Introduction
The intermediate-filament supergene family makes cytoskeletal fibres that provide some of the most resilient structures in metazoan cells. Lamins are type V intermediate-filament proteins and the evolutionary progenitors of the intermediate-filament supergene family. Lamins can be categorized into two sub-families: A-type, which are expressed in most differentiated somatic cells; and B-type, which are expressed in nearly all cells and are essential for cell viability. Recent reports reveal that mutations in A-type lamins, in addition to causing several other diseases, promote premature ageing syndromes termed 'progerias'. Individuals suffering from progerias age about eight times faster than normal and exhibit a wide range of degenerative pathologies, mainly affecting mesenchymal tissues, and generally die from complications of atherosclerosis in their teens. Over a decade ago, the term 'guardian of the genome' was proposed for p53, due to its critical role in regulating apoptosis in cancerous cells. In this review, we discuss the case for describing A-type lamins as guardians of the soma.
Lamins
Intermediate-filament proteins form 10–13-nm filaments in the cytoplasm and nucleus. Lamins are the intermediate-filament family members that are the principle components of the nuclear lamina. They are fairly typical of intermediate-filament proteins, but possess important unique features. They are organized around a central
-helical rod comprising three coiled–coil domains separated by flexible linker regions and globular head and tail domains (Fig. 1). The coiled–coil domains are organized around heptad repeats, and the lamins contain an extra 42 residues (six heptads) within coil 1b, when compared with vertebrate cytoplasmic intermediate-filament proteins1, 2. In addition, the tail domain of the lamins harbours a nuclear-localization signal3 and, in most cases, a carboxy-terminal CaaX box, which is a target for isoprenylation and carboxymethylation4, 5, 6. These motifs ensure that lamins enter the nucleus and are located at the inner nuclear membrane (INM). Lamins form obligate parallel coiled-coil dimers, which assemble end-to-end to eventually form filaments 10–13 nm in diameter. In frog oocytes, these filaments line the nucleoplasmic face of the INM, where they form a twodimensional lattice interconnecting nuclear pores7. The B-type lamins (lamins B1, B2 and B3) are the initial building blocks of the lamina8 and at least one B-type lamin must be expressed in a cell to ensure its viability9. A-type lamins (A, A
10, C and C2), all alternatively spliced products of the LMNA gene, are probably built into this structure once it has formed8. Although A-type lamins have essential roles in maintaining lamina stability (see below), they also seem to have additional roles in regulating transcription factors10.
Figure 1: Generalized structure of cytoplasmic intermediate-filament proteins compared to lamins.
Intermediate-filament proteins have a conserved domain structure consisting of a variable globular head domain, a central
-helical coiled-coil dimerization domain consisting of four coiled coils, based on heptad repeats, interrupted by flexible linker domains and a variable globular tail domain. The coiled-coil domains are termed 1A, 1B, 2A and 2B, respectively. The linker domains are non-helical. The major differences between lamins and vertebrate cytoplasmic intermediate filaments are that: first, the head domains are very short (about 33 amino acids); second, there is a six-heptad extension of coil 1B; and third, the globular tail domain is usually characterized by possession of a nuclear localization signal sequence and a site for carboxyl methylation, farnesylation and proteolytic cleavage (CaaX).
The lamina as a tensegrity element
The lamina has been described as a 'tensegrity element', which resists forces of deformation and protects chromatin from physical damage11. This hypothesis is supported by a number of investigations. For example, in cell-free extracts of Xenopus laevis oocytes, either physical or functional depletion of lamins leads to the assembly of very small and fragile nuclei12, 13, 14, 15. Similarly, RNA interference knock-down of C-lam in Caenorhabditis elegans leads to abnormally shaped nuclei that constantly alter their morphology16. In mouse spermatocytes, nuclei are hook shaped rather than ovoid or spherical and these cells express a germ-line-specific lamin, lamin B3. When lamin B3 is expressed ectopically in somatic cells, their nuclei adopt a hook-shaped morphology17. Finally, expression of a dominant lamin B1 mutant lacking four fifths of the rod domain causes massive deformation of the nuclear envelope in somatic cells18. Thus, lamins determine the size, shape and strength of the nuclear envelope and therefore possess the key features of a tensegrity element11. However, the properties of this tensegrity element are probably determined not only by the properties of individual lamins but also by the way that they interact with components of the cytoskeleton.
Recently, a novel family of spectrin-repeat proteins has been described, some members of which are integral proteins of the INM, and others may specifically localize to the outer nuclear membrane. Variously termed nesprins19, NUANCE20 or synes21, these proteins are characterized by the giant size of some alternatively spliced variants (they can have relative molecular masses in excess of 800,000). They possess multiple clustered spectrin repeats throughout the core of the protein, amino-terminal calponin homology domains and a conserved C-terminal singlepass membrane domain. NUANCE binds to actin and its distribution is influenced by the actin cytoskeleton20. In vitro, nesprin 1
— a relatively ocus small nesprin isoform — binds to lamins A and C and to emerin, a type 2 integral membrane protein of the INM22. Its localisation at the INM is dependent on the expression of lamins A and C23. Nesprins also self associate19. In addition, emerin directly binds to lamins A/C in vitro and is localized to the INM by a complex of these two proteins24, 25; furthermore, it is also an actin-binding protein26. Thus, lamins A and C can assemble a complex of proteins at the nuclear envelope that have characteristics of cytoskeleton linker proteins27 and potentially link the actin cytoskeleton to the lamina. One model proposes that integral proteins of the INM that bind to lamins A and C interact, either directly or indirectly through proteins in the perinuclear space, with larger nesprin isoforms in the outer nuclear membrane, which in turn bind to cytoskeletal elements. This would mean that when cells express A-type lamins at the point of their differentiation they gain contacts between the lamina and the cytoskeleton.
Given these connections, the tensile properties of the lamina would radiate through the cytoskeleton to the plasma membrane (Fig. 2), generating a mechanotransduction signalling capability in the cell that links the extracellular matrix to the inside of the nucleus28. There is now direct evidence to support this hypothesis. Fibroblasts from Lmna-/- mice were subjected to mechanical strain, and nuclear deformation and strain-induced signalling simultaneously measured29. Nuclei of fibroblasts from Lmna-/- mice have notably greater deformation than fibroblasts from Lmna+/+ littermates, with a resultant decreased viability when subjected to mechanical strain. In addition, NF-
B signalling is attenuated in response to either mechanical or cytokine stimulation29. In a complementary study, a bespoke biomechanical device was used to investigate how fibroblasts from the same Lmna-/- mice resist forces of compression30. The device was capable of simultaneously measuring forces of decompression generated by single cells subjected to compression, while observing how the cells and their nuclei were deformed. Nuclei of fibroblasts from Lmna-/- mice are notably less able to resist compression compared with nuclei of fibroblasts from Lmna+/+ mice. In addition, the nuclei of Lmna-/- mice exhibit isotropic deformation when compressed, whereas Lmna+/+ mice exhibit anisotropic deformation. This finding implies that nuclei of cells lacking A-type lamins no longer respond to the polarity of the cell and that this lack of response occurrs because all elements of the cytoskeleton are disorganized around the nuclear envelope30. Thus, in mesenchymal cells at least, mechano-induced signalling is propagated by linking the lamina to the cytoskeleton.
Figure 2: Model of the lamina connected to the actin cytoskeleton.
Model of the lamina as a tensegrity structure in which lamin filaments (bars) form a cage-like structure reminiscent of a geodesic dome. The cage is connected to the actin cytoskeleton (long wavy filaments) by nesprins (blue elements) and emerin (red spots), both of which are anchored to the lamina. These connections constrain the shape of the lamina to the polarity of the cell.
Full size image (50 KB)Lamins as regulators of transcription
A number of reports link lamins to transcription and post-transcriptional processing. In Xenopus laevis oocytes, the germ-line specific lamin Liii associates with RNA polymerase II (RNA pol II) and when it is prevented from forming filaments using dominant-negative lamin mutants, RNA pol II activity is inhibited31. In somatic cells, lamin B1 is reported to bind to the POU domain repressor protein Oct-1, which regulates the expression of collagenase genes32. In addition, LAP2
, an integral protein of the INM, together with B-type lamins, forms functional complexes with the transcription factors germ-cell-less (GCL) and DP to repress E2F33. Thus, B-type lamins seem to have negative and positive influences on transcription.
A-type lamins also have roles in RNA metabolism. A-type lamins are distributed throughout the nuclear interior and associate not only with the lamina but also with a range of nuclear bodies. This implies that they are involved in transcription and RNA processing34, 35, 36. In addition, A-type lamins associate with specific transcription factors and at least one of those associations is functional37. A-type lamins are been reported to associate with retinoblastoma protein (Rb)37, sterol response element binding protein 1 (SREPB1)38 and MOK2 (ref. 39), whereas emerin binds to GCL40. The C-terminal region of lamins A and C also interacts with DNA41.
The function of lamin-A– or lamin-C–Rb interactions is now becoming clear. In fibroblasts, lamin A/C and the nucleoskeleton protein LAP2
form a complex with Rb that tethers unphosphorylated forms of the protein within the nucleus42. It had previously been shown that tethering of Rb in the nucleus is necessary for its function, and mutant forms of Rb that cannot be tethered promote cancer43. Importantly, these mutants do not interact with lamin-A– or lamin-C–LAP2
42. Direct evidence for the role of lamins A and C in Rb function has now been reported. In cells from Lmna-/- mice, Rb is depleted and targeted for destruction by the proteosome10. Moreover, the fibroblasts exhibit growth and size characteristics typical of fibroblasts from Rb-/- mice. Thus, the purpose of the lamin-A– or lamin-C–Rb interaction seems to be to prevent targeting of Rb for destruction by the proteosome, thereby regulating cell growth and division (Fig. 3).
Figure 3: Lamin-A–LAP2
complexes are required for Rb function.
A complex of lamin A/C (green) and LAP2
(blue) tethers Rb (red) and thereby prevents its destruction by the proteosome (yellow). In the presence of lamin A/C and LAP2
and the absence of Rb phosphorylation, E2F activity is inhibited and cells remain in G1. When lamin-A/C–LAP2
function is disrupted, Rb is targeted for destruction by the proteosome, thereby de-repressing E2F and allowing cell-cycle progression.
The laminopathies: Human diseases caused by mutations in LMNA raise new questions
In the early 1990s, human geneticists set out to identify the gene responsible for X-linked Emery-Dreifuss muscular dystrophy (X-EDMD) using a positional cloning approach. They identified a new gene encoding an integral membrane protein they named 'emerin'44. They named the protein after Alan Emery, who along with Fritz Dreifuss first described the clinical condition with contractures of the elbows, achilles tendons and posterior neck, slow progressive muscle wasting and cardiomyopathy with atrioventricular conduction block45. Subsequently, emerin was shown to localize to the nuclear envelope46, 47 and to interact with A-type lamins25, 48. Hence, this was the first human disease identified that was related to an INM protein.
A condition identical to X-EDMD is inherited in an autosomaldominant manner (AD-EDMD). Using positional cloning, another group showed that AD-EDMD is caused by mutations in LMNA49. Subsequently, mutations in LMNA were found to cause related autosomal- dominant striated muscle diseases with predominant cardiomyopathy and little-to-no skeletal muscle involvement50 or cardiomypathy with a limb-girdle distribution of skeletal muscle involvement51. Before these discoveries, the structural organization of the lamin-A/C-coding portion of LMNA had been determined52. Most mutations in LMNA causing these striated muscle diseases were found to lead to small deletions or amino-acid substitutions throughout lamins A and C, with rare mutations causing haploinsufficiency49, 53.
Dunnigan-type familial partial lipodystrophy (FPLD), or lipoatrophic diabetes, is an autosomal-dominantly inherited disorder characterised by loss of fat from the extremities, excess fat in the face, neck and trunk and insulin resistance54. In many aspects, FPLD mimics the metabolic syndrome, which occurs in aging individuals in developed regions of the world and is characterized by abdominal obesity, elevated serum lipids, high blood pressure, insulin resistance, a proinflammatory state and a prothrombotic state55. The responsible gene for FPLD was linked to chromosome 1q21–1q22 in 1998 (ref. 56), a chromosomal region to which LMNA had been mapped previously57. Subsequently, researchers using a candidate gene approach58, and others using positional cloning59, 60, showed that mutations in LMNA cause FPLD. About 90% of mutations are missense mutations localized to exon 8. The amino-acid substitutions alter the charge of a solvent-exposed surface in an immunoglobulin fold in the tail of lamins A and C61, 62. In contrast, amino-acid substitutions in the same region of the molecules that cause striated muscle disease lead to an overall disruption of three-dimensional fold structure. Hence, mutations that cause FPLD lead to subtle changes in lamin A/C molecular structure, whereas mutations causing striated muscle disease result in more dramatic structural alterations. FPLD-causing A-type lamin mutants may function in a dominantly interfering manner, as Lmna-/- mice do not develop partial lipodystrophy63.
More recent research has shown that mutations in A-type lamins not only cause disorders of striated muscle and adipose but a complicated and perplexing expanding array of diseases. Some homozygous mutations in LMNA cause an autosomal-recessive axonal peripheral neuropathy in humans, and Lmna-/- mice have pathologically similar peripheral nerve abnormalities64. Homozygous mutations in LMNA also cause mandibuloacral dysplasia (MAD), a rare disorder characterized by postnatal growth retardation, skull and facial anomalies, skeletal malformations, mottled skin pigmentation, partial lipodystrophy and signs of premature aging65. MAD is also caused by mutations in the ZMPSTE24 protease involved in the processing of pre-lamin A to lamin A, and Zmpste24-/- mice have phenotypic similarities to humans with MAD66, 67, 68. Recently, heterozygous mutations in both LMNA and ZMPSTE24 have been associated with restrictive dermopathy, also know as 'tight-skin contracture syndrome', another rare disorder mainly characterized by intra-uterine growth retardation, tight and rigid skin with erosions, facial abnormalities, bone-mineralization defects and early neonatal lethality69. Autosomalrecessive mutations in LMNA also cause a syndromatic condition with generalized fat loss, insulin-resistant diabetes mellitus, disseminated raised skin lesions, fatty liver and cardiomyopathy70.
Perhaps the most dramatic phenotype caused by mutations in LMNA is progeria. Hutchinson-Gildford progeria syndrome (HGPS) is a rare autosomal-dominant condition with growth retardation, abnormal facial development, loss of subcutaneous fat and premature atherosclerosis leading to death in the second decade of life71. In 2003, two groups72, 73 reported a de novo dominant mutation in exon 11 of LMNA in subjects with HGPS. This mutation leads to the creation of an abnormal RNA splice-donor site in exon 11 that results in expression of a truncated pre-lamin A protein with 50 amino acids deleted from its C terminus. This mutation is found in the majority of subjects with HGPS, but other LMNA mutations may be the cause in some subjects72, 73, 74. 'Knock-in' mice with an RNA-splicing mutation, different than the one in human subjects with HGPS, also develop signs of progeria75. Mutations in LMNA have also been reported in subjects with atypical premature aging disorders not meeting the diagnostic criteria for HGPS or other specific syndromes76, 77.
That a broad spectrum of disease phenotypes arises from mutations in A-type lamins has raised new questions about their functions. It suggests that these proteins have different roles in different somatic cells. Some mutants, such as those causing progerias, affect multiple cell types. Others mutants, such as those causing EDMD or FPLD, seem to exert effects in a relatively cell type-specific way. The efforts of many cell biologists are now focused on deciphering the cell-specific functions of A-type lamins and determining how mutations cause the range of so-called laminopathies.
The 'structural' and 'gene expression' hypotheses to explain laminopathies
Two working hypotheses have been proposed to explain how laminopathies arise. The 'structural' hypothesis proposes that mutations in A-type lamins or emerin give rise to a weakened nuclear envelope, which is predisposed to damage. In striated muscle particularly, damage to the nucleus is thought to promote myocyte death and cause replacement with fatty and fibrotic tissue8, 25, 78. The 'gene expression' hypothesis proposes that, as A-type lamins are important regulators of gene expression, mutations in these proteins will alter their interactions with various gene regulatory proteins and thereby promote disease in different tissues78, 79, 80. There is now evidence to support each hypothesis, with a consensus emerging that structural weakness and altered gene expression both contribute to pathogenesis.
The case for the structural hypothesis
Studies of fibroblasts from Lmna-/- mice show that an absence of A-type lamins results in a structurally compromised nuclear envelope that is susceptible to stress-induced damage leading to loss of cell viability29, 30. However, although Lmna-/- mice do exhibit a phenotype similar to human EDMD25, 81, human diseases caused by mutations in LMNA, including EDMD, do not arise from null genotypes. Nevertheless, there is evidence that mutant A-type lamins promote structural weakness. When lamin A or lamin C mutants are expressed in model cell lines, they often fail to associate appropriately with the nuclear envelope and form aggregates in the nucleoplasm82, 83. Moreover, these mutant lamins can promote structural abnormalities in the nuclear envelope that are commonly observed in fibroblasts from human subjects with laminopathies72, 73, 65, 84, 85, 86, 87. It seems that this structural weakness does translate into physical damage. Ultrastructural investigations of skeletal muscle from subjects with ADEDMD and X-EDMD88, 89 and cardiac muscle from Lmna-/- mice81 both reveal widespread breakage of nuclear envelopes and leakage of chromatin into the cytoplasm. This damage correlates with cell death. Consistent with these findings, nuclear envelope proteins of fibroblasts from subjects with EDMD exhibit notably increased solubility properties compared with those of normal individuals90. Thus, weakness of the nuclear envelope seems to promote cell death. One interesting and unresolved issue is the absence of mutations in LMNA and emerin in about half of subjects with a clinical diagnosis of EDMD53. If the tensegrity model is correct, new candidate genes could include cytoskeleton linker proteins, such as the nesprins or components of the cytoskeleton.
The case for the gene expression hypothesis
Clear evidence is now available for the potential involvement of Rb in laminopathies. Rb is not only involved in cell proliferation but its activity is also required for the differentiation of mesenchymal cells, including skeletal muscle91 and adipocytes92. It is now clear that loss of expression of lamins A and C impairs Rb function and leads to a loss of growth control10. Expression of lamin A mutants that cause AD-EDMD in C2C12 myoblasts inhibits their differentiation and promotes apoptotic cell death93. These findings correlate with a recent report showing that growth of early-passage fibroblasts from individuals with HGPS is characterized by rapid proliferation but high rates of apoptosis85. Therefore, loss of Rb function could be responsible for these phenotypes. Indeed, as an independent investigation has confirmed that the expression of lamin A mutants causing AD-EDMD in C2C12 myoblasts does prevent their differentiation by disrupting LAP2
–Rb complexes leading to loss of expression of Rb isoforms94. One problem with these findings is that they imply that skeletal muscle differentiation should fail at an early stage in the development of Lmna-/- mice. However, the mice actually develop normally but experience muscle fibre degeneration after birth25. Because C2C12 myoblasts are derived from satellite cells, they better represent a model for regeneration rather than development. Therefore, we propose that the presence of A-type lamin mutants in adult stem cells may compromise their ability to regenerate some tissues, such as striated muscle, because certain differentiation programmes require Rb function. In the most severe examples, laminopathies may arise because certain cells are susceptible to degeneration through fragility of the nuclear envelope, whilst at the same time adult stem cells may be unable to regenerate damaged tissues through a loss of Rb function.
SREBP1, which has a critical role in adipocyte differentiation, interacts with lamin A in vitro38. Mutations that cause FPLD but not other laminopathies decrease the efficiency of lamin-A–SREBP1 interactions. Therefore, FPLD may be promoted by impairment of SREBP1 function through loss of binding to lamin A. To validate this hypothesis, a more detailed investigation of how lamin A influences SREBP1 activity is required.
A-type lamins as guardians of the soma
We have highlighted two essential functions of A-type lamins and the nuclear lamina. First, they have a role as a load-bearing structure at the centre of the cell, which resists compression and is required for the propagation of stress-induced signalling pathways. Second, lamin complexes have a role in regulating transcriptional repressors and, in particular, lamin A has a role in maintaining Rb function. Loss of A-type lamin function compromises both pathways and leads to degeneration and possibly a lack of normal regeneration of a variety of tissues and, in some cases, premature death. Thus, A-type lamins seem to be essential for survival to old age and should be considered guardians of the soma.

