All known intermediate filament (IF) proteins display -8 -4 -1 a consensus sequence TYRKLLEGE at the carboxyl end of the rod domain. To analyse the contribution of this sequence to the formation of IF we have changed two of the invariant positions of this motif by site-directed mutagenesis. We produced three mutant keratins, each containing a single point mutation. Tyrosine at position -8 was changed to alanine in keratin K8 (K8Y----A-8) and keratin K18 (K18Y----A-8) and leucine at position -4 was changed to glycine in keratin K18 (K18L----G-4). Mutant keratins were expressed in Escherichia coli, purified and analysed for their filament-forming capacity in vitro using either the complementary wild-type keratin or the corresponding mixture of mutant keratins. In standard filament buffer (50 mM Tris-HCl, pH7.5), assembly involving any of the mutants leads to large electron-dense aggregates instead of normal IF. In order to explain this effect, we studied the process of filament formation in more detail. Whereas the formation of tetramers in buffers containing 4M urea is unaffected, the elongation process seems slowed down. In buffer of lower ionic strength (10 mM Tris-HCl, pH7.5) mutant keratins K8Y----A-8 plus K18Y----A-8 become able to form long filaments, although short filaments and protofilamentous material are still detected. The filaments formed differ from normal keratin IF by their remarkable tendency to aggregate into thick cables. Assemblies involving K18L----G-4 can only form short IF lengths. The dense aggregates formed in standard filament buffer are able to dissociate into IF and their fragments upon dialysis into 10 mM Tris-HCl, pH7.5. The results show that the consensus sequence is needed for IF formation under normal conditions and that already one mutation per heterodimer affects the assembly.

Nearly all intermediate filament (IF) proteins share two sequence motifs located at the N- and the C-terminal ends of their helical rod domain (\'coil 1a\' and \'coil 2b\', respectively). To examine the structural role of the coil 2b motif, we have performed in vitro assembly studies and in vivo microinjection experiments employing two site-specific reagents: (a) a 20-residue synthetic peptide (C-2) representing the conserved motif itself and (b) a monoclonal antibody (anti-IFA) that recognises an epitope within the conserved coil 2b sequence. We demonstrate here that vimentin protofilaments, when induced to assemble in the presence of C-2 or anti-IFA, show a lower propensity to polymerise and yield various abberant structures. The few filaments that are formed under these conditions appear much shorter than normal IFs and are unravelled or aggregated. Furthermore, when preformed vimentin filaments are exposed to C-2 or anti-IFA, most of the normal IFs are converted into shorter filamentous forms that possess an abberant morphology. None of these effects is seen when vimentin subunits are coincubated with control peptides. Microinjection of anti-IFA into the cytoplasm of interphasic 3T3 cells provokes collapse of vimentin IFs into a juxtanuclear mass and formation of numerous amorphous aggregates distributed throughout the cytoplasm. These two effects are not seen when the anti-IFA is microinjected into the cell nucleus. Our results provide experimental evidence supporting previous suggestions for a role for the conserved coil 2b sequence in filament assembly. We propose that this region is interacting with other sites along the vimentin molecule and that these interactions are essential for proper protofilament-protofilament alignment and filament stability.

All intermediate filament (IF) proteins share a highly conserved sequence motif at the COOH-terminal end of their rod domains. We have studied the influence of a 20-residue peptide, representing the consensus motif on filament formation and stability. Addition of the peptide at a 10-20-fold molar excess over keratins K8 plus K18 had a severe effect on subsequent IF assembly. Filaments displayed a rough surface and variable diameters with a substantial amount present in unravelled form. At higher peptide concentration (50-100-fold molar excess), IF formation was completely inhibited and instead only loose aggregates of \"globular\" particles were formed. The peptide also influenced performed keratin IF in a dose-dependent manner. While a three-fold molar excess was sufficient to cause partial fragmentation of IF, a 50-fold molar excess caused complete disassembly within 5 min. Loosely associated protofibrils, short needlelike IF fragments, and aggregates of globular particles were detected. The motif peptide also caused the disassembly of filaments formed by desmin, a type III IF protein. Peptide concentrations and incubation times required for complete disassembly were somewhat higher than for the filaments containing K8 plus K18. A 50-fold molar excess was sufficient to cause complete disassembly within 1 h. Peptides unrelated in sequence to the motif did not interfere with filament formation or stability even when present for more than 12 h at a 100-fold molar excess. The results suggest that the motif sequence normally binds to a specific acceptor site for which the motif peptide can successfully compete. Taken together with current models of IF structure the results indicate that normal binding of the motif sequence to its acceptor must play an essential role in IF formation, possibly by directing the proper alignment of neighboring tetramers or protofilaments. Finally we show that in vitro formed IF are much more sensitive and dynamic strutures than previously thought.