Over the past few years a convergence of clinical and basic research has highlighted an unexpected link between nuclear envelope proteins and a variety of human hereditary diseases. Initially, a single disease, the Emery–Dreifuss muscular dystrophy, captured the interest since two differently transmitted forms of the disease have been demonstrated to be due to mutations of emerin or lamin A/C, two structural proteins located at the nuclear envelope (Emery, 2000). In a second time, attention has been progressively extended to an increasing number of diseases, affecting a variety of tissues and organs, caused by mutations of the gene coding for lamin A/C. These diseases, collectively referred to as laminopathies (Wilson, 2000), received a significant amount of attention, since their study could contribute to gain insight not only in their pathogenesis, but also in key nuclear activities involving interactions with other nuclear proteins.
In this review, we first consider the peculiar characteristics of the diseases due to mutations of genes coding for nuclear structural proteins that are normally located at the nuclear envelope level. Since now fourteen diseases have been identified; ten of them are due to mutations of the gene coding for lamin A/C, while four involve either nuclear envelope-associated proteins (emerin or lamin B receptor), or enzymes involved in lamin A processing. Then, we consider the molecular mechanisms possibly involved in the pathogenesis of these diseases affecting a variety of tissues, and finally, we speculate about new players involved in nuclear functions that can account for unsuspected connections between chromatin arrangement, transcriptional control, gene regulation and disease.
EMERY–DREIFUSS MUSCULAR DYSTROPHY AND OTHER NUCLEAR ENVELOPE DISEASES
The very early beginning of the laminopathy saga started just 10 years ago. As in all sagas, the future development of the story was not predictable at the beginning. In fact, in 1994 it was demonstrated that a rare form of myopathy, the X-linked Emery–Dreifuss muscular dystrophy (XL-EDMD), was caused by mutations in the EMD gene that encodes for emerin (Bione et al., 1994), a protein localized at the inner nuclear membrane (Manilal et al., 1996; Nagano et al., 1996), and connected with the nuclear matrix (Squarzoni et al., 1998). The localization of a protein involved into the pathogenesis of a muscular dystrophy not at the plasma membrane, as in Duchenne/Becker, limb-girdle, and congenital muscular dystrophies (Straub and Campbell, 1997), but at the nuclear envelope, was quite unexpected. This raised the question of how the absence of a nuclear envelope protein could cause a progressive wasting of some skeletal muscles, early tendon contractures, and cardiac conduction defects (Morris and Manilal, 1999).
The first “coup de théatre” took place when the autosomal form of EDMD (AD-EDMD), characterized by phenotypic alterations almost undistinguishable from those of the X-linked EDMD, was demonstrated to depend on mutation in the LMNA gene, coding for A-type lamins (Bonne et al., 1999). An animal model of EDMD has been developed in which mice homozygous for a targeted disruption of the Lmna gene developed a syndrome quite similar to human AD-EDMD (Sullivan et al., 1999). The autosomal recessive form of Emery–Dreifuss muscular dystrophy (AR-EDMD) was also demonstrated to be due to mutations in the LMNA gene (Raffaele di Barletta et al., 2000). An increased attention to mutations in LMNA revealed that two other myopathies did have the same origin. The first was the dilated cardiomyopathy with conduction system disease (DCM-1A), a syndrome characterized by ventricular dilation and rhythmic repeated systolic contraction, with rare skeletal muscle defects (Fatkin et al., 1999; Brodsky et al., 2000). The second one was the limb-girdle muscular dystrophy with atrio-ventricular conduction disturbances (LGMD-1B), in which both muscular and cardiac defects were present, but not early tendon contractures, as in EDMD (Muchir et al., 2000). The finding that mutations in two nuclear proteins located at the nuclear envelope (emerin and lamin A/C) were involved in the pathogenesis of a group of disorders, classified as nuclear envelopathies, stressed the importance of the role of structural proteins in unsuspected nuclear functions (Nagano and Arahata, 2000).
The second and major “coup de théatre” occurred when a specific lamin A/C R482Q mutation was demonstrated to be involved in a rare autosomal disease, the Dunningan-type familial partial lipodystrophy (FPLD), characterized by a partial absence of adipose tissue (Cao and Hegele, 2000). In this disease, the subcutaneous fat is progressively lost from extremities, trunk and gluteal region of FPLD-patients after the onset of puberty, whilst excess fat is present at face and neck. Interestingly, a profound insulin resistance with diabetes can develop, with coronary heart disease, secondary to metabolic disturbances (Speckman et al., 2000). Mutation in lamin A/C has been recently described in a patient showing a generalized lipodystrophy, insulin-resistant diabetes, leukomelanodermic papules, liver steatosis and hyperthrophic cardiomyopathy (LIRLLC). This syndrome presents some symptoms of the FPLD associated with cardiac defects (Caux et al., 2003). The unexpected finding of LMNA-linked non-muscle diseases suggested that several human diseases, collectively defined laminopathies, could have a feature in common, since the affected tissues derived from the mesenchymal stem cell (Wilson, 2000).
The number of diseases due to mutation of the LMNA gene further increased in the last 2 years; interestingly, not all the tissues affected were of mesenchymal origin. Charcot-Marie-Tooth Disorder Type 2 (CMT2) is a heterogeneous group of peripheral neuropathies. Autosomal recessive CMT2 (AR-CMT2) is caused by a homozygous C892T mutation in exon 5 of LMNA (De Sandre-Giovannoli et al., 2002). The main symptoms of AR-CMT2 are symmetrical muscle weakness and wasting in the distal lower limbs, walking difficulties associated with reduced or absent tendon reflexes, due to loss of myelin in peripheral nerves without regenerative attempts. Interestingly, severe reduction of axon density and presence of nonmyelinated axons were detected in the sciatic nerve of LMNA null mice (Sullivan et al., 1999). Therefore, LMNA gene seems to have a fundamental role also in the regulation of axon development and/or survival, as suggested by the decrease of lamin A/C expression with the progression of the neuronal differentiation state (Pierce et al., 1999).
Recent studies showed that also some human progeroid syndromes are due to mutations of the LMNA gene. The first discovered syndrome with progeroid tracts was Mandibulo acral dysplasia (MAD) (Novelli et al., 2002). MAD is a rare autosomal recessive disorder characterized by mandibular and clavicular hypoplasia, acroosteolysis and delayed closure of the cranial suture (skeletal malformations), joint contractures, mottled cutaneous pigmentations, loss of subcutaneous fat in the extremities and fat accumulation in the face, trunk and occipit, and post-load hyperinsulinemia, indicating insulin resistance (Simha and Garg, 2002). Interestingly, these last characteristics are the main features of FPLD. A homozygous missense mutation (R527H) in the LMNA gene was considered to be causative of MAD (Novelli et al., 2002). Skin fibroblasts from MAD patients showed nuclei with abnormal lamin A/C distribution and a dysmorphic envelope, alterations observed in other laminopathies. It is particularly significant that a heterozygous R527P mutation causes AD-EDMD (Bonne et al., 1999). Patients with MAD who bear the R527H mutation do not display muscle wasting or weakness, nor any cardiac symptoms (Novelli et al., 2002).
Werner's syndrome (WS) is an autosomal recessive segmental progeroid syndrome, in which multiple aspects of aging phenotypes are entailed, along with lipodystrophy tracts. The typical disorder is caused by mutations in WRN, coding for a DNA helicase. In individuals with atypical (non-WRN) Werner's syndrome, some missense mutations in LMNA have been found. Interestingly, these mutations altered relatively conserved residues within lamin A/C (Chen et al., 2003).
Five different mutations in LMNA have been reported in Hutchison–Gilford progeria syndrome (HGPS), the most severe laminopathy so far discovered, by three distinct studies (Cao and Hegele, 2003; De Sandre-Giovannoli et al., 2003; Eriksson et al., 2003). A typical mutation, G608G, results in an in-frame deletion of 50 amino acids in the C-terminal region of LMNA, which is involved in lamin A maturation (Sinensky et al., 1994). The splicing defect leads to accumulation of a mutated protein isoform termed progerin (Csoka et al., 2004).
The development of mice carrying a homozygous Lmna mutation, which represent an animal model of the HGPS, could contribute to clarify the role of lamin A/C mutation in premature aging (Mounkes et al., 2003). In fact, Lmna mutation resulted in nuclear alterations in mouse cells, similar to those described in human laminopathies. The HGPS phenotype of homozygous mice included hypoplasia and/or atrophy of myocardiocytes and mild to moderate degeneration of several skeletal muscles. However, typical dystrophic features such as centrally located nuclei and fiber size variability were not observed. A defect in prelamin A processing, as a consequence of inactivation of the Zmpste24 metalloproteinase gene, has been reported to cause a phenotype in mice resembling the clinical features observed in HGPS and MAD patients (Pendas et al., 2002). An independent study in Zmpste24−/− mice, confirming the defects in prelamin A processing, suggests that the most striking features are multiple spontaneous bone fractures, similar to those occurring in mouse models of osteogenesis imperfecta (Bergo et al., 2002). These findings suggest that human disorders could be associated to mutations in genes coding for enzymes involved in lamin A/C processing. In fact, it has been reported that some cases of MAD are not due to mutations in the LMNA gene, but in the gene coding for ZMPSTE24. Interestingly, ZMPSTE24 mutations cause impairment of prelamin A processing (Agarwal et al., 2003).
Furthermore, other nuclear proteins, besides lamin A/C and emerin, are involved in human diseases.
The lamin B receptor (LBR) is a member of the sterol reductase family, and like emerin, targets lamins and heterochromatin to the inner nuclear membrane (Ye et al., 1997). Several mutations in LBR have been demonstrated to produce alterations in the granulocyte nuclear morphology, in the autosomal dominant Pelger–Huët anomaly (PHA). PHA affected individuals show hyperlobulated nuclei with coarse chromatin, as well as varying degrees of developmental delay, epilepsy, and skeletal abnormalities (Hoffmann et al., 2002). Finally, LBR mutations have been demonstrated to cause a deficiency of the cholesterol biosynthetic enzyme 3β-hydroxylsterol δ(14)-reductase in patients affected by Greenberg skeletal dysplasia (HEM); the truncated protein resulting from a substitution in exon 13 of the LBR gene, in fact, represents the defective sterol δ(14)-reductase (Waterham et al., 2003).
It is therefore conceivable that other diseases due to mutations of genes coding for other nuclear envelope associated proteins will be identified in the near future.
PATHOGENETIC HYPOTHESES ON LAMINOPATHIES
Laminopathies represent a group of disorders that affect almost all the tissues. It is therefore mandatory to understand in which way do LMNA mutations lead to so many diseases (Hutchison et al., 2001). Since 1999 phenotype–genotype correlations were attempted to gain insight in the pathogenesis of the different disorders caused by mutations of proteins of the nuclear envelope (Toniolo and Minetti, 1999). The hypothesis that any defect of the nuclear membrane could interfere with satellite cell function, and thereby skeletal muscle regeneration, was advanced (Fairley et al., 1999). It has been also suggested that an increased apoptosis may play a pathogenetic role, at least in EDMD (Morris, 2000; Cohen et al., 2001; Haraguchi et al., 2004).
The finding that emerin and lamin A/C mutations were responsible for myopathies suggested, that in skeletal muscle and heart, interactions between nuclear envelope components were necessary to maintain skeletal and cardiac muscle function and/or that the loss of integrity of nuclear envelope was an underlying cause of muscular dystrophy (Manilal et al., 1999; Morris and Manilal, 1999). The mechanical stress hypothesis was based on the assumption that skeletal and cardiac muscle cells are particularly subjected to mechanical stress; it has been therefore suggested that emerin may be part of a nucleo-cytoplasmic skeleton that protects cell from stress (Tsuchiya et al., 1999). An interesting analogy between lamin-associated diseases and keratin-associated diseases has been proposed taking into account the fact that lamins form a subtype of the intermediate filament (IF) family of proteins (Moir and Spann, 2001). Because the common feature of keratin-associated diseases is the fragility of epithelial cells due to a reduced resistance to mechanical stress, by analogy, the mutant lamins may act as disruptor of lamin organization, supporting the idea that a loss of cytoskeletal integrity ultimately leads to tissue degeneration. Along this line Nikolova et al. (2004) recently showed altered desmin network in Lmna null mouse heart, and altered desmin distribution was also found in LMNA-linked dilated cardiomyopathy (Sebillon et al., 2003).
An alternative hypothesis on the pathogenetic mechanism of laminopathies has been advanced (Mounkes et al., 2001) in which LMNA mutations, resulting in changes in the lamina and/or nuclear membrane composition, may have dominant effects on the structure and function of the peripheral endoplasmic reticulum (ER). The implication is that the tissue specificity associated with laminopathies may be ascribed to impaired function of the ER, such as cholesterol regulation and fatty acid synthesis or Ca2+ release. Alternatively, accumulation of nuclear envelope proteins, such as emerin, in the ER could affect intracellular signaling pathways with downstream effects on gene expression and cell viability. It has been also suggested that the muscle A-kinase anchoring protein (mAKAP), normally targeted to muscle-cell nuclear envelope (Kapiloff et al., 1999), could be mislocalized following changes in INM protein composition (Burke and Stewart, 2002).
A crucial turning point was represented by a commentary appeared in 2000 (Wilson, 2000), whose title: “The nuclear envelope, muscular dystrophy and gene expression” clearly indicated a new pathogenetic hypothesis. The starting point of this hypothesis was that nuclear fragility alone cannot explain the selective effects of EDMD on cardiac muscle, and all the more reason, of FPLD on adipose tissue. The alternative possibility was that alterations in the expression of nuclear envelope proteins might result in subtle defect in gene expression, as previously suggested (Östlund et al., 1999). The finding that the layer of heterochromatin is thinned or absent in cells from EDMD patients (Ognibene et al., 1999; Sabatelli et al., 2001) suggested that emerin or lamina A/C-deficient nuclei have difficulty attaching heterochromatin to the envelope or stabilizing heterochromatin structure. Similar disorganization of heterochromatin at the nuclear lamina was also observed in Lmna-null mice (Sullivan et al., 1999). Emerin and lamin A/C may have a crucial role in maintaining a correct organization of the nucleus based on the interaction between the nuclear envelope and the peripheral heterochromatin (Yuan et al., 1991). The heterochromatin is mainly repressed or silenced by the interactions with the nuclear envelope, dynamically mediated by nuclear envelope associated proteins, including lamins, LAP2 and LBR (Stuurman et al., 1998; Holaska et al., 2002). Among these proteins, LAP2 has been demonstrated to interact with the DNA-binding protein barrier-to-autointegration factor BAF, that plays a crucial role in the chromosome architecture (Furukawa, 1999). Emerin too, that interacts with lamins A and C, was predicted to interact with BAF (Wilson, 2000); this interaction was thereafter demonstrated (Holaska et al., 2003). Interestingly, lamin A did interact with nucleosomal DNA in vitro with higher affinity with respect to lamin B (Yuan et al., 1991). Moreover, a non-sequence specific interaction between DNA and lamin A/C globular domain was demonstrated (Stierlé et al., 2003). Other chromatin-associated proteins that function as transcriptional regulators, including Hp1, YA, YT521B, Rb (Wilkinson et al., 2003) and Btf (Haraguchi et al., 2004), interacting with lamins and/or nuclear envelope associated proteins, may depend for their activity on the proper localization at the nuclear envelope. Emerin binds GCL (germ-cell-less) a repressor of E2F-DP regulated genes (Holaska et al., 2003) and Btf, a repressor involved in programmed cell death (Tzur et al., 2002). Emerin influences splice site selection by interacting with YT521-B in vivo (Wilkinson et al., 2003). These findings, as well as the misregulation of a set of genes observed by DNA microarray expression profiles in X-EDMD (Tsukahara et al., 2002), imply that emerin mutations could interfere with transcription.
Emerin and lamin A/C may have multiple functions (Hegele, 2001), such as the interaction with tissue-specific transcription factors (Dreuillet et al., 2002), or factors involved in DNA replication (Moir et al., 2000), cell survival (Steen and Collas, 2001), or chromatin remodeling (Maraldi et al., 2002). A detailed report on the functional evolution of the nuclear lamina proteins, that pointed out that several transcription factors (Oct-1, Rb, GCL, IPF-1/PDX-1), most of which are repressors, localize at the nuclear lamina, provided a strong support to the hypothesis that nuclear lamina proteins play active roles in transcriptional repression (Cohen et al., 2001). A further evidence supporting the gene-expression hypothesis has been advanced (Lloyd et al., 2002). A-type lamins, indeed, can interact with sterol response element binding protein 1 (SREBP1), a transcription factor involved in adipocyte differentiation (Horton, 2002); furthermore, lamin A carrying the FPLD-linked R482W mutation shows a slightly reduced affinity for SREBP1, while a more evident reduction of binding affinity was found in an EDMD mutation (Lloyd et al., 2002). SREBP1 nuclear translocation is affected in Lmna null mouse cardiomyocytes (Nikolova et al., 2004) suggesting a role for lamin A or C in the nuclear import of the transcription factor. However, adipocyte nuclear import of SREBP1 is not affected, indicating a tissue-specific mechanism.
The paradox of how can the mutant LMNA alleles give rise to diverse, tissue-restricted phenotypes, has been accounted by the analysis of different pathogenetic models that can be not mutually exclusive (Burke and Stewart, 2002). A result has been recently reported that unifies the mechanical stress hypothesis and the transcription impairment hypothesis; in fact, it has been demonstrated that the transactivation activity of NF-kB transcription factor, in response to mechanical strain, is impaired in lamin A/C null mice fibroblasts (Lammerding et al., 2004). As a general consideration, the fact that nuclear alterations are present in both muscle cells and fibroblasts from EDMD patients (Ognibene et al., 1999; Sabatelli et al., 2001) and in fibroblasts from FPLD patients (Vigouroux et al., 2001; Capanni et al., 2003), would indicate that these ultrastructural defects are not an effect of mechanical damage, but possibly contribute to the pathogenetic mechanism. Some lamin A mutants found in EDMD, LGMD-1B, DCM-1A, and FPLD have been found to be aberrantly localized within the nucleus and to partially disrupt the nuclear lamina (Capanni et al., 2003; Muchir et al., 2003). Other mutants localized normally even in transfected cells (Östlund et al., 2001; Holt et al., 2003). Moreover, mutations that cause skeletal and cardiac muscle disease are spread throughout the LMNA gene and no clear correlation between the severity of the disease and the mutation site has been found (Bonne et al., 2000). On the contrary, FPLD is due almost exclusively to mutations clustering in exon 8 and 11, and mainly to codon 482 (Vigouroux et al., 2001). This suggests that the region encodes a domain important for a specific interaction with a protein possibly involved in adipocyte survival or function. It was then established that mutations causing FPLD do not alter the 3D structure of the C-terminal of lamin A/C, but might prevent the interaction with lamin A/C protein targets (Dhe-Paganon et al., 2002), thus causing a specific gain of function of type A lamins. On the contrary, mutations found in EDMD, DCM-1A, and LGMD-1B perturb the structure of the lamin nuclear filaments, causing a global loss of function of the A-type lamins (Krimm et al., 2002).
Finally, it has been proposed that LMNA-linked myopathies (AD-EDMD, AR-EDMD, DCM-1A, and LGMD-1B) represent variations on a single clinical continuum that could be influenced by modifying genes or environmental factors (Burke and Stewart, 2002). The existence of a multi-system dystrophy syndrome has been described in two families presenting missense mutations in the amino-terminal head and α-helical rod domains of the lamin A/C gene; these mutations (R28W and R62G) had evidence of myocardial dystrophy, and mild muscular dystrophy in patients with FPLD (Garg et al., 2002). A detailed analysis of the known mutations in LMNA involved in laminopathies and their relationship with the clinical phenotype indicated that the clinical expression of the same mutation often varies in severity among members of the same family and that EDMD patients may show lipodystrophy (Brodsky et al., 2000). Thus, the mechanism by which a mutation in LMNA may cause the disease could depend on the alteration in the affinity of lamin A for a cell-type specific binding partner (Goldman et al., 2002). A further possibility is that lamin A/C distinct functional domains could have a specific role in the maintenance of distinct tissues. Whether these domains can be divided neatly into those involved in interactions with TFs, in mechanical anchorage, and in maintaining chromatin structure, has yet to be determined (Hutchison, 2002).
THE SIGNALING HYPOTHESIS: THE EMERGING ROLE OF ACTIN IN CHROMATIN REMODELING
To date, the known muscular dystrophies are mapped to 29 different loci that give rise to 34 distinct disorders; as more genes are found, it has become evident that signaling processes crucial for muscle functions are altered in disease (Dalkilic and Kunkel, 2003). The hypothesis that signaling processes could be affected in EDMD has been initially proposed by our group in the first workshop of the Myo-Cluster project Euromen (European muscle envelope nucleopathies) in September 2000 (Bonne et al., 2002). In particular, attention has been focused on nuclear envelope related mechanisms that modulate the chromatin arrangement through chromatin remodeling complexes (CRCs), responding to nuclear signaling systems constituted by inositol lipids (Maraldi et al., 2002). In higher eukaryotes CRCs are multisubunit protein complexes, among which those called BAFs (Brahma-related gene associated factors) have been extensively characterized (Wang et al., 1996). The BAF complex in human lymphocytes is constituted by BRG1, β-actin, and the actin-related protein BAF53. The complex is induced to associate to the nuclear matrix/chromatin few minutes after antigen receptor stimulation. The activation of the BAF complex, leading to an impressive chromatin decondensation, is triggered by intranuclear increase of the PI(4,5)P2 levels, that conceivably modulate actin polymerization in the BAF complex, by displacing actin-binding protein BAF53 (Zhao et al., 1998). These findings confirmed a direct interface between chromatin arrangement regulation and signal transduction at the nuclear level (Maraldi et al., 1999). It has therefore been proposed that mutations in lamin A/C and emerin can affect gene expression through a signaling mechanism capable of modulating the chromatin arrangement. This mechanism necessarily requires proteins having dynamic properties, such as actin.
A large body of interest has been recently grown on the role of actin and actin-related proteins (Arps) in the nucleus. Beside cytoplasmic and cytoskeletal localization, actin has been detected in the nucleus of differentiated muscle cells, as well as of other cell types (Gonsior et al., 1999; Rando et al., 2000). The use of a specific anti-epitope antibody, generated against a complex between actin and Arps, demonstrated that nuclear actin presents a specific conformation (Gonsior et al., 1999). Nuclear actin does not appear to form filaments, at least in myogenic cells, suggesting that it is probably in its monomeric, or G-actin form, as indicated by a punctate nuclear staining pattern. Co-localization experiments indicated that nuclear actin is associated with mRNA transcripts of the Balbiani rings of the polytene chromosomes (Percipalle et al., 2001) and with pre-messenger ribonucleoprotein (RNP) complexes in rat liver (Percipalle et al., 2002). Interestingly, actin antibodies are able to inhibit nuclear export of RNAs (Hofmann et al., 2001), while actin has been demonstrated to anchor the tumor suppressor protein p53 to the nuclear matrix (Okorokov et al., 2002). The significance of these findings was further recognized once nuclear actin has been determined to be a nuclear matrix associated protein that could provide the nucleus of the required dynamic properties. To perform these activities, actin-binding proteins are required to be constitutively present in the nucleus, such as the nuclear matrix associated protein 4.1, that plays a role in RNA splicing (Lallena et al., 1998; Krauss et al., 2003), or they have to be translocated to the nucleus in a regulated manner, such as CapG and cofilin (Onoda and Yin, 1993; Nebl et al., 1996). Cofilin may be involved in nuclear actin-mediated control of gene transcription (Sotiropoulos et al., 1999). Profilin antibodies form a speckle-like pattern and co-localize with small nuclear ribonucleoprotein particles (snRNPs); following actinomycin D treatment, profilin and snRNPs reorganize into larger aggregates, suggesting that profilin might play a role in pre mRNA processing and that actin–profilin complexes may be functional components of the spliceosome (Skare et al., 2003).
The identification of actin within the nucleus appeared particularly significant in the light of the presence in the nucleus of physiological factors that regulate actin dynamics (Rando et al., 2000). Among the INM proteins recently identified, nesprins appear to play a crucial role since these actin-associated proteins may serve structural or signaling roles inside the nucleus (Zhang et al., 2001). In fact, nesprins are characterized by the presence of spectrin repeats that are found in many cytoskeletal proteins, including dystrophin (Mislow et al., 2002).
Nuclear actin and Arps are also integral components of complexes involved in modifying or remodeling chromatin (Olave et al., 2002). The chromatin remodeling BAF complex contains β-actin and the actin-related protein BAF53. The complex uses the energy of ATP hydrolysis to increase the accessibility to nucleosomal DNA of transcription factors (Boyer and Peterson, 2000; Rando et al., 2000). BAF53 shows a close homology to Arp3 and Arp4; this last has been detected in colocalization with the heterochromatin protein HP1 (Frankel et al., 1997). The association of actin-related proteins with chromatin remodeling complexes suggests that higher-order chromatin structures may be regulated through actin and actin-related protein interactions.
The presence of profilin, the actin-binding protein involved in actin polymerization, at nuclear functional sites such as speckles (Skare et al., 2003), in association with PI(4,5)P2 (Boronenkov et al., 1998; Maraldi et al., 1999), that is involved in the control of actin polymerization (Holt and Koffer, 2001), suggests that actin–profilin complexes are components of the spliceosome. Inositol phosphates, in association with actin and profilin, could also regulate chromatin remodeling complexes, and in turn, modulate gene expression (Shen et al., 2003; Steger et al., 2003). The involvement of the mammalian BAF complex in gene expression is indirectly suggested by studies on knock-out mice lacking some components of the BAF complex. While the complex appears not to be required for cell viability of differentiated cells, it is essential for embryonic development (Reyes et al., 1998). Furthermore, the inhibition by latrunculin B of actin polymerization prevents SRF-dependent gene activation, reducing ATPase activity of the BAF complex (Sotiropoulos et al., 1999).
As far as the molecular mechanisms involved in the pathogenesis of laminopathies, it has been suggested that alterations in either emerin or lamin A/C might affect the relationships between the nuclear envelope and the nuclear matrix-associated actin, thus affecting the CRC-controlled chromatin decondensation (Maraldi et al., 2002). The failure to correctly sequester transcriptionally inert chromatin at the nuclear periphery (heterochromatin detachment from the nuclear envelope has been observed in several cell types obtained from patients affected by laminopathies) might contribute to the pathogenesis of these diseases by perturbing gene expression in crucial moments of cell differentiation (Maraldi et al., 2002).
The possibility that actin-based and lamin-based signaling complexes within the nucleus could be involved in the pathogenesis of laminopathies has been also advanced (Holaska et al., 2002). This implies that lamin and actin interact inside the nucleus (Shumaker et al., 2003). Experimental evidence about the actual interaction of both emerin and lamin A/C with nuclear actin and of its regulation along myoblast differentiation has been recently provided (Lattanzi et al., 2003).
It is thus conceivable that the BAF complex is regulated via actin and Arps that can serve as motor proteins within the nucleus. Nuclear actin interacts with both lamina A and emerin (Maraldi et al., 2002; Pederson and Aebi, 2002); interestingly, actin–emerin interactions might be regulated by phosphorylation in differentiating myoblasts (Lattanzi et al., 2003), possibly through PKA activation (unpublished results), suggesting that nuclear actin is a biologically relevant partner for emerin and lamin A during myogenesis. Furthermore, emerin has the potential to stabilize actin polymers at the nuclear envelope (Bengtsson and Wilson, 2004), while actin binds directly to two regions of lamin A/C tail (Sasseville and Langelier, 1998; Zastrow et al., 2004). Therefore, it is conceivable that actin polymers constitute architectural partners for lamins, thus influencing the arrangement of chromatin, and directly or indirectly, gene regulation.
Emerin-deficient nuclei, or nuclei containing altered lamin A/C, might present defects arising from an altered chromatin arrangement that does not necessarily affect gene expression in all cell types (Maraldi et al., 2003). Some cell types, indeed, present long quiescent periods with sudden activation phases; such changes require deep chromatin remodeling and the reprogramming of the whole nuclear size and shape. The cells affected in laminopathies, those of mesenchymal origin, such as cardiac and skeletal muscle cells, and non mesenchymal cells, such as adipocytes and motor neurons, are usually non proliferating cells. It is conceivable that subtle alterations in chromatin arrangement affecting gene expression might negatively affect mainly long-lasting cells (Maraldi et al., 2003).
DISEASE MECHANISMS AFFECTING GENE REGULATION: HETEROCHROMATIN REORGANIZATION BY LAMIN A
Relationships between epigenetic changes and the pathogenesis of several monogenic diseases have been established to involve alterations in chromatin remodeling (Huang et al., 2003). This process, that involves DNA modifications, covalent histone modifications and relocation, controls chromatin structure and thereby gene expression. It is conceivable that perturbation of chromatin remodeling and of gene expression can explain the multi-system interest and the phenotypic complexity of laminopathies.
Lamin A/C provides scaffolds for a variety of proteins that regulate gene expression (Cohen et al., 2001). Rb protein, a transcriptional regulator that controls both the cell-cycle progression and apoptosis (Vlcek et al., 2002), is part of a functional complex that includes lamins, LAP2α and GCL (Markiewicz et al., 2002; Holaska et al., 2003) and could tether chromatin remodeling complexes (Neely and Workman, 2002). The transcriptional repressor MOK2 links specific binding sites of A-type lamins (Dreuillet et al., 2002), while another gene repressor that binds lamin A/C is Barrier to autointegration factor (Wang et al., 2002), that also interacts with emerin.
It is particularly compelling to note that the gene regulators found to bind A-type lamins are repressors and that a common feature of laminopathic cells is a rearrangement of peripheral chromatin (Fidzianska et al., 1998; Ognibene et al., 1999; Sullivan et al., 1999; Raharjo et al., 2001; Sabatelli et al., 2001; Capanni et al., 2003; Favreau et al., 2003), which is considered to be repressed in normal cells (Mattout-Drubezki and Gruenbaum, 2003).
A hint for the understanding of lamin role in the modulation of chromatin arrangement came from an unexpected result we obtained while investigating the role of lamin A precursor protein. Accumulation of pre-lamin A can be obtained by treatment with farnesyltransferase inhibitors (Sasseville and Raymond, 1995) which impair subsequent processing of the precursor protein by endoproteases such as the metalloprotease ZMPSTE24 (Agarwal et al., 2003). ZMPSTE24 gene mutations give rise to MAD (Agarwal et al., 2003), strongly suggesting that altered lamin A processing may be causative of laminopathic features. In fact, we obtained evidence, that in myoblasts, pre-lamin A accumulation by farnesyltransferase inhibitor treatment is associated with formation of nuclear lamina invaginations protruding in the nuclear interior. Moreover, chromatin organization is strikingly affected (Fig. 1). Re-organization of methylated histone H3 within pre-lamin accumulating nuclei is evident (Fig. 1A) and peripheral heterochromatin areas are widely increased in nuclei with high pre-lamin A amounts (Fig. 1B), suggesting a direct link between the rate of lamin A processing and the amount of peripheral heterochromatin. In other cell types, our preliminary results indicate a different degree of chromatin reorganization and a more relevant effect on the overall nuclear morphology. Thus, at least in MAD patients bearing ZMPSTE24 mutations, altered lamin processing may cause chromatin disorganization. This further suggests that chromatin remodeling is a key event in the cascade of epigenetic events causing laminopathies.
However, a direct evidence that alteration of chromatin organization occurring in laminopathies also influence gene expression is required to sustain this pathogenetic model. Inhibition of RNA polymerase II-dependent transcription has been observed in dominant-negative lamin A mutants (Spann et al., 2002), while RNA polymerase II inhibitors induce lamin A to aggregate within the nucleus (Kumaran et al., 2002). Altered lamin A/C organization at the nuclear periphery and/or an abnormal accumulation at intranuclear foci may occur in laminopathic cells, and this event could also affect RNA transcription. A more direct evidence of this mechanism has been obtained in FPLD cells carrying a R482L mutation in lamin A/C. These cells, in fact, presented large nuclear lamin A/C aggregates, mainly in G1 and G2 cells, and not in serum-starved or in S-phase cells. Moreover, by measuring BrU incorporation, the cells bearing lamin A/C aggregates showed a reduced RNA transcription rate (Capanni et al., 2003). These data provide the first evidence that RNA transcription might be related to altered lamin A/C localization or to failure of mutated lamin A/C to interact with chromatin constituents.
Finally, it remains to fully demonstrate, as a general rule, that laminopathic cells are affected by developmental disorders that result is an accelerated ageing, as observed in some human progeroid syndromes (Novelli et al., 2002; Chen et al., 2003; De Sandre-Giovannoli et al., 2003; Eriksson et al., 2003; Mounkes et al., 2003). It has been suggested that progeria is a developmental disease in which some tissues present an accelerated or improper development; many of these tissues arise from the mesenchymal stem cell lineage, such as heart, muscle, bone, and subcutaneous adipose tissue. It is also conceivable that terminally differentiated tissues with an altered nuclear lamina may be unable to maintain the silenced heterochromatin organization capable to preserve a state of terminal differentiation, resulting in improper dedifferentiation. This hypothesis is particularly attractive since our recent data show involvement of lamin A precursor protein in heterochromatin dynamics (Fig. 1 and unpublished results). Moreover, altered pre-lamin A processing features have been reported in Hutchinson–Gilford progeria (Eriksson et al., 2003) and ZMPSTE24-associated MAD (Agarwal et al., 2003). The confirmation of this hypothesis will be of particular interest not only for understanding the pathogenesis of laminopathies but also to consider nuclear envelope proteins as possible modulators of aging mechanisms.