sign in sign up
<<
Home , Dynactin, Dynein, Fimbrin, Intermediate filament, Keratin 8, Kinesin

Review

Intermediate filament assembly: dynamics to disease

Lisa M. Godsel, Ryan P. Hobbs and Kathleen J. Green

Department of Pathology, Northwestern University Feinberg School of Medicine, 303 E. Chicago Avenue, Chicago IL 60611, USA

Intermediate filament (IF) belong to a large and providing mechanical integrity that is cru- diverse family with broad representation in cially important for tissue function. This is highlighted tissues. Although considered the ‘toughest’ by a growing list of >75 human genetic diseases caused cytoskeletal fibers, studies in cultured cells have revealed by deficiencies in this network, including skin fragility that IF can be surprisingly dynamic and highly regulated. and epidermolytic disorders, , myopathies, This review examines the diversity of IF assembly beha- neuropathies, cataracts and premature aging [6,7] viors, and considers the ideas that IF proteins are co- or (Human Mutation Database, post-translationally assembled into oligomeric precur- http://www.interfil.org). Notably, an emerging set of sors, which can be delivered to different subcellular com- mutations in nuclear (reviewed elsewhere) com- partments by or actomyosin and associated prise a large proportion of human diseases attributable to motor proteins. Their interaction with other cellular IFs [8,9]. This article focuses on cytoplasmic IFs, which elements via IF associated proteins (IFAPs) affects IF are now recognized as players in signaling, growth, dynamics and also results in cellular networks with prop- epithelial polarity, and apoptosis in erties that transcend those of individual components. We addition to providing the cell with resilience to environ- end by discussing how mutations leading to defects in IF mental stress [2,10]. assembly, network formation or IF–IFAP association com- These broad-ranging functions derive from the diversity promise in vivo functions of IF as protectors against of IFs coupled with their unique mechanical and bio- environmental stress. chemical properties. IFs are the most flexible of the bio- logical filaments. Furthermore, unlike MFs and MTs, a Introduction single IF can withstand stretching to more than three Intermediate filaments (IF) are flexible, rod-shaped fibers times its resting length before breaking [11]. Although averaging 10 nm in diameter, a size that is ‘intermediate’ integration with other filament systems is necessary to between microfilaments (MF; 7–8 nm) and microtubules create the final viscoelastic properties of the cytoplasm, it (MT; 25 nm) [1,2]. Of the three non-muscle cytoskeletal is thought that IFs contribute the tensile strength necess- fibers, IFs are the most diverse and are encoded by an ary for maintaining cell integrity (Box 1) [1,11,12]. IFs are estimated 70 IF in the (Human also biologically stable structures. However, a recent con- Intermediate Filament Mutation Database; http://www. vergence of in vitro and in vivo explorations has shown the interfil.org). IFs are classified into five major families IF to be a malleable and dynamic system that expressed in cell-, tissue-, differentiation- and developmen- can be structurally and functionally tailored to suit cells’ tal-specific patterns (Table 1). Families I–IV are localized changing needs. In this review we explore how recent to the cell cytoplasm whereas the type V nuclear lamins are trends have shaped our understanding of IF function, important organizers of the and karyo- organization and assembly properties in the test tube plasm. IF family members share a common blueprint built and in living cells. We have yet to fully understand how from a central a-helical coiled-coil rod flanked by flexible, these properties are translated into physiologically highly variable N- and C-termini that lead to exceptional relevant in vivo situations. However, the work discussed structural diversity among IFs [3]. This diversity presents here provides insight into how human disease phenotypes many opportunities for tailoring IF networks to cell type- might arise from fundamental defects in IF assembly and specific functions in contrast to the broadly conserved integration with IF-associated proteins (IFAPs) into func- functions of MT and MF. tionally competent networks. In most vertebrate cells cytoplasmic IFs are tethered to the nucleus and extend into the cytoplasm where they IF structure and in vitro assembly properties provide a scaffold for mitochondria, the Golgi complex, IF proteins exhibit an extended secondary structure built organizing centers (MTOCs) and other cyto- from a conserved a-helical rod domain of 310–350 amino skeletal elements (Figure 1) [1,2,4,5]. In the periphery acids flanked by divergent non-helical N- and C-termini IFs associate with plasma membrane specializations (Figure 2) [1]. The rod domain drives the formation of such as , and focal adhe- parallel a-helical coiled-coil dimers through long-range sions. The resulting network integrates and organizes the heptad repeats organized as shown in Figure 2, each with a characteristic pattern of apolar residues in the first (a) Corresponding author: Godsel, L.M. ([email protected]). and fourth (d) positions. These dimers constitute the

28 0962-8924/$ – see front matter ß 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.tcb.2007.11.004 Review TRENDS in Vol.18 No.1

Table 1. Members of the IF superfamily Type Cell types Localization Proteins Disease associations I Epithelia Cytoplasm, Acidic (pI < 5.7) K14 – epidermolysis bullosa simplex diseases 17 human epithelial keratins; K9–K28 K10, K16, K14 – disorders Hair 11 human hair keratins; K31–40 (Ha1–8) K12 – Meesmann corneal dystrophy K13 – white nevus of cannon K16 – type I K17 – pachyonychia congenita type II II Epithelia Cytoplasm, hair Basic keratins (pI  6.0) K5 – simplex diseases 20 human epithelial keratins; K1–8, K71–80 K1, K9, K2 – keratoderma disorders 6 human hair keratins: K81–86 (Hb1–6) K3 – Meesmann corneal dystrophy K4 – of cannon K6a – pachyonychia congenita type I K6b – pachyonychia congenita type II Hair K81 (Hb1), K83 (Hb3), K86 (Hb6) – K85 (Hb5) – pure hair- type ectodermal dysplasia III Muscle Cytoplasm Desmin – desmin related myopathy, 1I, familial restrictive cardiomyopathy 2 Mesenchymal Peripherin – amyotrophic lateral sclerosis and GFAP GFAP – IV Neurons Cytoplasm NF-L NF-L, M and H – amyotrophic lateral sclerosis Neurons NF-M NF-L – Charcot-Marie-Tooth diseases Neurons NF-H NF-M – Parkinson disease Neurons a- NF-H – neuronal IF disease Muscle a Muscle Synemin b (desmuslin) Muscle Neuroepithelia V Ubiquitous Lamins A/C Lamins – large number of disorders, including lipodystrophies, muscular dystrophies, neurological disorders and premature aging B1 B2 Orphan Eye lens Cytoplasm Phakinin (CP49) CP49 – autosomal dominant cataract disease Filensin (CP115) CP115 – autosomal recessive cataract disease elemental building blocks of IFs and depending on the IF predicted that the head domains of type I and type II type these can be hetero- (e.g. type I and II keratins and epidermal keratins and type III IFs exhibit a flexible type IV neurofilament chains) or homodimeric (e.g. type III structure and could interact with sites in the rod domain. vimentin and desmin). In both cases these N-termini are required for in vitro A hypothetical model based on the in vitro behavior of the filament assembly [13,15–17]. Deletion of the vimentin type III vimentin provides a useful platform for C-terminus did not block filament formation, but did result understanding how individual polypeptides might be in an increase in their mass-per-length [3]. However, assembled into an apolar filament (Figure 2). According mutations affecting the K5 tail found in patients with to this model, filament assembly comprises several major epidermolysis bullosa simplex (EBS) dramatically steps starting with the formation of parallel, in-register impaired IF assembly and network formation [18]. Impor- dimers. Dimers then associate into tetramers, thought to tantly, exposure of the N- and C-termini on the filament be organized primarily in a mode termed A11, in which the surface also leaves them free to associate with other fila- 1B subunits of the rods overlap in an anti-parallel manner ments and cellular structures. [13,14]. Tetramers aggregate into higher order oligomers to We are far from understanding the specific mechanism of form unit length filaments (ULF) 60 nm long, which assembly for all IF family members, but it is clear that undergo reorganization and elongation by longitudinal differences exist, underscored by the observed rapid kinetics annealing to form immature IF. Within the , other of polymerization and the completely distinct anti-parallel arrangements A12,A22 and ACN can also occur, assembly behavior of nuclear lamins, which form long corresponding to associations between the 1B and 2B sub- head-to-tail arrays that associate laterally [16]. However, domains, between the 2B subdomains, or between the IF family members all share certain attributes that dis- C- and N-termini. The final step is radial compaction of tinguish them from MTs and MFs, including their apolar the filament from 16 nm to a diameter of 10–12 nm. This structure, their inability to bind or hydrolyze nucleotides step has been proposed to occur via lateral rearrangements and their capacity for undergoing cycles of assembly and of protein subunits such that compaction occurs without disassembly simply by altering the ionic strength and pH of losing mass or increasing the length of the filament [1,3,13]. solution [1,16]. By contrast, MFs and MTs are polar fila- The rod domains form the IF core and the N- and ments that use cofactors to facilitate polymerization and C-termini are displayed on the filament surface. It is conformational changes, as well as nucleotide hydrolysis to

29 Review TRENDS in Cell Biology Vol.18 No.1

Figure 1. IFs are integrators of cyto- and tissue architecture. IFs emanate from the nucleus, extending throughout the cytoplasm and out to contact points at the plasma membrane. (a) Immunofluorescence microscopy reveals the keratin network in epithelial cells (red) linking to the (colocalization shown in yellow) via interactions with the desmosomal protein, (green). In this way the IF networks of adjacent cells are linked together giving the tissue mechanical

30 Review TRENDS in Cell Biology Vol.18 No.1

Figure 2. Schematic model of filament formation from dimer to mature filament, based on in vitro assembly data for the type III IF protein vimentin. Intermediate filaments are made up of an a-helical coiled-coil domain flanked by two non-helical N- and C-termini (which vary considerably among IF types). IF assembly begins with the formation of parallel a-helical coiled-coil dimers mediated by four heptad repeat-containing regions designated 1A, 1B, 2A and 2B, separated by short, flexible non-coiled linkers, L1, L12 and L2 [1,3,17]. 1B and 2B harbor motifs important for ‘triggering’ and stabilizing the coiled-coil structure and for keratin IF heterodimerization [91]. Local variations in the heptad repeat regions, such as a three residue deletion in 2B, also contribute to IF structure and assembly [17]. Dimers associate laterally into anti-parallel, half- staggered tetramers or protofilaments in which the 1B subdomains are aligned – shown as a cylinder in this Figure [13]. ULFs subsequently anneal longitudinally to form filaments that are loosely arranged and larger in caliber than mature filaments. During a subsequent compaction phase, IF are compressed radially from 16 nm to 10 nm resulting in formation of a mature filament [1,3,13]. One hypothesis based on data acquired from vimentin studies suggests that the strands form a cylinder with a visible open or low-density central core when observed in cross-section [3,13]. This figure presents a general model; however the number of tetramers varies when filaments are viewed in cross-section and the extent of variation is dependent on IF type and assembly conditions [3,16]. regulate polymer dynamics. IF have not traditionally been regulate polymerization of IF in vitro, but how these rules thought to require co-factors. However, recently it has been apply to IF assembly and remodeling in cells remains shown that a neurofilament light chain (NF-L)-binding largely unknown. protein called NUDEL (nuclear distribution element-like), One characteristic that is maintained in cells is the which regulates neuronal migration, facilitates assembly of apolar nature of IFs and it has been speculated that IFs IF. This observation raises the possibility that there might might over or remodel in part through a process of be a more important role for co-factors in IF function than intrafilament exchange. The possibility that IF turnover originally thought [19]. can occur by exchange within a network has been con- sidered for a number of years, based in part on the observed IF dynamics in cells: initiation and remodeling of fluorescence recovery after photobleaching (FRAP) along IF networks vimentin IF networks in cultured cells [29]. Further stu- Although IF polypeptides typically have long half-lives and dies will be necessary to determine whether IF subunits or are biochemically stable, IF networks routinely undergo oligomers can exchange along individual filaments in cells rearrangements involving disassembly and reassembly and tissues, or if exchange occurs at filament ends, or during processes such as cell spreading, wound healing whether both scenarios occur. Mechanisms that govern and , and in response to environmental stres- in vivo assembly and remodeling are also beginning to ses such as stretching and shear flow [20–28]. As described emerge. It has been suggested that of here, progress is being made in elucidating the rules that the vimentin end domains might regulate IF assembly strength. Scale bar, 20 mm. (b) The details of IF interactions with the nucleus, intracellular and cell-surface junctional structures are highlighted. IF link through and plakin family members to the nucleus, and extend into the cytoplasm where they associate with mitochondria, Golgi membranes [e.g. formiminotransferase cyclodeaminase (FTCD)], and the endolysosomal system [e.g. adaptor -3 (AP-3)] [1,5,89]. IF might also position microtubule organizing centers (MTOCs), perhaps through interactions with the g- ring complex (gTuRC) protein GCP6 [90], thus influencing MT organization and proper trafficking and distribution of membrane proteins in polarized epithelial cells [5].

31 Review TRENDS in Cell Biology Vol.18 No.1

pendent fashion, leaving the stationary particle behind Box 1. Intermediate filament mechanical properties [42]. Once separated from the mRNA, these particles are MTs, MFs and IFs all contribute to cell structure, however IFs exhibit also capable of rapid MT-dependent translocation. Evi- properties that make them uniquely suited for maintaining cell dence for post-translational precursor formation in fila- integrity in the face of mechanical and environmental stresses. IFs are more flexible than MTs or MFs. This flexibility might be because ment turnover also exists. One example in support of a of the short persistence length of IFs compared with or MTs; pre-existing pool of IF subunits comes from the observation that is, IFs curve along their length more than MTs and MFs [12]. that type II keratins are produced earlier in embryogenesis Compared with MTs and MFs, which rupture with small amounts of than their type I partners, and are stored in aggregates strain, , desmin and some keratins have been until heterodimerization takes place in the correct observed to extend to two to three times their original length, becoming thinner and increasing in stiffness in response to the temporal sequence [50]. Thus it is possible that both co- strain placed on the filament. This extensibility might be due in part and post-translational mechanisms for precursor assembly to intrafilament sliding, a property unique to IFs because of the are used in cells, providing multiple modes for regulating longitudinal annealing of filament subunits [1,11,12,92,93]. Remark- IF network formation. ably, unlike the other networks, IFs can recover from large amounts Although various subcellular locations throughout the of strain, thus the intrafilament sliding and resultant thinning is to some extent reversible. Interactions among the IF, MT and MF cell have been proposed as IF organizing centers, the cell can increase the resultant stiffness of the combined periphery has been identified as a ‘hot spot’ for initiation of networks. For example, the stiffness reported for vimentin and actin IF assembly. Fluorescent-labeled keratin probes have been mixtures is greater than that observed for either cytoskeleton alone shown to incorporate into small particles in the actin-rich when subjected to the same force [94]. periphery of epithelial cells and move towards the cell center, converting into rodlets and eventually integrating in vivo by altering the equilibrium constant of subunit into the existing IF network [45]. The idea that the IF exchange towards a higher off-rate [25,30]. Post-transla- network can be replenished primarily by addition of pre- tional modifications such as O-linked glycosylation and cursors at the network periphery was supported by FRAP interactions with IFAPs might also have an impact on analysis. Together, these observations suggest another the dynamics of IF remodeling [2,25,31,32]. IFs thus have way in which existing networks might be remodeled in the potential to be highly variable in structure and readily cells. Recently, both vimentin and keratin precursors have remodeled in a dynamic, spatially regulated manner in been shown to form in close proximity to focal complexes at vivo. sites of close apposition to the cell substrate [43,47,51] and The possible exchange of IF polypeptides into existing they seem to move along associated actin fibers [47]. networks could help explain how remodeling of networks Intriguingly, -deficient focal adhesions do not support might occur in vivo, but it doesn’t tell us how an IF network cortical keratin precursor formation and dynamics. forms de novo inside a cell. Recent studies using fluor- Although it seems possible that indirect effects owing to escent-tagged reporter proteins have identified non-fila- alterations in focal contact-mediated adhesion or signaling mentous ‘particles’ and small, filamentous forms of IFs might contribute to impaired IF behavior, collectively the called ‘squiggles’ in cells, which can incorporate into the data support an important role for the cell periphery in IF polymerizing network [20,22,23,33,34]. Although the dynamics [47]. The ability to initiate precursor formation possibility that these ectopically expressed probes could from multiple cellular locations might be important for influence normal IF dynamics should be kept in mind, customizing the IF network with high spatial resolution to these live-cell imaging studies have nevertheless been allow structural and signaling functions to be targeted to instructive in exposing potential mechanisms of assembly specific cell compartments. inside a cell. The formation of putative IF precursors seems to be a general property of IF because this has been IF precursors: on the move observed for neurofilament proteins (NFs) [35–40], the Unlike MTs and MFs, IFs do not seem to serve as tracks for type III proteins peripherin [41,42] and vimentin movement of membrane vesicles and other cellular traffic. [22,23,33,43] and some keratins [20,21,24,34,44–47]. IF particles are themselves cargo that can be moved Particles seem to merge to form squiggles, however the around the cell by various motor proteins. The direction, precise structure of the precursors observed in cells is rate of movement and pause frequency vary as a function of unknown [23]. It is hypothesized that the particles and IF type, cell type and size of IF precursor (Figure 3). Type squiggles contain aggregates of intermediates, such as IV NFs and type III vimentin and peripherin have several ULFs, and short filaments comprising polymerized ULFs, dynamic behaviors in common. They have each been respectively [14,44,48,49]. observed to move 60% of the time in the anterograde Several ideas have been put forward to explain how and and 40% of the time in the retrograde direction where IF precursors arise in cells. Precursors might form [35,36,40,41]. MT motors in the family have been de novo from newly synthesized protein, or they might implicated in anterograde translocation of vimentin assemble from a pre-existing soluble pool of protein, or [22,33,48,52,53] and NF [37,54–58] whereas and both. Support for a co-translational mechanism comes from are part of a retrograde complex, which in recent experiments carried out in rat PC12 cells, in which neurons might be mediated by an interaction between type III peripherin mRNA and its protein product were the neurofilament medium chain (NF-M) and the dynein imaged simultaneously and were shown to assemble into intermediate chain [40,41,57–60]. presumptive IF precursor particles. After particle for- Vimentin and peripherin exhibit particle speeds - mation, the associated mRNAs moved away in a MT-de- ging between 0.3 and 0.4 mm/s consistent with trafficking

32 Review TRENDS in Cell Biology Vol.18 No.1

Figure 3. Schematic representation of the interaction between IF precursors with microtubule- and actin-associated motor complexes. Type III and type IV IF precursors are observed in two forms, non-filamentous particles and small filament squiggles. Particles are formed cotranslationally in the cytoplasm and the mRNP moves away from the precursor, which then can associate with the network at the site of synthesis or can be moved to distant incorporation sites via a [42]. mRNPs appear to travel along MTs, but the motor interaction is not yet known. Kinesin presumably directs anterograde movements of IF particles and squiggles, whereas dynein and dynactin mediate retrograde mobility [48]. No motors have yet been implicated in keratin particle and squiggle translocation, however the precursors seem to use MT and actin. Actin also seems to have a role in NF translocation as Va binds to NF-L, and such associations with MF might regulate the frequency of NF precursor movements [68]. This schematic highlights interactions between IF and motor complexes and is not drawn to scale. by conventional MT motors, whereas squiggle movement Keratin precursors have been reported to move more can be much slower, between 0.06 and 0.3 mm/s frequently in the retrograde direction than type III and IV [22,23,33,41,61]. Particles and squiggles of all IF types IFs [20–23,33,47,61]. In addition, keratin precursor can pause during translocation, sometimes up to 80% of kinetics can be divided into two classes, one with slow the time [21,35,36,41]. For example, NF precursors kinetics (0.001–0.01 mm/s) requiring actin and one with were once believed to be translocated by a mechanism more rapid kinetics (>0.01 mm/s) requiring MTs [46].In distinct from the faster, MT-based transport observed for support of a physiological role for actin in IF dynamics, other proteins, but live cell imaging has recently shown keratins were shown to move in concert with F-actin in that they actually move in bursts of speed at >1 mm/s Xenopus extracts. Furthermore, IF polymerization in these interspersed with long, frequent pauses [35–37,58,62,63]. extracts, as well as in activated oocytes, required F-actin, The underlying basis of the observed differences in speed perhaps for proper IF precursor alignment and exchange and pause times between IF types is not well understood. into the polymerizing filament [67]. A role for a specific It has been speculated that transient associations with myosin motor has been demonstrated in mice harboring a MT- or IF-associated proteins might put a drag on the myosin Va mutation, as the animals exhibit a higher particles, or that a change in the phosphorylation state of density of NF in the cytoplasm associated with a modest IFs regulates their translocation properties [25,58] per- increase in NF protein levels [68]. Likewise, myosin Va is haps by affecting interactions with the translocation aberrantly distributed in mice deficient in NF-L, a binding machinery. In support of this hypothesis, NF C-terminal partner and potential cargo for this myosin, and it has been phosphorylation can modulate precursor dynamics proposed that an NF–myosin Va interaction regulates the [54,57,58,64–66]. frequency of NF precursor movements [68].

33 Review TRENDS in Cell Biology Vol.18 No.1

IFs and IFAPs: partners in regulation of cytoskeletal contact [74,77]. The precise molecular mechanism of networks translocation is not known, but it has been shown to be IFAPs are emerging as important regulators of IF network actin-dependent. Based on previous data raising the remodeling and function, particularly members of the possibility that DP can co-assemble with IF in vivo, plakin family. Members of this family have been described possibly by forming co-oligomers, it seems possible that as structural linkers between IF and cell–cell or cell–sub- DP could co-assemble with and piggy-back on IF particles strate adhesions and as crosslinkers with other cyto- or squiggles during the assembly process and be translo- skeletal elements [69]. The linkages are facilitated by cated by some mechanism requiring actin. Alternatively, the modular structure of the family members, each of DP particles enmeshed in the IF filament network might which contains a ‘mix-and-match’ collection of different be brought into proximity with the cell borders by the building blocks, including IF-binding, actin-binding contraction or remodeling of cortical actomyosin, which (ABD), coiled-coil rod, repeat-containing and is coupled to the assembly of a related cell–, MT-binding domains, as well as domains that target cer- the adherens junction [78,79]. tain members to junctions such as desmosomes or hemi- desmosomes. IFs and disease: relationship with dynamics and One plakin, known as , is a large family member assembly containing domains that enable it to associate with and Demonstration that IFs provide mechanical integrity to crosslink all three cytoskeletal systems. These domains tissues first came from pioneering studies showing that include an N-terminal ABD, built from two N-terminal mutations in K5 and 14 led to epidermal fragility in mice calponin homology (CH) domains and a C-terminal IF- and epidermolysis bullosa simplex (EBS) in humans [6]. binding domain. Highlighting its importance in organizing Similar mutations (although distributed differently along IF, plectin-deficient cells lose proper orthogonal IF cross- the length of the polypeptide chain) have since then been linking making them more susceptible to stress-induced found in other IF family members, leading to identification disruption of the IF cytoskeleton and accompanying of a plethora of disease phenotypes, at least some of which changes in cell signaling and cell migration [70]. are due to fragility because of loss of IF network function Plectin might also regulate IF polymerization and (Table 1) [10] (Human Intermediate Filament Mutation dynamics. In addition to its C-terminus, a second IF-bind- Database, http://www.interfil.org). ing site was recently discovered within the plectin N- Early in vitro studies supported the idea that keratin terminus in the first CH domain. However, this site cannot mutations, which were frequently seen to be clustered in bind filamentous vimentin, which prompted investigators the highly conserved rod end domains, led to defects in to suggest that it might instead regulate the dynamics of IF polymer assembly (reviewed in [6]). Supporting the idea polymerization [71]. Such a function might contribute to that mutations in type III IFs also affect polymer assembly, the observed nucleation of vimentin at matrix adhesions desmin mutations that result in severe myopathies were [43,51]. Other CH domain-containing proteins such as shown to alter filament assembly at different stages of ULF fimbrin, which colocalizes with vimentin in podosomes, formation [80], mutations in glial fibrillary acidic protein and calponin, which interacts with desmin, are thought (GFAP) impair assembly in vitro and in cells lacking other to bind tetrameric forms of IF with a higher affinity than IF [81], and NF-L mutations linked to Charcot-Marie- polymerized filaments. These observations highlight a Tooth disease result in defects in assembly and network potential conserved function for CH domain-containing formation [82]. At least some keratin mutant proteins act proteins in regulating IF assembly [4]. in a dominant negative way, increasing the soluble pool of Recent additions to the plakin family – and subunits and leading to defects in the IF network and the epiplakin – might also contribute to IF organization and formation of cytoplasmic aggregates (reviewed in [2,10]). assembly state. Epiplakin deficiency resulted in destabi- Not all IF mutations behave in this way. For example, lization of simple epithelial networks in cultured cells and GFAP mutants that cause Alexander disease, which is acceleration of keratinocyte migration in mice [72,73]. characterized by cytoplasmic inclusions in astrocytes Interestingly, siRNA-mediated knock-down of the family called Rosenthal fibers, are resistant to detergent extrac- member periplakin inhibited proper keratin organization tion [81,83]. and also inhibited wound closure in vitro [32]. Although Recently there has been discussion regarding whether further work will be needed to understand how these and impaired assembly and/or network formation is a hallmark other associated proteins control IF assembly at a molecu- of all IF mutations, particularly in the presence of a wild- lar level, these observations collectively support a general type partner. For example, the simple epithelial keratin K18 role for IFAPs in regulating IF networks. mutation R89C leads to network disruption in vivo, render- Similarly, plakin might be regulated ing hepatocytes fragile and predisposing these cells to apop- by their interactions with IFs, as is the case with desmo- tosis. By contrast, the presence of a G61C mutation in the plakin (DP) during its assembly into desmosomes [74]. This partner protein K8 does not result in detectable changes in plakin protein has been shown to be required for anchoring IF networks or hepatocyte fragility, yet still predisposes IF to the desmosomal cadherin complex at the plasma these cells to apoptosis [84]. Similarly, although some dis- membrane of epithelial cells [75,76]. However, its associ- ease-causing neurofilament mutations result in assembly ation with IF has also been shown to regulate the trajectory defects, a G336S variant in the rod domain of the human NF- and efficiency of DP precursor particle movement from the M protein present in a patient with Parkinson’s disease did cytoplasm into nascent desmosomes following cell–cell not [85]. Furthermore, desmin mutants co-polymerize with

34 Review TRENDS in Cell Biology Vol.18 No.1 wild-type protein in physiologically relevant concentrations As we continue to probe the structural basis for disease [86] and keratin mutants have also been shown to form phenotypes it is also becoming clear that the complexity of filaments in vitro without detectable defects [15].Even genotype–phenotype relationships is increasing with the though mutant IF and IF networks might be perceived as emergence of non-mechanical functions for IFs [2,25,63]. superficially ‘normal’ in vitro or in cells, it seems possible In the future it will be interesting to determine the fre- that structural defects might be unmasked by physical quency with which the various functions of IFs are corre- stress applied in vitro or experienced in vivo. lated with their assembly state, and whether mutations These observations underscore the emerging realization that shift the ratio of different oligomeric forms of IF in a that mutations can contribute to disease in ways that go cell preferentially affect mechanical versus non-mechan- beyond formation of the polymer. For instance, it was ical functions. shown that a K14 ‘hotspot’ mutation impairs the ability of mutant filaments to bundle and reduces resilience of the Conclusion and future challenges network [87]. This defect in bundling could contribute to The studies reviewed here provide some hints as to the the observed collapse of the filament system seen in vivo. richness and diversity of IF dynamics but we have far to go Mutation of binding sites for other network constituents before these mechanisms are completely unraveled, even could also affect IF network formation. For example, an for a single type of IF. Advances in fluorescence microscopy I451M mutation within the type III desmin tail (associated have enhanced the resolution of visualization such that with cardiomyopathy) that exhibited only partial impair- single molecules can now be tracked in vivo, enabling ment of IF network formation was recently shown to lead to researchers to put the current IF assembly model to the a lack of binding to the junctional plaque protein desmo- test in physiological situations. How IFAPs contribute to plakin in vitro [88]. Mutations in the KI tail lead to severe IF initiation and polymerization, as well as the formation palmoplantar resulting from uncoupling and mechanical properties of integrated cytoskeletal net- with the cornified envelope protein loricrin (reviewed in works involving actin and MTs, are questions that will [6]). Interestingly, a similar ichthyosis results from loricrin increasingly guide experiments in the coming years. mutations and cardiomyopathies can arise as a result of Although our understanding of MT and MF dynamics mutations in the desmoplakin-binding domain of desmin. has been facilitated by high-resolution electron microscopy These observations have led to the concept of ‘pheno- and X-ray crystallographic analysis, the 3D structure of typic convergence’ in which similar phenotypes arise from IFs remains elusive [1]. However, by applying emerging mutations in an IF family member or its associated techniques such as cryo-electron tomography and atomic proteins [2]. Other examples of this sort of convergence force microscopy (AFM), along with the continued use of IF are: (i) mutations in plectin lead to a type of EBS with fragments for X-ray crystallographic analysis, it is muscular dystrophy; (ii) mutations in desmin-associated expected that the structure and nanomechanics of IF aB-crystallin results in cardiomyopathy; and (iii) type 2 polypeptides will continue to be unraveled. In the future, Charcot-Marie-Tooth can be caused by mutations in NF-L, production of co-crystals of IF fragments and their associ- A/C, or the IF-associated heat shock protein hsp27 ated proteins, coupled with analytical biochemistry, will [2] (Human Intermediate Filament Mutation Database; enable us to determine how these interactions affect the http://www.interfil.org). Finally, the NF-L-binding protein polymerization dynamics of IF and lead to the assembly of NUDEL might emerge as another example of this concept. novel fibrous structures such as junctional plaques. This protein destabilizes NF-L in neuronal cells resulting in NF defects and morphological changes similar to those Acknowledgements observed in [19]. The authors are grateful to colleagues in the field for their contributions Just as live-cell imaging has revealed dynamic beha- to the work discussed here and apologize to those whose work we were unable to cite owing to lack of space. The authors also thank P. Coulombe viors of wild-type IFs in cells, it has also suggested and B. Omary for helpful discussions. K.J.G. is supported by grants additional ways in which disease mutations might impair R01AR43380, R01AR41836, R01CA122151, project #4 of P01DE12328 IF function. Imaging of a yellow fluorescent protein-tagged and the J.L. Mayberry Endowment. K14 R125C mutant found in patients with EBS showed that this mutant participates in the formation of highly References dynamic, short-lived aggregates containing both mutant 1 Herrmann, H. et al. (2007) Intermediate filaments: from cell and wild-type protein. These seem morphologically similar architecture to nanomechanics. Nat. Rev. Mol. Cell Biol. 8, 562–573 2 Kim, S. and Coulombe, P.A. (2007) Intermediate filament scaffolds to aggregates observed in vivo. Aggregates in the periphery fulfill mechanical, organizational, and signaling functions in the of the cell cytoplasm flow inward via an actin-dependent cytoplasm. Genes Dev. 21, 1581–1597 mechanism, are transient in nature, and are in equilibrium 3 Parry, D.A. et al. (2007) Towards a molecular description of with a soluble pool of increased magnitude [34,45]. The intermediate filament structure and assembly. Exp. Cell Res. 313, 2204–2216 aggregates are mostly segregated from the centrally 4 Green, K.J. et al. (2005) Intermediate filament associated proteins. located filamentous network (thought to contain a higher Adv. Protein Chem. 70, 143–202 proportion of wild-type protein that the aggregates) that 5 Toivola, D.M. et al. (2005) Cellular integrity plus: -related and co-exists in these cells and that seems to be less dynamic protein-targeting functions of intermediate filaments. Trends Cell Biol. and more susceptible to environmental stress than control 15, 608–617 6 Magin, T.M. et al. (2004) Emerging functions: diseases and animal networks. It was suggested that the zone in which aggre- models reshape our view of the cytoskeleton. Exp. Cell Res. 301, 91–102 gates form might be equivalent to the zone of lysis in the 7 Omary, M.B. et al. (2004) Intermediate filament proteins and their basal layer of epidermis. associated diseases. N. Engl. J. Med. 351, 2087–2100

35 Review TRENDS in Cell Biology Vol.18 No.1

8 Mattout, A. et al. (2006) Nuclear lamins, diseases and aging. Curr. a contribution of actin to the transport of keratin subunits. Mol. Biol. Opin. Cell Biol. 18, 335–341 Cell 15, 990–1002 9 Worman, H.J. and Bonne, G. (2007) Laminopathies: a wide spectrum of 35 Roy, S. et al. (2000) Neurofilaments are transported rapidly human diseases. Exp. Cell Res. 313, 2121–2133 but intermittently in : implications for slow . 10 Magin, T.M. et al. (2007) Structural and regulatory functions of J. Neurosci. 20, 6849–6861 keratins. Exp. Cell Res. 313, 2021–2032 36 Wang, L. et al. (2000) Rapid movement of axonal neurofilaments 11 Kreplak, L. and Fudge, D. (2007) Biomechanical properties of interrupted by prolonged pauses. Nat. Cell Biol. 2, 137–141 intermediate filaments: from tissues to single filaments and back. 37 Yabe, J.T. et al. (1999) Kinesin-mediated transport of neurofilament Bioessays 29, 26–35 protein oligomers in growing axons. J. Cell Sci. 112, 3799–3814 12 Wagner, O.I. et al. (2007) Softness, strength and self-repair in 38 Yan, Y. and Brown, A. (2005) Neurofilament polymer transport in intermediate filament networks. Exp. Cell Res. 313, 2228–2235 axons. J. Neurosci. 25, 7014–7021 13 Sokolova, A.V. et al. (2006) Monitoring intermediate filament assembly 39 Wang, L. and Brown, A. (2001) Rapid intermittent movement of axonal by small-angle x-ray scattering reveals the molecular architecture of neurofilaments observed by fluorescence photobleaching. Mol. Biol. assembly intermediates. Proc. Natl. Acad. Sci. U. S. A. 103, 16206–16211 Cell 12, 3257–3267 14 Mucke, N. et al. (2004) Molecular and biophysical characterization of 40 Shah, J.V. et al. (2000) Bidirectional translocation of neurofilaments assembly-starter units of human vimentin. J. Mol. Biol. 340, 97–114 along microtubules mediated in part by dynein/dynactin. Mol. Biol. 15 Herrmann, H. et al. (2002) Characterization of early assembly Cell 11, 3495–3508 intermediates of recombinant human keratins. J. Struct. Biol. 137, 41 Helfand, B.T. et al. (2003) Rapid transport of neural intermediate 82–96 filament protein. J. Cell Sci. 116, 2345–2359 16 Herrmann, H. and Aebi, U. (2004) Intermediate filaments: molecular 42 Chang, L. et al. (2006) Assembling an intermediate filament network structure, assembly mechanism, and integration into functionally by dynamic cotranslation. J. Cell Biol. 172, 747–758 distinct intracellular scaffolds. Annu. Rev. Biochem. 73, 749–789 43 Tsuruta, D. and Jones, J.C. (2003) The vimentin cytoskeleton regulates 17 Parry, D.A. (2005) Microdissection of the sequence and structure of focal contact size and adhesion of endothelial cells subjected to shear intermediate filament chains. Adv. Protein Chem. 70, 113–142 stress. J. Cell Sci. 116, 4977–4984 18 Gu, L.H. and Coulombe, P.A. (2005) Defining the properties of the 44 Liovic, M. et al. (2003) Observation of keratin particles showing fast nonhelical tail domain in type II : insight from a bullous bidirectional movement colocalized with microtubules. J. Cell Sci. 116, disease-causing mutation. Mol. Biol. Cell 16, 1427–143818 1417–1427 19 Nguyen, M.D. et al. (2004) A NUDEL-dependent mechanism of 45 Windoffer, R. et al. (2004) Identification of novel principles of keratin neurofilament assembly regulates the integrity of CNS neurons. filament network turnover in living cells. Mol. Biol. Cell 15, 2436–2448 Nat. Cell Biol. 6, 595–608 46 Woll, S. et al. (2005) Dissection of keratin dynamics: different 20 Windoffer, R. and Leube, R.E. (1999) Detection of dynamics contributions of the actin and microtubule systems. Eur. J. Cell by time-lapse fluorescence microscopy in living cells. J. Cell Sci. 112, Biol. 84, 311–328 4521–4534 47 Windoffer, R. et al. (2006) Focal adhesions are hotspots for Yoon, K.H. et al. (2001) Insights into the dynamic properties of keratin filament precursor formation. J. Cell Biol. 173, 341–348 intermediate filaments in living epithelial cells. J. Cell Biol. 153, 503– 48 Chang, L. and Goldman, R.D. (2004) Intermediate filaments mediate 516 cytoskeletal crosstalk. Nat. Rev. Mol. Cell Biol. 5, 601–613 22 Prahlad, V. et al. (1998) Rapid movements of vimentin on microtubule 49 Kirmse, R. et al. (2007) A quantitative kinetic model for the in vitro tracks: kinesin-dependent assembly of intermediate filament assembly of intermediate filaments from tetrameric vimentin. J. Biol. networks. J. Cell Biol. 143, 159–170 Chem. 282, 18563–18572 23 Yoon, M. et al. (1998) Motile properties of vimentin intermediate 50 Lu, H. et al. (2005) Type II keratins precede type I keratins during early filament networks in living cells. J. Cell Biol. 143, 147–157 embryonic development. Eur. J. Cell Biol. 84, 709–718 24 Windoffer, R. and Leube, R.E. (2001) De novo formation of cytokeratin 51 Gonzales, M. et al. (2001) Structure and function of a vimentin- filament networks originates from the cell cortex in A-431 cells. Cell associated matrix adhesion in endothelial cells. Mol. Biol. Cell 12, Motil. Cytoskeleton 50, 33–44 85–100 25 Omary, M.B. et al. (2006) ‘Heads and tails’ of intermediate filament 52 Kreitzer, G. et al. (1999) Detyrosination of tubulin regulates the phosphorylation: multiple sites and functional insights. Trends interaction of intermediate filaments with microtubules in vivo via a Biochem. Sci. 31, 383–394 kinesin-dependent mechanism. Mol. Biol. Cell 10, 1105–1118 26 Sahlgren, C.M. et al. (2001) Mitotic reorganization of the intermediate 53 Liao, G. and Gundersen, G.G. (1998) Kinesin is a candidate for cross- filament protein nestin involves phosphorylation by cdc2 kinase. bridging microtubules and intermediate filaments. Selective binding of J. Biol. Chem. 276, 16456–16463 kinesin to detyrosinated tubulin and vimentin. J. Biol. Chem. 273, 27 Chou, Y.H. et al. (2007) The motility and dynamic properties of 9797–9803 intermediate filaments and their constituent proteins. Exp. Cell Res. 54 Yabe, J.T. et al. (2000) Phospho-dependent association of 313, 2236–2243 neurofilament proteins with kinesin in situ. Cell Motil. Cytoskeleton 28 Ridge, K.M. et al. (2005) phosphorylation by protein kinase C 45, 249–262 delta regulates shear stress-mediated disassembly of keratin 55 Theiss, C. et al. (2005) Impairment of anterograde and retrograde intermediate filaments in alveolar epithelial cells. J. Biol. Chem. neurofilament transport after anti-kinesin and anti-dynein 280, 30400–30405 microinjection in chicken dorsal root ganglia. Eur. J. Cell Biol. 84, 29 Vikstrom, K.L. et al. (1992) Steady state dynamics of intermediate 29–43 filament networks. J. Cell Biol. 118, 121–129 56 Xia, C.H. et al. (2003) Abnormal neurofilament transport caused by 30 Eriksson, J.E. et al. (2004) Specific in vivo phosphorylation sites targeted disruption of neuronal kinesin heavy chain KIF5A. J. Cell determine the assembly dynamics of vimentin intermediate Biol. 161, 55–66 filaments. J. Cell Sci. 117, 919–932 57 Motil, J. et al. (2006) Dynein mediates retrograde neurofilament 31 Izawa, I. and Inagaki, M. (2006) Regulatory mechanisms and functions transport within axons and anterograde delivery of NFs from of intermediate filaments: A study using site- and phosphorylation perikarya into axons: regulation by multiple phosphorylation events. state-specific . Cancer Sci. 97, 167–174 Cell Motil. Cytoskeleton 63, 266–286 32 Long, H.A. et al. (2006) Periplakin-dependent re-organisation of 58 Sihag, R.K. et al. (2007) Role of phosphorylation on the structural keratin cytoskeleton and loss of collective migration in keratin-8- dynamics and function of types III and IV intermediate filaments. Exp. downregulated epithelial sheets. J. Cell Sci. 119, 5147–5159 Cell Res. 313, 2098–2109 33 Martys, J.L. et al. (1999) Intermediate filaments in motion: 59 Helfand, B.T. et al. (2002) A requirement for cytoplasmic dynein and observations of intermediate filaments in cells using green dynactin in intermediate filament network assembly and organization. fluorescent protein-vimentin. Mol. Biol. Cell 10, 1289–1295 J. Cell Biol. 157, 795–806 34 Werner, N.S. et al. (2004) Epidermolysis bullosa simplex-type 60 Wagner, O.I. et al. (2004) The interaction of neurofilaments with the mutations alter the dynamics of the keratin cytoskeleton and reveal microtubule motor cytoplasmic dynein. Mol. Biol. Cell 15, 5092–5100

36 Review TRENDS in Cell Biology Vol.18 No.1

61 Helfand, B.T. et al. (2004) Intermediate filaments are dynamic and 78 Zhang, J. et al. (2005) Actin at cell-cell junctions is composed of two motile elements of cellular architecture. J. Cell Sci. 117, 133–141 dynamic and functional populations. J. Cell Sci. 118, 5549–5562 62 Shea, T.B. (2000) Microtubule motors, phosphorylation and axonal 79 Yamada, S. and Nelson, W.J. (2007) Localized zones of Rho and Rac transport of neurofilaments. J. Neurocytol. 29, 873–887 activities drive initiation and expansion of epithelial cell . 63 Brown, A. (2000) Slow axonal transport: stop and go traffic in the . J. Cell Biol. 178, 517–527 Nat. Rev. Mol. Cell Biol. 1, 153–156 80 Bar, H. et al. (2005) Severe muscle disease-causing desmin mutations 64 Jung, C. et al. (2000) Hypophosphorylated neurofilament subunits interfere with in vitro filament assembly at distinct stages. Proc. Natl. undergo axonal transport more rapidly than more extensively Acad. Sci. U. S. A. 102, 15099–15104 phosphorylated subunits in situ. Cell Motil. Cytoskeleton 47, 120–129 81 Hsiao, V.C. et al. (2005) Alexander-disease mutation of GFAP causes 65 Jung, C. et al. (2000) C-terminal phosphorylation of the high molecular filament disorganization and decreased solubility of GFAP. J. Cell Sci. weight neurofilament subunit correlates with decreased neurofilament 118, 2057–2065 axonal transport velocity. Brain Res. 856, 12–19 82 Perez-Olle, R. et al. (2004) Phenotypic analysis of neurofilament light 66 Ackerley, S. et al. (2003) Neurofilament heavy chain side arm gene mutations linked to Charcot-Marie-Tooth disease in cell culture phosphorylation regulates axonal transport of neurofilaments. models. Hum. Mol. Genet. 13, 2207–2220 J. Cell Biol. 161, 489–495 83 Quinlan, R.A. et al. (2007) GFAP and its role in Alexander disease. Exp. 67 Weber, K.L. and Bement, W.M. (2002) F-actin serves as a template for Cell Res. 313, 2077–2087 cytokeratin organization in cell free extracts. J. Cell Sci. 115, 1373– 84 Ku, N.O. and Omary, M.B. (2006) A disease- and phosphorylation- 1382 related nonmechanical function for keratin 8. J. Cell Biol. 174, 115–125 68 Rao, M.V. et al. (2002) Myosin Va binding to neurofilaments is essential 85 Perez-Olle, R. et al. (2004) The G336S variant in the human for correct myosin Va distribution and transport and neurofilament neurofilament-M gene does not affect its assembly or distribution: density. J. Cell Biol 159, 279–289 importance of the functional analysis of neurofilament variants. 69 Sonnenberg, A. and Liem, R.K. (2007) in development and J. Neuropathol. Exp. Neurol. 63, 759–774 disease. Exp. Cell Res. 313, 2189–2203 86 Bar, H. et al. (2007) Assembly defects of desmin disease mutants 70 Osmanagic-Myers, S. et al. (2006) Plectin-controlled keratin carrying deletions in the alpha-helical rod domain are rescued by cytoarchitecture affects MAP kinases involved in cellular stress wild type protein. J. Struct. Biol. 158, 107–115 response and migration. J. Cell Biol. 174, 557–568 87 Ma, L. et al. (2001) A ‘hot-spot’ mutation alters the mechanical 71 Sevcik, J. et al. (2004) Actin-binding domain of mouse plectin. Crystal properties of keratin filament networks. Nat. Cell Biol. 3, 503–506 structure and binding to vimentin. Eur. J. Biochem. 271, 1873–1884 88 Lapouge, K. et al. (2006) New insights into the molecular basis of 72 Jang, S.I. et al. (2005) Characterization of human epiplakin: RNAi- desmoplakin- and desmin-related cardiomyopathies. J. Cell Sci. 119, mediated epiplakin depletion leads to the disruption of keratin and 4974–4985 vimentin IF networks. J. Cell Sci. 118, 781–793 89 Styers, M.L. et al. (2005) Intermediate filaments and vesicular 73 Goto, M. et al. (2006) Elimination of epiplakin by gene targeting results membrane traffic: the odd couple’s first dance? Traffic 6, 359–365 in acceleration of keratinocyte migration in mice. Mol. Cell. Biol. 26, 90 Oriolo, A.S. et al. (2007) GCP6 binds to intermediate filaments: a novel 548–558 function of keratins in the organization of microtubules in epithelial 74 Godsel, L.M. et al. (2005) Desmoplakin assembly dynamics in four cells. Mol. Biol. Cell 18, 781–794 dimensions: multiple phases differentially regulated by intermediate 91 Wu, K.C. et al. (2000) Coiled-coil trigger motifs in the 1B and 2B rod filaments and actin. J. Cell Biol. 171, 1045–1059 domain segments are required for the stability of keratin intermediate 75 Vasioukhin, V. et al. (2001) Desmoplakin is essential in epidermal filaments. Mol. Biol. Cell 11, 3539–3558 sheet formation. Nat. Cell Biol. 3, 1076–1085 92 Kreplak, L. et al. (2005) Exploring the mechanical behavior of single 76 Huen, A.C. et al. (2002) Intermediate filament-membrane attachments intermediate filaments. J. Mol. Biol. 354, 569–577 function synergistically with actin-dependent contacts to regulate 93 Storm, C. et al. (2005) Nonlinear elasticity in biological gels. Nature intercellular adhesive strength. J. Cell Biol. 159, 1005–1017 435, 191–194 77 Loranger, A. et al. (2006) Keratin 8 modulation of desmoplakin 94 Esue, O. et al. (2006) A direct interaction between actin and vimentin deposition at desmosomes in hepatocytes. Exp. Cell Res. 312, 4108– filaments mediated by the tail domain of vimentin. J. Biol. Chem. 281, 4119 30393–30399

37