sign in sign up
<<
Home , Dystonin, Intermediate filament, Keratin 6A

93

Intermediate filaments and their associated : multiple dynamic personalities Megan K Houseweart∗ and Don W Cleveland²

A fusion of mouse and human genetics has now proven that regulated by , the recent identification of intermediate filaments form a flexible scaffold essential for several kinases involved strengthens the argument that structuring in a variety of contexts. In some these dynamics are true in vivo events. The growing cases, the formation of this scaffold is achieved through a list of intermediate-filament-associated proteins (IFAPs) newly identified family of intermediate-filament-associated surely expands the repertoire of possible interactions proteins that form cross-bridges between intermediate permitted to IFs and may also add another layer of filaments and other cytoskeletal elements, including and regulatory complexity. Indeed, the discovery of several . human and mouse diseases caused by mutations in the IFAP proteins and neuronal antigen (BPAG1n) illustrates the importance of these Addresses molecules in providing links between components of the ∗Ludwig Institute for Cancer Research and Division of Cellular and Molecular Medicine, University of California at San Diego, 9500 cellular . Most would agree that intermediate Gilman Drive, La Jolla, CA 92093, USA filaments can no longer be thought of as the least dynamic ²Ludwig Institute for Cancer Research, Division of Cellular and components of the cell cytoskeleton. This review focuses Molecular Medicine, and Department of Medicine and Neuroscience, on the recent advances in understanding IFs and their University of California at San Diego, 9500 Gilman Drive, La Jolla, CA associated proteins. 92093, USA; e-mail: [email protected] Current Opinion in 1998, 10:93–101 A growing family of vital intermediate http://biomednet.com/elecref/0955067401000093 filament cross-linkers  Current Biology Ltd ISSN 0955-0674 Perhaps the most exciting new development in the IF field has been the recognition that IFs have binding Abbreviations partners and that these partners that have important ALS amyotrophic lateral sclerosis BPAG bullous pemphigoid antigen functions in structuring the three-dimensional cytoplasm. GFAP glial fibrillary acidic The IF-associated proteins, or IFAPs as they have come IF intermediate filament to be called, are steadily gaining attention as more diseases IFAP IF-associated protein in humans and mice are shown to arise from mutations MD-EBS muscular dystrophy with NF neurofilament in these proteins. The ability of IFAPs to link various NLS nuclear localization signal components of the cytoskeleton in many different cell SOD1 superoxide dismutase 1 types suggests several potential functions as dynamic regulators of cytoskeletal assembly and maintainers of IF network integrity. Support for this notion came recently Introduction from studies that utilized peptides corresponding to Intermediate filaments (IFs) have long been thought a conserved helix region of IF proteins that, when of as fixed structural bystanders around whom the injected into fibroblast cells, disrupted the IF, , •• lively activity of the cell is distributed. More recently, and microfilament networks [1 ]. The authors proposed however, IFs and their associated proteins have been that the introduced peptides effectively competed for firmly established as constituents of deformable cellular IFAPs that would normally link the filament networks latticeworks, imparting integrity and strength to tissues together and in this way caused the rapid disassembly throughout the body. Long known to extend throughout of the cytoskeleton. Currently, the list of IFAPs includes the cytoplasm, possibly positioning the nucleus within the members that interact with IFs from all five IF subtypes, cell, IF networks have been shown to reversibly link the but as new IFs (including the lens-specific beaded • plasma membrane to other cytoskeletal components to filaments [2 ]) are identified, the discovery of new linking modulate cell shape and confer resistance to mechanical partners will surely follow. stress. As dynamic components of the cytoplasmic and nuclear , IFs are thought to contribute to Plectin: an essential linker between intermediate cellular structural rearrangements that occur during cell filaments, microtubules, , and actin division. Evidence for pools of soluble IF subunits that can One of the most thoroughly characterized IFAPs is plectin, exchange along the entire length of assembled filaments an abundant and extremely large (>500 kDa) cytoskeletal has grown and now helps to explain the basis for the cross-linker. Plectin is expressed in many cell types and rapid changes in IF organization that observed in response was initially shown by solid phase binding analysis to to such stimuli as heat shock and application of growth interact with a wide variety of other proteins such as factors. As many examples of IF dynamics appear to be , glial fibrillary acidic protein, , B, 94 Cytoskeleton

microtubule-associated proteins, α-, and all three attachment of epidermal cells to the basement membrane. neurofilament proteins [3,4]. More recently, immunoelec- That the strength provided by plectin is an essential tron microscopy studies [5••] to evaluate plectin-binding feature of cellular architecture has been demonstrated by interactions revealed that a) plectin forms cross-bridges the recent discovery of mutations in the for plectin between IFs and microtubules, b) vimentin filaments are as a cause of the human disease muscular dystrophy decorated with plectin projections, c) plectin links IFs with epidermolysis bullosa (MD-EBS) [13••–15••]. This and actin filament bundles, and d) plectin associates with inherited disease is characterized by muscle degeneration myosin filaments in cultured cells. The extraordinarily and skin blistering due to a failure to anchor the cellular IF clear images of plectin cross-bridges between various network to the plasma membrane via . filament networks supports many of the previously Three of the four known mutations, an eight base postulated plectin interactions and is further corroborated pair insertion [13••], an eight deletion [14••], by biochemical and domain sequence data. Inspection of and a single nucleotide deletion [15••], occur within the rat and human plectin sequences has demonstrated the same region of plectin and result in premature the existence of a putative amino-terminal actin-binding termination approximately one third of the way through domain [6]. the >500 kDa polypeptide. The fourth known mutation, a nine nucleotide deletion [15••], removes three amino acids Sequence inspection and mutagenesis studies have mapped from within a 23 stretch that was shown to be a nuclear localization signal (NLS) within plectin’s carboxy- identical between human and rat plectin. terminal vimentin binding domain motif [7••]. Beyond its role as a versatile cross-linker in interphase, the Tissues from patients harboring the three truncation presence of a plectin NLS [6], the fact that plectin mutations are typically devoid of plectin, whereas the binds lamin B in vitro [4], and the established pattern fourth mutation apparently encodes an otherwise full of plectin disassociation from vimentin upon entry into length protein and results in reduced plectin levels. mitosis [8•] suggest several possible roles for plectin during Patients with any of the four mutations typically display cell division. One scenario predicts that separating from reduced levels of BPAG1e, muscle fiber abnormalities, cytoplasmic IFs would free plectin and collapse the IF and hemidesmosomes without an inner attachment plate network into the typical cagelike IF structure seen during [13••–15••]. mitosis. Thus freed, the uncovered plectin NLS could sequester NLS binding proteins. Alternatively, plectin’s In order to determine more directly which aspects association with lamin B may act to disassemble the of plectin cross-linking activities are essential in vivo, nuclear matrix during breakdown, or transgenic mice were engineered that lack the plectin perhaps the interaction really functions in the opposite protein [16••]. These plectin-null mice die two to three manner, that is, to promote nuclear reassembly (as argued days after birth and display pathological features typical in detail [9]). Although these ideas are as yet only of MD-EBS, with a few notable differences. For example, speculative, the protein kinase p34cdc2 has been shown the keratinocytes of MD-EBS skin rupture at the basal to phosphorylate plectin and cause its dissociation from level, whereas the keratinocytes of vimentin during mitosis [8•]. Similarly, phosphorylation plectin-null mice rupture at all levels and contain ul- of plectin by protein kinase A or C can inhibit the trastructurally normal hemidesmosomes and . binding of plectin to lamin B [4], providing the means by Plectin-deficient cells in mice were often which plectin could accomplish the mitosis-specific roles necrotic and disrupted were prevalent, whereas proposed above. cardiomyocytes displayed abnormally arranged sarcomeres and disintegrating intercalated discs. Additionally, the Comparison of plectin’s structural domains with those distribution and expression levels of selected cytoskeletal of other cytoskeletal components has yielded intriguing components were shown to be altered in the mice lacking similarities. Plectin shares a high degree of sequence plectin, suggesting the involvement of plectin in some homology and similar domain organization with the aspect of their function. The existence of muscle and neuronal protein BPAG1n/ [6], the epidermal skin tissue in these newborn mice indicate that plectin is isoforms of BPAG [10], mACF7 [11•], and the desmosomal not necessary for formation and assembly of IFs in these proteins , I, and desmoplakin II. tissues, but is required to provide the stability to withstand Because of these similarities, this group of proteins has subsequent mechanical stresses during life. recently been dubbed the ‘ family’ [12]. Plectin, the and cadherins are linker components The loss of plectin protein expression resulting from of desmosomes in cells that experience mechanical stress the previously described mutations in humans and in and function to link IF networks to the plasma plectin-null mice accounts for the skin fragility, muscle membrane, thereby imparting mechanical strength to degeneration and neurodegeneration typically seen in the individual cell and to the entire tissue. Similarly, MD-EBS. One important question that remains unad- plectin, BPAG2e, BPAG1e, and the α6β4 integrin link dressed concerns the aberrant muscle phenotype: is it the the components of hemidesmosomes and mediate the loss of plectin’s linkage between the IF network and the Intermediate filaments and their associated proteins Houseweart and Cleveland 95

plasma membrane, or is it the loss of plectin’s binding sensory neurons in the BPAG1n-deficient mice are the to actin fibers, that accounts for the fragile muscle cells most severely affected of all the neurons when it is of MD-EBS patients and plectin-deficient mice? Overall, likely that motor neurons should have similar needs it appears safe to conclude that plectin is a true linker for linker proteins [20•,21]. One explanation for this of multiple cytoskeletal components, providing flexible apparent paradox is that some other protein can perform tensile strength to the three-dimensional cytoplasm within the necessary linking function when BPAG1n/dystonin many different cell types. is missing or not expressed in a particular neuronal cell type. One such candidate protein is the most recently • BPAG1n/dystonin: an essential component of sensory discovered plakin family member, mACF7 [11 ]. Although neurons the analysis of the mACF7 gene is at an early stage, this As mentioned earlier, the epidermal forms of bullous protein shows significant homology to BPAG1n/dystonin, pemphigoid antigen and the neuronal form, BPAG1n, are contains an actin-binding domain, and is expressed at members of a sequence-related family of cytoskeletal-link- appreciable levels in the nervous system, making it an ing proteins that also includes plectin, envoplakin, the attractive potential linker protein. desmoplakins, mACF7 [11•], and a 450 kDa plectin-like antigen [17]. Like the other members of this family, Neuronal intermediate filaments: the BPAG members are large proteins with a central understanding normal function and role in α-helical coiled-coil flanked by a globular head and disease repeating tail segments. On the basis of sequence analysis Neurofilaments (NFs) are the predominant type of and its position within epithelial hemidesmosomes, it intermediate filament in most adult neurons of both the was initially proposed that BPAG1e would help to central and the peripheral nervous system. The NFs of form the connections between keratin IF networks and mature myelinated are composed of an NF-L core, the basement membrane. To address this possibility, with NF-H and NF-M subunits incorporated into the mice bearing disruptions in the BPAG1e gene were fiber allowing their tails to extend laterally (reviewed in developed [18]. The basal epidermal cells of BPAG1e-null depth in [22]). Several observations have led to the idea mice contained hemidesmosomes that were disengaged that NFs control the increases in axonal diameter that from the keratin network and cytolysis occurred at occur following synapse formation. As neurons mature, this level upon mechanical stress, demonstrating the expression of NFs is increased and accumulation of importance of BPAG1e in maintaining IF contacts with the NFs corresponds directly with increased diameter of hemidesmosome. Surprisingly, these mice also displayed developing axons, a key determinant of conduction severe neurologic defects characteristic of the dystonia velocity. NF investment into axons was initially shown musculorum (dt/dt) mouse mutant [18]. The explanation to be essential for the establishment of proper for this additional neuronal phenotype was unclear until diameter by the study of a quail mutant that lacks NFs positional cloning studies revealed that the targeted [23]. This finding has been confirmed both in transgenic in the BPAG1e-null mouse also codes for an alternatively mice expressing an NF-H gene that blocks filament spliced isoform, BPAG1n/dystonin [19]. transport into axons [24], and, more recently, in mice lacking NF-L [25••]. These latter mice lack axonal NFs (or Subsequent characterization of the previously unknown NF-L, NF-M, and NF-H) and display decreased axonal neuronal isoform BPAG1n/dystonin demonstrated that outgrowth, delayed regeneration after nerve injury, and it contains a carboxy-terminal neurofilament-binding do- have a 15–20% loss of motory and sensory axons at 2 main, found in common with the epidermal BPAG months of age. isoforms, and an amino-terminal actin-binding domain that is found only in the neuronal isoform. Upon expression Studies using transgenic mice engineered to overexpress in tissue culture cells lacking IFs, BPAG1n/dystonin was different combinations of NF subunits have proven shown to co-align neurofilaments with actin filaments, that NF-dependent radial growth of axons requires the suggesting that such a linking property is likely to be presence of NF-L to drive filament assembly as well physiologically important in neurons [7••]. The appearance as properties provided by NF-M and NF-H [26•,27•]. of abnormal neurofilamentous networks in degenerating The latter two are thought to organize axoplasm in a BPAG1n-deficient neurons supports this possibility [20•]. three-dimensional cross-linked array that is capable of One problem with this scenario is the finding that, supporting axonal expansion. However, a separate increase although the expression pattern of BPAG1n/dystonin in either NF-M or NF-H levels reduces the number generally corresponds to the affected areas of the dt/dt of axonal filaments (by trapping NFs in the cell body), and BPAG-null nervous system, there are instances where while increasing the levels of either NF-M or NF-H this is not the case [20•,21]. For example, some neurons in the presence of increased NF-L stimulates radial that would normally express BPAG1n/dystonin do not growth without changing the nearest neighbor spacing of degenerate in the dt/dt and BPAG-null mice, questioning neurofilaments [26•,27•]. Challenging this finding is the the universal requirement for such a linking protein demonstration of increased filament number and closer in axons. In addition, it is unclear why the primary filament spacing in the smaller central nervous system 96 Cytoskeleton

axons of mice overexpressing the human NF-M gene introduction of a single point mutation into the mouse [28]. Here, the elevated levels of NF-M that provoke NF-L gene (modeled after keratin mutations frequently increased accumulations of NF-L and reduced levels of found in epidermolysis bullosa simplex) produces mice phosphorylated NF-H may have reduced the amount of with selective spinal motor neuron degeneration and NF-H subunits available to form filaments with NF-L. death, NF accumulations, and skeletal muscle atrophy Both studies seem to agree that increased radial growth [32]. requires a carefully balanced ratio of NF-L subunits and of the cross-bridge-forming components NF-M and NF-H, The ability of these varied NF alterations to selectively but do not fully resolve the question of how an NF array cause neurodegenerative disease in mice provoked a specifies axon diameter. It is clear that, in the largest search for mutations in NFs as the primary cause of ALS caliber axons of mice, when very different ratios of the in humans. These efforts revealed no mutations in any three subunits are expressed and very different amounts of NF gene from 100 individuals with familial ALS [33], axonal NF are made, the nearest neighbor spacing of these and none in the NF-H KSP region of 117 familial ALS filaments is unchanged. This must indicate the existence patients [34]. Nevertheless, a previous effort identified of attractive forces between adjacent NFs (these forces either of two small deletions within a KSP-repeating are quite apparent in mice with few axonal NFs), but domain of the NF-H gene in 5 of 356 patients with also that the cross-linkers crucial for radial growth must sporadic ALS [35]. Despite these disappointing figures, provide longer-range interactions either between NFs that ∼15% of familial ALS patients have been shown to bear are not nearest neighbors or between NFs and other axonal mutations in the antioxidant metalloenzyme superoxide components. dismutase 1 (SOD1) [36]. Although the exact mechanism by which mutant SOD1 causes ALS is unknown, a These NF transgenic mice have also been used to show series of experiments has shown that the SOD1 defects that expression of NF-H at four times the endogenous are not caused by a loss of SOD1 activity, but instead level selectively slows filament transport through axons may result from a gained toxic property. This proposed [26•]. It has been suggested that this overabundance of toxic activity may cause the most damage to long-lived NF-H leads to the assembly of more NFs than can be and abundant neuronal components such as NFs. Strong effectively transported, resulting in the NF accumulation evidence for this proposal comes from the recent discovery typically seen in cell bodies and swollen axons of these of neurofilamentous inclusions in individuals with SOD1 mice. Despite the dual insults of slowed transport and mutations [37•,38]. In summary, the data available to widespread NF accumulations, these mice with high levels date clearly demonstrate that NFs themselves can be the of wild-type mouse NF-H do not display phenotypic primary cause of ALS-like motor neuron disease in mice, abnormalities or a loss of neurons. This is in sharp but whether the same is true for humans remains to be contrast to mice expressing lower amounts of wild-type proven. human NF-H, which develop symptoms characteristic of the disease amyotrophic lateral sclerosis (ALS), such : an essential role in cardiac, skeletal, as neurologic defects, muscle atrophy, and neuronal NF and accumulations amid a general slowing of slow axonal The muscle specific IF desmin is expressed in all transport [29]. The finding strongly indicates that the three muscle tissue types: skeletal; cardiac; and smooth human NF-H protein acts as a mutant protein in mice muscle. In developing mammalian muscles, desmin is [26•]. initially co-expressed with vimentin, but upon terminal differentiation vimentin is downregulated and desmin The discovery that neuronal accumulations of NFs accumulates around the Z discs of the maturing cells. are a common pathological hallmark of several human This localization led to suggestions that desmin functions neurodegenerative diseases, including ALS, has sparked a to maintain the adult contractile apparatus by aligning great deal of effort aimed at understanding whether NFs striated myofibrils laterally via their Z discs and also by are themselves the cause of neuron dysfunction, or, at linking myofibrils to the nucleus, T tubules, mitochondria, the other extreme, whether they are innocent bystanders. and sarcolemma. Another possible function of desmin that ALS is a late-onset neurodegenerative disease in which has received more attention lately is a potential role in the selective loss of motor neurons in the brain and muscle differentiation and morphogenesis. spinal cord leads to progressive muscle weakness, followed by paralysis and death. Several transgenic mouse lines To test these predictions, two groups used gene dis- expressing either mutant NFs or high levels of NFs ruption to engineer mice lacking desmin. Both obtained have been created that faithfully mimic the symptoms viable, fertile animals with strikingly similar phenotypes of ALS. Elevation of wild-type mouse NF-L levels [30] [39••,40••]. Specifically, skeletal, cardiac, and smooth or human NF-H levels [31] to approximately four times muscle developed normally, but displayed widespread cell the normal amounts resulted in NF accumulations in the architecture defects such as misaligned muscle fibers, cell bodies and axons of motor neurons, muscle atrophy, abnormal sarcomeres, swollen mitochondria, and calcium and axonal degeneration. Even more compelling, the deposits in cardiac muscle tissue. The desmin-deficient Intermediate filaments and their associated proteins Houseweart and Cleveland 97

mice developed weaker skeletal muscles with less en- abnormalities in myelination by oligodendrocytes and a durance and force generation capabilities than normal loss of white matter in aged mice [46•]. In addition, mice [41•]. Moreover, although all early stages of muscle although various electrophysiological parameters such as differentiation and cell fusion occurred normally, it was basic synaptic transmission were found to be normal in only after birth that myofibers were ruptured during the GFAP-null mice, enhancement of hippocampal LTP contraction and underwent an aberrant repair process that (long-term potentiation) [45] and deficits in cerebellar led to the final pathological state of the muscle [41•]. LTD (long-term depression) were also demonstrated [47•]. The possibility that another IF protein such as vimentin The severe morphological and functional abnormalities may partially compensate for the absence of GFAP seems observed in the most active muscle types of both desmin- plausible given the discovery of GFAP filament assembly null animals mentioned above underscores the importance defects in mice without vimentin [48•]. It seems likely that of desmin in maintaining the structural integrity of muscle a cross between a GFAP-null mouse and the vimentin-null cells, and lends support to the structural model of desmin mouse [49] would definitively settle the question of function. These findings prove that desmin itself is whether vimentin compensates for a lack of GFAP. not required for muscle commitment or differentiation and, when combined with the lack of evidence for Nuclear : emerging dynamic functions any compensatory increase in vimentin levels during Although related structurally to the cytoplasmic IFs, muscle development [39••,40••,41•] and regeneration nuclear lamin IFs are found exclusively in the nucleus in desmin-null mice [41•], or muscle formation in a and are the major constituents of the . double mutant mouse lacking both desmin and vimentin The nuclear lamina underlies the inner nuclear membrane [41•], it is apparent that IFs are not required early in and is thought not only to provide mechanical support muscle development. Although no human myopathies or to the nucleus, but also to aid in nuclear membrane cardiomyopathies have been attributed to a complete lack reassembly following mitosis [50]. A recent method used of desmin, there are several accounts of myopathies with to test the potential roles of lamins and other putative excess desmin in the form of granular and filamentous cell cycle proteins involves a nuclear assembly system aggregates [39••]. made from Xenopus laevis egg extracts which, upon entry into an interphase state, assembles many physiological Glial fibrillary acidic protein: deletion in mice features of the mitotic nucleus [51•]. Addition of a produces subtle, but measurable, defects dominant truncated human lamin A protein to disrupt the The IF family member glial fibrillary acidic protein endogenous nuclear lamin structure blocked the formation (GFAP) is expressed in of the central nervous of a normal nuclear lamina and resulted in increased system, the enteric , and in myelin-forming Schwann fragility of the nuclei, aggregation of endogenous and cells of the peripheral nervous system. Most developing mutant lamin A into nucleoplasmic spheriods, a decreased astrocytes initially express vimentin, but later switch ability to replicate DNA, and the redistribution of chain to express their final adult IF type, GFAP, as they elongation factors from to the abnormal lamin mature. Astrocytes are thought to modulate neuronal aggregates [51•]. These results lend support to arguments function, help form the blood brain barrier, provide that nuclear lamins are integral to nuclear function and structural and nutritional support for adult neurons, and contribute to the formation of some aspects of the DNA maintain glial processes as paths for migrating neurons replication machinery. during development. An early study removed GFAP from an cell line using antisense transfection and Keratins: more telling mutations demonstrated an inhibition of glial process extension in Keratin proteins constitute the largest and most complex the presence of neurons, implying that GFAP is necessary class of intermediate filaments. They are expressed in for this important developmental event [42]. The in vivo epidermal cells throughout the body where they form a validity of this result has been diminished in light of structural network that spans the cell cytoplasm, linking the recent findings from a series of GFAP-null mice the plasma membrane, nucleus, and other cytoskeletal [43–45,46•]. components. Keratin deletions in both humans and mice have proven that the keratin network is crucial for All four independently derived GFAP-deficient mice maintaining the physical integrity and diverse connec- exhibit normal behavior, motor activity, growth, reproduc- tions required of various epithelial tissues [52]. Keratin tion, and life span. This does not mean that GFAP is filaments are obligate heteropolymers, meaning that they completely dispensable, however. Despite the absence of naturally consist of a 1:1 ratio of type I to type II gross structural aberrations, missing astrocyte populations keratin monomers. The 12 type I keratins and 8 type II and a loss of blood brain barrier integrity [43–45,46•], keratins all share the typical IF family structure, which are examples of more subtle phenotypes that have is characterized by an amino-terminal, non-helical, head emerged as more sophisticated functional tests have been domain followed by a central α-helical rod domain and a performed and older mice have been examined. For non-helical tail domain at the carboxyl terminus. Certain example, one group has reported important late-onset defined combinations of keratin monomer pairing occur 98 Cytoskeleton

in a tissue-specific and developmentally regulated fashion, late-onset disorder characterized by fragility of the corneal thereby expanding the properties of keratin filaments [57]. Also, the description of a point mutation to suit the requirements of various epithelial cell types in the type II cortex keratin, hHb6, in two unrelated (reviewed in detail in [52]). families with the disease is the first to provide direct evidence for involvement of keratins in In the past few years, rapid progress has resulted from inherited hair disease [58•]. Like many mutations found in investigators’ ability to link a great number of skin epidermal keratin forms, both the cornea-specific and the disorders to specific keratin mutations, thereby gaining a hair-specific keratin mutations found to date were confined better understanding of both the etiology of the particular to the ends of the central rod domain, demonstrating disease and the function of individual keratins. One the universal importance of these regions across keratin unifying theme emerging from the wealth of data on subtypes. At the last count, 14 of the 20 epithelial keratin mutations is that mutations within the keratin keratin and one of the ten hard α-keratin genes genes generally cluster at the ends of the keratin rod had been shown to harbor mutations causing human domain in severe forms of dominantly inherited disease, genetic disorders. At the current rate, it seems likely that but milder forms of disease result from more tolerated additional mutations will be discovered in the remaining changes in the less conserved head and tail regions [52]. keratins that may, in turn, help determine what aspect of Important conclusions about requirements for filament each abnormal keratin actually induces the cell fragility alignment and higher-order assembly can be drawn from that is characteristic of these diseases. these observations. As expected, the number of skin disease causing keratin mutations has continued to grow In order to directly study the in vivo cellular sequence of and has led to the identification of other non-skin diseases events that leads to mechanosensitivity due to abnormal caused by keratin mutations. keratin, it would be helpful to begin with a normal animal model cell population and be able to induce production Previous studies had shown that all three major subtypes of an altered keratin gene of interest into a defined of the skin blistering disease epidermolysis bullosa set of cells at any time. Such a system was developed simplex (EBS) can result from mutations in using the human K6a promoter and shows promise both K14 or its binding partner (K5). Recently, a as a tool to study keratin function and as a way to rare subtype of EBS, EBS with mottled pigmentation, deliver foreign gene products to humans via inducible, was also shown to be caused by a point mutation in the transgenic skin grafts [59••]. This technique makes use head domain of K5 [53•]. Many mutations that cause the of the fact that expression of the K6a gene is spatially severest forms of epidermolytic hyperkeratosis (EH) had restricted and can be induced in response to the topical traditionally been found in the rod domains of the K1 and application of various chemicals. It was shown that the K10 genes, with less severe forms of the disease caused expression of a gene of interest (in this case lac Z or by head domain mutations, but a recent report was able hK6a) could be reproducibly induced in the stratified to show that a severe form of EH can result from a head epithelia and of mice upon mechanical stress domain mutation as well [54]. To date, only about half or treatment with retinoic acid or the phorbol ester PMA of the epidermolytic palmoplantar keratoderma (EPPK) (phorbol-12-myristate-13-acetate). In a similar fashion, the patients examined have mutations in the K9 gene, raising human K14 promoter was harnessed to overproduce a the possibility that other genes will soon be found that growth hormone in the skin cells of a donor mouse that may contribute to this disorder. The genetic basis for were then successfully grafted to a recipient mouse. The icthyosis bullosa of Siemens (IBS) had previously been recipient mice did display elevated levels of the hormone ascribed to mutations within the rod domain of Ke2; in the bloodstream, indicating the potential utility of using the oral/esophageal mucosal disorder keratin promoters and skin grafts manipulated in vitro (WSN) has been documented to arise from K4 and K13 to secrete specific compounds as drug delivery systems rod domain mutations; and pachyonychia congenita (PC) [60••]. results from mutations in K17, K16 and K6a (other diseases caused by mutations in keratins are reviewed in [52,55]). Conclusions and prospects Most recently, a mutation in K18 has been found in one The collective findings from the past few years have gone patient with a liver cirrhosis of unknown etiology, raising a long way in advancing our understanding of intermediate speculation that similar keratin mutations may either cause filaments and their associated proteins. In particular, or predispose individuals to liver disease [56•]. the widespread use of transgenic and gene deletion mice has helped prove that IFs and their cross-linking In addition to the numerous mutations of epithelial proteins, including plectin and BPAG1n, structure the keratins found to underlie many serious skin disorders, cell cytoplasm by forming flexible, reversible arrays that novel mutations in the cornea-specific keratins of the provide essential resistance to environmental stresses. The eye and the hard α-keratins of the hair and have pace at which new human disorders are found to be come forth. Specifically, K3 and K12 missense mutations associated with, or directly caused by, aberrant/missing were found to cause Meesmann’s corneal dystrophy, a IF proteins and faulty IF connections has revealed an Intermediate filaments and their associated proteins Houseweart and Cleveland 99

efficient way to learn more about the effects of IFs in and localization (8q24). Proc Natl Acad Sci USA vivo. The continued identification of novel IFs and their 1996, 93:4278-4283. binding partners in different cell types will also lead to a 11. Bernier G, Mathieu M, de Repentigny Y, Vidal SM, Kothary R: • Cloning and characterization of mouse ACF7, a novel member better awareness of what role IFs normally play in cellular of the dystonin subfamily of actin binding proteins. Genomics dynamics and how these proteins can cause disease in 1996, 38:19-29. This preliminary characterization of the newest plakin family member, mACF7, humans. reveals that the three mACF7 isoforms have considerable homology to BPAG1n/dystonin and also contain actin binding domain regions. It seems likely that this is the first of many such linking proteins to be found using Acknowledgements homology with other proteins known to connect intermediate filaments to The authors wish to thank their fellow investigators who sent preprints of other cytoskeletal components. work in progress, and apologize to those whose work was not cited because The plakin family: versatile organizers of of space considerations. 12. Ruhrberg C, Watt FM: cytoskeletal architecture. Curr Opin Genet Dev 1997, 7:392-397. 13. Smith FJ, Eady RA, Leigh IM, McMillan JR, Rugg EL, Kelsell References and recommended reading •• DP, Bryant SP, Spurr NK, Geddes JF, Kirtschig G et al.: Plectin deficiency results in muscular dystrophy with epidermolysis Papers of particular interest, published within the annual period of review, bullosa. Nat Genet 1996, 13:450-457. have been highlighted as: The demonstration that mutations in a cytoskeletal linker protein such as plectin can cause a human disease characterized by a lack of cell integrity • of special interest •• was a clear indication of plectin’s essential function in different cell types of outstanding interest and argued strongly for the importance of other linker proteins. See also [14••–16••]. 1. Goldman RD, Khuon S, Chou YH, Opal P, Steinert PM: •• The function of intermediate filaments in cell shape and 14. McLean WH, Pulkkinen L, Smith FJ, Rugg EL, Lane EB, Bullrich •• Loss of cytoskeletal integrity. J Cell Biol 1996, 134:971-983. F, Burgeson RE, Amano S, Hudson DL, Owaribe K et al.: plectin causes epidermolysis bullosa with muscular dystrophy: The use of injected intermediate filament (IF) mimetic peptides that rapidly cDNA cloning and genomic organization. disrupt all three filamentous networks of the cytoskeleton suggests a central Genes Dev 1996, 10 role for filaments in maintaining cellular architecture and hints that IF-asso- :1724-1735. ciated proteins may act as adaptors between the different network types. The demonstration that mutations in a cytoskeletal linker protein such as plectin can cause a human disease characterized by a lack of cell integrity 2. Georgatos SD, Gounari F, Goulielmos G, Aebi U: To bead or not was a clear indication of plectin’s essential function in different cell types • to bead? Lens specific intermediate filaments revisited. J Cell and argued strongly for the importance of other linker proteins. See also Sci 1997, 110:2629-2634. [13••,15••,16••]. This thorough discussion of lens-specific beaded intermediate filaments (IFs) 15. Pulkkinen L, Smith FJ, Shimizu H, Murata S, Yaoita H, Hachisuka describes the current state of the field and highlights the work that led to •• our present understanding of this distinct class of IFs. H, Nishikawa T, McLean WH, Uitto J: Homozygous deletion mutations in the plectin gene (PLEC1) in patients with 3. Foisner R, Leichtfried FE, Herrmann H, Small JV, Lawson D, Wiche epidermolysis bullosa simplex associated with late-onset G: Cytoskeleton-associated plectin: in situ localization, in muscular dystrophy. Hum Mol Genet 1996, 5:1539-1546. vitro reconstitution, and binding to immobilized intermediate The demonstration that mutations in a cytoskeletal linker protein such as filament proteins. J Cell Biol 1988, 106:723-733. plectin can cause a human disease characterized by a lack of cell integrity 4. Foisner R, Traub P, Wiche G: Protein kinase A- and protein was a clear indication of plectin’s essential function in different cell types and argued strongly for the importance of other linker proteins. See also kinase C-regulated interaction of plectin with lamin B and •• •• •• vimentin. Proc Natl Acad Sci USA 1991, 88:3812-3816. [13 ,14 ,16 ]. 5. Svitkina TM, Verkhovsky AB, Borisy GG: Plectin sidearms 16. Andra K, Lassmann H, Bittner R, Shorny S, Fassler R, Propst F, •• mediate interaction of intermediate filaments with •• Wiche G: Targeted inactivation of plectin reveals essential microtubules and other components of the cytoskeleton. J Cell function in maintaining the integrity of skin, muscle, and heart Biol 1996, 135:991-1007. cytoarchitecture. Genes Dev 1997, in press. The immunoelectron microscopy images presented in this work convincingly By disrupting the plectin gene in mice, these authors demonstrate the impor- document the association of plectin with multiple cellular partners (intermedi- tance of the abundant cytoskeletal cross-linking protein plectin for reinforce- ate filaments, microtubules, actin filaments, and myosin), confirming previous ment of several tissues. The defects displayed by these mice are most similar to the human disease muscular dystrophy with epidermolysis bullosa, which biochemical data and establishing plectin as a true organizer of intracellular •• space. was also recently shown (in 14 ]) to result from a loss of plectin. 6. Nikolic B, Mac Nulty E, Mir B, Wiche G: Basic amino acid 17. Fujiwara S, Kohno K, Iwamatsu A, Naito I, Shinki H: Identification residue cluster within nuclear targeting sequence motif is of a 450kDa autoantigen as a new member of the plectin essential for cytoplasmic plectin-vimentin network junctions. family. J Invest Dermatol 1996, 106:1125-1130. J Cell Biol 1996, 134:1455-1467. 18. Guo L, Degenstein L, Dowling J, Yu QC, Wollmann R, Perman 7. Yang Y, Dowling J, Yu QC, Kouklis P, Cleveland DW, Fuchs B, Fuchs E: Gene targeting of BPAG1: abnormalities in •• E: An essential cytoskeletal linker protein connecting actin mechanical strength and cell migration in stratified epithelia microfilaments to intermediate filaments. Cell 1996, 86:655- and neurologic degeneration. Cell 1995, 81:233-243. 665. The discovery that the neuronal protein BPAG1n/dystonin can co-align neu- 19. Brown A, Bernier G, Mathieu M, Rossant J, Kothary R: The mouse rofilaments with actin filaments in cultured cells strengthens its claim as a dystonia musculorum gene is a neural isoform of bullous cytoskeletal cross-linking protein and provides an explanation for the defects pemphigoid antigen 1. Nat Genet 1995, 10:301-306. seen in mice disrupted in this locus (see also [18]). 20. Dowling J, Yang Y, Wollmann R, Reichardt J, Fuchs E: 8. Foisner R, Malecz N, Dressel N, Stadler C, Wiche G: M-phase- • Developmental expression of BPAG1-n: insights into the • specific phosphorylation and structural rearrangement of spastic ataxia and gross neurologic degeneration in dystonia the cytoplasmic cross-linking protein plectin involve p34cdc2 musculorum mice. Dev Biol 1997, 187:131-142. kinase. Mol Biol Cell 1996, 7:273-288. A more thorough examination of BPAG1n/dystonin expression in normal The identification of a kinase responsible for the redistribution of the cross- mice and the pathological changes caused by disrupting the BPAG locus in linking protein plectin from an insoluble vimentin-bound state in interphase transgenic mice showed a less restricted expression pattern and distribution cells to a more soluble vimentin-independent state during mitosis strongly of neuronal abnormalities than was previously reported [18]. suggests that plectin acts to modulate cell architecture in many stages of 21. Bernier G, Brown A, Dalpe G, de Repentigny Y, Mathieu M, the cell cycle. Kothary R: Dystonin expression in the developing nervous 9. Foisner R: Dynamic organization of intermediate filaments system predominates in the neurons that degenerate in and associated proteins during the cell cycle. Bioessays 1997, dystonia musculorum mutant mice. Mol Cell Neurosci 1995, 17:297-305. 6:509-520. 10. Liu CG, Maercker C, Castanon MJ, Hauptmann R, Wiche G: 22. Lee MK, Cleveland DW: Neuronal intermediate filaments. Annu Human plectin: organization of the gene, sequence analysis, Rev Neurosci 1996, 19:187-217. 100 Cytoskeleton

23. Yamasaki H, Itakura C, Mizutani M: Hereditary hypotrophic familial amyotrophic lateral sclerosis with posterior column axonopathy with neurofilament deficiency in a mutant strain of involvement. J Neuropathol Exp Neurol 1996, 55:481-490. the Japanese quail. Acta Neuropathol (Berl) 1991, 82:427-434. 39. Milner DJ, Weitzer G, Tran D, Bradley A, Capetanaki Y: Disruption 24. Eyer J, Peterson A: Neurofilament-deficient axons and •• of muscle architecture and myocardial degeneration in mice perikaryal aggregates in viable transgenic mice expressing a lacking desmin. J Cell Biol 1996, 134:1255-1270. neurofilament-beta-galactosidase fusion protein. Neuron 1994, This work showed directly that desmin imparts the strength and structural 12:389-405. integrity that are necessary for all three muscle types. A lack of desmin in mice does not preclude muscle formation, but does result in mitochondrial 25. Zhu Q, Couillard-Despres S, Julien J-P: Delayed maturation of •• regenerating myelinated axons in mice lacking neurofilaments. disorganization, myofiber misalignment, and degeneration. Exp Neurol 1997, in press. 40. Li Z, Colucci-Guyon E, Pincon-Raymond M, Mericskay M, Pournin Disruption of the neurofilament-L gene in mice resulted in the production of •• S, Paulin D, Babinet C: Cardiovascular lesions and skeletal axons without neurofilaments, reduced axonal diameter, and axonal regener- myopathy in mice lacking desmin. Dev Biol 1996, 175:362-366. ation defects. These animals are another example supporting the view that Together with [39 ••,41•], this work with desmin-deficient mice demonstrates filament number is a key determinant of axonal caliber. the importance of desmin in maintaining the structural integrity of highly used muscle types and proves that desmin is not necessary for the formation of 26. Marszalek JR, Williamson TL, Lee MK, Xu Z-S, Crawford TO, muscle during development. • Hoffman PN, Cleveland DW: Neurofilament subunit NF- H modulates axonal diameter by affecting the rate or 41. Li Z, Mericskay M, Agbulut O, Butler-Borwne G, Carlsson L, neurofilament transport. J Cell Biol 1996, 135:711-724. • Thronell L-E, Babinet C, Paulin D: Desmin is essential for With a series of transgenic mice expressing completely wild-type mouse NF- the tensile strength and integrity of myofibrils but not for H at up to four times the normal amount, this work demonstrates a selective myogenic commitment, differentiation, and fusion of skeletal slowing of NF transport, but the absence of any neuronal degeneration under muscle. J Cell Biol 1997, 139:1-16. these conditions. This indicates that that motor neuron disease developed Together with [39••,40••], this work with desmin-deficient mice demonstrates by mice expressing human NF-H must arise from the human protein acting the importance of desmin in maintaining the structural integrity of highly used like a mutant in mice. muscle types and proves that desmin is not necessary for the formation of muscle during development. 27. Xu Z-S, Marszalek JR, Lee MK, Wong PC, Folmer J, Crawford TO, • Hsieh S-T, Griffin JW, Cleveland DW: Subunit composition of 42. Weinstein DE, Shelanski ML, Liem RK: Suppression by antisense neurofilaments specifies axonal diameter. J Cell Biol 1996, mRNA demonstrates a requirement for the glial fibrillary acidic 133:1061-1069. protein in the formation of stable astrocytic processes in By mating transgenic mice expressing elevated levels of murine NF-L, NF-M response to neurons. J Cell Biol 1991, 112:1205-1213. or NF-H, to produce mice with elevated synthesis of any pair of subunits, this work demonstrates that NF-dependent stimulation of growth in axonal 43. Gomi H, Yokoyama T, Fujimoto K, Ikeda T, Katoh A, Itoh T, Itohara Mice devoid of the glial fibrillary acidic protein develop diameter requires NF-L to drive assembly and NF-M and NF-H to mediate S: normally and are susceptible to scrapie prions. interactions in the axoplasm. Neuron 1995, 14:29-41. 28. Tu PH, Elder G, Lazzarini RA, Nelson D, Trojanowski JQ, Lee VM: Overexpression of the human NFM subunit in transgenic mice 44. Pekny M, Leveen P, Pekna M, Eliasson C, Berthold C-H, modifies the level of endogenous NFL and the phosphorylation Westermark B, Betsholtz C: Mice lacking glial fibrillary acidic protein display astrocytes devoid of intermediate filaments but state of NFH subunits. J Cell Biol 1995, 129:1629-1640. develop and reproduce normally. EMBO J 1995, 14:1590-1598. 29. Collard JF, Cote F, Julien JP: Defective axonal transport in a transgenic mouse model of amyotrophic lateral sclerosis. 45. McCall MA, Gregg RG, Behringer RR, Brenner M, Delaney Nature 1995, 375:61-64. CL, Galbreath EJ, Zhang CL, Pearce RA, Chiu SY, Messing A: Targeted deletion in astrocyte intermediate filament (GFAP) 30. Xu Z, Cork LC, Griffin JW, Cleveland DW: Increased expression alters neuronal physiology. Proc Natl Acad Sci USA 1996, of neurofilament subunit NF-L produces morphological 93:6361-6366. alterations that resemble the pathology of human motor neuron disease. Cell 1993, 73:23-33. 46. Liedtke W, Edelmann W, Bieri PL, Chiu F-C, Cowan NJ, • Kucherlapati R, Raine CS: GFAP is necessary for the integrity 31. Cote F, Collard JF, Julien JP: Progressive neuropathy in of CNS white matter architecture and long-term maintenance transgenic mice expressing the human neurofilament heavy of myelination. Neuron 1996, 17:607-615. gene: a mouse model of amyotrophic lateral sclerosis. Cell More detailed examination of older mice lacking GFAP revealed defects in 1993, 73:35-46. axon myelination and white matter abnormalities that were not apparent in earlier efforts [43,44]. With [47•], it demonstrates the surprisingly subtle 32. Lee MK, Marszalek JR, Cleveland DW: A mutant neurofilament subunit causes massive, selective motor neuron death: effects of removing the primary intermediate filament constituent, GFAP, from implications for the pathogenesis of human motor neuron support cells of the CNS. disease. Neuron 1994, 13:975-988. 47. Shibuki K, Gomi H, Chen L, Bao S, Kim JJ, Wakasuki H, Fujisaki • T, Fujimoto K, Katoh A, Ikeda T et al.: Deficient cerebellar long 33. Vechio JD, Bruijn LI, Xu Z, Brown RH Jr, Cleveland DW: term depression, impaired eyeblink conditioning, and normal Sequence variants in human neurofilament proteins: absence motor coordination in glial fibrillary acidic protein mutant mice. of linkage to familial amyotrophic lateral sclerosis. Ann Neurol Neuron 1996, 16:587-599. 1996, 40:603-610. Mice devoid of glial fibrillary acidic protein were found to be deficient in 34. Rooke K, Figlewicz DA, Han FY, Rouleau GA: Analysis of the a form of motor learning that is thought to be mediated by long-term de- KSP repeat of the neurofilament heavy subunit in familiar pression, supporting the view that astrocytes can modulate higher neuronal amyotrophic lateral sclerosis. Neurology 1996, 46:789-790. functions. 35. Figlewicz DA, Krizus A, Martinoli MG, Meininger V, Dib M, Rouleau 48. Galou M, Colucci-Guyon E, Ensergueix D, Ridet JL, Gimenez GA, Julien J-P: Variants of the heavy neurofilament subunit • Y, Ribotta M, Privat A, Babinet C, Dupouey P: Disrupted glial are associated with the development of amyotrophic lateral fibrillary acidic protein network in astrocytes from vimentin sclerosis. Hum Mol Genet 1994, 3:1757-1761. knockout mice. J Cell Biol 1996, 133:853-863. This report demonstrated that mature astrocytes normally co-expressing vi- 36. Rosen DR, Siddique T, Patterson D, Figlewicz DA, Sapp P, Hentati mentin and glial fibrillary acidic protein (GFAP) require the presence of vi- Mutations in A, Donaldson D, Goto J, O’Regan JP, Deng HX: mentin for normal assembly of a GFAP network, whereas other astrocytes Cu/Zn superoxide dismutase gene are associated with familial that only expressed vimentin during development form GFAP structures in amyotrophic lateral sclerosis. 362 Nature 1993, :59-62. the absence of vimentin. This represents the first step in understanding the 37. Rouleau GA, Clark AW, Rooke K, Pramatarova A, Krizus differential IF requirements of even highly related cell types. • SOD1 mutation A, Suchowersky O, Julien JP, Figlewicz D: 49. Colucci-Guyon E, Portier MM, Dunia I, Paulin D, Pournin S, is associated with accumulation of neurofilaments in Babinet C: Mice lacking vimentin develop and reproduce amyotrophic lateral sclerosis. 39 Ann Neurol 1996, :128-131. without an obvious phenotype. Cell 1994, 79:679-694. This work demonstrates that neurofilament misaccumulation, itself capable of causing motor neuron disease, is a downstream consequence of a mutation 50. Moir RD, Spann TP, Goldmann RD: The dynamic properties in superoxide dismutase 1 that causes amyotrophic lateral sclerosis. and possible functions of nuclear lamins. Int Rev Cytol 1995, 162B:141-181. 38. Shibata N, Hirano A, Kobayashi M, Siddique T, Deng HX, Hung WY, Kato T, Asayama K: Intense superoxide dismutase-1 51. Spann TP, Moir RD, Goldman AE, Stick R, Goldman RD: immunoreactivity in intracytoplasmic hyaline inclusions of • Disruption of nuclear lamin organization alters the distribution Intermediate filaments and their associated proteins Houseweart and Cleveland 101

of replication factors and inhibits DNA synthesis. J Cell Biol 57. Irvine AD, Corden LD, Swensson O, Swensson B, Moore JE, 1997, 136:1201-1212. Frazer DG, Smith FJ, Knowlton RG, Christophers E, Rochels R et Use of dominant-negative mutant lamin proteins to disrupt nuclear lamin al.: Mutations in cornea-specific keratin K3 or K12 genes cause networks provides evidence for the ability of lamins to modulate nuclear Meesmann’s corneal dystrophy. Nat Genet 1997, 16:184-187. processes such as DNA replication and nuclear reassembly. 58. Winter H, Rogers MA, Langbein L, Stevens HP, Leigh IM, Labreze 52. Albers KM: Keratin biochemistry. Clin Dermatol 1996, 14:309- • C, Roul S, Taieb A, Kreig T, Schweizer J: Mutations in the 320. hair cortex keratin hHb6 cause the inherited hair disease 53. Uttam J, Hutton E, Coulombe PA, Anton-Lamprecht I, Yu QC, monilethrix. Nat Genet 1997, 16:372-374. • Gedde-Dahl T Jr, Fine JD, Fuchs E: The genetic basis of This first description of a mutation in a hard α-keratin that causes an inherited epidermolysis bullosa simplex with mottled pigmentation. Proc hair disease demonstrates the universal requirement for keratins in the varied Natl Acad Sci USA 1996, 93:9079-9084. cell types they inhabit. This report is the first to identify a genetic mutation responsible for a rare 59. Takahashi K, Coulombe PA: A transgenic mouse model with an subtype of epiermolysis bullosa simplex (EBS), EBS with mottled pigmenta- •• tion. Also of note, this mutation occurs within the head domain of keratin 5, inducible skin blistering disease phenotype. Proc Natl Acad Sci whereas most other disease-causing keratin mutations have been shown to USA 1996, 93:14776-14781. reside in the more conserved ends of the rod domain. This work used the keratin 6 promoter inducibly to express a heterologous in the skin of mice. The utility of the system was demonstrated 54. Yang JM, Nam K, Park KB, Kim WS, Moon KC, Koh JK, Steinert by producing mice that developed a skin blistering disease most similar to PM, Lee ES: A novel H1 mutation in the chain in IBS (icthyosis bullosa of Siemens) upon induction of a mutation. epidermolytic hyperkeratosis. J Invest Dermatol 1996, 107:439- The work sets the stage for delivery of foreign gene products via transgenic 441. skin grafts. See also [60••]. The cytoskeleton and disease: genetic disorders of 55. Fuchs E: Transgenic studies intermediate filaments. 30 60. Wang X, Zinkel S, Polonsky K, Fuchs E: Annu Rev Genet 1996, :197-231. •• with a keratin promoter-driven growth hormone transgene: 56. Ku NO, Wright TL, Terrault NA, Gish R, Omary MB: Mutation of prospects for gene therapy. Proc Natl Acad Sci USA 1997, • human in association with cryptogenic cirrhosis. J 94:219-226. Clin Invest 1997, 99:19-23. The keratin 14 promoter was used to drive expression of a growth hormone These investigators show for the first time that a mutation within the keratin in the skin of mice. Skin grafts from such donor mice could be transferred to pair K8/K18 may account for human liver disease of unknown etiology, in recipient mice to sustain increased levels of the hormone in the blood stream, accordance with previous work using K8-null mice which die during embryo- demonstrating the possible utility of this system for human gene therapy by genesis as a result of liver hemorrhage. delivering products secreted by grafted skin cells. See also [59••].