of the animals. Lap2α-deficient mice were generated by deleting Lap2α-specific exon 4 in the Lap2 gene (also known as thymopoietin, Tmpo) using the Cre/loxP system (Naetar et al., 2008). Lmna^ΔK32/ΔK32 mice were generated by a knock-in strategy replacing the wild-type Lmna exon 1 with an exon 1 where the lysine at position 32 (delAAG) is deleted (Bertrand et al., 2012). In order to obtain double mutants and littermate controls, mice heterozygous for both Lap2α and Lmna ΔK32 (Lap2α^+/–, Lmna^+/ΔK32) were crossed. All experiments were performed in a mixed genetic background (C57BL/6, B6129F1, BALB/c) on postnatal day 15–18. Animals were killed by cervical dislocation. For genotyping, genomic DNA was prepared from tail tips and PCR analyses were performed using puRETaq Ready-To-Go PCR beads (GE Healthcare Biosciences, NJ, USA) in a PTC-200 Peltier Thermo Cycler (TJ Research). The used primers were:

Lap2α: Exon 4: 5′-CACAATCCCTAGAGGACTTCACTT-3′, Intron 4: 5′-CTGTGACTTTGCTGGCCTTCCAGTCTA-3′ and Exon 3: 5′-CAGGGAACTGAATCGAGATCCTCTAC-3′; Lmna: intron 2 forward: 5′-CAAAGTGCGTGAGGAGTTCA-3′ and intron 2 reverse: 5′-TGACAGCATAGGCCCTGTCAC-3′.

Following isolation, organs were immediately dipped into pre-chilled 2-methyl-butane and snap frozen in liquid nitrogen, embedded in TBS™ tissue freezing medium (Triangle Biomedical Sciences, Durham, NC, USA), and 5 µm sections were cut using a Cryostat HM500 OM at −23°C. Alternatively, tissues were fixed in 4% formaldehyde (Rotifix from Roth, Karlsruhe, Germany), dehydrated, cleared, embedded in paraffin and sectioned using a Leica RM 2155 microtome. Hematoxylin and Eosin staining was done according to standard protocol using an automated Ass-1 staining unit. Immunostainings for embryonic myosin heavy chain were performed using mouse monoclonal antibody against embryonic myosin (F1.652; DSHB, University of Iowa, Iowa City, USA) and biotinylated anti-mouse antibody. Stains were developed using DAB (Vector Laboratories, Burlingame, CA, USA), and nuclei were counterstained with Hematoxylin. The sections were dehydrated, mounted in Entellan (Merck, Darmstadt, Germany), and analyzed using a Zeiss Axio Imager.M1 microscope equipped with a Zeiss AxioCam MRc5 and images processed by AxioVision Rel. 4.5 software.

For immunofluorescence microscopy, cryosections were fixed either in 3.7% formaldehyde (Merck, Darmstadt, Germany) in PBS or in ice-cold acetone. Paraffin sections were incubated in xylene for 20 minutes, isopropanol for 10 minutes, 96%, 80%, 70% and 60% ethanol for 2 minutes each, and in double distilled water for 5 minutes. Rehydrated sections were incubated for 60 minutes in citrate buffer (1.8 mM citric acid and 8.2 mM sodium citrate) at 100°C. After washing in PBS, sections were incubated in 0.1% Triton X-100/PBS for 30 minutes, blocked with goat serum (Vectastain; Vector Labs, Burlingame, CA, USA), and incubated with antibodies and Hoechst dye as described previously (Naetar et al., 2007). Samples were viewed in a Zeiss Axiovert 200 M microscope equipped with a Zeiss LSM510META confocal laser-scanning unit, an alpha Plan-Fluor 100×/1.45 oil and a Plan-Apochromat 63×/1.40 oil DIC MC27 objectives (Zeiss). Images were prepared with Adobe Photoshop software.

The following antibodies were used: goat polyclonal anti-lamin A/C antibody N18 (Santa Cruz Biotechnology Inc., Heidelberg, Germany), rabbit antiserum to LAP2α (Vlcek et al., 2002), mouse monoclonal antibody to LAP2α (15/2) (Dechat et al., 1998), rabbit serum NCL-KI67p (Novocastra Lab., New Castle UK) and rabbit antiserum to desmin (ab8592; Abcam, Cambridge, UK). DNA was stained with DAPI (#32670, Sigma, St Louis, MO, USA).

Primary fibroblasts were isolated 1–3 days after birth from back skin of newborn mice as described previously (Andrä et al., 1998). Cells were cultivated in high glucose DMEM, 10% fetal calf serum (FCS), 50 U/ml penicillin, 50 µg/ml streptomycin and 0.2 µM L-glutamine (all from Invitrogen, Carlsbad, CA, USA) at 37°C and 5% CO₂. Experiments were performed between passage 1 and 4. For immunofluorescence, cells were seeded on coverslips and processed as previously described using the following antibodies: goat polyclonal anti-lamin A/C antibody N18 (Santa Cruz), mouse monoclonal anti-lamin A/C antibody, clone 4C11, provided by E. Ogris (Roblek et al., 2010), rabbit antiserum to LAP2α (Vlcek et al., 2002), goat polyclonal anti-lamin B antibody (C20, Santa Cruz).

Fluorescence intensity measurements were made using the profile tool in Zeiss LSM Image Browser software, version 4.2.0.121. Ratios of nucleoplasmic to peripheral mean A-type lamin fluorescence intensities were calculated for 25 or 30 fibroblasts of each genotype.

Primary myoblasts were obtained from de-skinned front and hind limbs of neonatal mice (2-days old) as described previously (Gotic et al., 2010). To induce muscle differentiation, proliferation medium (20% FCS/2.5 ng/ml basic FGF/Ham's F-10/penicillin and streptomycin) was replaced with DMEM containing 5% horse serum and penicillin and streptomycin. All cells were kept on collagen-coated dishes in a humidified atmosphere at 37°C and 5% CO₂.

Mice were killed and single fibers from particular muscles (tibialis anterior extensor, digitorum longus, soleus, gastrocnemius, quadriceps, triceps and biceps brachii) were isolated according to the method of Shefer and Yablonka-Reuveni and modified as described in Gotic et al. (Shefer and Yablonka-Reuveni, 2005; Gotic et al., 2010). In brief, muscles were collected in PBS and subsequently incubated in sterile 0.2% collagenase I (Gibco Life Technologies, Carlsbad, CA, USA)/DMEM solution (3 ml/50 mg of tissue) for 1.5–2 hours in a shaking water bath at 37°C. Muscle digestion was stopped by transferring samples to a series of DMEM-containing Petri dishes coated with filtered horse serum. Single fibers were released by gentle trituration, collected in DMEM and pelleted by centrifugation at 17 g for 5 minutes. After washing twice in PBS, fibers were resuspended in 0.01% collagenase II (Gibco Life Technologies)/0.15 U/ml dispase II (Roche Applied Science, Mannheim, Germany)/PBS and incubated for 30 minutes at 37°C with shaking. Samples were filtered through 40 µm pore cell strainers and cells were pelleted by 5 minutes centrifugation at 210 g. After two washes in PBS, samples were stained on ice for 30 minutes in 2% FCS/PBS containing the following antibody cocktail: anti-CD45, APC-conjugated anti-mouse CD45 (Ly-5); anti-Sca1, PerCP–Cy5.5-conjugated anti-mouse Sca-1 (Ly-6A/E); anti-Mac1, APC-conjugated anti-mouse CD11b; and anti-β1-integrin, PE anti-mouse/rat CD29 (all from eBioscience, Frankfurt, Germany; and anti-CXCR4, FITCS rat anti-mouse CD184 (CXCR4) (BD Pharmingen™Heidelberg, Germany).

Cells were washed and analyzed using a FASCAria equipped with DIVA acquisition software (BD Biosciences) as described previously (Cerletti et al., 2008). In brief, viable cells were first analyzed on the basis of size and granulation. Subsequently, a population of cells homogenous in size was tested for the expression of CD45, Mac1 and Sca1 surface markers and a subpopulation of CD45−/Sca1−/Mac1− cells was selected for CXCR4 and β1-integrin expression analysis. The number of CD45− Sca1− Mac1− CXCR4+ β1-integrin+ cells in each mouse sample was presented as percentage within the parent CD45− Sca1− Mac1− population.

Cells and tissues were lysed and analyzed by SDS-PAGE and immunoblotting as previously described (Naetar et al., 2008) using the following antibodies: antiserum to LAP2α (Vlcek et al., 2002), goat polyclonal anti-lamin A/C serum N18 (Santa Cruz), anti-lamin B (C-20, Santa Cruz), anti-emerin (NCL-Emerin, Novoscastra), rabbit polyclonal actin antiserum A-2066 (Sigma), anti-γ-tubulin (B-5-1-2, Sigma) and rabbit polyclonal antiserum to histone 3, (Abcam, Cambridge, MA, USA). Quantification of protein levels was performed with LICOR Odyssee Infrared Imaging System, application software version 2.1.12. Band intensities of lamins A and C were combined and normalized to the band intensity of actin or γ-tubulin or histone 3 as loading control.

Total RNA was isolated from muscle tissue and cultured cells using TRIzol® reagent (Invitrogen, Carlsbad, CA, USA) or RNeasy® Plus Micro Kit (Qiagen, Hilden, Germany). cDNA was synthesized by First Strand cDNA Synthesis Kit for RT-PCR (Roche Applied Science, Mannheim, Germany) according to manufacturer's instructions and specific sequences were subsequently amplified on an Mastercycler® ep realplex (Eppendorf, Hamburg, Germany) using MESA GREEN qPCR MasterMix Plus for SYBR Assay I TTP (Eurogentec, Liege, Belgium) for quantitative PCR. Specific primers are listed in supplementary material Table S1 (see also Gotic et al., 2010; Ozawa et al., 2006; Usami et al., 2003). Data were documented using Mastercycler® ep realplex software (Eppendorf, Hamburg, Germany) and processed using Microsoft Excel XP. Endogenous levels of Hprt (hypoxanthine-guanine phosphoribosyltransferase) in quantitative PCR were used for data normalization according to the Pfaffl method (Pfaffl, 2001).

The ΔK32 mutation was introduced into prelamin A cDNA, in the pET24-LA construct (Goldman et al., 2004) by in vitro mutagenesis using a QuikChange™ Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA), using the following primers: forward: 5′-AGGAGGAGGACCTGCAGGAGCTCAATG-3′, reverse: 5′-AGGTCCTCCTCCTGCAGCCGGGTGA-3′. The construct was sequenced before use. Wild-type and ΔK32 prelamin A were expressed in bacteria as described previously (Dechat et al., 2000) and resolved using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Purified rat vimentin was used as a negative control (Foisner et al., 1988). Nitrocellulose membranes were stained with PonceauS, washed in PBST (PBS, 0.05% Tween 20) and incubated in overlay buffer (10 mM Hepes, pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 2 mM EGTA, 0.1% Triton X-100, 1 mM DTT) for 1 hour with three changes. After blocking with 2% BSA in overlay buffer, membranes were probed overnight at 4°C with in vitro-translated, radioactively labeled FLAG-tagged LAP2α, diluted 1∶50 in overlay buffer plus 1% BSA (w/v) and 1 mM PMSF. For this, a plasmid containing FLAG-tagged LAP2α cDNA (pSV5) (Vlcek et al., 1999) was in vitro translated using the TnT® T7 Quick Coupled Transcription/Translation System (Promega, Mannheim, Germany) according to the manufacturer's instructions using ³⁵S-labelled methionine (Hartmann Analytic, Braunschweig, Germany). After extensive washing in overlay buffer, nitrocellulose was air dried, and bound proteins were detected by autoradiography.

For solubility assays, 1×10⁷ mutant or wild-type fibroblasts were lysed by sonication in 3 ml lysis buffer (20 mM Hepes, pH 7.4, 78 mM KCl, 42 mM NaCl, 10 mM EGTA, 8.4 mM CaCl₂, 4 mM MgCl₂, 1 mM DTT) supplemented with protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). Following addition of 0.5% Triton X-100 and 0.1% SDS, soluble fractions were obtained by centrifugation at 1700 g for 10 minutes. Total cell lysates and soluble fractions were used for immunoblotting and protein amounts were quantified by ImageJ.

Immunoprecipitation of ΔK32 lamin A/C was done from soluble cell fractions (see above) as described previously (Dechat et al., 2000) using mouse monoclonal 3A6-4C11 anti-lamin A/C antibody (Active Motif, Carlsbad, CA) and protein-A–Sepharose beads (Sigma).

Data are presented as the mean of n individual experiments (n being indicated in each figure), the error bars denote standard errors. Log-Rank tests for survival curves, Chi-squared test, Student's t-test and one-way ANOVA were applied using Microsoft Excel HP. Statistical significance was assumed at a P-value <0.05, and is indicated on graphs using an asterisk (*** indicates P<0.01).

We thank Thomas Sauer, MFPL, Vienna, for valuable help in FACS analysis of SMPCs. The monoclonal antibody F1.652, developed by Helen Blau, was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52242. We thank Bob Goldman, Northwestern University Chicago, for the generous gift of lamin A plasmids and Egon Ogris, Medical University Vienna, for anti-lamin A/C monoclonal antibody.