Search

MACF1 (microtubule actin crosslinking factor), also called ACF7 (actin crosslinking family 7) is a cytoskeletal linker protein that can associate with both actin filaments and microtubules. We have identified a novel alternatively spliced isoform of MACF1. We named this isoform MACF1b and renamed the original isoform MACF1a. MACF1b is identical to MACF1a, except that it has a region containing plakin (or plectin) repeats in the middle of the molecule. MACF1b is ubiquitously expressed in adult tissues with especially high levels in the lung. We studied the subcellular localization of MACF1b proteins in mammalian cell lines. In two lung cell lines, MACF1b was chiefly localized to the Golgi complex. Upon treatments that disrupt the Golgi complex, MACF1b redistributed into the cytosol, but remained co-localized with the dispersed Golgi ministacks. MACF1b proteins can be detected in the enriched Golgi fraction by western blotting. The domain of MACF1b that targets it to the Golgi was found at the N-terminal part of the region that contains the plakin repeats. Reducing the level of MACF1 proteins by small-interfering RNA resulted in the dispersal of the Golgi complex.

The human Gas2-related gene on chromosome 22 (hGAR22) encodes two alternatively spliced mRNA species. The longer mRNA encodes a protein with a deduced molecular mass of 36.3 kDa (GAR22alpha), whereas the shorter mRNA encodes a larger protein with a deduced molecular mass of 72.6 kDa (GAR22beta). We show that both hGAR22 proteins contain a calponin homology actin-binding domain and a Gas2-related microtubule-binding domain. Using rapid amplification of cDNA ends, we have cloned the mouse orthologue of hGAR22, mGAR22, and found its protein products to be extremely well conserved. We also report the cDNA cloning of a human Gas2-related gene on chromosome 17 (hGAR17). hGAR17 also encodes two protein isoforms. The overall cytoskeletal binding properties of the hGAR17 and hGAR22 proteins are remarkably similar. hGAR17 mRNA expression is limited to skeletal muscle. Although hGAR22 and mGAR22 mRNAs are expressed nearly ubiquitously, mGAR22 protein can only be detected in testis and brain. Furthermore, only the beta isoform is present in these tissues. GAR22beta expression is induced in a variety of cultured cells by growth arrest. The absolute amounts of GAR22beta protein expressed are low. The beta isoforms of hGAR17 and hGAR22 appear to be able to crosslink microtubules and microfilaments in transfected cells. This finding suggests that the physiological functions of these proteins may involve integration of these two components of the cytoskeleton.

Neurofilaments (NFs) are the major intermediate filaments (IFs) of mature neurons. They play important roles in the structure and function of axons. Recently, two mutations in the neurofilament light (NFL) subunit have been identified in families affected by Charcot-Marie-Tooth (CMT) neuropathy type 2. We have characterized the effects of these NFL mutations on the formation of IF networks using a transient transfection system. Both mutations disrupted the self-assembly of human NFL. The Q333P mutant in the rod domain of NFL also disrupted the formation of rat and human NFL/NFM heteropolymers. The phenotypes produced by the P8R mutation in the head domain of NFL were less severe. The P8R mutant NFL co-polymerized with NFM to form bundled filaments and, less often, aggregates. Our results suggest that alterations in the formation of a normal IF network in neurons elicited by these NFL mutations may contribute to the development of Charcot-Marie-Tooth neuropathy.

MACF (microtubule actin cross-linking factor) is a large, 608-kDa protein that can associate with both actin microfilaments and microtubules (MTs). Structurally, MACF can be divided into 3 domains: an N-terminal domain that contains both a calponin type actin-binding domain and a plakin domain; a rod domain that is composed of 23 dystrophin-like spectrin repeats; and a C-terminal domain that includes two EF-hand calcium-binding motifs, as well as a region that is homologous to two related proteins, GAR22 and Gas2. We have previously demonstrated that the C-terminal domain of MACF binds to MTs, although no homology was observed between this domain and other known microtubule-binding proteins. In this report, we describe the characterization of this microtubule-binding domain of MACF by transient transfection studies and in vitro binding assays. We found that the C-terminus of MACF contains at least two microtubule-binding regions, a GAR domain and a domain containing glycine-serine-arginine (GSR) repeats. In transfected cells, the GAR domain bound to and partially stabilized MTs to depolymerization by nocodazole. The GSR-containing domain caused MTs to form bundles that are still sensitive to nocodazole-induced depolymerization. When present together, these two domains acted in concert to bundle MTs and render them stable to nocodazole treatment. Recently, a study has shown that the N-terminal half of the plakin domain (called the M1 domain) of MACF also binds MTs. We therefore examined the microtubule binding ability of the M1 domain in the context of the entire plakin domain with and without the remaining N-terminal regions of two different MACF isoforms. Interestingly, in the presence of the surrounding sequences, the M1 domain did not bind MTs. In addition to MACF, cDNA sequences encoding the GAR and GSR-containing domains are also found in the partial human EST clone KIAA0728, which has high sequence homology to the 3' end of the MACF cDNA; hence, we refer to it as MACF2. The C-terminal domain of mouse MACF2 was cloned and characterized. The microtubule-binding properties of MACF2 C-terminal domain are similar to that of MACF. The GAR domain was originally found in Gas 2 protein and here we show that it can associate with MTs in transfected cells. Plectin and desmoplakin have GSR-containing domains at their C-termini and we further demonstrate that the GSR-containing domain of plectin, but not desmoplakin, can bind to MTs in vivo.

Type IV neuronal intermediate filament proteins consist of alpha-internexin, which can self-assemble into filaments and the neurofilament triplet proteins, which are obligate heteropolymers, at least in rodents. These IF proteins therefore provide good systems for elucidating the mechanism of intermediate filament assembly. To analyze the roles of the head domains of these proteins in contributing to their differential assembly properties, we generated chimeric proteins by swapping the head domains between rat alpha-internexin and either rat NF-L or NF-M and examined their assembly properties in transfected cells that lack their own cytoplasmic intermediate filament network. Lalphaalpha and Malphaalpha, the chimeric proteins generated by replacing the head domain of alpha-internexin with those of NF-L and NF-M, respectively, were unable to self-assemble into filaments. In contrast, alphaLL, a chimeric NF-L protein generated by replacing the head domain of NF-L with that of alpha-internexin, was able to self-assemble into filaments, whereas MLL, a chimeric NF-L protein containing the NF-M head domain, was unable to do so. These results demonstrate that the alpha-internexin head domain is essential for alpha-internexin's ability to self-assemble. While coassembly of Lalphaalpha with NF-M and coassembly of Malphaalpha with NF-L resulted in formation of filaments, coassembly of Lalphaalpha with NF-L and coassembly of Malphaalpha with NF-M yielded punctate patterns. These coassembly results show that heteropolymeric filament formation requires that one partner has the NF-L head domain and the other partner has the NF-M head domain. Thus, the head domains of rat NF-L and NF-M play important roles in determining the obligate heteropolymeric nature of filament formation. The data obtained from these self-assembly and coassembly studies provide some new insights into the mechanism of intermediate filament assembly.

In order to study the dynamic behavior of intermediate filament networks in living cells, we have prepared constructs fusing green fluorescent protein to intermediate filament proteins. Vimentin fused to green fluorescent protein labeled the endogenous intermediate filament network. We generated stable SW13 and NIH3T3 cell lines that express an enhanced green fluorescent protein fused to the N-terminus of full-length vimentin. We were able to observe the dynamic behavior of the intermediate filament network in these cells for periods as long as 4 hours (images acquired every 2 minutes). In both cell lines, the vimentin network constantly moves in a wavy manner. In the NIH3T3 cells, we observed extension of individual vimentin filaments at the edge of the cell. This movement is dependent on microtubules, since the addition of nocodazole stopped the extension of the intermediate filaments. Injection of anti-IFA causes the redistribution or 'collapse' of intermediate filaments. We injected anti-IFA antibodies into NIH3T3 cells stably expressing green fluorescent protein fused to vimentin and found that individual intermediate filaments move slowly towards the perinuclear area without obvious disassembly. These results demonstrate that individual intermediate filaments are translocated during the collapse, rather than undergoing disassembly-induced redistribution. Injections of tubulin antibodies disrupt the interactions between intermediate filaments and stable microtubules and cause the collapse of the vimentin network showing that these interactions play an important role in keeping the intermediate filament network extended. The nocodazole inhibition of intermediate filament extension and the anti-IFA microinjection experiments are consistent with a model in which intermediate filaments exhibit an extended distribution when tethered to microtubules, but are translocated to the perinuclear area when these connections are severed.

The roles of the head and tail domains of alpha-internexin, a type IV neuronal intermediate filament protein, in its self-assembly and coassemblies with neurofilament triplet proteins, were examined by transient transfections with deletion mutants in a non-neuronal cell line lacking an endogenous cytoplasmic intermediate filament network. The results from the self-assembly studies showed that the head domain was essential for alpha-internexin's ability to self-assemble into a filament network and the tail domain was important for establishing a proper filament network. The data from the coassembly studies demonstrated that alpha-internexin interacted differentially with the neurofilament triplet protein subunits. Wild-type NF-L or NF-M, but not NF-H, was able to complement and form a normal filament network with the tailless alpha-internexin mutant, the alpha-internexin head-deletion mutant, or the alpha-internexin mutant missing the entire tail and some amino-terminal portion of the head domain. In contrast, neither the tailless NF-L mutant nor the NF-L head-deletion mutant was able to form a normal filament network with any of these alpha-internexin deletion mutants. However, coassembly of the tailless NF-M mutant with the alpha-internexin head-deletion mutant and coassembly of the NF-M head-deletion mutant with the tailless alpha-internexin mutant resulted in the formation of a normal filament network. Thus, the coassembly between alpha-internexin and NF-M exhibits some unique characteristics previously not observed with other intermediate filament proteins: only one intact tail and one intact head are required for the formation of a normal filament network, and they can be present within the same partner or separately in two partners.

All intermediate filament proteins consist of an alpha-helical rod domain flanked by non-helical N-terminal head and C-terminal tail domains. The roles of the non-helical domains of various intermediate filament proteins in the assembly and co-assembly of higher-order filamentous structures have been studied by many groups but with quite contradictory results. Type III intermediate filaments are unique in that they can form homopolymers both in vitro and in vivo. The expression and assembly characteristics of carboxy- and amino-terminal deletion mutants of glial fibrillary acidic protein (GFAP), an astrocyte-specific type III intermediate filament protein, were examined by transient transfections of either vimentin-positive or vimentin-negative variants of human adrenocarcinoma-derived SW13 cell lines. Whereas complete deletion of the C-terminal tail domain of GFAP results in the formation of polymorphic aggregates, both intranuclear and cytoplasmic in self-assembly experiments, efficient co-assembly of these tail-less GFAP mutants with vimentin can be achieved as long as the KLLEGEE sequence at the C-terminal end of the rod domain is preserved. Up to one-fifth of the C-terminal end of the tail domain can be deleted without affecting the capability of GFAP to self-assemble. The highly conserved RDG-containing motif in the tail domain may be important for self-assembly but is not sufficient. The entire head domain seems to be required for self-assembly. All N-terminal deletion mutants of GFAP share the same phenotype of diffuse cytoplasmic staining when expressed in vimentin-negative SW13 cells. Although co-assembly with vimentin can still be achieved with completely head-less GFAP, preservation of some of the head domain greatly enhanced the efficiency. Our results form the basis for further, more detailed mapping of the essential regions in filament assembly of GFAP and other type III IFs.

Using a novel PCR approach, we have cloned a cDNA encoding the entire high molecular weight tau molecule from rat dorsal root ganglia. The resulting 2080 bp cDNA differs from low molecular weight rat brain tau by the insertion of a novel 762 bp region (exon 4a) between exons 4 and 5. This cDNA clone is identical in sequence with a high molecular weight tau (HMW) cDNA from rat PC12 tumor cells and is closely related to a HMW tau cDNA from mouse N115 tumor cells. In vitro transcription/translation produces a protein that migrates on SDS-PAGE with the same apparent molecular weight as HMW tau purified from rat sciatic nerve. The HMW tau protein is generated from an 8 kb mRNA, which can be detected by northern blots in peripheral ganglia, but not in brain. A more sensitive assay using PCR and Southern blot analysis demonstrates the presence of exon 4a in spinal cord and in retina. In combination with immunohistochemical studies of spinal cord, these data suggest that HMW tau, though primarily in the peripheral nervous system, is also expressed in limited areas of the central nervous system, although its presence cannot be detected in the cerebral cortices.

Using epitope mapping we have demonstrated that a high molecular weight protein (Mr approximately 115 x 10(3)) present in brain and spinal cord is a member of the tau family of microtubule-associated proteins. Antibodies directed against the amino-terminal, middle and carboxyl-terminal portions of tau recognize this protein. A limited survey of neuronal tissues has shown that this high molecular weight tau protein is present in brain, spinal cord, dorsal root ganglia, dorsal and ventral roots and peripheral nerves. High molecular weight tau protein is expressed at higher levels in spinal cord than in brain and is the only form of tau detected in the adult peripheral nervous system.

The expression and assembly characteristics of carboxyl- and amino-terminal deletion mutants of rat neurofilament low Mr (NF-L) and neurofilament middle Mr (NF-M) proteins were examined by transient transfection of cultured fibroblasts. Deletion of the carboxyl-terminal tail domain of either protein indicated that this region was not absolutely essential for co-assembly into the endogenous vimentin cytoskeleton. However, deletion into the alpha-helical rod domain resulted in an inability of the mutant proteins to co-assemble with vimentin into filamentous structures. Instead, the mutant proteins appeared to be assembled into unusual tubular-vesicular structures. Additionally, these latter deletions appeared to act as dominant negative mutants which induced the collapse of the endogenous vimentin cytoskeleton as well as the constitutively expressed NF-H and NF-M cytoskeletons in stably transfected cell lines. Thus, an intact alpha-helical rod domain was essential for normal IF co-assembly whereas carboxyl-terminal deletions into this region resulted in dramatic alterations of the existing type III and IV intermediate filament cytoskeletons in vivo. Deletions from the amino-terminal end into the alpha-helical rod region gave different results. With these deletions, the transfected protein was not co-assembled into filaments and the endogenous vimentin IF network was not disrupted, indicating that these deletion mutants are recessive. The dominant negative mutants may provide a novel approach to studying intermediate filament function within living cells.