Cyclic AMP and cyclic GMP (cAMP and cGMP) are important second messengers in key signal-transduction pathways that mediate various physiological functions in bacteria and eukaryotes. Adenylyl Cyclases (ACs) and Guanylyl Cyclases (GCs) cyclize ATP and GTP to produce cAMP and cGMP, respectively. Though most nucleotide cyclises show exquisite specificity for their substrates, there are instances where ACs were observed to have low GC activity as well, and vice versa. To understand structural basis of substrate (ATP or GTP) recognition, discrimination and binding by an adenylyl cyclase, we have taken up Ma1120, an AC from Mycobacterium avium, for our studies. Work presented in the thesis includes crystal structures of Ma1120 in the presence of substrate (ATP or GTP), by-product pyrophosphate and ATP analogue 2′,5′-dideoxyd-3′-adenosine triphosphate (2′,5′-dd-3′-ATP). A triple mutant of Ma1120 (K101→E, D157→G, A167→Y) was generated to increase specificity of Ma1120 towards GTP by mutation in the substrate specifying residues, but the enzyme showed equal specificity for ATP as well as for GTP. Ma1120 exists as a monomer in solution and crystallized as a monomer in the absence of substrate or inhibitor. The substrate specifying lysine residue plays a dual role of interacting with the substrate and stabilizing the dimer. The dimerization loop region harbouring the second substrate specifying residue, an aspartate, shows significant differences in conformation and position between the monomeric and dimeric structures. Thus, this study has not only revealed that significant structural transitions are required for the interconversion of the inactive and the active forms of the enzyme, but also provided precise nature of these transitions. ATP bound to Ma-Cat has two different conformations, one with C2′-endo and the other with C3′-endo puckering for the ribose. C3′-endo conformation is favourable for catalysis as it brings 3′-OH group of ribose and free oxygen of α-phosphate closer to each other. The crystal structure of GTP bound to Ma-Cat showed a novel mode of GTP binding to AC. This is the first report of GTP bound to AC. ATP bound to Ma-Cat-KDA→EGY forms non-cognate substrate complex and ATP is stabilized by stacking of adenines over each other with Tyr167 flanking on both sides of adenines. Ma-Cat-KDA→EGY+GTP complex is the first report of GTP bound to a guanylyl cyclase. GTP is bound in reverse orientation when compared to ATP bound to AC. Reverse orientation of GTP is attained to stabilize the guanine in highly electronegative guanine binding pocket. Also, O3' of GTP is placed in opposite orientation as compared to ATP bound to Ma-Cat. Therefore, during cyclization reaction guanine and ribose changes their orientation to bring O3' atom of ribose closer to α-phosphate, after cleavage of the bond between α- and β-phosphates. Thus, this study has revealed novel modes of binding of ATP and GTP to catalytic domains of Ma1120 and its triple mutant, mechanism of substrate discrimination and residual activity for the non-cognate substrate.

Structural investigations on two mycobacterial enzymes were carried out. Tuberculosis still remains a major threat to mankind even though drugs against it have been in use for many decades. The emergence of drug resistant strains of the bacteria calls for the identification of new targets based on which new drugs can be developed to combat the disease. A thorough understanding of the functioning of the target molecules is essential for this approach. We have taken up the structural studies on two such molecules, aspartic proteinases and adenylyl cyclases, of Mycobacterium tuberculosis with a view to obtain insights into their mechanisms of action at the atomic level. The work presented in the thesis includes (i) the identification, cloning, expression, purification and structure determination of a putative aspartic proteinase domain of M. tuberculosis and (ii) the crystal structure of an adenylyl cyclase of M. tuberculosis and its mutant; and also of an adenylyl cyclase from M. avium. Chapter 1 presents an overview of aspartic proteinases and nucleotide cyclases with an emphasis on their structural features. The methods employed during the course of the present work are described in Chapter 2. Work on the putative aspartic proteinase domain identified in M. tuberculosis is presented in Chapter 3. The structure of the aspartic proteinase domain is the first structural report of such domain from any bacteria. A search in the genome of M. tuberculosis showed a weak similarity to the HIV aspartic proteinase sequence. This region corresponds to the C-terminal domain of a PE family protein in M. tuberculosis. The presence of two signature motifs, DTG and DSG, of aspartic proteinases in the full sequence of this domain encouraged us to take up further studies on this domain. Previous reports identifying the protein as a surface antigen and our findings on the occurrence of similar domains in two other PE proteins of M. tuberculosis and also in other pathological strains of Mycobacteria indicated that these domains probably play an important role in infecting the host. The crystal structure of one of the domains showed that it has a pepsin-like fold and the catalytic site architecture of known aspartic proteinases. However, no proteolytic activity was detected. The size of the molecule is intermediate to eukaryotic pepsins and the homodimeric retroviral pepsins. A close examination of the binding site revealed subtle differences when compared to the active enzyme structures. Appropriate mutations of some of the residues in this region to convert it to an active enzyme did not make it active. Once the in vivo function of these putative aspartic proteinase domains is established, their potential to act as drug targets can be probed as the PE proteins are present exclusively in pathogenic Mycobacteria. As part of an ongoing project on adenylyl cyclases of Mycobacteria, we have taken up the structure analysis of the catalytic domains of two adenylyl cyclases; Rv1625c from M. tuberculosis and Ma1120 from M. avium. This work is described in Chapter 4. The wild-type of Rv1625c crystallized as a domain swapped head to head inactive dimer even though it is an active dimer in solution and expected to have a head to tail arrangement as in the previously reported structures of the active forms of the enzyme. Mutation of a phenylalanine residue presumed to occur at the subunit interface of the active dimeric structure of the enzyme to an arginine residue, a conserved residue of guanylyl cyclases, resulted in reduced adenylyl cyclase activity. This mutant crystallized as a monomer though it was expected to be an active dimer. Similarly, Ma1120 also has a monomeric structure in the crystal in spite of showing activity in solution. Though our aim was to capture the active dimers in the crystalline state we did not succeed in this effort in any of the three cases. The catalytic reaction probably takes place very rapidly with the formation of a transient active form of the dimer which cannot be easily crystallized. However, the analysis revealed new structures which are likely to represent the stable states of the enzyme when it is required to stay inactive in certain conditions. We have also established that the N-terminal segments of the enzyme, a loop at the dimeric interface and external factors like pH are involved in determining the oligomeric status of the enzyme thereby regulating its function. Publications 1 Crystal structure of a putative aspartic proteinase domain of the Mycobacterium tuberculosis cell surface antigen PE_PGRS16; Deivanayaga V. Barathy and K. Suguna; FEBS Open Bio (In Press) 2 New structural forms of mycobacterial adenylyl cyclases (in preparation)