Emery-Dreifuss muscular dystrophy (EDMD) is an X-linked muscular disease first described in the early 1960s and characterized by early contracture of the elbows, Achilles tendons and post-cervical muscles and slow progressive muscle wasting and weakness, with humeroperoneal distribution in the early stages of the disease, and cardiomyopathy usually presenting as a heart block (1 ). The cardiomyopathy manifests as a cardiac conduction defect and the associated heart block is a frequent cause of death. Provided that diagnosis is made sufficiently early, the insertion of a cardiac pacemaker can be life saving. In no case has mental retardation, intellectual defect or involvement of other organs been described.
The EDMD gene was identified among a large number of candidates in a very gene rich region where the disease had been mapped (2 ). The EDMD gene is ubiquitously expressed and encodes a novel ubiquitous protein of 254 amino acids, emerin, localized at the nuclear rim of most cell types analyzed (2 -5 ). Emerin showed limited sequence similarity with LAP2, a ubiquitous integral membrane protein involved in association between the nuclear lamina and nuclear envelope (6 ). The similarities between the two proteins suggested that emerin is a member of the nuclear lamina-associated protein family (6 ). This finding was quite unique, as most muscle or cardiac disorders are caused by alterations in molecules specifically or highly expressed in muscle tissues.
We have further investigated emerin localization and we have conclusively established that emerin is indeed an integral membrane protein of the inner nuclear membrane. In this paper we report that emerin is, however, not exclusively localized at the nuclear membrane but that it is also found associated with cytoplasmic membranes. In heart its specific localization to desmosomes and fasciae adherentes of the intercalated discs could account for the characteristic conduction defects described in patients.
From the sequence analysis emerin appeared to possess multiple phosphorylation sites for a range of kinases (2 ). In vivo labeling of HeLa cells with [32P]orthophosphate confirmed that emerin is phosphorylated in vivo (Fig. 1 a). The antibodies immunoprecipitated two phosphorylated bands of 34 and 36 kDa of the same size as those commonly seen in Western blots. Treatment of HeLa cells with the phosphatase inhibitor okadaic acid (7 ) caused a relative increase in the 36 kDa form over the more abundant 34 kDa form and demonstrated that the two forms correspond to different levels of phosphorylation of emerin (Fig. 1 b).
Emerin is a small hydrophilic protein with a hydrophobic tail of 21 amino acids, 11 residues from the C-terminus. To establish whether the C-terminal hydrophobic domain of emerin was important for nuclear rim localization, mutants were constructed carrying different short deletions spanning the entire protein and an HA epitope tag at the N-terminus (Fig. 1 c). Upon transfection into COS cells the wild-type protein localized to the nuclear envelope (Fig. 1 d), as did proteins lacking the conserved region at the N-terminus (2 ) or with any of the internal deletions. Only the mutant lacking the last 27 amino acids spanning the C-terminal putative transmembrane domain (HA-[Delta]Trans) was unable to localize to the nuclear envelope and was concentrated inside the nuclei (Fig. 1 d). To confirm that the hydrophobic domain is essential for membrane localization, the transmembrane domain was fused to the C-terminus of the reporter protein GFP (GFP-Trans). The 27 C-terminal amino acids were sufficient to drive most of the transfected chimeric protein to the nuclear periphery (Fig. 1 d).
Electron microscopy immunogold studies of ultrathin frozen sections of skeletal muscle, heart and HeLa cells showed that emerin is associated with the inner nuclear membrane. In skeletal muscle sections most gold granules localized to the nuclear lamina or the inner nuclear membrane (Fig. 1 ) at positions where the two membranes are well separated by enlargements of the intramembrane cisterna. Similar results were obtained in heart (Fig. 1 f) and HeLa cells (not shown).
Emerin appears to be an integral membrane protein localized in the inner nuclear membrane and the main determinant of this localization resides in the C-terminal hydrophobic stretch, which can be considered a genuine transmembrane domain. A nuclear localization is, however, insufficient to ensure a normal phenotype and the stability and/or the specificity of emerin binding to other proteins must involve determinants in the rest of the protein. This is demonstrated by the behavior of a mutated emerin found in one patient (LB1520) presenting a frameshift mutation in the C-terminal moiety (2 ). In this patient a shorter protein was predicted, deleted of the emerin sequence starting from amino acid 169 but possessing a new hydrophobic tail compatible in amino acid composition and length with a nuclear envelope transmembrane domain (8 ; Fig. 2 a). Such a shorter emerin was indeed synthesized (5 ) and immunohistochemical analysis of a muscle biopsy of the patient demonstrated that a reduced but still detectable amount of emerin was at the nuclear rim (Fig. 2 b).
As previously shown by Manilal et al. (4 ), in brain tissue emerin was not exclusively found in the nuclear envelope. Cell fractionation of several cell types by mechanical shearing in hypotonic buffer separated emerin into nuclear and cytoplasmic `pools'. In HeLa cells (Fig. 3 a) the nuclear emerin (lane 2) was found in the lamina-associated fraction, resistant to nuclease digestion, high salt and detergent (lane 10). The cytoplasmic emerin (lane 3) co-localized with membranes (lane 4) but, unlike that in the nuclear fraction, it was readily solubilized in the presence of detergents (lines 5 and 6). Correct fractionation of lamins in nuclei and vimentin in cytoplasm excluded major cross-contamination between the fractions. The relative abundance of emerin in the cytoplasmic membrane-associated fraction varied slightly in different cell types: fibroblasts, myoblasts, HepG2 and neuroblastoma cells were analyzed and examples are shown in Figure 3 b. No great differences were found in myoblasts compared with myotubes (not shown) and between human, rat and simian cells, though a smaller amount of cytoplasmic emerin was usually found in myoblasts, especially from rat.
Immunohistochemical analysis of cells and tissues did not show any significant association of emerin with specific cytoplasmic structures. The only notable exception was the heart. In unfixed cryosections of human heart the antibodies also decorated heart intercalated discs (Fig. 4 a). Three different antisera, two raised against the human (5 ) and one against the mouse protein, gave identical results in human, mouse and rat heart. Identical results were obtained with the antisera and with affinity-purified antibodies. No reaction was observed with preimmune serum and all labeling was titrated away when excess recombinant antigen was added to the incubation (not shown). Serial 6 µm thick sections of an entire adult mouse heart demonstrated that emerin stained nuclei and intercalated discs throughout the organ (not shown).
The data presented in this paper confirm the suggestion that emerin, the product of the X-linked EDMD gene is a ubiquitous highly phosphorylated integral membrane protein of the nuclear envelope, tightly associated with the nuclear lamina. As predicted, emerin is indeed phosphorylated in vivo and the different phosphorylated forms may play important roles in interactions with the nuclear membrane and other nuclear proteins, during the cell cycle or in different cell types (12 ,13 ). Our data showed that the main determinant for inner nuclear membrane targeting is the C-terminal hydrophobic domain, which can be considered a genuine transmembrane domain. In the absence of the C-terminal domain in transfected COS cells most of the protein was localized in the nucleus, suggesting that a nuclear localization signal exists in the remaining portion of the protein. In vivo synthesis of a mutated emerin having a new C-terminal hydrophobic domain with an amino acid composition compatible with nuclear membrane insertion was sufficient for nuclear envelope localization.
In many cell types and tissues subcellular fractionation experiments demonstrated that a relevant fraction of emerin is also associated with cytoplasmic membranes. The finding of this fraction in tissues (see Results; 4 ) as well as in dividing cells indicates that it may not merely represent newly synthesized emerin on its way to the nuclear membrane and suggests different and possibly specific roles for emerin. In heart the cytoplasmic membrane-associated fraction appears to be part of the highly specialized heart adhesive junctions, desmosomes and fasciae adherentes of the intercalated discs. The evidence came primarily from immunohistochemical analysis using three different affinity-purified antibodies. Two were previously described (5 ), raised against a human protein lacking the C-terminal domain (amino acids 1-168). The third was raised in rabbits against a similar protein (amino acids 1-168) deduced from a mouse emerin cDNA sequence presenting 75% amino acid identity to the human sequence (Cartegni et al., in preparation). In Western blots the three antibodies reacted almost exclusively with emerin and did not show heart-specific bands. This result was confirmed by electron microscopic analysis of heart sections, by immunohistochemistry of cardiomyocyte cultures and subcellular fractionation experiments. Moreover, confocal analysis of cardiomyocyte cultures demonstrated that cardiac emerin and not emerin from fibroblasts could be recruited to newly formed focal adhesions, strongly indicating that emerin is part of a heart-specific adhesion complex. Nagano et al. (3 ) had previously shown that one of their polyclonal antibodies raised against a synthetic peptide (ED1, amino acids 173-188) showed positive immunostaining of the surface membrane of control skeletal and cardiac muscle. The staining in this case was very likely non-specific, as EDMD patients with a null mutation still had a positive immunoreaction at the cell surface. We did not have samples from patient hearts to study, but we never observed staining of the sarcolemma or of the cardiac cell outer membrane with our antibodies, which specifically stained only intercalated discs. The two reagents are, however, also very different, as both the antibodies raised by Nagano et al. were against C-terminal peptide sequences not contained in our antigen and may not recognize the same epitope(s).
Desmosomes and fasciae adherentes anchor desmin-containing intermediate filaments and the bundles of sarcomeric myofilaments respectively (10 ). They consist of transmembrane adhesive glycoproteins, members of the cadherin superfamily, and of cytoplasmic proteins such as vinculin, catenins and actin binding proteins (10 ). Different assortments of the same or similar proteins in desmosomes, fasciae adherentes, focal adhesions and other adhesive junctions seem to confer specific functions to ensure cell-cell communication and tight adhesion between cells and to the extracellular matrix. The role of this complex assortment of proteins is best demonstrated by the existence of many genetic diseases that perturb adhesion and in heart by the dramatic consequences of plakoglobin ([gamma]-catenin) knock out (14 ): plakoglobin -/- mice die at mid-gestation due to rupture of the ventricles.
The cardiac conduction defect in EDMD patients is the most severe and life threatening clinical manifestation of the disease. Cardiac alterations have also been described in female carriers (1 ) in the absence of any skeletal muscle abnormality, suggesting a prominent role in cardiac conduction for emerin. Localization of emerin in heart adhesive junctions suggests that lack of emerin in heart may have a direct effect on cardiac conduction abnormalities. We propose that lack or a decreased amount of emerin in heart alters cardiomyocyte adhesion and/or communication between adjacent cells and is responsible for arrhythmia and slow pacing and eventually heart block. Whether this is due to a general failure in cardiac conduction or to a more specific effect at the level of the heart pacemaker cannot be established with the present knowledge of the disease. On the other hand, in favor of our general model is the notion that emerin, with myotonic dystrophy kinase (15 ), is the second protein localized to the intercalated discs and responsible for a genetic disorder presenting a severe heart conduction defect.
Our results suggest that emerin may associate directly with membranes through its hydrophobic tail and we expect that it may do the same at the intercalated discs. Its cytoplasmic hydrophilic portion should interact with protein(s) of the adhesive junctions. In the emerin sequence a serine-rich stretch of 12 amino acids (PVSASR-SSL-DLS), 42 amino acids from the transmembrane domain, presented striking similarity to the consensus sequence (PV/ICFSRXSSLSS/DLS) of the 20 amino acid repeats of the adenomatous polyposis coli (APC) tumor suppressor gene (16 ), involved in binding [beta]-catenin. It also showed significant though lower similarity to the minimal [beta]-catenin binding region of the cadherins (17 ). Thus emerin in heart may also bind [beta]-catenin or a novel protein presenting similar sequence. It is noteworthy that this part of the protein is deleted in patient LB1520, who still retains some emerin capable of correct localization.
We have no evidence yet of a specific emerin localization in skeletal muscle that may account for the muscular dystrophy of EDMD patients. The finding of emerin in cytoplasmic membranes suggests a general and novel role for emerin in membrane attachment to the cytoskeleton and this could actually represent the most relevant function of this protein in both skeletal muscle and heart.
Cells were fractionated essentially as described by Manilal et al. (4 ) in the presence of protease inhibitors. After low speed centrifugation purified nuclei were digested with nucleases (18 ) and extracted with 1 M NaCl to obtain nucleoplasm. The pellet was washed with 1% Triton X-100 to separate detergent wash and insoluble lamina-associated proteins. The cytoplasmic fraction was centrifuged at 100 000 g to obtain soluble and insoluble fractions. The pellet was solubilized with 1 M NaCl and a 1% Triton X-100 wash to obtain extractable and non-extractable fractions.
Heart tissue was processed as described by Kuehl et al. (19 ) with minor modifications. A crude homogenate of 500 mg human heart in 25 mM Tris-HCl, pH 7.1, 2.5 mM MgCl2, 0.25 M sucrose, 0.6% (w/v) Triton X-100 and protease inhibitors was centrifuged at low speed. The cellular pellet was resuspended in 25 mM Tris-HCl, pH 7.1, 2.5 mM MgCl2, 2.3 M sucrose and protease inhibitors and centrifuged for 75 min at 82 000 g through a 2.3 M sucrose cushion to separate the soluble cytoplasmic fraction, an insoluble fraction which accumulated at the interface and a pellet containing purified nuclei. These were treated as above to obtain the nuclear lamina-associated fraction.
The anti-human emerin antisera have been described previously (5 ). Rabbit polyclonal antisera were raised against a bacterial fusion protein expressing a fragment of mouse recombinant emerin corresponding to amino acids 1-168. The mouse sequence was deduced from a mouse cDNA obtained in the laboratory (Cartegni et al., in preparation) and the PCR amplification product was cloned in vector pGEX 2T and controlled by sequencing as described for the human cDNA. Specific antibodies were isolated from the sera by affinity purification using antigen immobilized on nitrocellulose filters (20 ).
Indirect immunofluorescence and Western blots were performed as described previously (5 ), with the anti-emerin antibodies used at 1:100 or 1:1000 dilutions respectively. The 12CA5 mouse anti-HA mAb (Boehringer), the anti-vimentin, anti-vinculin and anti-cadherin (CH-19) mAb (Sigma, St Louis, MO) and the mouse anti-lamin A/C mAb (Ylem, Avezzano, Italy) were used as indicated by the suppliers.
Electron microscopy analysis was performed as described previously (21 ) on normal skeletal muscle samples obtained from a non-dystrophic patient during surgical treatment of osteosarcoma or heart samples taken from routine control biopsies on a heart transplant patient. The tissue fragments were fixed with 0.001% glutaraldehyde, 2% paraformaldehyde in 0.1 M NaPO4 buffer, pH 7.2, for 1 h at 4°C, cryoprotected in 2.3 M sucrose overnight at 4°C, frozen in liquid nitrogen and sectioned with a Reichert FC4 cryosectioning unit at -95°C (21 ). The anti-emerin antibodies were used diluted 1:50. Cryosectioning was the only method among several tested that allowed emerin immunostaining.
Emerin deletion mutants were prepared by PCR, inserted in-frame in a vector containing the HA epitope, sequenced and cloned in pMT2 (22 ). Alternatively, PCR-amplified fragments of emerin were cloned in vector pEGFP-C1 (Clontech). COS cells were transiently transfected by the standard CaPO4 method and analyzed 24 h after transfection.
We thank R.Sitia and P.C.Marchisio for many suggestions and discussions, S.Schiaffino and S.Ausoni for help with the cardiomyocyte cultures, G.Camerino, S.Bione, P.D'Adamo and C.Manzini for advice and critical reading of the manuscript and M.Gatti for technical assistance. This work was supported by grants from Telethon Italy to D.T. and F.C.
Human Molecular Genetics
Pages
Introduction
Results
Emerin is a phosphorylated protein
The C-terminal hydrophobic domain of emerin is the determinant for nuclear membrane localization
Emerin is localized to the inner nuclear membrane
Nuclear localization is insufficient for emerin function
Subcellular distribution of emerin
An alternative localization of emerin in heart
Discussion
Materials And Methods
Cell and tissue fractionation
Production of antisera
Immunocytochemistry and Western blots
Constructs and transfections
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Sat Oct 18 13:29:50 BST 1997. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1997