Intermediate filament structure: 3. Analysis of sequence homologies

Abstract

Homology profiles have been calculated for each chain classification (Types I–V) using all available IF sequence data and detailed analyses of the results have been made. Features of structural relevance which have been noted include the following: (i) The linear distribution of homology scores for the acidic and the basic residues in segments 1B and 2 includes the same periods as those previously observed in the linear distribution of the same residue types in individual chains. This confirms that the spatial distribution of the ionic residues has been maintained during evolution, probably as a consequence of its importance in specifying the aggregation of molecules in IF. (ii) The maintenance of highly conserved sequences at the ends of the rod domain appears crucial to IF structure in general but significant differences in the location of regions of very high (or very low) homology are apparent when one chain type is compared to another. Furthermore, although differences exist between the various types of IF chains, the pronounced homologous features argue that intermediate filaments are a polymorphic family of closely related structures. (iii) The high degree of homology in the L2 link segment in all chain types suggests a highly conserved structure, probably α-helical; a limited degree of homology in all chain types for segment L12 is consistent with a partially conserved structure, possibly β; the highly variable length of link L1, together with the low level of homology present, suggests an irregular conformation. (iv) On the basis of sequence homology the Type I and Type II sequences can each be subdivided. Type Ia and IIa chains constitute those from the ‘hard’ α-keratins and Type Ib and IIb chains constitute those from the ‘soft’ epidermal keratins. (v) The neurofilament chains are confirmed as Type IV chains and not Type III. (vi) The NF-L, NF-M and NF-H neurofilament chains may be clearly distinguished within the Type IV classification on the basis of homology within both the rod domain and the H1 and H2 subdomains. (vii) The extended sequence data base now available has revealed that both the H1 and H2 subdomains differ significantly in extent. For example, the H1 subdomain encompasses the entire N-terminal domain of Type IV chains (both NF-L and NF-M) and the H2 subdomain encompasses the entire C-terminal domain of Type III chains. In contrast the Type I IF chains appear to have negligible H subdomains.