Intermediate filament subunits consist of a central a-helical domain flanked by non-α-helical sequences of varying length and composition. The α-helical domain confers a rod-like structure to these proteins and is largely responsible for their assembly into 10-nm filaments. However, our understanding of the functions of the non-α-helical end-domains has been slow to emerge. The N- and C-termini of type I (keratin K14), and type II (keratin K5), and type III (vimentin) intermediate filaments were fused to green fluorescent protein (GFP) to determine whether these sequences play a role in sub-cellular localisation. When transfected into keratinocytes, the type II head-GFP fusion protein displayed a cytoplasmic filamentous localisation. The type III GFP-tail specified a cytoplasmic location in fibroblast and simple epithelial cells but did not appear filamentous in these cells. The type II tail, and the type I head and tail domains fused to GFP were present both in the nucleus and cytoplasm and their distribution was similar to GFP alone. The type III head domain however, localised to the nuclear compartment but not to the cytoplasm. Together these observations became the focus of my thesis and have revealed a novel role for the non-helical end domains of intermediate filament subunits.
The studies reported in this thesis show that the type II K5 head domain interacts with the K5/K14 filament network. Preliminary studies suggest that the type II K5 head is preferentially associating with the K14 subunit but further studies are required to confirm this observation. Deletion analysis has identified 19 residues from a conserved region within VI that are critical for this interaction, and has also shown that both El and VI subdomains contribute to this association. Also important is that the type II tail and type I head and tail domains do not associate (at least not tightly) with other components in the cytoplasm. These studies have confirmed that the type II K5 head domain is also associated with other complexes at the cell periphery and suggest a role for phosphorylation of specific residues in this association as well as in cytoplasmic localisation. Further studies are required to confirm this role.
Disruption of microtubule with nocodazole clearly demonstrated that microtubules are not involved in the localisation and organisation of the type 11 head, or keratin filaments themselves. This in turn rules out a requirement for a microtubule-dependent motor for organising the type II head or keratin filaments in the cytoplasm. In support of this, the present study found no evidence for an interaction between type II head-GFP and the microtubule-dependent motor, conventional kinesin. In addition, photobleaching studies revealed that the interaction between K5 head-GFP and K5/K14 filaments is dynamic. These findings challenge published reports that place a far greater role on microtubules for the formation and maintenance of intermediate filaments in the cytoplasm.
Disruption of actin microfilaments with cytochalasin D demonstrated that both the keratins and the type II head are closely associated with actin-containing aggregates but it was not possible to distinguish whether this association was through the type II head domain sequences. A noteworthy feature of cells transfected with the H1 fusion protein, is a distribution of fluorescence reminiscent of actin filaments found at ruffled borders.
The head domain of the type III subunit (vimentin) localised in the nuclear compartment while the tail domain in cells also expressing vimentin, was located in the cytoplasm. The same cellular location for vimentin has been reported by Lowrie et al, (2000) in their study of the GFP-tagged type III end-domains and therefore validates the approach used in this thesis. A closer examination of the N-terminus showed that the beginning of the head domain, including the highly conserved 'SSYRRIFGG' motif, is not sufficient to direct nuclear localisation.
The kinesin studies found no evidence for conventional kinesin in the transport of vimentin intermediate filament subunits as reported by Prahlad et al., (1998). Preliminary photobleaching data on the vimentin tail-GFP fusion protein in living PtK2 cells suggests that vimentin GFP-tail, like K5 head-GFP, is highly mobile. These studies however remain to be completed. Vimentin filaments do behave differently from keratin filaments when treated with nocodazole and their mobility rates as polymers are different (Yoon et al., 2001). It will be interesting to determine and compare the dissociation constants for the vimentin tail and the K5 head fusion proteins. In light of findings of the present study, the interpretation of work on vimentin needs to be reassessed.
In conclusion, based on the studies described in this thesis I propose that sequences within the type II head domain direct diffusion of keratin precursors to the keratin filament and facilitate their sliding along the filament to sites of assembly or repair. This two-dimensional diffusion of keratin intermediate filament precursors would seem to be a more efficient and faster mechanism for the formation of new filaments and for the maintenance of pre-existing keratin intermediate filaments.