Solution structures of the Bacillus cereus metallo-β-lactamase BcII and its complex with the broad spectrum inhibitor R-thiomandelic acid

Metallo-β-lactamases, enzymes which inactivate β-lactam antibiotics, are of increasing biological and clinical significance as a source of antibiotic resistance in pathogenic bacteria. In the present study we describe the high-resolution solution NMR structures of the Bacillus cereus metallo-β-lactamase BcII and of its complex with R-thiomandelic acid, a broad-spectrum inhibitor of metallo-β-lactamases. This is the first reported solution structure of any metallo-β-lactamase. There are differences between the solution structure of the free enzyme and previously reported crystal structures in the loops flanking the active site, which are important for substrate and inhibitor binding and catalysis. The binding of R-thiomandelic acid and the roles of active-site residues are defined in detail. Changes in the enzyme structure upon inhibitor binding clarify the role of the mobile β3–β4 loop. Comparisons with other metallo-β-lactamases highlight the roles of individual amino-acid residues in the active site and the β3–β4 loop in inhibitor binding and provide information on the basis of structure–activity relationships among metallo-β-lactamase inhibitors.

to N, CA and C in the case of amino acids) which ensure the connectivity of the new residue with the generic linker residues was essential since thiomandelate is not an amino acid derivative.
A connecting system such as the one used in the creation of proxy residues [2] was used successfully.

Simulated annealing
It has been reported, in the case of structural calculations performed with the program ARIA, that an increase in the number of simulated annealing steps can prove to be highly beneficial in achieving convergence, especially large proteins where chemical shift degeneracy is a significant issue [3,4]. In the present study we found that the use of extended torsion angle dynamics (TAD) steps (15 times the normal) can indeed be beneficial in terms of convergence and target function values. However, these calculations tended to become biased towards similar conformations, as reflected by the unusually low RMSD values observed even in some regions expected to be unstructured. This suggests that extended TAD steps can be valuable, providing converged and essentially correct structures and reducing exhaustive peak list analysis, but must be used with caution.

Dihedral angle constraints
We found that too general a use of dihedral angle constraints derived from the TALOS database may introduce erroneous features in regions of unusual or complex backbone structure distribution, for example around the zinc co-ordination sites. Consequently, the structures reported in the present study were calculated both with and without dihedral angle constraints and we focus on the structures obtained without these constraints. Table 1 in the main text gives the structural statistics for the structures calculated without dihedral angle constraints. Table S2 gives these statistics for structures calculated with these constraints. The structures calculated with and without dihedral angle constraints had very similar RMSD values (0.35-0.38 for backbone atoms) and similar Ramachandran statistics (>98 % of residues in the core and allowed regions), but the structures calculated with dihedral angle constraints showed significantly more distance and angle violations.   The bundles of structures shown were determined solely from the NOE-based constraints, before the introduction of zinc restraints in refinement. Structures are shown for the free enzyme (slate blue) and the complex (orange). The positions of the protein ligands after refinement are shown in blue (free enzyme) and red (complex) and the positions of the zinc atoms as blue (free enzyme) and orange (complex) spheres.

Figure S5 Conformations of Lys 171 and Asn 180 in the BcII-thiomandelate complex
The side chains of Lys 171 and Asn 180 are shown in yellow and thiomandelate in red. The constraint network defining the positions of Lys 171 and Asn 180 is depicted with red lines and the side chains of the residues involved in these constraints are shown in blue. The distances between the NZ atom of Lys 171 and the oxygen atoms of the carboxylate group of the inhibitor and the distances between the HD21 and HB3 atoms of the Asn 180 and the QR pseudo atom of the aromatic ring of the inhibitor are indicated. The zinc atoms are shown as cyan spheres and the protein backbone is coloured grey.

Table S1 Metal co-ordination constraints used (lower and upper limits) and corresponding distances on the X-ray structure (PDB code 1BVT) and both NMR structures
Bundles of 20 NMR structures were used for the distance calculations. RTM, R-thiomandelate.