The major focus of the research in our laboratory is the investigation of the role of the nitrogenase component, theMoFe protein, in the catalytic mechanism of biological nitrogen fixation. This dissertation research centers on the role(s) of the P cluster, one of the two unique FeS clusters of the MoFe protein, in the electron transfer mechanism of nitrogenase. Prior to the solution of the x-ray crystal structure of the Azotobacter vinelandii MoFe protein, it was had been determined which of the highly conserved cysteinyl residues of this protein were likely P cluster ligands. After elucidation of the crystal structure, it became evident that cysteine-88 of the a-subunit (a-Cys88) and cysteine-95 of the b-subunit (b-Cys95) could play important roles in maintaining and/or perturbing the conformation of the double-cubane structure by virtue of their bridging positions. It was found that three out of ten bacterial strains with substitutions at the a-Cys88 ligand retained significant catalytic activity. We investigated the effects of these substitutions on the overall structural, kinetic and spectroscopic parameters. The results of prior studies suggested a role for the P clusters in accepting, storing, and then delivering the electrons received from the Fe protein. Therefore, we asked whether a-Cys88 substitution resulted in perturbed functioning of the overall catalytic mechanism and more importantly what these differences reveal about the normal mechanism of nitrogenase. Alterations in the bridging cysteine a-88 affected the rate of substrate reduction which can be explained in part by production of cluster-less MoFe protein. Electron flux, NaCl concentration, and reductant concentration titration assays revealed a significant uncoupling of the ATP hydrolysis rate from the substrate reduction rate in the a-88Cys-to-Gly MoFe protein. Rapid kinetic analysis revealed decreased electron tranfer rates in all three of the a-Cys88 altered MoFe proteins when compared to the wild-type MoFe protein. The intermolecular electron transfer rate was lowered in the a-88Cys-to-Asp MoFe protein, while the intramolecular electron transfer rate was limiting in the a-88Cys-to-Gly and a-88Cys-to-Thr MoFe proteins. These results indicated a role for the a-88 position in controlling electron flow through the P cluster. Another significant finding centers on the spectroscopic signals derived from one of these a-88Cys substituted MoFe proteins. The a-88Cys-to-Gly MoFe protein possesses a unique S=1/2 EPR signal in the native, dithionite-reduced state that was shown to be due to a one-electron-oxidized P cluster. This new paramagnetic center was evidence for the dramatic perturbation of the electromagnetic properties of the P cluster by the a-88Cys-to-Gly substitution. Additionally, both Mössbauer and magnetic circular dichroism spectroscopies have also demonstrated significant changes in the electromagnetic environments of the P clusters of these a-88Cys altered MoFe proteins and that each substitution affected the P cluster differently. The novel EPR signal was exploited in order to follow the sequence of electron transfer events in the nitrogenase reaction. Finally, altered nitrogenase component proteins were combined and analyzed in an attempt to distinguish which particular step(s) are perturbed in the overall enzymatic reaction. / Ph. D.

The flagellated soil-dwelling bacterium Sinorhizobium meliloti is known for its symbiotic relationship with several leguminous genera. The symbiosis between the bacterium and its host plants facilitates fixation of atmospheric nitrogen and ultimately replenishment of nitrogen to the soil. However, before nitrogen fixation can occur, the bacterial cells must actively travel to the plant's roots and successfully induce formation of a plant organ called a root nodule. To initiate the nodulation process, the bacterium needs to be in direct contact with the root hairs. This requires free living bacterial cells within the soil to sense the presence of their host plant and travel to its roots. S. meliloti is able to do this through a process called chemotaxis. Chemotaxis is the ability to respond to chemical gradients within the environment by directed movement. It is facilitated by Methyl-accepting Chemotaxis Proteins (MCPs) as part of a two-component signal transduction system. These receptor proteins are able to bind ligands and influence the state of the signal transduction system, ultimately controlling flagellar behavior. The chemotaxis system of Escherichia coli has been well characterized and serves as a useful point of comparison to that of S. melilot throughout this work. Within this work we have determined the stoichiometry of all chemotaxis proteins of S. meliloti by means of quantitative immunoblotting. Chapter 2 addresses the stoichiometry of MCPs and the histidine kinase CheA. The eight MCPs were grouped by total abundance within the cell, in high abundance (McpV), low abundance (IcpA, McpU, McpX, and McpW), and very low abundance (McpY, McpZ and McpT). The approximate cellular ratio of these three receptor groups is 300:30:1. The chemoreceptor-to-CheA ratio is 22.3:1, highly similar to the 23:1 ratio known for Bacilius subtiltis. Chapter 3 continues the investigation of the protein stoichiometry, expanding to all chemotaxis proteins. We compare ratios of S. meliloti chemotaxis proteins to those of E. coli and B. subtilis. We address the possible reasons for the high ratio of CheR / CheB to the total amount of receptors. Proteins again can be grouped by abundance: CheD, CheY1, and CheY2 are the most abundant. CheR and CheB appear in lower amounts, CheS and CheT appear to be auxiliary proteins, and finally CheW1 and CheW2, which directly interact with the receptors and CheA. Chapter 4 focuses on altered receptor abundance in S. meliloti due to the fusion of common epitope tags to the C-terminus. The fusion of these tags promotes greater cellular abundance of many receptors including McpU. The fusion of charged residues to the C-terminus promotes a greater increase in McpU abundance thanthe addition of single amino acid residues. Truncations of McpU were made to investigate the presence of a protease recognition site near the C-terminus. These truncations resulted in an increase in abundance similar to those resulting from epitope tag fusions. As epitope tags are widely used in protein studies to help determine protein stoichiometry, this study obviates a potential stumbling block for future experimenters. The function of CheT, a small protein (13.4 kDa) encoded by the last gene in S. meliloti's major chemotaxis operon, is the subject of chapter 5. A cheT deletion strain is chemotactically deficient compared to the S. meliloti wild-type strain. Through two separate experiments (a glutaraldehyde cross-linking assay and co-purification) we demonstrated that CheT interacts with the methyltransferase CheR. We also investigate its possible role in CheR's methylation activity through a series of methylation assays. This work contributes to our understanding of Sinorhizobium meliloti's chemotaxis signal transduction system. We have discovered evidence for new a protein-protein interaction within our system and have revealed the abundance of all chemotaxis proteins within the cell. We also showed that fusions of epitope tags to various chemotaxis proteins can dramatically influence their abundance. We shed light on the possible function of a previously uncharacterized protein, although more work is required to determine its exact role. / Doctor of Philosophy / Sinorhizboium meliloti is a bacterium that lives in the soil and forms a symbiotic relationship with many plants including alfalfa, a commonly grown cover crop. This symbiotic relationship is important because it allows for nitrogen to be replenished into the soil without the use of artificial fertilizer. However, to form this relationship the bacterial cells in the soil must be able to colonize the plant roots. The soil is a complex environment with many different kinds of chemical molecules and sources of nutrients. Like many other types of bacteria, S. meliloti uses flagella (long helical structures that rotate much like a propeller) to move through the soil. Control of the flagella falls to what is known as a chemotaxis signal transduction system, which can be thought of as a navigation system for each bacterial cell. The system has proteins that act as receptors to sense different chemical molecules. The bacterial cells can sense signals for the plant and move towards their host. This work shows the abundance of each type of receptor and other components within the cell. It also examines the function of a previously unknown protein, CheT, within the chemotaxis system.

Restricting dietary protein may reduce endogenous acid load. Horses were fed diets with 10% supplemental corn oil in experiment one and either 0% or 10% supplemental corn oil in experiment two. Also, low protein (7.5% crude protein, LP) fortified with .5% lysine and .3% threonine or high protein (14.5% crude protein, HP) was fed. Horses underwent similar interval training and standard exercise tests. In experiment two, horses performed an SET prior to conditioning. The SET consisted of warm ups at the walk and trot followed by six repeated sprints and concluding with a 30 minute recovery at the walk. All sprints were at 10 m/s except the SET prior to conditioning in experiment two, which were at 7 m/s. Blood samples were taken every two weeks and as part of SETs. Samples were analyzed for pH, pCO₂, pO2, Na⁺, K⁺, Cl⁻, lactate, total protein (TP), albumin, creatinine and plasma urea-N (PUN). Bicarbonate, strong ion difference (SID) and total weak acids (Atot) were calculated. Plasma urea-N concentrations were higher in the HP group. Plasma creatinine was not different in experiment one but was higher in the LPHF group in experiment two. Also, the LPHF group had a low body condition score and the same weight therefore had a higher lean body mass. Plasma albumin was not different in either experiment and TP was not different in experiment one. Total protein was higher in the HF groups in experiment two. Protein moderated acid-base responses to SETs in both experiments. The LP group had higher pH and bicarbonate levels as well as a tendency for a higher SID in experiment one and in the SET prior to conditioning for experiment two. Lower lactate levels were observed in the LP group in experiment one. Following conditioning in experiment two, the LP group had higher pH and bicarbonate levels but only combined with HF. Restricting dietary protein can increase pH and bicarbonate levels and high fat has been shown to improve fatty acid oxidation and spare muscle glycogen. Therefore, restricting dietary protein especially in combination with high fat may be beneficial for the exercising horse. / Ph. D.

Teff (Eragrostis tef (Zucc.)) is an annual, warm-season cereal crop most notable for its gluten-free, nutrient-packed seed. Experiments were conducted in two regions of Virginia (Blacksburg and Steeles Tavern) in 2010 and 2011 to determine the grain production potential of two teff varieties (brown and white). Additionally, commercially purchased teff flour was evaluated for its suitability in producing a satisfactory baked product. Teff varieties were planted in early June and July at a seeding rate of 6 kg PLS ha??. Nitrogen fertilizer was applied at planting in the form of urea at a rate of 56 kg ha??. The experimental design was a randomized complete block with a two-way factorial treatment structure (variety and planting date) and four replications. Grain yield and nutritive value, straw yield and quality, and plant height were evaluated for each variety and planting date at Steeles Tavern in 2010. Due to failure in crop establishment and difficulties involved in threshing and processing the harvested crop, no data is available in 2010 or 2011 for Kentland or in 2011 for Steeles Tavern. In 2010 at Steeles Tavern, grain yield was significantly higher for the brown variety (367 kg ha??) compared to the white variety (97 kg ha??) for both planting dates. There was no significant difference in straw yield between varieties or planting dates with straw yield averaging 2645 and 2475 kg DM ha?? for brown and white varieties, respectively. Precipitation accumulation at Steeles Tavern was higher in 2010 (greater than 10 cm) during June and July compared to 2011 and the historic average. This may explain why the plots in 2010 were able to successfully establish and out compete weeds. In the lab, four types of baked products were tested to determine the suitability of teff for baked goods. Cakes, cookies, biscuits and bread were tested with varying treatments of teff: control (100% wheat flour) and 10, 20, 30, 40 and 100% teff flour. Each treatment was replicated three times for each product. Generally, bread and cake volumes decreased as the percent of teff increased. Teff flour was best suited for use in cookie and biscuit products compared to cakes and breads since cookies and biscuits require less leavening. Overall, both experiments (field and laboratory) demonstrated the potential of teff as an alternative grain crop in Virginia. However, additional research is needed to overcome problems associated with establishment, harvest, threshing and processing. / Master of Science

There are twenty standard amino acids that are the structural units of biomolecules and biomaterials such as proteins and peptide amphiphiles (PAs). The focus of this study was to develop accurate transferable coarse-grained (CG) models of those amino acids. In CG models, several atoms are represented together as a single pseudo-atom or "bead," which can allow the modeling of processes like self-assembly of biomolecules and biomaterials through reduction of degrees of freedom and corresponding increased computational speed. A 2:1 to 4:1 mapping scheme, in which a CG bead is comprised of two to four heavy atoms, respectively, and associated hydrogens, has been employed to represent functional groups in the amino acids. The amino acid backbone atoms are modeled as two beads while the side chains are modeled with one to three beads, and each terminus is modeled as one bead. The bonded parameters for the CG models were obtained from bond, angle, and dihedral distributions from all-atom molecular dynamics (MD) simulations of dipeptides. Non-bonded parameters were optimized using the particle swarm optimization (PSO) method to reproduce experimental properties (heat of vaporization, surface tension, and density) of analogues of the side chains, termini, and backbone groups of the amino acids. These CG models were used to study the self-assembly pathways and mechanisms of the PA c16-AHL3K3-CO2H in the presence of explicit CG water. / Master of Science / In this study, models of the amino acids were developed for computer simulations. In these models, the amino acids are represented as a collection of two or more “beads” bonded together rather than as a collection of atoms bonded together. The beads were created in such a way that their characteristics reflect those of the molecules and atom groups that they represent. This was accomplished in part by selecting parameters for each bead that approximately reproduce experimental properties (density, heat of vaporization, and surface tension) and structures (bonds and angles) of the molecules and atom groups of which they are representative. Amino acids can link together to form short segments, known as peptides, or longer chains that form proteins. The bead models that were developed in this study can be linked together in the same way. They can also be linked with other beads that represent other atom groups—carbon groups of a carbon chain, for example. Certain types of molecules known as peptide amphiphiles (PAs) are often composed of amino acids and a carbon chain. The amino acid bead models were created especially to study these molecules, so once the models were developed, they were used in computer simulations to represent PAs. Many types of PAs can automatically assemble into structures that resemble fibers, and it is this behavior in particular that was studied. By using these models in computer simulations, we are able to see things that cannot be seen in a lab with a microscope or other lab tools. This may help with future efforts to study and design molecules such as PAs which show promise for medical applications like drug delivery.

Native states of globular proteins typically show stabilization in the range of 5 to 15 kcal/mol with respect to their unfolded states. There has been a considerable progress in the area of protein stability and folding in recent years, but increasing protein stability through rationally designed mutations has remained a challenging task. Current ability to predict protein structure from the amino acid sequence is also limited due to the lack of quantitative understanding of various factors that defines the single lowest energy fold or native state. The most important factors, which are considered primarily responsible for the structure and stability of the biological active form of proteins, are hydrophobic interactions, hydrogen bonding and electrostatic interactions such as salt bridges as well as packing interactions. Several studies have been carried out to decipher the importance of each these factors in protein stability and structure via rationally designed mutant proteins. The limited success of previous studies emphasizes the need for comprehensive studies on various aspect of protein stability. An integrated approach involving thermodynamic and structural analysis of a protein is very useful in understanding this particular phenomenon. This approach is very useful in relating the thermodynamic stability with the structure of a protein. A survey of the current literature on thermodynamic stability of protein indicates that the majority of the model proteins which have been used for understanding the determinants of protein stability are small, monomeric, single domain globular proteins like RNase A, Lysozyme and Myoglobin. On the other hand large proteins often show complex unfolding transition profiles that are rarely reversible. The major part of this thesis is focused on studying potential stabilizing/destabilizing interactions in small and large globular proteins. These interactions have been identified and characterized by exploring the effects of various rationally designed mutations on protein stability. Spectroscopic, molecular biological and calorimetric techniques were employed to understand the relationships between protein sequence, structure and stability. The experimental systems used are Leucine binding proteins, Leucine isoleucine valine binding protein (LIVBP), Maltose binding protein (MBP), Ribose binding protein (RBP) and Thioredoxin (Trx). The last section of the thesis discusses thermodynamic properties of molten globule states of the periplasmic protein LBP, LIVBP, MBP and RBP. The amino acid Pro is unique among all the twenty naturally occurring amino acids. In the case of proline, the Cδ of the side chain is covalently linked with the main chain nitrogen atom in a five membered ring. Therefore, Pro lacks amide hydrogen and it is not able to form a main chain hydrogen bond with a carbonyl oxygen. Hence Pro is typically not found in the hydrogen bonded, interior region of α-helix. There have been several studies which showed that introduction of the Pro residue into the interior of an α-helix is destabilizing. Although, it is not common to find Pro residue in the interiors of an α-helix, it has been reported that it occurs with appreciable frequency (14%). The thermodynamic effects of replacements of Pro residue in helix interiors of MBP were investigated in Chapter 2 of this thesis. Unlike many other small proteins, MBP contains 21 Pro residues distributed throughout the structure. It contains three residues in the interiors of α-helices, at positions 48, 133 and 159. These Pro residues were replaced with an alanine and serine amino acids using site directed mutagenesis. Stabilities of all the mutant and wild type proteins have been studied via isothermal chemical denaturation at pH 7.4 and thermal denaturation as a function of pH ranging from pH 6.5 to 10.4. It has been observed that replacement of a proline residue in the middle of an α-helix does not always stabilize a protein. It can be stabilizing if the carbonyl oxygen of residue (i-3) or (i-4) is well positioned to form a hydrogen bond with the ith (mutated) residue and the position of mutation is not buried or conserved in the protein. Partially exposed position have the ability to form main chain hydrogen bonds and Ala seems to be a better choice to substitute Pro than Ser. Unlike other amino acids, the pyrolidine ring of Pro residue imposes rigid constraints on the rotation about the N---Cα bond in the peptide backbone. This causes conformational restriction of the φ dihedral angle of Pro to -63±15º in polypeptides. Therefore, introduction of a rigid Pro residue into an appropriate position in a protein sequence is expected to decrease the conformational entropy of the denatured state and consequently lead to protein stabilization. In Chapter 3 of this thesis, the thermodynamic effects of Pro introduction on protein stability has been investigated in LIVBP, MBP, RBP and Trx. Thirteen single and two double mutants have been generated in the above four proteins. Three of the MBP mutants were characterized by X-ray crystallography to confirm that no structural changes had occurred upon mutation. In the remaining cases, CD spectroscopy was used to show the absence of structural changes. Stability of all the mutant and wild type proteins was studied via isothermal chemical denaturation at neutral pH and thermal denaturation as a function of pH. The mutants did not show enhanced stability with respect to chemical denaturation at room temperature. However, six of the thirteen single mutants showed a small but significant increase in the free energy of thermal unfolding in the range of 0.3-2.4 kcal/mol, two mutants showed no change and five were destabilized. In five of the six cases, the stabilization was because of a reduced entropy of unfolding. Two double mutants were constructed. In both cases, the effects of the single mutations on the free energy of thermal unfolding were non-additive. In addition to the hydrogen bond, hydrophobic and electrostatic interactions, other interactions like cation-π and aromatic-aromatic interactions etc. are also considered to make important contributions to protein stability. The relevance of cation-π interaction in biological systems has been recognized in recent years. It has been reported that positively charged amino acids (Lys, Arg and His) are often located within 6 Å of the ring centroids of aromatic amino acids (Phe, Tyr and Trp). The importance of cation-π interaction in protein stability has been suggested by previous theoretical and experimental studies. We have attempted to determine the magnitude of cation-π interactions of Lys with aromatic amino acids in four different proteins (LIVBP, MBP, RBP and Trx) in Chapter 4 of the thesis. Cation-π pairs have been identified by using the program CaPTURE. We have found thirteen cation-π pairs in five different proteins (PDB ID’s 2liv, 1omp, 1anf, 1urp and 2trx). Five cation-π pairs were selected for the study. In each pair, Lys was replaced with Gln and Met. In a separate series of experiments, the aromatic amino acid in each cation-π pair was replaced by Leu. Stabilities of wild type (WT) and mutant proteins were characterized by similar methods, to those discussed in previous chapters. Gln and Aromatic → Leu mutants were consistently less stable than the corresponding Met mutants reflecting the non-isosteric nature of these substitutions. The strength of the cation-π interaction was assessed by the value of the change in the free energy of unfolding (ΔΔG0=ΔG0 (Met) - ΔG0(WT)). This ranged from +1.1 to –1.9 kcal/mol (average value – 0.4 kcal/mol) at 298 K and +0.7 to –2.6 kcal/mol (average value –1.1 kcal/mol) at the Tm of each WT. It therefore appears that the strength of cation-π interactions increases with temperature. In addition, the experimentally measured values are appreciably smaller in magnitude than the calculated values with an average difference |ΔG0expt -ΔG0calc|avg of 2.9 kcal/mol. At room temperature, the data indicate that cation-π interactions are at best weakly stabilizing and in some cases are clearly destabilizing. However at elevated temperatures, close to typical Tm’s, cation-π interactions are generally stabilizing. In Chapter 5, we have attempted to characterize molten globule states for the periplasmic proteins LBP, LIVBP, MBP and RBP. It was observed that all these proteins form molten globule states at acidic pH (3 - 3.4). All these molten globule states showed cooperative thermal transitions and bound with their ligand comparable to (LBP and LIVBP) or with lower (MBP and RBP) affinity than the corresponding native states. Trp, ANS fluorescence and near-UV CD spectra for ligand bound and free forms of molten globule states were found to be very similar. This shows that molten globule states of these proteins have the ability to bind to their corresponding ligand without conversion to the native state. All four molten globule states showed destabilization relative to the native state. ΔCp values indicate that these molten globule states contain approximately 29-67% of tertiary structure relative to the native state. All four proteins lack prosthetic groups and disulfide bonds. Therefore, it is likely that molten globule states of these proteins are stabilized via hydrophobic and hydrogen bonding interactions.

BACKGROUND: A long-standing challenge in the post-genomic era of Bioinformatics is the prediction of protein-protein interactions, and ultimately the prediction of protein functions. The problem is intrinsically harder, when only amino acid sequences are available, but a solution is more universally applicable. So far, the problem of uncovering protein-protein interactions has been addressed in a variety of ways, both experimentally and computationally. MOTIVATION: The central problem is: How can protein complexes with solved threedimensional structure be utilized to identify and classify protein binding sites and how can knowledge be inferred from this classification such that protein interactions can be predicted for proteins without solved structure? The underlying hypothesis is that protein binding sites are often restricted to a small number of residues, which additionally often are well-conserved in order to maintain an interaction. Therefore, the signal-to-noise ratio in binding sites is expected to be higher than in other parts of the surface. This enables binding site detection in unknown proteins, when homology based annotation transfer fails. APPROACH: The problem is addressed by first investigating how geometrical aspects of domain-domain associations can lead to a rigorous structural classification of the multitude of protein interface types. The interface types are explored with respect to two aspects: First, how do interface types with one-sided homology reveal convergently evolved motifs? Second, how can sequential descriptors for local structural features be derived from the interface type classification? Then, the use of sequential representations for binding sites in order to predict protein interactions is investigated. The underlying algorithms are based on machine learning techniques, in particular Hidden Markov Models. RESULTS: This work includes a novel approach to a comprehensive geometrical classification of domain interfaces. Alternative structural domain associations are found for 40% of all family-family interactions. Evaluation of the classification algorithm on a hand-curated set of interfaces yielded a precision of 83% and a recall of 95%. For the first time, a systematic screen of convergently evolved motifs in 102.000 protein-protein interactions with structural information is derived. With respect to this dataset, all cases related to viral mimicry of human interface bindings are identified. Finally, a library of 740 motif descriptors for binding site recognition - encoded as Hidden Markov Models - is generated and cross-validated. Tests for the significance of motifs are provided. The usefulness of descriptors for protein-ligand binding sites is demonstrated for the case of "ATP-binding", where a precision of 89% is achieved, thus outperforming comparable motifs from PROSITE. In particular, a novel descriptor for a P-loop variant has been used to identify ATP-binding sites in 60 protein sequences that have not been annotated before by existing motif databases.

5 Abstract Ca2+/calmodulin-dependent protein kinase kinase 2 (CaMKK2) belongs to the serine/ threonine protein kinase family, which is involved in the calcium signaling pathway. The increase of intracellular calcium concentration induces the activation of calmodulin (CaM), which then activates its binding partners including CaMKII, CaMKIII, CaMKK1 and CaMKK2. CaMKK2 activates CaMKI, CaMKIV and AMP-dependent kinase, AMPK, by phosphorylation. CaMKK2 is naturally present in cells in an autoinhibited state, which is caused by the steric hindrance of the active site by the autoinhibitory domain. When calmodulin binds to the calmodulin-binding domain, the autoinhibitory domain is removed and the active site becomes accessible. Upon activation, CaMKK2 undergoes autophosphorylation, which increases its enzyme activity. Negative regulation of CaMKK2 is mediated by cAMP-dependent protein kinase A (PKA)- and GSK3-dependent phosphorylation. Sites phosphorylated by PKA have been identified for both CaMKK1 and CaMKK2. Two of them are also motifs recognized by scaffolding 14-3-3 proteins. Previous studies have shown that the 14-3-3 protein binding maintains phosphorylated CaMKK2 in an inhibited state by blocking the dephosphorylation of S495, which prevents the binding to calmodulin. However, it is unclear if it is the...

BACKGROUND: A long-standing challenge in the post-genomic era of Bioinformatics is the prediction of protein-protein interactions, and ultimately the prediction of protein functions. The problem is intrinsically harder, when only amino acid sequences are available, but a solution is more universally applicable. So far, the problem of uncovering protein-protein interactions has been addressed in a variety of ways, both experimentally and computationally. MOTIVATION: The central problem is: How can protein complexes with solved threedimensional structure be utilized to identify and classify protein binding sites and how can knowledge be inferred from this classification such that protein interactions can be predicted for proteins without solved structure? The underlying hypothesis is that protein binding sites are often restricted to a small number of residues, which additionally often are well-conserved in order to maintain an interaction. Therefore, the signal-to-noise ratio in binding sites is expected to be higher than in other parts of the surface. This enables binding site detection in unknown proteins, when homology based annotation transfer fails. APPROACH: The problem is addressed by first investigating how geometrical aspects of domain-domain associations can lead to a rigorous structural classification of the multitude of protein interface types. The interface types are explored with respect to two aspects: First, how do interface types with one-sided homology reveal convergently evolved motifs? Second, how can sequential descriptors for local structural features be derived from the interface type classification? Then, the use of sequential representations for binding sites in order to predict protein interactions is investigated. The underlying algorithms are based on machine learning techniques, in particular Hidden Markov Models. RESULTS: This work includes a novel approach to a comprehensive geometrical classification of domain interfaces. Alternative structural domain associations are found for 40% of all family-family interactions. Evaluation of the classification algorithm on a hand-curated set of interfaces yielded a precision of 83% and a recall of 95%. For the first time, a systematic screen of convergently evolved motifs in 102.000 protein-protein interactions with structural information is derived. With respect to this dataset, all cases related to viral mimicry of human interface bindings are identified. Finally, a library of 740 motif descriptors for binding site recognition - encoded as Hidden Markov Models - is generated and cross-validated. Tests for the significance of motifs are provided. The usefulness of descriptors for protein-ligand binding sites is demonstrated for the case of "ATP-binding", where a precision of 89% is achieved, thus outperforming comparable motifs from PROSITE. In particular, a novel descriptor for a P-loop variant has been used to identify ATP-binding sites in 60 protein sequences that have not been annotated before by existing motif databases.

Indiana University-Purdue University Indianapolis (IUPUI) / The discovery and isolation of naturally occurring fluorescent proteins, FPs, have provided much needed tools for molecular and cellular level studies. Specifically the cloning of green fluorescent protein, GFP, revolutionized the field of biotechnology and biochemical research. Recently, a red fluorescent protein, DsRed, isolated from the Discosoma coral has further expanded the pallet of available fluorescent tools. DsRed shares only 23 % amino acid sequence homology with GFP, however the X-ray crystal structures of the two proteins are nearly identical. DsRed has been subjected to a number of mutagenesis studies, which have been found to offer improved physical and spectral characteristics. One such mutant, DsRed-Monomer, with a total of 45 amino acid substitutions in native DsRed, has shown improved fluorescence characteristics without the toxic oligomerization seen for the native protein. In our laboratory, we have demonstrated that DsRed proteins have a unique and selective copper-binding affinity, which results in fluorescence quenching. This copper-binding property was utilized in the purification of DsRed proteins using copper-bound affinity columns. The work presented here has explored the mechanism of copper-binding by DsRed-Monomer using binding studies, molecular biology, and other biochemical techniques. Another focus of this thesis work was to demonstrate the applications of DsRed-Monomer in biochemical studies based on the copper-binding affinity and fluorescence properties of the protein. To achieve this, we have focused on genetic fusions of DsRed-Monomer with peptides and proteins. The work with these fusions have demonstrated the feasibility of using DsRed-Monomer as a dual functional tag, as both an affinity tag and as a label in the development of a fluorescence assay to detect a ligand of interest. Further, a complex between DsRed-Monomer-bait peptide/protein fusion and an interacting protein has been isolated taking advantage of the copper-binding affinity of DsRed-Monomer. We have also demonstrated the use of non-natural amino acid analogues, incorporated into the fluorophore of DsRed-Monomer, as a tool for varying the spectral properties of the protein. These mutations demonstrated not only shifted fluorescence emission compared to the native protein, but also improved extinction coefficients and quantum yields.