Three-dimensional structure of a type III glutamine synthetase by single particle reconstruction
Van Rooyen, Jason Macrae
MetadataShow full item record
This study represents the first structural investigation of any type III glutamine synthetase (GS). The GS, GlnA, from the medically important opportunistic human pathogen Bacteroides fragilis was studied with a view to better understanding its structure/functioning in relation to the extensively characterised GSIs. GSIIIs are the most recently discovered family of GSs and are the most phylogenitically distant GSs from the GSIs. Images (160) of negatively stained rGlnA, expressed in E.coli YMC11 (glnA-), were recorded at 50K magnification using a Leo 912 operating at 120kV with energy filtering coupled to a 4 megapixel CCD camera. An angular refinement based reconstruction strategy was adopted using SPIDER. A reconstruction based on an ab initio starting model, derived by a common-lines based simultaneous minimization of rotationally invariant K-mean clustered class averages, converged to the same structure as a reconstruction based on a GSI starting model to a resolution of 2.1nm as determined by Fourier shell correlation). In contrast to preliminary EM observations, which identified GlnA as a hexamer, this work has revealed a dodecameric structure, with subunits (82.8KDa) arranged in two opposing hexagonal rings with distinct handedness. This is similar to the quaternary structure of GSIs and GlnTs except that the complex is 50% longer and the two rings are not symmetrically related. They differ not only in diameter (16.5 or 15.0nm) but also the degree of separation of subunits and as such the particle possesses only C6 and not D6 symmetry. The finding that particles lie in a preferred orientation, with the larger ring in contact with the carbon support, accounted for this asymmetry, through partial staining. Hexameric views, with similar overall arrangement but larger size in comparison to GSI, were also observed. However, it was uncertain whether these were true hexamers resulting from dissociation of the dodecamers or were a consequence of partial staining. Homology modelling was also undertaken in an attempt to predict the structure of GlnA based on GSI, with a view to interpreting the low resolution EM structure. Due to the failures of state of the art algorithms in detecting the distant homologies between GS families, manual profile-based alignment strategies, incorporating structural information, were employed. Through the first full length alignments of GS sequences from all four families, conservation of all active site residues, core active site αβ barrel fold motifs, and additional previously unreported regions was demonstrated. Docking of these homology models into the 3D structure confirmed the presence of the αβ barrel fold predicted by the bioinformatic analysis of the sequences alone, thus, identifying the indentations between subunits in the volume as putative active sites. In addition to providing unequivocal proof that GlnA is a GS and confirming the presence of putative αβ barrel active site folds, this work has made steps towards understanding the regulation of this enzyme. It is hypothesised that GlnA occurs as both active hexamer and an inactive dodecamer, the interconversion of which, is thought to represent a means of reversible post-translational regulation.