Studies on the structure and function of protein kinase G, a virulence factor of "Mycobacterium tuberculosis"

The genome of M. tuberculosis comprises eleven serine/threonine protein kinases which carry out various functions, e.g. in cell division, metabolism and pathogenicity. Nine of these eleven kinases are membrane-bound and have a C-terminal extracellular domain and an intracellular Hanks-type kinase domain located at the N-terminus. Protein kinase G (PknG) differs from these kinases, because it is predicted to be a cytosolic protein since it lacks a transmembrane domain. The kinase domain is preceded by a long N-terminal stretch containing two CXXC motifs. Moreover, PknG possesses a tetratricopeptide repeat within its C-terminal domain. PknG, expressed by pathogenic mycobacteria upon infection, has been shown to be secreted into the macrophage cytosol where it modulates and prevents phagosome-lysosome fusion, finally resulting in survival of the bacilli. A potent inhibitor, termed AX20017, was identified which specifically inhibited the catalytic activity of PknG in vitro. Infection of J774 macrophages with mycobacteria treated with AX20017 led to an increase rate of lysosomol delivery events. The same effect was observed when macrophages were infected with mycobacteria expressing a kinase-dead-version of PknG which strongly supports the idea that PknG activity is required for mycobacterial survival. A broad-ranged kinase assay screen including 25 kinases representatively chosen from the six major kinase families showed that the inhibitor was highly selective by inhibiting only PknG with high efficiency. Normal cellular processes within host cells were not affected. To elucidate the structural basis of inhibition of PknG and to learn more about the function of the kinase, the structure of PknG in complex with its inhibitor was solved. Since PknG full length was shown to be unstable, limited analysis was performed and a truncated version of PknG missing 8 kDa of the N-terminus was constructed. A purification protocol was established that allowed the purification of pure and stable protein. Crystallization assays were performed to screen for optimal conditions. The structure of dimeric PknG in complex with its specific inhibitor AX20017 was solved at a resolution of 2.4 Å using SIRAS. Three domains of PknG were defined: a) The N-terminal region encompassing the two CXXC motifs, which were shown to complex iron, thus forming a rubredoxin domain b) The kinase domain displaying a closed configuration, with the inhibitor occupying the nucleotide-binding pocket c) and the C-terminal domain with the tetratricopeptide repeat mediating dimerization of the two PknG molecules. The structure explained the high specificity of inhibition by AX20017. In total, 15 polar and non-polar interactions between the inhibitor and the kinase- or N-terminal domain were determined. Sequence alignments with all 518 human kinases derived from six major kinase families revealed that 6 of these interactions were not detected in any other kinase underlining the high specificity of inhibition. By expressing proteins with mutated inhibitor binding sites, these sites of interaction were further confirmed. Moreover, the molecular model of the kinase allowed insights into the regulation of PknG. Site-directed mutagenesis on the four cysteines forming the rubredoxin motif completely abolished PknG activity. This, in combination with the observed interactions between the rubredoxin domain and the lobes of the kinase domain, suggested that PknG activity might be regulated by the N-terminal globular domain. PknG undergoes autophosphorylation, similar to most mycobacterial kinases studied thus far. However, in-depth analysis of PknG autophosphorylation revealed significant differences. Whereas most mycobacterial kinases display a conserved autophosphorylation pattern on the activation loop, phosphorylated residues in PknG were exclusively identified at the Nterminus. For PknG, classified as the unique mycobacterial non-RD kinase due to a missing arginine residue in the catalytic loop, absence of autophosphorylation in the activation loop was suggested. By performing kinase assays using constructs lacking the potential Nterminal phosphorylation sites, this assumption was confirmed. Moreover, by analyzing these mutants on an endogenous substrate, it was demonstrated that autophosphorylation is not a prerequisite for activating the catalytic activity of PknG. Infection experiments with mycobacteria expressing mutated unphosphorylated PknG protein indicated that autophosphorylation is crucial for the role of PknG in preventing lysosomal delivery. Preliminary results of survival assays showed that mycobacteria which express PknG devoid of autophosphorylation sites were rapidly transferred to lysosomes and degraded. To further understand the activity of PknG within host macrophages, PknG was localized in eukaryotic cells. Upon transfection, PknG was detected in the nucleus of HeLa, Mel JuSo and HEK cells. Transfection assays with different truncated PknG constructs point to the Cterminal TPR containing domain of PknG for being responsible for nuclear translocation. In addition, Ag84, a downstream target of PknA, was studied. To address its function, Ag84 was overproduced in mycobacteria. Overexpression resulted in drastic changes in cell morphology, particularly in cell size and shape. Localization studies revealed that wildtype Ag84 localized equally to the cell poles, whereas overexpressed Ag84 was asymmetrically distributed at the poles leading to unbalanced cell wall extension. Furthermore, overexpressed Ag84 was found to affect cell division by preventing septum formation adjacent to the cell pole with higher Ag84 concentrations. Biochemical analysis revealed that Ag84 is a cytosolic protein. As an alternative explanation for the polar localization, oligomerization of Ag84 was analyzed. Chemical crosslinking and size-exclusion chromatography analysis showed that Ag84 is able to form oligomers of >6 molecules which might act as a complex scaffold fitting to the curvature of the cell poles. This complex structure then might allow recruiting other proteins involved in processes such as cell wall synthesis. Ag84 is the first effector protein, for which a role in mycobacterial cell shape control has been described. In addition, Ag84 was found to be involved in septum formation and therefore has a similar role as B. subtilis DivIVA, which in Bacillus however interacts with the MinCD system that is absent in mycobacteria. To conclude, the results presented in this thesis contribute to a better understanding of mycobacterial virulence. The knowledge of the structure of PknG might be particular useful with regard to the design of more potent drugs and the biochemistry data obtained for PknG allowed important insights into function and regulation of this eukaryotic-like serine/threonine protein kinase. Taken together, the results add more information to the complex network of mycobacteria-host interactions.