A target-protection mechanism of antibiotic resistance at atomic resolution: insights into FusB-type fusidic acid resistance

Antibiotic resistance in clinically important bacteria can be mediated by proteins that physically associate with the drug target and act to protect it from the inhibitory effects of an antibiotic. We present here the first detailed structural characterization of such a target protection mechanism mediated through a protein-protein interaction, revealing the architecture of the complex formed between the FusB fusidic acid resistance protein and the drug target (EF-G) it acts to protect. Binding of FusB to EF-G induces conformational and dynamic changes in the latter, shedding light on the molecular mechanism of fusidic acid resistance.

All proteins were purified by nickel affinity chromatography followed by overnight dialysis and gel filtration at 4 ºC as previously described. 1 Protein samples were prepared in 20 mM trisHCl, 300 mM NaCl, 1 mM DTT, 10 %D 2 O, pH 8.0 except for PRE samples for which the DTT was omitted.

NMR Spectroscopy
Backbone assignments of EF-G C3 were determined from analysis of TROSY-HNCA, HNCO, HN(CO)CA, HN(CA)CO, HNCACB and HN(CO)CACB spectra. For assignment of EF-G C3 bound to FusB, spectra were limited to the HNCO, HNCA and HN(CO)CA coupled to comparison with the apo state assignments due to technical limitations caused by the molecular size and the limited sample concentration available, higher concentrations promoting aggregation. These assignment spectra were supplemented using selectively unlabelled 3 samples in which 15 N EF-G C3 was enriched with a single non-isotopically enhanced amino acid (alanine, asparagine, lysine, valine and phenylalanine). Assignments of FusB bound to EF-G C3 were transferred from the apo spectra by visual inspection with the use of spectra of selectively unlabelled lysine, leucine, phenylalanine, valine and asparagine 15 N FusB bound to EF-G C3 .
3 1 H-15 N chemical shift perturbation analysis for EF-G C3 was performed using only residues assigned in both the apo and FusB bound spectra. Conservative chemical shift perturbation analysis to compare EF-G C3 and EF-G was performed by finding the closest peak in the EF-G spectrum to the assigned peaks in the EF-G C3 spectrum. 4 For FusB binding to EF-G C3 , analysis was performed as for EF-G C3 for residues assigned in both spectra and then using the conservative closest peak method for the remainder of the apo FusB assignments. The chemical shift change was calculated using the

Computational approaches
Where PALES fitting of RDC data indicated realignment of elements of the protein structure in the bound state, the crystal structure was refined to better fit the RDC data using Xplor-NIH. 6 To maintain a compact protein structure, solvent PRE restraints were included using the method of Wang et. al.. 7 For realignment of the domain IV helices, the structure of all three domains excepting the helices in question and the loops joining them to the remainder of the structure was fixed while the helices were allowed to move as a rigid body. Docking of the structure of EF-G C3 with realigned domains IV and V and that of FusB with realigned domains was performed using HADDOCK. 8 Interaction surfaces were defined by ambiguous interaction restraints (AIRs) determined from those residues showing significant chemical shift perturbation on binding that were solvent exposed in the crystal structures. Orientational information for the two proteins was provided by the inclusion of RDCs from EF-G C3 domains IV and V and full length FusB. NOE style distance restraints were included using intermolecular PRE data, with residues showing an I ox /I red greater than 0.9 defined as 25 Å or greater from the MTSL tag.
Residues with an I ox /I red less than 0.1 were defined as 15 Å or less from the MTSL tag. The numbers of each distance restraint included in structure refinement and docking are shown in supplementary  show a better fit to the observed data than those calculated from the crystal structure.