Intermediate filaments (IF) are the ubiquitous constituents of the cytoskeletons of eukaryote cells. They consist of at least 6 different types, of which the most numerous and complex are the type I and type II keratins that are widely expressed in epithelia. We are interested in not only the structure, function and expression of keratin IF of human skin and their roles in keratinopathy diseases, but also of the related IF of other cell types in order to understand their roles in biology. Ongoing structural studies While the roles of the keratins in many genetic diseases are now well understood, further structural studies are necessary to develop rational approaches to therapy. We have initiated two types of structural/functional studies. In the first, we have developed synthetic peptides corresponding to the beginning of the 1A or end of the 2B rod domain segments. Several of these synthetic peptides have been injected into living cells to explore their dynamic behavior. Most function as very specific reagents for the disruption of all types of IF, but do not interfere with the state of assembly of microtubules or microfilaments. However, they do affect the supramolecular organization of them, thus indicating that all three components of the cytoskeleton function cooperatively in cells. On the other hand, we have discovered that an H1 peptide, a sequence which is specifically found only in type II keratin chains, is a very specific poison for the organization of keratin IF in epithelia. In those cultured cells which express both keratin IF and vimentin IF, only the keratin IF organization is disrupted. In a second experimental approach, we have synthesized a series of peptides of sequence corresponding to keratin chains which are involved in the important overlap regions in assembled filaments, for biophysical structural studies, solution nmr, and for X-ray crystallography. We have now determined that 2B peptides derived from the type III protein vimentin show simple dimerization, are essentially monodisperse in solution, and possess >90% a-helix, and form crystals that are suitable for X-ray diffraction analyses. Thus attempts to obtain atomic-level resolution structural information will continue. Similarly, we shall continue to make peptide constructs of the 1A rod domain segment that will be suitable for X-ray crystallography. The role of ionic interactions in IF structure We have previously discovered that a Glu106Asp substitution in the 2B rod domain segment of the keratin 1 chain causes a moderately severe form of epidermolytic hyperkeratosis. This has prompted us to re-evaluate the role of ionic interactions in the stability of the keratin chains in IF. We note that there are three potential pairs of charged residues which occupy e-g positions of the heptad repeat that have been precisely conserved in all IF chain types. Accordingly, using the type I/II keratin 5/14 paradigm, we have expressed in bacteria a large number of mutant chains bearing substitutions at a variety of charged residue positions in each of the 1A, 1B and 2B rod domain segments. For assays, we have used visual examination of IF formed in vitro following negative staining, coiled-coil molecule stabilities as assessed by urea dissolution experiments, and isolation and characterization of a-helix-enriched proteolytic fragments to examine molecular alignments. In addition, we have coupled GFP to the keratin 14 chain and explored the consequences of substitutions on KIF assembly and organization in living cells in vivo. We have found that the Glu106 position is absolutely required for the formation of a two-chain coiled-coil. Moreover, our data suggest that the Glu106 position is also absolutely required to stabilize the A22 mode of alignment, but the Asp106 substitution fails to form this alignment mode. This may provide a molecular explanation for the disease. Using similar techniques, we have explored the Arg10 position of the 1A rod domain segment of the keratin 14 chain. Our data show that while it is not essential for the formation of a two-chain coiled-coil molecule, it is essential for stabilization of the A11 mode of alignment. Ongoing crosslinking studies In explorations of the structure and organization of the cornified cell envelope, we havediscovered a large number of peptides involving crosslinks between a keratin chain and a variety of other envelope proteins. Notably, the vast majority of crosslinks involve a very specific and precisely conserved lysine residue located in the V1 region of the head domain of the type II keratins K1, K5 or K6. Interestingly, we have previously identified a case of non-epidermolytic palmaplantar keratoderma in which this conserved lysine residue of the K1 chain was substituted by isoleucine. Detailed ultrastructural analyses of the patient tissue at the level of the electron microscope revealed a severe discoordination between the cell periphery and the keratin IF cytoskeleton in the upper granular cells of the epidermis, proximal to the formation of the cornified cell envelope. Therefore, we believe this residue is critically involved in the structural organization of the cytoskeleton with the cornified cell envelope in terminally differentiated epidermis, and otherrelated stratified squamous epithelia. Loss of this critical mode of organization results in a diminished barrier function for the epidermis. Future studies will involve attempts to ablate this lysine residue to create a mouse model for this disease and to further study the connection between cytoskeletal-cornified cell envelope coordination and barrier function. In addition, we have identified other crosslinks which reveal that the keratin IF are attached to the desmoplakin component of desmosomes indirectly through a series of related intermediate filament associated proteins. Further work will be directed to understand the complexity of these apparent connections and role in disease. The organization of molecules in various IF types We have previously demonstrated by detailed crosslinking experiments that pairs of epidermal keratin molecules are aligned in three basic modes termed A11, A22 and A12. When assimilated into IF, pairs of molecules in the same axial row adopt a fourth mode termed ACN, in which the end of one molecule overlaps the beginning of the adjacent molecule by about 1 nm. Interestingly, almost all known keratinopathy mutations/substitutions reside in this overlap window. We also have shown that the molecules of type III IF adopt the same basic 4 modes, but the alignments of the former three are slightly offset. This adequately explains why type III and types I/II keratin chains cannot and do not coassemble in vivo or in vitro. However, previous studies have indicated that the molecular alignments in hair/wool keratin IF should be different, based on X-ray diffraction data. In order to begin to address the question of how and why hair keratin IF are different, we have expressed in and purified from bacteria full length representative type I and II mouse hair keratin chains. Recently, we have determined optimal conditions for their assembly into native-type IF in vitro, and will now use these for additional crosslinking studies to determine their molecular alignments. In certain important ways, this issue of how hair keratin IF chains are aligned resembles the situation in developing neuronal cells. During neuronal development, stem cells first express the type VI IF protein nestin. Later type III vimentin is expressed. As neuronal differentiation proceeds, the type IV a-internexin chain is expressed, followed eventually by the other neurofilament triplet proteins. By crosslinking experiments, we have shown that both a- internexin molecules and a-internexin-vimentin heterodimer molecules do indeed have the same dimensions and adopt the same molecular alignments in IF. Such data adequately explain how dynamic exchange/replacement processes may occur during development and differentiation. Together, these data establish that there are at least two different ways in which molecules may be packed as in the IF of the epidermal type I/II keratins and type III/IV IF. Our new experiments with the hair keratin IF will attempt to determine whether hair keratin IF molecules are aligned the same way as epidermal keratin IF molecules, or in a third different way.Searches for high molecular weight IF associated proteins When type III vimentin IF are passaged through cycles of assembly and disassembly in vitro, certain high molecular weight proteins always co-cycle. We have purified one of these from BHK-21 fibroblasts, and by amino acid sequencing, have now shown that it is in fact type VI nestin. By use of RT-PCR and RACE methods, we have determined the full length sequence of hamster nestin and showed that it possesses a very long carboxy-terminal tail of 1750 amino acids configured in quasi-11 residue repeats. The nestin can form homodimers in vitro, but prefers to form heterodimers with vimentin. While nestin does not assemble into IF by itself, it cancoassemble with vimentin, providing the amount of nestin does not exceed about 25%. These data establish that nestin is a high molecular weight protein which plays an important role in the supramolecular organization of vimentin IF in cells. Further studies will investigate the role of nestin in the known dynamic behavior of IF in cells. Likewise, in keratinocytes, individual keratin IF exist as loosely-arranged bundles or tonofibrils. Ongoing experiments are designed to address the question as to whether keratin IF are bundled together by an analogous organizing protein.
