Dissolution dynamic nuclear polarization (DNP) provides several orders of magnitude of NMR signal enhancement by converting the much larger electron spin polarization to nuclear spin polarization. Polarization occurs at low temperature (1.4K) and is followed by quickly dissolving the sample for room temperature NMR detection. DNP is generally applicable to almost any small molecules and can polarize various nuclei including 1H, 19F and 13C. The large signal from DNP enhancement reduces the limit of detection to micromolar or sub-micromolar concentration in a single scan. Since DNP enhancement often provides the only source for the observable signal, it enables tracking of the polarization flow. Therefore, DNP is ideal for studying chemical processes. Here, quantitative tools are developed to separate kinetics and spin relaxation, as well as to obtain structural information from these measurements. Techniques needed for analyzing DNP polarized sample are different from those used in conventional NMR because a large, yet non-renewable hyperpolarization is available. Using small flip angle pulse excitation, the hyperpolarization can still be divided into multiple scans. Based on this principle, a scheme is presented that allows reconstruction of indirect spectral dimensions similarly to conventional 2D NMR. Additionally, small flip angle pulses can be used to obtain a succession of scans separated in time. A model describing the combined effects of the evolution of a chemical process and of spin-lattice relaxation is shown. Applied to a Diels-Alder reaction, it permitted measuring kinetics along with the effects of auto- and cross-relaxation. DNP polarization of small molecules also shows significant promise for studying protein-ligand interaction. The binding of fluorinated ligands to the protease trypsin was studied through the observation of various NMR parameter changes, such as line width, signal intensity and chemical shift of the ligands. Intermolecular polarization transfer from hyperpolarized ligand to protein can further provide information about the binding pocket of the protein. As an alternative to direct observation of protein signal, a model is presented to describe a two-step intermolecular polarization transfer between competitively binding ligands mediated through the common binding pocket of the protein. The solutions of this model relate the evolution of signal intensities to the intermolecular cross relaxation rates, which depend on individual distances in the binding epitope. In summary, DNP provides incomparable sensitivity, speed and selectivity to NMR. Quantitative models such as those discussed here enable taking full advantage of these benefits for the study of chemical processes.

Design of self-assembling ABC-type collagen triple helical heterotrimers is challenging due to the number of competing species that can be formed in ternary mixture of peptides with a high propensity to fold into triple helices and the fact that well understood rules for pair-wise amino acid stabilization of the canonical collagen triple helix have remained elusive. Given the required one amino acid stagger between adjacent peptide strands in this fold, a ternary mixture of peptides can form as many as 27 triple helices with unique composition or register. Previously we have demonstrated that electrostatic interactions can be used to bias the helix population towards a desired target but the presence of competing states in mixtures has remained an outstanding problem. In this work we use high-resolution structural biology techniques to do a detailed study of stabilizing pair-wise interactions between positively and negatively charged amino acids in triple helices. Two types of contacts with distinct sequence requirements depending on the relative stagger of the interacting chains are observed: axial and lateral. Such register-specific interactions are crucial for the understanding of the registration process of collagens and the overall stability of proteins in this family. Using this knowledge we developed distinct design strategies to improve the specificity of our designed systems towards the desired ABC heterotrimeric target state. We validate our strategies through the synthesis and characterization of the designed sequences and show that they self-assemble into a highly stable ABC triple helices with control over composition in the case of the rational approach and with control over both composition and register in the case of the computational approach.

Proteins are important building blocks of all living organisms. They are composed of a defined sequence of different amino acids, and fold into a specific three-dimensional, ordered structure. The three-dimensional structure largely determines the function of the protein, but protein function always requires motion. Small movements within the protein structure govern the functional properties, and this thesis aims to better understand these discrete protein movements. The motions within the protein structure are governed by thermodynamics, which therefore is useful to predict protein interactions. Nuclear magnetic resonance (NMR) is a powerful tool to study proteins at atomic resolution. Therefore, NMR is the primary method used within this thesis, along with other biophysical techniques such as Fluorescence spectroscopy, Circular Dichroism spectroscopy and in silico modeling. In paper I, NMR in combination with molecular engineering is used to show that the folding of the catalytical subdomains of the enzyme Adenylate kinase does not affect the core of the protein, and thus takes a first step to linking folding, thermodynamic stability and catalysis. In paper II, the structure of the primary allergen from Brazil nut, Ber e 1, is presented along with biophysical measurements that help explain the allergenic potential of the protein. Paper III describes the need for a specific Brazil nut lipid fraction needed to induce an allergenic response. NMR and fluorescence spectroscopy is used to show that there is a direct interaction between Ber e 1 and one or several components in the lipid fraction.

Forkhead box Class O (FOXO), one subfamily of the Forkhead box (Fox) family, which is featured by the Forkhead (FH) DNA-binding domain, includes four human transcription factors: FOXO1, FOXO3a, FOXO4, and FOXO6. The tumor suppressor FOXO3a is involved in multiple physiological and pathological processes, such as breast cancer and acute myeloid leukemia, and is related to human longevity. It plays essential role in metabolism, cell cycle arrest, DNA repair, and apoptosis. Besides the FH domain, FOXO3a contains three other regions (CR1-3), conserved within FOXO subfamily. It specifically binds a consensus Forkhead response element (FRE) DNA sequence through the FH domain, and recruits transcriptional coactivator CBP/p300 to activate gene transcription. FOXO3a functions through interacting with other proteins as well. FOXO3a binds p53 through the FH domain and the CR3 region, which are also engaged in an intramolecular interaction, and the solution structure of the former one was determined. This intramolecular interaction regulates coactivator recruitment and is disrupted by FRE DNA. A novel transactivation domain (TAD) CR2C was identified in addition to the known TAD CR3, both of which promiscuously associate with the KIX domain of CBP/p300 in equilibrium between two conformational states, the structures of which were determined by NMR spectroscopy. These two TADs of FOXO3a form additional multivalent binding to the TAZ1 and TAZ2 domains of CBP/p300, further increasing the promiscuity and complexity of the interaction. The coactivator recruitment is modulated by AMPK phosphorylation, which enhances the multivalent interaction between FOXO3a and CBP/p300, and thus the transactivation. These results indicate the significance of intrinsically disordered regions (IDRs) of FOXO3a in transcriptional activation and protein interaction, provide insight of the role of FOXO3a in gene transcription and apoptosis under various conditions, and potentially contribute to the cancer therapy.

RNA - protein interactions constitute a vital part of numerous biochemical processes. In the HIV life cycle, the interaction of the viral protein Rev and the Rev Responsive Element (RRE), a part of unspliced HIV RNA, is crucial for the propagation of infectious virions. Intervention of this interaction disrupts the viral life cycle. Rev - RRE interaction initially occurs at a high affinity binding site localized to a relatively small stem loop structure called RREIIB. This binding event has been well characterized by a variety of biochemical, enzymatic and structural studies. Our collaborators have previously demonstrated the efficacy of zinc finger proteins, generated by phage display, in the specific targeting of RREIIB. We have shown that the binding of these zinc finger proteins is restricted to the bulge in stem loop IIB that Rev also targets. Currently these proteins bind RREIIB with dissociation constants in the nanomolar range. We have employed a wide assortment of biophysical techniques such as gel shift assays, circular dichroism, isothermal titration calorimetry and NMR structural studies to further investigate this interaction. Several mutants of the zinc finger protein and the RNA were also studied to delineate the parts of the protein secondary structure as well as the role of specific side chains in this interaction. We have generated a solution structure of the RREIIBTR RNA bound zinc finger protein, ZNF29G29R, which displayed the highest affinity to this RNA. This has allowed us to shed further light on the molecular basis of this RNA - protein interaction and provides input for further refinement in our structure guided phage display.

Calmodulin is an essential EF-hand protein with a helix-loop-helix calcium binding motif. Understanding Ca(II) dependent activation of calmodulin and other EF-hand proteins is limited by Ca(II)-induced conformational change, multiple and cooperative binding of Ca(II) ions, and interactions between the paired EF-hand motifs. The goal of this research project is to probe key determinants for calcium binding properties and pairing interactions at the site specific level using a grafting approach and high resolution NMR. An individual Ca(II) binding site of the EF-hand motifs of calmodulin was grafted into a non-calcium dependent protein, CD2, to bypass limitations associated with natural EF-hand proteins and peptide fragments. Using high resolution NMR, we have shown that the grafted EF-loop III of calmodulin in the host protein retains its native conformation with a strong loop and β-conformation preference. Grafted ligand residues in the engineered protein are directly involved in binding of Ca(II) and La(III). The NMR studies support our hypothesis that both ligand arrangement and dynamic properties play essential role in tuning Ca(II) binding affinities. Using pulse-field diffusion NMR and protein engineering, we further demonstrated that grafted EF- loop remains as a monomer. Although the EF-loop with flanking helices dimerizes in the presence of Ca(II). Additionally, removal of conserved hydrophobic residues at the flanking helices of the EF-hand motif leads to be monomer in the absence and presence of metal ions. Our results suggest that conserved hydrophobic residues are essential for the pair-paired interaction in the coupled EF-hand protein. We have shown that our developed grafting approach can be applied to probe intrinsic Ca(II) binding affinities of different Ca(II) binding sites.

Copper is one of the most vital trace elements in ruminant nutrition. It is required for several metabolic activities and it is also an essential component of several physiologically important metalloenzymes. Thus copper deficiency in ruminants results in distinctive pathologies, and hence in significant economic losses to farmers. Copper deficiency results from very low copper in diet (primary copper deficiency) and interference with Cu absorption in the animal due to Mo and S in food or water (secondary copper deficiency). The molybdenum-induced copper deficiency that affects ruminants can be attributed to the formation of thiomolybdates (TMs)from molybdate and sulfide in the rumen. The TMs formed then react irreversibly with copper to form insoluble Cu-TM complex which ultimately end up being excreted, thus reducing copper bioavailability to the ruminant. <p>In this study, an attempt has been made to use computer simulations to model speciation of copper in rumen fluid in the presence of TMs with the aim of understanding the extent to which TMs affects the levels of copper in the rumen. <p>This was done by initially refining the computer model of copper speciation with respect to low molecular mass (LMM) ligands in bovine rumen with the aim of correcting the discrepancy that was observed during experimental validation of the computer model in a previous study. To this end, mass balance equations which describes the distribution of Cu(II) amongst the different ligands were encoded into a spreadsheet to calculate equilibrium concentration of all species. Formation constants obtained from literature as well as those obtained from studies in our group were used as input values in the spreadsheet. Results show that at average ruminal pH, the metal would be present mostly as carbonate and phosphate complexes. The results obtained from the computer model in the present study were validated using 1H NMR experiments on simulated rumen fluid as well as actual rumen fluid containing Cu(II); using acetic acid chemical shift as the probe for monitoring the speciation pattern. Excellent agreement was observed between the computer model and experimental results. Discrepancy was however observed upon introduction of copper lysine as copper source into the model. Incorporation of a mixed ligand complex of Cu(II), acetate and lysine into the computer model gave an excellent agreement between the computer model and experimental results. <p>The study was extended to include glycine, histidine, methionine and EDTA complexes as the copper source in both rumen saliva (McDougalls solution) and rumen fluid. Results show that only the histidine and EDTA complexes persist to any significant extent, in spite of the large number of competing ligands present in these matrices.<p>In this study, success has also been achieved in the integration of the slow (kinetically controlled) formation of TMs and copper-tetrathiomolybdate (TM4) complexation into the previously developed model for the rapidly equilibrating copper-ligand speciation. To simulate the formation of the TMs and Cu-TM4 complex with respect to time, the differential equations representing rate expressions for each chemical species were solved to obtain an analytical solution using the Laplace transform method. The analytical solutions obtained were encoded in a spreadsheet and calculated as function of time to obtain time dependent concentrations of TMs and Cu-TM4 complex. This was then integrated with previously developed model for the rapidly equilibrating copper-ligand speciation in the rumen. The kinetic data used in the simulation of the formation thiomolybdates was obtained fron literature wheras that for Cu-TM4 complexation was obtained from our lab using Cu(II) - Ion Selective Electrode. The results show that that in the presence of TM4 the, Cu(II) bound to low molecular ligands in the rumen is drastically reduced confirming the effect TM4 on Cu(II) observed in several in vitro studies.<p>The study shows that in thiomolybdate contaminated rumen environment, the bioavailability of copper is considerably reduced. Though metal bioavailabilities are hard to predict this approach could help better our understanding of this process.

Forkhead box Class O (FOXO), one subfamily of the Forkhead box (Fox) family, which is featured by the Forkhead (FH) DNA-binding domain, includes four human transcription factors: FOXO1, FOXO3a, FOXO4, and FOXO6. The tumor suppressor FOXO3a is involved in multiple physiological and pathological processes, such as breast cancer and acute myeloid leukemia, and is related to human longevity. It plays essential role in metabolism, cell cycle arrest, DNA repair, and apoptosis. Besides the FH domain, FOXO3a contains three other regions (CR1-3), conserved within FOXO subfamily. It specifically binds a consensus Forkhead response element (FRE) DNA sequence through the FH domain, and recruits transcriptional coactivator CBP/p300 to activate gene transcription. FOXO3a functions through interacting with other proteins as well. FOXO3a binds p53 through the FH domain and the CR3 region, which are also engaged in an intramolecular interaction, and the solution structure of the former one was determined. This intramolecular interaction regulates coactivator recruitment and is disrupted by FRE DNA. A novel transactivation domain (TAD) CR2C was identified in addition to the known TAD CR3, both of which promiscuously associate with the KIX domain of CBP/p300 in equilibrium between two conformational states, the structures of which were determined by NMR spectroscopy. These two TADs of FOXO3a form additional multivalent binding to the TAZ1 and TAZ2 domains of CBP/p300, further increasing the promiscuity and complexity of the interaction. The coactivator recruitment is modulated by AMPK phosphorylation, which enhances the multivalent interaction between FOXO3a and CBP/p300, and thus the transactivation. These results indicate the significance of intrinsically disordered regions (IDRs) of FOXO3a in transcriptional activation and protein interaction, provide insight of the role of FOXO3a in gene transcription and apoptosis under various conditions, and potentially contribute to the cancer therapy.

The Cre sequence (ACGT) is a site responsible for the binding of specific transcription factors that determine the activation of genes. Due to its major role in gene transcription, it has become a subject of immense research. The binding of transcription factors to the Cre binding site has been determined to be dependent on DNA conformation. In this study, the effects of flanking sequence around the Cre binding site on the conformation and the dynamics of DNA were investigated. The Cre binding site was studied in its native form with differing flanking sequences to determine the BI/BII profile (conformation) and the magnitude of the energy transition barrier (dynamics) between the BI and BII conformations of each phosphate step of the following three dodecamer sequences: CreACAG, CreGGAG, and CreTATA. In order to obtain the BI/BII profile of each phosphate step, 2D 31P-NMR NOESY and HSQC experiments at various temperatures were utilized. Based of the basic principles of kinetics, the lower the energy barrier between the two conformations, the easier the transition between the BI and BII conformation. Therefore, it was hypothesized that low and high %BII character lead to a large energy barrier (high ∆G‡ values), whereas average %BII character leads to a small energy barrier (low ∆G‡ values). The results of the 2D 31P-NMR experiments of the three dodecamer sequences confirmed this relationship between the %BII character and the magnitude of the energy barrier (∆G‡). However, further conformation and dynamics studies must be conducted to further understand the correlation.

The HIV-1 transactivator of transcription (Tat) is a protein essential for both viral gene expression and virus replication. Tat is an RNA-binding protein that, in cooperation with host cell factors cyclin T1 and cyclin-dependent kinase 9, regulates transcription at the level of elongation. Tat also interacts with numerous other intracellular and extracellular proteins, and is implicated in a number of pathogenic processes. The Tat protein is encoded by two exons and is 101 residues in length. The first exon encodes a 72-residue molecule that activates transcription with the same proficiency as the full-length protein. The physico-chemical properties of Tat make it a particularly challenging target for structural studies: Tat contains seven cysteine residues, six of which are essential for transactivation, and is highly susceptible to oxidative cross-linking and aggregation. In addition, a basic segment (residues 48-57) gives the protein a high net positive charge of +12 at pH 7, endowing it with a high affinity for anionic polymers and surfaces. In order to study the structure of Tat, both alone and in complex with partner molecules, we have developed a system for the bacterial expression and purification of polyhistidine-tagged and isotopically enriched (in 15N and 15N /13C) recombinant HIV-1 Tat1-72 (BH10 isolate) that yields large amounts of protein. These preparations have facilitated the assignment of 95% of the non-proline backbone resonances using heteronuclear 3-dimensional nuclear magnetic resonance (NMR) spectroscopy. Analysis by mass spectrometry and NMR demonstrate that the cysteine-rich Tat protein is unambiguously reduced and monomeric in aqueous solution at pH 4. NMR chemical shifts and coupling constants suggest that it exists in a disordered conformation. Line broadening and multiple peaks in the cysteine-rich and core regions suggest that transient folding occurs in two of the five sequence domains. NMR relaxation parameters were measured and analysed by spectral density and model-free approaches both confirming the lack of structure throughout the length of the molecule. The absence of a fixed conformation and the observation of fast dynamics are consistent with the ability of the Tat protein to interact with a wide variety of proteins and nucleic acid lending further support to the concept that Tat exists as an intrinsically disordered protein. / October 2007