Allosteric regulation of proteins by reversible ligand binding is essential for regulation of fundamental biological processes. The mechanism by which a binding event alters the function of a distant site in a protein is only poorly understood. In this thesis, I use the Tetracycline Repressor (TetR) as a model system to study ligand induced allostery. The transcription of genes encoding the resistance to the antibiotic, tetracycline (Tc), is repressed by TetR, which is a homodimeric alpha-helical protein possessing a small N-terminal DNA binding domain (DNB domain) and a larger C-terminal tetracycline binding and dimerization domain (TBD domain). Based on previous structural and thermodynamic studies, the DNB domains are thought to exist in two stable, distinct conformations. One conformation is able to bind the Tc resistance operator sequence (tetO) with high affinity, while the other, which is induced by Tc binding, binds very weakly. While most previous studies on TetR have focused on the effects of Tc binding on the DNB domain conformation, here I have investigated the role of the DNB domain in modulating Tc binding. By introducing destabilizing mutations into the DNB domain I ascertained that the conformation and stability of the DNB domain plays an important role in determining Tc binding affinity. I also discovered that in the absence of ligand, the DNB domain exists in an unstable and flexible state with respect to the TBD domain. However, Tc binding to the TBD domain stabilizes the DNB domain, causing it to fold cooperatively with the TBD domain. I have discovered that the behavior of previously isolated non-inducible mutants is caused by the inability of Tc to stabilize the DNB domain in these mutants. Furthermore, reverse TetR mutants, which bind DNA better in the presence of Tc have an unfolded DNB domain that is only partially stabilized by Tc binding. My work suggests a new comprehensive, Tc induced stabilization and domain cooperativity model that can describe the mechanism of allostery in TetR and previously unexplainable mutants. A practical outcome of this research is the creation of a Tc induced folding switch that can be exploited to control the in vivo degradation of a protein of interest.

abstract: Traditionally, allostery is perceived as the response of a catalytic pocket to perturbations induced by binding at another distal site through the interaction network in a protein, usually associated with a conformational change responsible for functional regulation. Here, I utilize dynamics-based metrics, Dynamic Flexibility Index and Dynamic Coupling Index to provide insight into how 3D network of interactions wire communications within a protein and give rise to the long-range dynamic coupling, thus regulating key allosteric interactions. Furthermore, I investigate its role in modulating protein function through mutations in evolution. I use Thioredoxin and β-lactamase enzymes as model systems, and show that nature exploits "hinge-shift'' mechanism, where the loss in rigidity of certain residue positions of a protein is compensated by reduced flexibility of other positions, for functional evolution. I also developed a novel approach based on this principle to computationally engineer new mutants of the promiscuous ancestral β-lactamase (i.e., degrading both penicillin and cephatoxime) to exhibit specificity only towards penicillin with a better catalytic efficiency through population shift in its native ensemble.I investigate how allosteric interactions in a protein can regulate protein interactions in a cell, particularly focusing on E. coli ribosome. I describe how mutations in a ribosome can allosterically change its associating with magnesium ions, which was further shown by my collaborators to distally impact the number of biologically active Adenosine Triphosphate molecules in a cell, thereby, impacting cell growth. This allosteric modulation via magnesium ion concentrations is coined, "ionic allostery''. I also describe, the role played by allosteric interactions to regulate information among proteins using a simplistic toy model of an allosteric enzyme. It shows how allostery can provide a mechanism to efficiently transmit information in a signaling pathway in a cell while up/down regulating an enzyme’s activity. The results discussed here suggest a deeper embedding of the role of allosteric interactions in a protein’s function at cellular level. Therefore, bridging the molecular impact of allosteric regulation with its role in communication in cellular signaling can provide further mechanistic insights of cellular function and disease development, and allow design of novel drugs regulating cellular functions. / Dissertation/Thesis / Doctoral Dissertation Physics 2020

Protein ubiquitination is a key regulator of both protein stability and activity, and is involved in the regulation of a vast variety of cellular pathways. The ubiquitination system therefore provides an exciting target for drug development aiming to regulate the function of specific proteins. Our understanding of ubiquitin signalling is far from complete; and if we are to exploit this system for the benefit of human health, it is important to gain a better understanding of this complex posttranslational modification system as well as the effect of ubiquitination on the target protein. The E3 ligases MDM2 and CHIP were implicated in the control of the two transcriptional activators (TAs) IRF-1 and p53, that normally function to maintain health at the cellular and organismal level. Research carried out as part of my PhD has focused on gaining a mechanistic understanding of the ubiquitination process in particular the relationship between the E3 ligase and its substrate. Broadly, the mechanisms of E3 ligase regulation have been linked to substrate specificity and then to the physiological outcome of site-specific ubiquitination of the DNA binding domain of the TAs IRF-1 and p53. More specifically I have; (i) identified a mechanism by which the E3 ligase activity of the CHIP U-box can be allosterically regulated by ligand binding to its TPR domain. (ii) Residues on IRF-1 that are targeted by MDM2 and CHIP have been mapped, revealing that both ligases modify sites exclusively in IRF-1's DNA binding domain (DBD). Furthermore, I showed that, in its DNA bound conformation, IRF-1 is neither bound nor ubiquitinated by the ligases, suggesting a mechanism by which IRF-1 ubiquitination and possibly degradation can be regulated through its DNA binding state. And lastly, (iii) I have shown that both IRF-1 and p53, which have ubiquitin acceptor lysines in their DBD, bind DNA more stably when ubiquitinated. Modelling suggests that interactions between a positively charged surface area of ubiquitin and the negatively charged DNA can stabilises the TA-ubiquitin complex. DBD ubiquitination sites are required for full transactivation potential of both TAs, supporting a role of ubiquitin in their activation. p53 is ubiquitinated in response to activation by IR or Nutlin-3 and these ubiquitinated forms of p53 are localised in the cell nucleus associated with chromatin and do not lead to protein degradation. Taken together, the data imply that p53 and IRF-1 DNA binding ability, and thereby activity, can be modulated by ubiquitin modification.

Many diseases correlate with abnormal glucose metabolism in cells and organisms. For instance, the human M2 isoform of the glycolytic enzyme pyruvate kinase (M2PYK) plays an important role in metabolic reprogramming of tumour cells whereby aerobic glycolysis or the ‘Warburg effect’ supports cell proliferation by accumulating necessary biomass. By contrast, gluconeogenesis may play an important role, as observed in certain types of trypanosomatid parasites (e.g. the amastigote form of Leishmania major) where anabolism is essential for infectious properties. Hence, these glucose metabolising enzymes are important potential drug targets for cancer and trypanosomiasis. However, many aspects of their regulatory mechanisms are still poorly understood. This thesis describes biochemical and structural studies on M2PYK and on L. major fructose-1,6-bisphosphatase (LmFBPase), providing insights into allosteric mechanisms and structure-based drug design for both enzymes. Human PYKs and LmFBPase were expressed and purified from Escherichia coli, and their kinetics were fully characterised. It was shown that certain amino acids regulate the activity of M2PYK allosterically, but in opposite ways, with some being inhibitors and others activators. X-ray crystallographic structures and biophysical data of M2PYK complexes with alanine, phenylalanine, serine or tryptophan reveal an R-/T-state oscillating model of M2PYK involving a 11° rotation of each subunit. In addition, M2PYK was demonstrated to be a redox-sensitive enzyme. Reducing reagents were shown to help maintain the tetramer and prevent its dissociation, and thereby to activate M2PYK, whereas oxidation and nitrosylation reagents functioned in the opposite sense. Nitrosylation assays showed that the main nitrosylated residue is Cys326 of M2PYK, which is located on the tetramer interface. Dynamics and modulator effects of PYKs were further studied by hydrogen–deuterium exchange by mass spectrometry. These observations highlight the important effects of amino acids on M2PYK regulation. M1PYK by contrast, was demonstrated to be a constitutively fully active pyruvate kinase, with minor effects from modulators. The gluconeogenic enzyme fructose-1,6-bisphosphatase (FBPase) is a potential drug target against leishmaniasis. Here we present biochemical and structural studies for LmFBPase, by characterising its activity in a metal-dependent reaction, as well as its inhibition by AMP. The crystal structure of LmFBPase is a homotetramer, composed of monomers with alternating α/β/α/β/α ‘club sandwich’ topologies. In comparison with previously revealed LmFBPase structures, the AMP-complexed structure shows a rotated form of the tetramer. Comparisons of the structures reveal an ‘unlock-androtate’ allosteric mechanism in which AMP binding causes a series of structural changes culminating in an incomplete and non-productive active site. The structure of the effector site of LmFBPase shows a different conformation from human FBPases, thereby offering a potential specific target for Leishmania.

Calcineurin (CaN), a heterodimeric Ca2+-calmodulin-dependent Ser/Thr phosphatase, regulates diverse pathways, from stress responses in yeast to T-cell activation and cardiac hypertrophy in humans. Calmodulin (CaM), an essential mediator of calcium–dependent signaling pathways, activates CaN in the presence of calcium by binding to an intrinsically disordered region of the enzyme and altering its conformation. My hydrodynamic studies have determined that CaM participates in a 1:1 complex with the CaM-binding domain of βCaN (CaNp, residues 400–423). To explore the molecular mechanism of CaM association with CaN, I have used spectroscopic methods to determine the calcium-dependent and domain–specific interactions of CaM with CaNp. These studies revealed that the affinity of CaM1–148 for CaNp was weak in the absence of calcium, and very high (Kd in the nM to pM range) in the presence of calcium. I have demonstrated that CaNp binding to CaM increases the calcium–binding affinity of each domain of CaM1–148 to a similar degree, thereby retaining the property of sequential calcium binding to the domains, with preference for sites in the C–domain. This allows the N–domain to lag in response to an increase in cellular calcium and perhaps contribute to the regulation of CaN in a manner distinct from that of the C–domain. NMR studies of calcium–saturated CaM1–148 demonstrated that the N–domain of CaM experienced a larger structural perturbation than the C–domain upon binding CaNp. Additional NMR studies revealed that CaNp adopts an anti–parallel orientation when bound to CaM, with the sole aromatic residue of CaNp contacting the N–domain of CaM. This contrasts with many CaM-target complexes in which the sole aromatic residue contacts the C–domain of CaM. Rigorous thermodynamic studies explored how mutations in the calcium-binding sites of mammalian CaM (mCaM) and mutations known to cause disruption of CaM–mediated ion channel regulation in Paramecia (PCaM) affected the allosteric interactions of the domains of CaM in the presence of CaNp. These studies demonstrated separable roles of the domains of CaM in recognition of CaNp. The consequences of a mutation depended upon its location within the complex. Collectively, research presented in this thesis provides insight into the mechanisms whereby the two domains of CaM contribute to recognition of CaN.

The aldolase-dehydrogenase complex catalyzes the last two steps in the microbial meta-cleavage pathway of various aromatic compounds including polychlorinated biphenyls (bph pathway) and cholesterol (hsa pathway). The aldolase, BphI, cleaves 4-hydroxy-2-oxoacids to produce pyruvate and an aldehyde. Linear aldehydes of up to six carbons long and branched isobutyraldehyde were directly channeled to the aldehyde dehydrogenase BphJ, via a molecular tunnel, with greater than 80% efficiency. The molecular tunnel is narrow in positions lined by Gly-322 and Gly-323 in the aldolase. BphI variants G322F, G322L and G323F were found to block aldehyde channeling. The replacement of Asn-170 in BphJ with alanine and aspartate did not substantially alter aldehyde channeling efficiencies, thus disproving a previous hypothesis that hydrogen bonding between the Asn-170 and the nicotinamide cofactor induces the opening of the exit of the tunnel. The H20A and Y290F BphI variants displayed significantly reduced aldehyde channeling efficiencies indicating that these residues control the entry and exit of substrates and products from the aldolase reaction. The BphI reaction was activated by NADH binding to BphJ in the wild-type enzyme and channel blocked variants. Activation of BphI by BphJ N170A, N170D and I171A was decreased by ≥ 3-fold in the presence of NADH and ≥ 4.5-fold when BphJ was undergoing turnover. These results demonstrate that the dehydrogenase coordinates catalytic activity of BphI through allostery rather than through faster aldehyde release from substrate channeling. HsaF, an ortholog of BphI from Mycobacterium tuberculosis could be expressed as a soluble dimer, however HsaF was inactive in the absence of HsaG, a BphJ ortholog. Acetaldehyde and propionaldehyde were channeled directly to HsaG with similar efficiencies as in the BphI-BphJ system. The HsaF-HsaG complex was crystallized and its structure solved to a resolution of 1.93 Å. Substitution of Ser-41 in HsaG with isoleucine or aspartate resulted in about 35-fold increase in Km for CoA but only 4-fold increase in Km for dephospho-CoA, confirming its importance in interacting with the 3’- ribose phosphate of CoA. A second gene annotated as 4-hydroxy-2-oxopentanoic acid aldolase (Rv3469c) from M. tuberculosis was expressed, purified and found to possess oxaloacetate decarboxylase and not aldolase activity.

The tumour suppressor p53 has evolved a MDM2-dependent feedback loop that has a dual role as either a stimulator of p53 protein translation through mRNA binding or a stimulator of p53 protein degradation through the ubiquitin-proteasome system. A unique pseudo-substrate motif or “lid” in MDM2 is adjacent to its N-terminal hydrophobic drug-binding pocket and we have evaluated whether the lid of MDM2 is a physiological regulator of this dual function of MDM2. Deletion of this flexible pseudosubstrate motif inhibits MDM2 indicating that this peptide stretch can function as a positive regulatory motif. Phospho-mimetic mutation in the pseudo-substrate motif at codon 17 (MDM2S17D) stabilizes the binding of MDM2 towards p53. Molecular modeling orientates the pseudo-substrate motif over a charged surface patch on the MDM2 surface at Arg97/Lys98 and mutation of these residues to the MDM4 equivalent reverses the activating effect of the phosphomimetic mutation. Transient or inducible low level expression of MDM2WT can promote an increase in p53 protein steady-state levels whilst the expression of MDM2S17D in cells results in p53 protein de-stabilization. Phospho-specific antibodies to the MDM2 lid demonstrate two physiological conditions that alter lid phosphorylation: (i) lid hypo-phosphorylation occurs after DNA damage where p53 protein is stabilized and (ii) lid hyper-phosphorylation occurs at high cell density under conditions where p53 protein is de-stabilized. Expression of MDM2S17D in cells also de-stabilizes hyperubiquitinated mutant p53 under conditions where MDM2WT has no effect on mutant p53 protein degradation. The lid functions as a flexible regulatory motif whose phosphorylation switches MDM2 from a synthesis mode to a degradation mode with implications for defining the physiological signals that control the MDM2-p53 feedback regulatory loop.

There is a growing body of interest to understand the regulation of allosteric proteins. Allostery is a phenomenon of protein regulation whereby binding of an effector molecule at a remote site affects binding and activity at the protein‟s active site. Over the years, these sites have become popular drug targets as they provide advantages in terms of selectivity and saturability. Both experimental and computational methods are being used to study and identify allosteric sites. Although experimental methods provide us with detailed structures and have been relatively successful in identifying these sites, they are subject to time and cost limitations. In the present dissertation, Molecular Dynamics Simulations (MDS) and Principal Component Analysis (PCA) have been employed to enhance our understanding ofallostery and protein dynamics. MD simulations generated trajectories which were then qualitatively assessed using PCA. Both of these techniques were applied to two important trypanosomatid drug targets and controlling enzymes of the glycolytic pathway - pyruvate kinase (PYK) and phosphofructokinase (PFK). Molecular Dynamics simulations were first carried out on both the effector bound and unbound forms of the proteins. This provided a framework for direct comparison and inspection of the conformational changes at the atomic level. Following MD simulations, PCA was run to further analyse the motions. The principal components thus captured are in quantitative agreement with the previously published experimental data which increased our confidence in the reliability of our simulations. Also, the binding of FBP affects the allosteric mechanism of PYK in a very interesting way. The inspection of the vibrational modes reveals interesting patterns in the movement of the subunits which differ from the conventional symmetrical pattern. Also, lowering of B-factors on effector binding provides evidence that the effector is not only locking the R-state but is also acting as a general heat-sink to cool down the whole tetramer. This observation suggests that protein rigidity and intrinsic heat capacity are important factors in stabilizing allosteric proteins. Thus, this work also provides new and promising insights into the classical Monod-Wyman-Changeux model of allostery.

Human glutathione synthetase (hGS) is a homodimeric enzymes that catalyzes the second step in the biological synthesis of glutathione, a critical cellular antioxidant. The enzyme exhibits negative cooperativity towards the γ-glutamylcysteine (γ-GC) substrate. In this type of allosteric regulation, the binding of γ-GC at one active site significantly reduces substrate affinity at a second active site over 40 Å away. The presented work explores protein-protein interactions, substrate binding, and allosteric communication through investigation of three regions of hGS: the dimer interface, the S-loop, and the E-loop. Strong electrostatic interactions across the dimer interface of hGS maintain the appropriate tertiary and quaternary enzymatic structure needed for activity. The S-loop and E-loop of hGS form walls of the active site near γ-GC, with some residues serving to bind and position the negatively cooperative substrate. These strong interactions in the active site serve as a trigger for allosteric communication, which then passes through hydrophobic interactions at the interface. A comprehensive computational and experimental approach relates hGS structure with activity and regulation. ATP-grasp enzymes, including hGS, utilize ATP in the nucleophilic attack of a carboxylic acid in a reaction thought to proceed through the formation of an acylphosphate intermediate. Small metal cations are known to chelate the terminal phosphates of actives site ATP, yet the role of these atoms remains unclear. In the presented work, a computational metal substitution study establishes the role these divalent cations in the catalysis of peptide bonds. The simple model is used to determine the impact of metal cations on the thermodynamics and kinetics, an important stepping stone in understanding the importance of metal cations in larger biological systems.

Cyclic-3',5 '-adenosine monophosphate (cAMP) is an ancient signaling molecule that is found in a variety of species from prokaryotes to eukaryotes and translates extra-cellular stimuli into tightly controlled intra-cellular responses. The two major mammalian cAMP sensors are protein kinase A (PKA), for the phosphorylation of the downstream effectors, and the exchange protein directly activated by cAMP (Epac ), for the guanine nucleotide exchange in the small GTPase Rap proteins. In this study, we investigated the intra-molecular cAMP dependent allosteric network of Epac cyclic nucleotide binding domain (CBD) via solution NMR spectroscopy. Epac proteins have been shown to employ an auto-inhibition strategy in the control of the equilibrium between the active and the inactive states. In the absence of cAMP, the periphery of the Rap recognition site is masked via an ionic interface provided by the N-terminus of the CBD. Binding of cAMP at the distal Phosphate Binding Cassette (PBC), results in weakening of this interface. Here we show that the cAMP binding signal is propagated to the sites important in Epac activation, i.e. the ionic interface, via two key allosteric spots within the CBD. We have also determined the dynamics as a key carrier of cAMP effects to the region forming the ionic interface (ionic latch). Hence entropic enhancements emerged as a key effector in the cAMP mediated ionic latch weakening. We have also provided initial evidence of a negative allosteric contribution from the C-terminal Hinge-Lid region (CHLR) on the cAMP induced Epac activation. In addition to these findings, we also observed critical differences in the mode of cAMP recognition and inter-subdomain communication between the Epac and PKA. A detailed understanding of these two ubiquitous systems, will aid in the development of agonists and antagonists that are relevant in the drug lead development for related diseases, such as Alzheimer's and diabetes. / Thesis / Master of Science (MSc)