Many important cellular functions are performed by large complexes whose constituents function in a coordinated manner as working parts of macromolecular machines. Complexes also play primarily structural roles as biomaterials in many tissues, including skin and muscle. The goals of this project are to elucidate the structures, assembly properties, and interactions of complexes of both kinds, with emphasis on the functional connotations of these observations. It consists of four subprojects. (1)Protein quality control is a vital function carried out by energy-dependent proteases - large complexes consisting of a peptidase and an ATPase with chaperone-like properties. The archetypal such protease is the proteasome which, among other activities, generates antigenic peptides for antibody production. Detailed mechanistic studies on the proteasome are hampered by the complex subunit composition of its ATPase. Our studies focus on the Clp proteases of E. coli whose ATPases are simply homomeric rings, as an attractive model system. In earlier work, we showed that the peptidase ClpP consists of two apposed heptameric rings and the cognate ATPase - ClpA or ClpX - consists of a hexameric ring that stacks axially on one or both faces of ClpP . These observations underpin the current paradigm, whereby the ATPase recognizes substrates, unfolds them, and feeds them into a digestion chamber inside the protease. In FY01, we sought to characterize the interactions of these proteases with protein substrates, distinguishing the steps of recognition, translocation, digestion, and dispersal of reaction products. We described the processing of substrates, RepA by ClpAP, and lambda-O by ClpXP. Both substrates bind to the distal surface of the ATPase and are then translocated into the digestion chamber of the peptidase. For ClpAP, we demonstrated that the translocation pathway is axial. We went on to study cooperativity in ClpXP with ATPase rings on both faces of the peptidase. ATPgS was found to support not only the assembly of ClpXP and its binding of lambda-O but also, unfolding and translocation, albeit ~ 100-fold more slowly than with ATP. We studied translocation by time-resolved EM, finding that translocation takes place from one end at a time, implying negative cooperativity. In ClpYQ, the ClpY ATPase has an """"""""intermediate"""""""" domain inserted into its ATPase domain. We demonstrated that this domain protrudes distal to the ring of ATPase domains. This orientation has since been confirmed in several crystallographic studies. We conclude that substrate processing by ClpXP and probably also other proteases is coordinated at the level of the complex as a whole. Translocation is the rate-limiting step, and proceeds from one side of the complex at a time. (2) The cornified cell envelopes (CEs) of terminally differentiated keratinocytes are lipoprotein layers covering their surfaces. CEs are resilient on account of covalent crosslinking of their proteins, principally loricrin, which confers physical resilience and impenetrability. We have investigated their biogenesis via a variety of EM approaches. Including compositional inferences based on mathematical modeling of their amino acid compositions, we developed a model of the CE as a layer of cross-linked loricrin molecules. Thus we envisage the CE as a composite biomaterial. This scenario allows for modulation of its biomechanical properties according to the requirements of different epithelia by adjusting the ratio of matrix to crosslinkers. In FY01, we finished off three studies of CE biogenesis in transgenic and knockout mice affected in loricrin synthesis. One of the transgenics emulates a frame-shift in loricrin encountered in human patients with Vohwinkel syndrome and Progressive Symmetric Erythrokeratoderma, genetic skin diseases whose symptoms include ichthyosis and autoamputation of digits. Surprisingly, knockout loricrin mice turned out to be essentially normal. Thin sections of their epidermis revealed corneocytes with normal-looking CEs, despite the absence of loricrin. Isolated CEs were normal in thickness and mass-per-unit-area, but they have an altered structure on their cytoplasmic surface, and altered protein compositions. The normal phenotype of loricrin knockout mice can be explained by a back-up system which supplies another protein(s) that is used to assemble normal-appearing CEs.We also studied epidermis from transgenic mice expressing a mutant form of loricrin that resembles, in its abnormal C-terminus, the protein produced in Vohwinkel?s syndrome. We detected unusual deposits of the mutant loricrin in the cytoplasm and nucleus of granular layer cells. In Vohwinkel transgenic mice, the mutant loricrin does not enter the CE but instead is transported into the nuclei of granulocytes where it appears to cause some generalized interference with nuclear function that results in the disease symptoms: i.e. the latter cannot be attributed to an alteration of the CE by incorporation by the mutant loricrin. (3) Yeast has non-Mendelian genetic elements that have been identified as prions. Their mode of cell-to-cell transmission is by cytoplasmic transmission of a polymeric form of the protein with an aberrant conformation (amyloid) resembling that of the mammalian prions implicated in neuropathies such as the spongiform encephalopathies. However, yeast prion phenotypes are manifested as lack of metabolic function rather than as cytopathic effects. In FY01, we sought to to establish a correlation between amyloid filament formation in vitro, and the protein present in infected yeast cells. We also measured the copy numbers of Ure2p in normal and overexpressing [URE3] and [ure-0] cells. Previous light microscopy studies showed that Ure2p is aggregated in [URE3] (prion-containing) cells. We found that [URE3] cells overexpressing Ure2p contain distinctive networks offilaments in their cytoplasm, and demonstrated by immunolabelling that they contain Ure2p. In [URE3] cell extracts, Ure2p is in aggregates that are only partially solubilized by boiling in SDS and urea. In these aggregates, the N-terminal prion domain (~ 80 residues long, and unusually rich in glutamine) is inaccessible to antibodies while the C-terminal nitrogen regulation domain is accessible. Our data support the concept that the prion domains stack to form the filament backbone, which is surrounded by C-terminal domains. The amount of Ure2p in normally-expressing cells is small, ~ 3000 molecules per cell, explaining why we have not detected filaments in them: i.e., the filaments are so few that they are unlikely to be sampled in random thin sections. (4) Bordetella pertussis, the pathogen responsible for whooping cough, adheres to the respiratory tract via adhesin molecules displayed on its outer surface. Filamentous hemagglutinin (FHA) is its most prominent adhesin, with multiple functionalities, and is a component of acellular vaccines.
We aim to establish a structural basis for understanding its adhesive and immunological properties. In earlier work, we devised two molecular models for FHA, which is a 50-nm monomeric rod: (1) a hairpin of two antiparallel beta-sheets; (2) a single beta-helix composed of three parallel beta-sheets. In FY00, we developed arguments in favor of the beta-helix based on sequence analysis and computational model-building. In FY01, we obtained experimental evidence in support of this model by EM of a truncated variant, and observed rods of the same width and about half the length of native FHA. These observations concur with the prediction of the beta-helical model. We have also expressing fragments in E. coli, anticipating that they may crystallize more readily than intact FHA.
Showing the most recent 10 out of 49 publications
