A new group of nanoporous materials, metal organic frameworks (MOFs), have emerged as a fascinating alternative to more traditional nanoporous materials for membrane based gas separations. Although hundreds of different MOF structures have been synthesized in powder forms, very little is currently known about the potential performance of MOFs as membranes since fabrication and testing of membranes from new materials require a large amount of time and resources. The purpose of this thesis is to predict the macroscopic flux of multi-component gas mixtures through MOF-based membranes with information obtained from detailed atomistic simulations. First, atomically detailed simulations of gas adsorption and diffusion in MOFs combined with a continuum description of a membrane are introduced to predict the performance of MOF membranes. These results are compared with the only available experimental data for a MOF membrane. An efficient approximate method based on limited information from molecular simulations to accelerate the modeling of MOF membranes is then introduced. The accuracy and computational efficiency of different modeling approaches are discussed. A robust screening strategy is proposed to screen numerous MOF materials to identify the ones with the high membrane selectivity and to direct experimental efforts to the most promising of many possible MOF materials. This study provides the first predictions of any kind about the potential of MOFs as membranes and demonstrates that using molecular modeling for this purpose can be a useful means of identifying the phenomena that control the performance of MOFs as membranes.

Novel, rationally-designed acrylate analogs of various known dipeptide substrates were found to be mechanism-based inactivators of the enzyme peptidylglycine alpha-amidating monooxygenase (PAM, EC 1.14.17.3). This enzyme is responsible for the rate-limiting and final bioactivation step, a C-terminal amidation of glycine-extended peptides, of a variety of peptide hormones including the potent pro-inflammatory compound Substance P. Protein-ligand docking studies, in tandem with in vitro kinetic analysis of these inactivators, indicated that the rational design of this class of compounds was successful in creating potent competitive inactivators of this enzyme. Pharmacological evaluation, via both acute and chronic models of inflammation in Sprague-Dawley rats, of these compounds indicates that they are highly potent anti-inflammatory agents which ameliorate both acute carrageenan-induced edema and the deleterious effects of chronic adjuvant-induced polyarthritis. Furthermore, these compounds were also able to induce a return toward a more normal phenotype in cancerous WB-Ras epithelial cells, via the interruption of the growth factor-stimulated pathway precipitated by Substance P. Finally, our modeling studies provide a structural basis for both the reaction and subsite stereospecificity of PAM toward its substrates, competitive inhibitors, and mechanism-based inactivators.

Not available / text

This thesis offers a detailed socio-historical exploration of the practice of molecular modelling in chemistry and allied disciplines. This exploration leads to observations and conclusions that are of relevance to the sociology of science in general and to the issue of representation in scientific practice, in particular. After a general introduction in chapter one, chapter two offers a review of the science studies' literature relevant to the topic, namely the literature on representation and on experimental practices in the sciences. The lack of sources dealing with the issue of physical modelling is highlighted. Chapter three, through specific cases studies, presents the argument that the circulation of structural concepts and accounts in chemistry has been historically associated with the circulation not only of graphical representations, but also of physical models. Implications of this argument for a semiotically-centred view of scientific practice are discussed. Chapter 4 discusses the role of models as a research tool. Through various historical cases, it shows how physical models have been central in the process of defining, exploring and intervening on the domain of molecular-structures as accessed through laboratory manipulations. The work of molecular modelling is presented in particular as the experimental articulation of structural constraints within clearly defined research settings. Chapters five and six analyse the development of three specific modelling systems from the late 1930s to the mid-1960s. These chapters show these developments as contingent processes based on the involvement of elite researchers and both private and public institutions. It shows in particular how these modelling systems were made relevant across individual research sites by providing not only a technical solution to the problem of assembling model structures, but also by incorporating specific structural constraints. Finally, it is argued that these systems are one of

The current work reports on structural studies of ribonucleoprotein complexes, Escherichia coli and Thermomyces lanuginosus ribosomes, and Pariacoto virus (PaV) using molecular modeling. Molecular modeling is the integration and representation of the structural data of molecules as models. Integrating high-resolution crystal structures available for the E. coli ribosome and the cryo-EM density maps for the PRE- and POST- accommodation states of the translational cycle, I generated two all-atom models for the ribosome in two functional states of the cycle. A program for flexible fitting of the crystal structures into low-resolution maps, YUP.scx, was used to generate the models. Based on these models, we hypothesize that the kinking of the tRNA plays a major role in cognate tRNA selection during accommodation. Secondly, we proposed all-atom models for the eukaryotic ribosomal RNA. This is part of a collaboration between Joachim Frank, Andrej Sali, and our lab to generate an all-atom model for the eukaryotic ribosome based on a cryo-EM density map of T. lanuginosus available at 8.9Å resolution. Homology modeling and ab initio RNA modeling were used to generate the rRNA components. Finally, we propose a first-order model for a T=3, icosahedral, RNA virus called Pariacoto virus. We used the structure available from x-ray crystallography as the starting model and modeled all the unresolved RNA and protein residues. Only 35% of the total RNA genome and 88% of the protein were resolved in the crystal structure. The generated models for the virus helped us determine the location of the missing N-terminal protein tails. The models were used to propose a new viral assembly pathway for small RNA viruses. We propose that the basic N-terminal tails make contact with the RNA genome and neutralize the negative charges in RNA and subsequently collapse the RNA/protein complex into a mature virus. This process is reminiscent of DNA condensation by positively charged ions.

Poly(2-ethyl-2-oxazoline) (PEOX) is a nonionic, synthetic polymer which is soluble in both a variety organic solvents and water. The negative entropy of mixing of this polymer in aqueous solution suggested that it adopts a rigid conformation such as a helix in aqueous solution. Hydrogen bonding between PEOX and water molecules is thought to facilitate a special conformation that is specific to aqueous solution. The intent of this work is to investigate the conformation of PEOX in aqueous solution and consequently propose the mechanism by which it would adsorb onto cellulose and make it a valuable additive in paper processing. This work ultimately contributes to the greater matter of understanding the mechanisms by which water solvates nonionic polymers. Viscometry measurements of PEOX in water show that its shape scales similar to a random coil and that its molecules collapse in the presence of sodium chloride. Investigation into the molecular structure of PEOX through molecular scale simulations have revealed that although a rigid helical conformation does not exist, the potential exists for PEOX to have secondary helical structure in both water and other solvents. Without the rigid predicted structure, however, it is not surprising that PEOX does not adsorb well on cellulose. Comparing this folded helical conformation to a random coil conformation reveals that the random coil produces a lower energy system in water.

No description available.

No description available.

No description available.

This work describes the installation of a Beowulf cluster at the University of KwaZulu-Natal / Thesis (Ph.D.-Eng)-University of KwaZulu-Natal, 2006.