Nuclear envelope defects associated with LMNA mutations cause dilated cardiomyopathy and Emery-Dreifuss muscular dystrophy

The nuclear branch of the intermediate filament protein family is represented by a group of at least seven proteins known as the nuclear lamins (Gerace and Burke, 1988; Stuurman et al., 1998). These proteins are localized largely at the nuclear periphery within the nuclear lamina, a protein meshwork that forms the nuclear face of the nuclear envelope. The nuclear lamins fall in to two broad sequence classes, A-type and B-type. In adult mammalian somatic cells there are two major A-type lamin species, lamins A and C. In addition, several minor A-type lamins may also be present. All of the mammalian A-type lamins are encoded by a single gene and arise through alternative splicing of the same primary transcript (Fisher et al., 1986; McKeon et al., 1986). Lamins A and C are identical for the first 566 amino acids, after which their sequences diverge. Although lamin A (74 kDa) has a unique C-terminal extension of 98 amino acids, that of lamin C (65 kDa) is considerably smaller at only six residues.

The B-type lamins in mammalian somatic cells are represented by lamins B1 and B2. In contrast to the A-type lamins, these two proteins are encoded by separate genes and, with molecular weights of about 67 kDa, are intermediate in size between lamins A and C. Both lamins B1 and B2 contain a C-terminal CaaX motif constituting a site for the addition of a farnesyl lipid group, a feature also shared by lamin A (Beck et al., 1990; Sinensky et al., 1994). However, proteolytic processing of the C-terminus of newly synthesized lamin A, soon after its incorporation into the nuclear lamina, results in the loss of its farnesyl lipid tail (Weber et al., 1989). Thus, while the mammalian B-type lamins are farnesylated proteins, the mature A-type lamins are not.

In interphase cells the nuclear lamina maintains intimate contact both with the nuclear face of the inner nuclear membrane (INM) and with underlying chromatin. In addition, nuclear pore complexes (NPCs), which span both the inner and outer nuclear membranes (OMNs) and which mediate macromolecular trafficking across the nuclear envelope, are anchored to the nuclear lamina, making the lamina an important determinant of interphase nuclear architecture (Gerace et al., 1984). The farnesyl lipid tails of lamins B1 and B2 are thought to mediate interactions of the nuclear lamina with the INM. Additional connections between the INM and both A- and B-type lamins involve a number of INM-specific integral membrane proteins. These include the lamina-associated protein (LAP) 1 and 2 families, the lamin B receptor (LBR) and emerin, a ubiquitously expressed 29 kDa LAP2-related protein (Wilson, 2000).

The functional interaction of the nuclear lamina with INM proteins may be considered reciprocal in nature. Although the integral INM proteins certainly have a central role in the attachment of the lamina to the INM, the lamina itself may also contribute to the appropriate localization of these proteins. Several studies have indicated that specific sorting signals are not required for membrane proteins to access the INM (Ellenberg et al., 1997; Ostlund et al., 1999; Powell and Burke, 1990). Instead, proteins that are mobile within the ER membrane may have relatively unrestricted access to the INM via the outer nuclear membrane (ONM), which forms part of the ER, and the membrane continuities joining the INM and ONM at the periphery of each NPC (Bergmann and Singer, 1983; Torrisi and Bonatti, 1985; Torrisi et al., 1987). Integral proteins concentrate in the INM only if they are able to form interactions with other nuclear or nuclear envelope proteins. In fact, the majority of INM proteins are known to interact with nuclear lamins, chromatin or both. In this way, sorting of INM proteins involves a mechanism of selective retention. Although yet to be conclusively demonstrated, it is likely that at least some INM proteins may have functions that go beyond simply lamina attachment. For example, LBR has recently been shown to have sterol reductase activity (Holmer et al., 1998; Prakash et al., 1999; Silve et al., 1998) and in fact is capable of replacing sterol C14-reductase in yeast.

Studies in a variety of systems have revealed that expression of nuclear lamins is developmentally regulated (Roeber et al., 1990; Roeber et al., 1989; Stewart and Burke, 1987). The theme that has emerged is that although B-type lamins as a family are found in all cell types, A-type lamins are absent from those of early embryos. In the mouse, expression of A-type lamins does not commence until the onset of organogenesis on embryonic day 9, about midway through gestation. Although the majority of adult cell types do express A-type lamins, a few, most notably cells of the hematopoietic system, never acquire these proteins. Taken together, these findings indicate that A-type lamins, at least at the single-cell level, are not essential. Nevertheless, as described below, these proteins are far from dispensable.

Investigations in several laboratories recently linked defects in the A-type lamin gene (LMNA) to four human diseases. These are the autosomal form of Emery-Dreifuss muscular dystrophy (Bonne et al., 1999), dilated cardiomyopathy (DCM) with conduction system defects (Fatkin et al., 1999), limb girdle muscular dystrophy 1B with atrioventricular conduction disturbances (LGMD1B) (Muchir et al., 2000) and Dunnigan-type familial partial lipodystrophy (FPLD) (Cao and Hegele, 2000; Shackleton et al., 2000). At the same time, Lmna gene targeting experiments in mouse ES cells has led to the derivation of Lmna null mice (Sullivan et al., 1999). Although these mice undergo normal embryonic development, their postnatal growth is profoundly retarded, resulting in death by eight weeks of age. The phenotype of the Lmna null mice features severe skeletal myopathy, as well as cardiomyopathy, which resembles human EDMD. Consequently, these mice are considered to represent a unique model for this human disorder.

At the ultrastructural level, cells derived from Lmna null mice exhibit striking nuclear abnormalities (Sullivan et al., 1999). In fibroblasts, the normally round or oval nuclei adopt highly irregular shapes with perturbations in the distribution of many NE components. In particular, withdrawal of B-type lamins, LAPs and NPCs from one pole of the nucleus is frequently observed. In addition, emerin, the INM protein linked to EDMD, is inefficiently retained at the nuclear periphery, and instead is partially mislocalized to the peripheral ER. These effects are accompanied by conspicuous disruption of heterochromatin at the nuclear margins. Taken together, these results clearly indicate that A-type lamins play an essential role in the nuclear organization of adult cells.

The fact that four classes of LMNA alleles can give rise to distinct tissue-restricted diseases is extremely puzzling given that the A-type lamins are expressed in the majority of adult cell types. It is not yet known whether the mutations associated with each disease result in assembly defects of the A-type lamins and whether they can induce the same sort of structural abnormalities that are seen in the Lmna null mice. Of particular interest are the effects of these LMNA mutations on emerin distribution as defects in emerin and A-type lamins can each independently result in EDMD. In this paper we have introduced a variety of point mutations into both lamins A and C and have expressed these in both wild-type and Lmna null cells. Our findings indicate that certain DCM- and EDMD-associated LMNA mutations do indeed result in misassembly of A-type lamins and give rise to a variety of nuclear structural abnormalities which may contribute to the progression of these diseases.

Lmna^(–/–) and Lmna^(+/+) mouse embryo fibroblasts (MEFs), as well as HeLa cells, were maintained in a 7.5% CO₂ atmosphere at 37°C. All of the cells were cultured in DMEM (Gibco BRL, Gaithersburg, MD) containing 10% fetal bovine serum (HyClone, UT), 100 μg/ml penicillin/streptomycin (Gibco BRL) and 2 mM glutamine (Gibco BRL).

The rabbit antipeptide antibody specific for human lamin A has been previously described (Burke, 1990). The anti-HA monoclonal antibody (12CA5) was obtained from BAbCO Inc. (Richmond, CA). A monoclonal antibody specific for lamin B1 was purchased from Zymed Inc. (San Francisco, CA). The monoclonal antibody E3, specific for lamin B2, was a generous gift from Eric Nigg (University of Geneva, Switzerland). The rabbit antiserum against emerin was a gift from Glenn Morris (NE Wales Institute, UK).

All of the expression constructs employed in this study were based upon pcDNA3.1(–) (Invitrogen Inc., Carlsbad, CA) and wild-type lamin A/C cDNAs. The latter were a gift from Frank McKeon (Harvard Medical School, MA). Lamin A or C cDNAs, tagged at the 5′ end with a double-stranded synthetic oligonucleotide encoding the 13 amino acid 12CA5 epitope of influenza HA (Field et al., 1988), were inserted into pcDNA3.1(–) between the Nhe1 and Afl2 sites (Sambrook et al., 1989). The junction between the HA and lamin sequences was formed by an Xho1 site, which exactly replaced the first two codons of the lamin A/C sequence. A consensus initiation codon spanned by an Nco1 site immediately preceded the HA sequence just 3′ of the Nhe1 site.

Point mutations were introduced into the lamin A/C cDNAs by two-stage PCR. In this procedure, overlapping mutagenic primers (see below) were employed in two separate PCR reactions. The sense mutagenic primer (below) was used in conjunction with an antisense full-length primer (AGGTCTTAAGTTACATGATGCTGCAGTTCTGG for lamin A and ATATCTTAAGTCAGCGGCGGCTACCAC for lamin C), which incorporated a 3′ (in the sense orientation) Afl2 site. Similarly, the antisense mutagenic primer was used in conjunction with a full-length sense primer (GGCCCTCGAGACCCCGTCCCAGCGGC), which included a 5′ Xho1 site. PCR was carried out for 25 cycles (95°C for 30 seconds, 55°C for 30 seconds, 72°C for 2 minutes) using the Pfx DNA polymerase (Gibco-BRL), employing the manufacturer’s recommended conditions. PCR products were resolved by electrophoresis on 1% agarose gels and DNA recovered from appropriate bands using the QIAquick Gel Extraction kit (Qiagen, Santa Clarita, CA). PCR products were then combined and subjected to a second PCR reaction employing the two full-length lamin primers. PCR products from this second reaction were purified using the QIAquick PCR Purification kit (Qiagen, Santa Clarita, CA), digested with Afl2 and Xho1 (New England BioLabs Inc., Watertown, MA) and then fractionated on a 1% agarose gel. DNA corresponding to full-length lamin A/C was recovered as described above and then ligated into similarly digested pcDNA3.1(–). All of the constructs were sequenced in their entirety by the University of Calgary DNA sequencing service. The mutagenic primers that we used are as follows: L85R (GCCTACGAGGCCGAGCGCGGGGATGCCCGCAA and CTTGCGGGCATCCCCGCGCTCGGCCTCGTAGGC), N195K (CGGGTGGATGCTGAGAAAAGGCTGCAGACC and CATGGTCTGCAGCCTTTTCTCAGCATCCACCCG), R482W (CCCTTGCTGACTTACTGGTTCCCACCAAAGTTC and GAACTTTGGTGGGAACCAGTAAGTCAGCAAGGG), L530P (AGCCTGCGTACGGCTCCCATCAACTCCACTGGG and CCCAGTGGAGTTGATGGGAGCCGTACGCAGGCT).

A full-length human emerin cDNA (Bione et al., 1994) was isolated by PCR using human placental cDNA (Clontech Inc., Palo Alto, CA) as template. PCR was carried out in two steps. In the first, 0.2 ng of human placental cDNA was amplified for 30 cycles (95°C for 30 seconds, 55°C for 30 seconds, 72°C for 45 seconds) employing the Pfx DNA polymerase (above) in conjunction with a pair of emerin-specific primers corresponding to 5′ and 3′ UTR sequences (CGCCTGAGCCCGCACCCGCCATGGAC and AGGGACCATGGAGGGCCTGAGTCTGCTGCC). The amplified product was purified using the QIAquick PCR Purification kit. Approximately 1.5% of this first-reaction product was subjected to a second round of PCR using primers that spanned the initiation and termination codons and which introduced an Xho1 site (CCGCCCTCGAGAACTACGCAGATCTTTCG) and an Afl2 site (AGACCCTTAAGGCTCCCTCTAGAAGGGGTTG) at the 5′ and 3′ ends, respectively, of the emerin cDNA. This second PCR product was recovered using the QIAquick PCR Purification kit, digested with Afl2 and Xho1 and, after gel purification, was ligated in pcDNA3.1(–) downstream of the N-terminal HA tag sequence described above. A version of emerin lacking its C-terminal membrane anchor sequence (EmΔC) and which was used for the in vitro translation experiments was also prepared by PCR using an alternative antisense primer, which introduced a stop codon and Afl2 site after codon 221 (CAGAGCTTAAGCTAGCGATCCTGGCCCAGCCC). This truncated emerin cDNA was also introduced into pcDNA3.1(–)HA.

Lamin and emerin cDNAs in pcDNA3.1 were transcribed and translated in vitro using the TnT® quick-coupled transcription/translation system from Promega Inc. (Madison, WI), according to the manufacturer’s instructions. Briefly, 50 μl of total transcription/translation mix, either with or without 20 μCi ³⁵S-methionine, was programmed with 2 μg of plasmid DNA. The translation mix was then incubated for 90 minutes at 30°C. To immunoprecipitate lamin-emerin complexes, 2 μl of combined reaction mixes in 1 ml of immunoprecipitation buffer (IPB, containing 10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM DTT, 10 μg/μl each of chymostatin, leupeptin, antipain and pepstatin (CLAP), 1mM PMSF and 0.1% Triton X-100), were incubated for 14 hours at 4°C, with 4 μl of Dynabeads (Dynal Biotech, Oslo, Norway) and 1 μl of rabbit anti-emerin serum or 1 μl of guinea pig anti-lamin A/C. The beads were subsequently washed five times by resuspension for 10 minutes in IPB lacking CLAP. After two final washes in a solution containing 10 mM Tris-HCl, pH 7.4, 1mM DTT, 1 mM PMSF, the beads were resuspended in 10 μl SDS in SDS-polyacrylamide gel sample buffer (Laemmli, 1970) and fractionated by electrophoresis on an SDS-polyacrylamide (10%) mini-gel. On completion of electrophoresis, gels were processed for fluorography employing Amplify® (Amersham Pharmacia Biotech Inc., Piscataway, NJ) and exposed to X-ray film.

Plasmids were propagated in E. coli XL-1 Blue (Stratagene Inc. La Jolla, CA) and were prepared for transfections using Qiagen Maxi Prep Kits (Qiagen, Santa Clarita, CA). The DNA was subsequently introduced into BHK cells, grown on glass coverslips, using the Superfect reagent exactly as described by the manufacturer (Qiagen). Cells were usually fixed and processed for immunofluorescence microscopy about 20-24 hours post transfection.

Cells grown on glass coverslips were fixed with formaldehyde and labeled with antibodies according to previously described procedures (Ash et al., 1977). In short, cells were fixed for 20 minutes at room temperature in 3% formaldehyde (prepared from paraformaldehyde dissolved at 80°C in phosphate-buffered saline). Following PBS washes, the fixed cells were permeabilized for 5 minutes at room temperature with 0.2% Triton X-100 in PBS and labeled with appropriate primary and secondary antibodies. In addition most samples were stained with the DNA-specific Hoechst dye #33258 to reveal the cell nuclei. Specimens were observed and photographed under appropriate illumination with a Leica DMRB microscope equipped with ×63 PL APO NA1.4 and ×100 FLUOTAR NA 1.32 objectives. Images were collected using a Princeton Instruments (Princeton, NJ) MicroMax cooled CCD camera linked to an Apple Macintosh G4 computer running IP Lab Spectrum software (version 1.3, Signal Analytics Inc). Figures were later composed using Adobe PhotoShop 5.5 and Deneba Canvas 7.0. For deconvolution microscopy, series of images acquired at 0.2 μm focal intervals were digitally processed using Micro-Tome version 3.1 (VayTek, Inc.) to yield stacks of confocal slices.

Cells grown on 35 mm tissue culture dishes were processed for immunoblot analysis 22-24 hours post transfection with appropriate plasmids. Cells were solubilized directly in SDS-polyacrylamide gel sample buffer (Laemmli, 1970) and fractionated by electrophoresis on an SDS-polyacrylamide (10%) mini-gel. On completion of electrophoresis, gels were blotted onto nitrocellulose filters (usually BA85 from Schleicher and Schuell, Keene, NH) (Burnette, 1981), employing a semi-dry blotting apparatus manufactured by Hoeffer Scientific Instruments Inc. (San Francisco, CA). Filters were blocked, labeled with the anti-HA 12CA5 monoclonal antibody and then developed with peroxidase-conjugated goat anti-mouse IgG (Biosource International, Camarillo, CA) exactly as previously described (Burke et al., 1982).

At least four human diseases have now been mapped to discrete mutations within the LMNA gene. The majority of these mutations are situated in exons common to both the lamin A and C messages and may modify the functions of both major A-type lamin proteins. From studies on mice deficient in A-type lamins, these proteins play a crucial role in the maintenance of interphase nuclear architecture. However, as A-type lamins are present in the majority of adult cell types, it is far from clear how mutations in the LMNA gene can give rise to distinct tissue-specific phenotypes involving pathological changes in cardiac and skeletal muscle, as well as in adipose tissue. To begin to address this question we have introduced a series of disease-linked point mutations into both lamins A and C cDNAs to establish whether any of these may induce overt alterations in nuclear lamina assembly, perhaps leading to changes in nuclear organization and the development of characteristic disease-related ultrastructural phenotypes.

Four point mutations were introduced into both lamin A and C cDNAs by two-stage PCR. The identity of these mutants is summarized in Fig. 1A. Two of these were mutations within the lamin alpha-helical rod domain (L85R and N195K) and have been linked to DCM (Fatkin et al., 1999). The remaining two are located within the nonhelical C-terminal domain and give rise to autosomal EDMD (L530P) (Bonne et al., 2000) and FPLD (R482W) (Shackleton et al., 2000). To facilitate detection in vivo an HA epitope tag (Field et al., 1988) was introduced at the N-terminus of each of the mutant proteins as well as that of wild-type lamins A and C.

Expression of each of the lamin mutants in HeLa cells was followed by immunoblot analysis of cell lysates and revealed proteins of the anticipated sizes, identical to those of wild-type controls (Fig. 1B). Each of the lamin A proteins migrated as a doublet, most likely corresponding to full-length (lamin A₀) and processed forms of the molecule. Lamin A processing involves proteolytic cleavage at Y646, resulting in loss of a C-terminal peptide containing the farnesyl lipid tail (Weber et al., 1989). Of the four lamin A mutants that we have examined, only lamin A L85R (LaA L85R) showed any obvious preponderance of lamin A₀ over mature lamin A (1.6-fold relative to wild-type). Although this likely represents a slight decline in processing efficiency relative to wild-type lamin A (Horton et al., 1992), we cannot at present formally rule out the possibility that mature LaA L85R is preferentially degraded. The lamin C proteins, both wild-type and mutant, which do not undergo any form of proteolytic processing, migrate as single bands, all with identical mobilities.

Immunofluorescence analysis of HeLa cells expressing HA-tagged forms of each lamin A allele revealed assembly defects associated with the DCM mutants LaA L85R and LaA N195K (Fig. 2; Fig. 3). Surprisingly, however, their microscopic phenotypes were quite distinct, despite the fact that they give rise to the same clinical symptoms. LaA L85R, although largely NE-associated (Fig. 2B,G), exhibits a distinctly punctate distribution across the nuclear surface (Fig. 3F,I) in 100% of transfected cells, and in this way appears subtly distinct from wild-type lamin A, which displays a more uniform distribution (Fig. 2A,F; Fig. 3E,H). LaA N195K localized almost exclusively to the nucleoplasm with little or none detectable at the nuclear envelope (Fig. 2C,H). 25% of cells expressing LaA N195K contained prominent spherical intranuclear aggregates or inclusions (Fig. 2C,H, inset; Fig. 3C), possibly as a result of higher levels of expression. These inclusions, which are most frequently concentrated at the nuclear periphery, were also positive for endogenous lamin B2 (Fig. 3D) indicating that LaA N195K can perturb B-type lamin organization. The LaA L530P mutant (Fig. 2E,J) exhibited a localization that was very similar to wild-type, although occasionally nucleoplasmic labeling was evident in a small percentage of cells. The last of the four lamin A mutants, LaA R482W, which is associated with FPLD, localized largely to the nuclear periphery and was indistinguishable from wild-type lamin A in terms of targeting and assembly. Identical results were obtained with each of these lamin A alleles in primary fibroblasts derived from wild-type and Lmna null mouse embryos (Fig. 5A-E) as well as in BHK cells and P19 embryonal carcinoma stem cells (which lack A-type lamins; data not shown). All of these results are very similar to those described by Östlund et al. (Östlund et al., 2001).

A strikingly different series of results were obtained when HeLa cells were transfected with the corresponding lamin C mutants (Fig. 2K-O). Although wild-type lamin C is targeted appropriately to the HeLa nuclear envelope (Fig. 2K), the LaC L85R, N195K and L530P mutants were found to be almost completely assembly incompetent and could be detected only within the nucleoplasm, often appearing in aggregates (Fig. 2L,M,O). This aggregation phenotype was most obvious with LaC L85R (Fig. 4B). Although these mutants showed little or no evidence of nuclear envelope association, they nevertheless exerted noticeable, albeit variable effects on endogenous lamin A localization (Fig. 4). This was determined in double label experiments employing a monoclonal antibody against the HA-tag on the recombinant lamin C alleles and a rabbit antibody against a lamin A-specific peptide, which would recognize only the endogenous molecule. All three lamin C mutants (L85R, N195K and L530P) diverted endogenous HeLa lamin A to the nucleoplasm (Fig. 4B-D,F-H) where it appeared in aggregates or inclusions of variable shapes and dimensions. This was most conspicuous in the case of LaC N195K where large lamin A aggregates could be seen in 63% of cells, and which were comparable in size to, but distinct from, nucleoli (Fig. 4G). Curiously, HA-LaC N195K was not detected in these aggregates (Fig. 4C,I). Whether this was due to inaccessibility of the HA-epitope or genuine absence of the mutant lamin from these aggregates is not clear, although the former would appear to be more likely. Clearly, each of these three lamin C alleles exert dominant effects on nuclear envelope organization insofar as they cause the mislocalization of other A-type lamin proteins. As with LaA R482W, the LaC R482W mutant exhibited a distribution that was indistinguishable from wild-type (Fig. 2N).

Analyses of the targeting and assembly properties of wild-type and mutant forms of lamin A yielded qualitatively similar results in HeLa cells (Fig. 2; Fig. 3; Fig. 4), Lmna^(+/+) fibroblasts and Lmna^(–/–) fibroblasts (Fig. 5A-E). In the case of the various forms of lamin C, however, significant differences were evident depending on whether these proteins were expressed in Lmna^(–/–) fibroblasts versus HeLa cells and Lmna^(+/+) fibroblasts. Although only the L85R, N195K and L530P mutant forms of lamin C failed to associate with the nuclear periphery in HeLa cells (Fig. 2), in Lmna^(–/–) fibroblasts none of the lamin C forms, neither wild-type nor mutant, showed any concentration at the nuclear envelope (Fig. 5G-K). Instead, these proteins accumulated in the nucleoplasm, often forming aggregates (Fig. 5H). This confirms earlier results where we showed that lamin C, in the absence of lamin A, fails to assemble efficiently at the nuclear envelope of P19 EC cells (Horton et al., 1992). However, co-transfection with nontagged wild-type lamin A restored the ability of HA-lamin C to associate with the nuclear envelopes of either Lmna null fibroblasts (Fig. 5F,L) or P19 EC cells (not shown). The consensus of the data displayed in Fig. 4 and Fig. 5 is that either A-type lamin can influence the distribution of the other.

X-linked EDMD is caused by mutations in the gene encoding emerin, an inner nuclear membrane protein (Bione, et al., 1994). We have previously shown that in Lmna^(–/–) fibroblasts, emerin is partly mislocalized to the peripheral ER, but that this aberrant localization could be rescued by overexpression of wild-type lamin A (Sullivan et al., 1999). Because each of the lamin mutants that we have constructed is associated with dominant forms of EDMD, DCM or FPLD, we examined the effect of overexpression of each of these mutants on emerin localization (Fig. 6). To accomplish this, transiently transfected HeLa cells, double labeled with rabbit anti-emerin and mouse monoclonal anti-HA, were scored for loss of nuclear-envelope-associated emerin. Nontransfected populations of HeLa cells exhibit a very uniform level of emerin expression with little variation in the labeling intensity of the NE. As shown in Fig. 6B, the EDMD mutants LaA L530P and LaC L530P both caused a partial decline in NE-associated emerin. This was found to occur in 38% of cells, a 2.2-fold increase over that observed with wild-type lamins A and C. This difference is considered statistically significant (P<0.001). The effect of LaA L530P was, however, more robust than that of LaC L530P when viewed at the single-cell level (Fig. 6A). Often, emerin that was displaced from the NE appeared in cytoplasmic aggregates (Fig. 6, arrowheads) rather than becoming uniformly distributed throughout the peripheral ER. The significance of this distribution is not clear. The two cardiomyopathy mutants LaA L85R and N195K caused a less dramatic, but nevertheless significant decline of NE-associated emerin (Fig. 6B). Of their two lamin C counterparts, only LaC N195K caused any detectable loss of emerin from the NE. The FPLD mutant forms of lamins A and C, R482W, had no discernible effect on emerin distribution when compared with wild-type lamins A and C. The consensus of these results indicate that DCM and EDMD mutants (L85R, N195K and L530P) but not the FPLD mutant (R482W) have a moderate dominant negative effect on emerin distribution. Qualitatively similar results were obtained in Lmna^(+/+) MEFs (data not shown).

We further determined whether any of these mutants had the capacity to rescue the aberrant emerin distribution that we have previously reported in Lmna^(–/–) MEFs (Sullivan et al., 1999). As shown in Fig. 7, only wild-type lamin A (Fig. 7A,B) and LaA R482W (Fig. 7G,H) when over-expressed restored the NE-localization of emerin. LaA L85R, N195K and L530P were incapable of altering emerin localization to any appreciable extent in these cells. In addition, none of the lamin C alleles, neither wild-type nor mutant, could restore the normal NE-localization of emerin (data not shown).

Because lamin A is known to interact with the nucleoplasmic N-terminal domain of emerin (Clements et al., 2000), we determined the effect of each of the A-type lamin mutations described here on this association. To accomplish this, we transcribed and translated both HA-tagged emerin N-terminal domain (EmΔC, residues 3-221) and lamin A mutant cDNAs in vitro (Fig. 8A). Emerin-lamin interactions were then assayed by co-immunoprecipitation employing either an anti-lamin antibody (Fig. 8) or an anti-emerin antibody (not shown) coupled to magnetic beads. As revealed in Fig. 8B, although wild-type lamin A and LaA R482W displayed comparable interactions with emerin, interaction of LaA L530P with emerin was virtually undetectable. This finding is consistent with the dominant negative effect that LaA/C L530P displays with respect to emerin localization in HeLa cells (Fig. 6). The LaA L85R and LaA N195K display reduced but nevertheless detectable interactions with emerin in vitro (Fig. 8B).

Taken all together, the results described here clearly demonstrate that the LaA/C L85R, N195K and L530P alleles each can cause structural changes at the nuclear periphery, which includes perturbation of A-type lamin interactions with at least one integral INM protein, emerin. By contrast, the FPLD mutants LaA/C R482W behave in a fashion that is identical to the wild-type proteins, at least with respect to assembly properties and interactions with emerin. These findings lead us to conclude that at least in the case of EDMD and DCM, structural perturbations of the nuclear envelope and changes in nuclear architecture may contribute to the progression of the two disorders.

Studies carried out in a number of laboratories have recently provided convincing evidence that nuclear lamins, both A-type and B-type, play key roles in the maintenance of nuclear envelope integrity (Lenz-Bohme et al., 1997). Indeed, depletion of the single lamin expressed in Caenorhabditis elegans is early embryonic lethal (Liu et al., 2000). Loss of peripheral heterochromatin in the nuclei of both hepatocytes and fibroblasts derived from Lmna null mice further implicate A-type lamins in the large-scale organization of interphase nuclear architecture (Sullivan et al., 1999). Given that A-type lamins are expressed in the majority of adult mammalian cells, it comes as no small surprise that diseases involving mutation or deletion of the LMNA gene, in both humans and mice, feature pathologies restricted to only a few tissue types, most notably including cardiac and skeletal muscle and adipose tissue.

To explore this paradox further we have examined the effects of various A-type lamin mutants on lamin assembly and nuclear organization in HeLa cells and both Lmna^(+/+) and Lmna^(–/–) mouse embryo fibroblasts. We reasoned that mutants characteristic of a particular disorder might display similar phenotypes when examined at the cellular and ultrastructural levels. Three of the mutants that we chose and which have been linked to DCM (L85R, N195K) and EDMD (L530P) involve changes in amino acid residues that are 100% conserved among both A-and B-type lamins from human to hydra (from a comparison of 16 lamin sequences deposited in GenBank) (B.B., unpublished)! The fourth mutant, R482W, which is linked to FPLD, involves a residue that exhibits only conservative changes (R to K) among the same species. In all cases, the conserved residue forms part of a larger conserved motif.

The two DCM mutations, L85R and N195K, lie within the lamin rod domain. Residue L85 is predicted to reside at the coiled-coiled interface within the first heptad repeat of coil 1b, whereas N195 is situated at the distal end of the same coil. Given the position that N195 occupies within the heptad repeat pattern, this residue would be predicted to face outwards from the lamin homodimer and therefore away from the coiled-coil interface. L85 and N195 are therefore not comparable in terms of their likely contributions towards the structure and organization of the lamin rod domain. Heald and McKeon (Heald and McKeon, 1990) have previously shown that mutations that affect residues located on the inner or interacting surface of the lamin A coiled-coil (positions 1, 4, 5 and 7 of each heptad repeat) invariably interfere with appropriate assembly into the nuclear lamina. In the case of LaA L85R (position 5), the effect is quite subtle and results in a speckled distribution of the mutant lamin across the nuclear surface but does not cause mislocalization to the nucleoplasm. However, the corresponding lamin C mutant is completely assembly incompetent. These data suggest to us that lamin A L85R is probably incapable of correct assembly into the lamina and that the NE association that we do see might simply result from hydrophobic interaction of the farnesyl lipid tail with protein or lipid components of the inner nuclear membrane.

Both LaA N195K and LaC N195K exhibit no detectable NE-association and instead accumulate within the nucleoplasm. In fact, LaA N195K partitions into intranuclear aggregates in about 25% of transfected cells. Clearly, these proteins are defective in their ability to assemble into the nuclear lamina. However, they cannot be written off as functionally dead molecules. The reasons for this are twofold. First, LaC N195K will cause the relocation of wild-type lamin A to the nucleoplasm, and in this way functions in a dominant negative fashion. Second, we know that loss of A-type lamin expression in humans, resulting from a nonsense mutation in codon 6, causes autosomal dominant EDMD (Bonne et al., 1999). If the N195K mutation truly resulted in a complete loss of lamin function we would expect that this mutation should give rise to EDMD rather than DCM.

The single EDMD mutation that we have examined, L530P, resides within a conserved sequence (TAL⁵³⁰IN) within the nonhelical C-terminal domain of both lamins A and C. Although this mutation has only a marginal effect on the localization of lamin A to the nuclear periphery, it virtually abolishes assembly of lamin C. As with the two DCM lamin C mutants, LaC L530P exerts a dominant negative effect on the localization of wild-type lamin A, indicating that it is able to perturb normal lamina organization. These results, in addition to those with the DCM lamin C mutants, support a model in which lamins A and C functionally interact within the nuclear lamina as recently suggested (Izumi et al., 2000). Such a notion is lent further credence by the fact that cotransfection with wild-type lamin A enhances the association of wild-type lamin C with the nuclear periphery of Lmna^(–/–) fibroblasts and P19 embryonal carcinomas.

There has so far been no comprehensive study of the fate of emerin in cells from patients suffering from autosomal dominant EDMD or DCM. The work that we have presented here, however, suggests that there may be some perturbation in the interaction of emerin with the nuclear lamina in patients carrying the L85R, N195K or L530P mutations and is consistent with findings that demonstrate a direct interaction between emerin and A-type lamins (Clements et al., 2000). It is perhaps significant that the one EDMD mutation that we have examined (L530P) exhibits a moderate dominant negative effect on emerin localization in HeLa cells, fails to rescue the aberrant emerin localization observed in Lmna^(–/–) fibroblasts and interacts only weakly, if at all, with emerin in vitro. Östlund et al. have observed similar moderate dominant effects of lamin mutants on emerin distribution (Östlund et al., 2001). It is important to appreciate, however, that A-type lamins can not be the only determinants of emerin localization. For instance, overexpression of LaA L530P is less effective at perturbing emerin distribution in Lmna^(+/+) MEFs than in HeLa cells, indicating possible cell-type-specific differences in emerin anchoring at the nuclear envelope. Moreover, although ablation of the Lmna gene results in the partial mislocalization of emerin to the peripheral ER in Lmna^(–/–) MEFs, some emerin still remains localized to the NE in these cells (Sullivan et al., 1999). This suggests that there are other factors besides LaA that may contribute to emerin localization. This notion is reinforced by the finding that P19 EC stem cells that do not express A-type lamins still localize emerin to the nuclear envelope. Not surprisingly, emerin localization in P19 cells is completely refractory to LaA L530P overexpression (data not shown). The implication of these findings is that there are additional nuclear or NE components with which emerin may interact and which are expressed in a cell-type-specific fashion.

Although easily distinguishable, the three DCM and EDMD lamin A/C mutants that we have examined do exhibit some mutual phenotypic properties. Most notable are the defects in lamin C assembly, as well as the dominant negative effects on wild-type lamin A localization and, to a lesser extent, on emerin localization (at least in certain cell types). When combined with the data of Östlund et al., who have analyzed a more comprehensive selection of lamin A alleles (Östlund et al., 2001), we may conclude that two out of four DCM-linked mutations and four out of eight EDMD-linked mutations give rise to lamin assembly defects. It is possible that those mutations that fail to dramatically perturb lamin A assembly (Östlund et al., 2001) may in fact interfere with lamin C behavior, as we have observed for L85R and L530P. Clearly, it will be valuable to examine this issue.

Because DCM and EDMD do have certain clinical features in common, our findings, and those of Östlund et al. (Östlund et al., 2001), would be consistent with the suggestion made by Bonne et al. that DCM and EDMD may represent a single disease entity with variable expression of symptoms that are perhaps modified by additional genetic factors (Bonne et al., 2000). The same may also be true of limb girdle muscular dystrophy with atrioventricular conduction disturbances (LGMD1B). All three of the LMNA mutations that have been mapped in this disorder (Muchir et al., 2000) would be predicted to compromise lamin A or C assembly, or both. The deletion of K208 at the distal end of coil 1b would alter the phase of the heptad repeats, thereby interfering with homodimerization. The R377H mutation at the distal end of coil 2 has previously been shown by Heald and McKeon (Heald and McKeon, 1990), in unrelated studies, to block lamin A assembly, and might therefore have the same general features as the N195K mutation. The frameshift at codon 536 would eliminate the lamin A C-terminus (including the CaaX motif), which is known to be essential for incorporation into the nuclear lamina (Holtz et al., 1989; Kitten and Nigg, 1991; Krohne et al., 1989). Indeed, Loewinger and McKeon (Loewinger and McKeon, 1988) showed that C-terminal deletion to E554 results in the formation of nucleoplasmic lamin filaments with gross changes in nuclear shape.

In contrast to the DCM and EDMD mutants, the single FPLD lamin A/C mutant that we have examined (R482W) is, in all respects, indistinguishable from wild-type lamin A/C in terms of targeting, assembly and interaction with emerin. Morris and colleagues have recently arrived at virtually identical conclusions (Holt et al., 2001). In both mouse and human cells LaA/C R482W induced no overt perturbations of emerin or endogenous lamins. Furthermore, LaA R482W, like wild-type lamin A, has the capacity to rescue mislocalized emerin when over expressed in Lmna^(–/–) fibroblasts. Clearly, the R482W mutation is in a class apart from the three that give rise to DCM and EDMD. However, Vigouroux et al. have found that fibroblasts from FPLD patients carrying either the R482W or R482Q mutations exhibit abnormal nuclear morphologies (Vigouroux et al., 2001). These frequently involved the formation of NE blebs lacking other NE components such as B-type lamins and NPCs. The resemblance to the ‘nuclear herniations’ that we have previously reported in fibroblasts from Lmna^(–/–) mice is striking (Sullivan et al., 1999). Consistent with our results, and those of others (Östlund et al., 2001; Holt et al., 2001), Vigouroux et al. report no obvious defects in A-type lamin targeting in FPLD fibroblasts (Vigouroux et al., 2001). Instead, the mutant lamins are able to assemble into the nuclear lamina but in some way perturb its normal organization as well as its interactions with other NE components such as NPCs, and with peripheral heterochromatin. Vigouroux et al. also note that emerin maintains its associations with A-type lamins in FPLD fibroblasts (Vigouroux et al., 2001). Again, this is consistent with our observations that LaA R482W interacts with emerin in vitro and that it is able to rescue the aberrant emerin localization seen in Lmna^(–/–) fibroblasts. The question of course arises as to why we have not observed similar NE abnormalities in HeLa cells transiently transfected with LaA/C R482W. The explanation may reside in differences between primary cells and a stable cell line. Vigouroux et al. point out that only 10% of early passage FPLD fibroblasts exhibit recognizable nuclear ultrastructural changes and that this figure rises to about 25% with increasing passage number (Vigouroux et al., 2001). It is possible that in HeLa cells, as in low-passage FPLD fibroblasts, the incidence of obvious nuclear abnormalities following transient transfection may be sufficiently infrequent as to go unremarked. It is also conceivable that the manifestation of this nuclear structure phenotype requires one or more cycles of nuclear disassembly and reformation as occurs during mitosis. If so, we would not see this effect because the majority of cells that we examined, typically 20 hours post transfection, had not undergone division.

The results described here clearly indicate that certain LMNA mutations associated with DCM and EDMD can cause changes in NE and nuclear architecture. However, how these contribute to the disease process is still far from clear. Some aspects of the DCM and EDMD phenotypes could arise from increased nuclear fragility as a result of compromised nuclear envelope integrity. This in turn could lead to an elevation in the incidence of mechanical damage to both cardiac and skeletal muscle nuclei, which may be unable to withstand the forces generated during repeated cycles of muscle contraction (Hutchison et al., 2001). In fact, NEs derived from Lmna^(–/–) fibroblasts are more fragile than those derived from corresponding wild-type cells (T.S., B.B. and C.S., unpublished). However, in light of the role that A-type lamins are considered to play in the organization of heterochromatin and their documented interactions with Rb, a key regulator of cell cycle-dependent transcription (Ozaki et al., 1994), it is distinctly possible that supplemental gene expression effects may also contribute to the DCM, EDMD and LGMD1B phenotypes. In the case of FPLD, it is hard to present any argument at all in favor of nuclear fragility contributing to the disorder. As we have shown, the LaA/C R482W mutants are indistinguishable from the wild-type proteins in terms of their assembly properties, although this mutant does induce changes in nuclear architecture in fibroblasts from affected patients. However, nuclei in adipose tissue are unlikely ever to be subjected to the degree of mechanical stress evident in muscle cells. Because muscle cells are unaffected in FPLD and, conversely, adipose tissue is unaffected in DCM, EDMD and LGMD1B, nuclear fragility simply cannot be invoked as a universal mechanism to account for all four diseases. The alternatives are that A-type lamins are required for either the differentiation or survival of adipocytes. This could involve the furnishing of a permissive environment for the expression of adipocyte-specific genes or perhaps the provision of binding sites at the nuclear periphery for additional proteins that protect the cells against apoptosis. These are issues that we shall be addressing in vivo employing mice bearing specific Lmna point mutations.

We thank Frank McKeon and Glenn Morris for the gifts of cDNAs and antibodies, respectively. We are also grateful to Howard Worman and Jean-Claude Courvalin for sharing their unpublished data. B.B. was supported by grants from the Canadian Institutes of Health Research and the Alberta Heritage Foundation for Medical Research.