Crystallographic and kinetic study of riboflavin synthase from Brucella abortus, a chemotherapeutic target with an enhanced intrinsic flexibility

Abstract

Riboflavin synthase (RS) catalyzes the last step of riboflavin biosynthesis in microorganisms and plants, which corresponds to the dismutation of two molecules of 6,7-dimethyl-8-ribityllumazine to yield one molecule of riboflavin and one molecule of 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione. Owing to the absence of this enzyme in animals and the fact that most pathogenic bacteria show a strict dependence on riboflavin biosynthesis, RS has been proposed as a potential target for antimicrobial drug development. Eubacterial, fungal and plant RSs assemble as homotrimers lacking C3 symmetry. Each monomer can bind two substrate molecules, yet there is only one active site for the whole enzyme, which is located at the interface between two neighbouring chains. This work reports the crystallographic structure of RS from the pathogenic bacterium Brucella abortus (the aetiological agent of the disease brucellosis) in its apo form, in complex with riboflavin and in complex with two different product analogues, being the first time that the structure of an intact RS trimer with bound ligands has been solved. These crystal models support the hypothesis of enhanced flexibility in the particle and also highlight the role of the ligands in assembling the unique active site. Kinetic and binding studies were also performed to complement these findings. The structural and biochemical information generated may be useful for the rational design of novel RS inhibitors with antimicrobial activity.

Keywords: 6,7-dimethyl-8-ribityllumazine; enzyme–ligand complex; inhibition by substrate and product; vitamin B2.

Figure 1

The main catalytic steps involved in the biosynthesis of riboflavin. 1, 5-Amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione; 2, 3,4-dihydroxy-2-butanone 4-phosphate; 3, 6,7-dimethyl-8-ribityllumazine; 4, riboflavin; DHBPS, 3,4-dihydroxy-2-butanone 4-phosphate synthase; LS, 6,7-dimethyl-8-ribityllumazine synthase; RS, riboflavin synthase (reaction highlighted on a green background).

Figure 2

(a) The RS-APO structure. Each chain is depicted in a different colour. N- and C-terminal barrels are indicated for each chain in their respective colours. The C-terminal three-helix bundle can clearly be seen at the centre of the molecule. An arrow marks the approximate location of the unique active site of the enzyme. (b) Structure of the RS monomer depicted in rainbow colours. The major structural elements are identified. The orientation highlights the pseudo-C ₂ symmetry observed between the N- and C-terminal barrels. A black oval capped with a tilde marks the approximate location of the pseudo-twofold rotation axis, which is perpendicular to the plane of the paper. The monomer depicted here corresponds to chain A from the RS-NRP complex, which bears two copies of the NRP ligand, as explained in the text. (c) Superposition of the N- and C-terminal barrels. Shown are residues 1–91 (N-terminal barrel, pink) and 102–188 (C-terminal barrel, cyan) belonging to chain A from the RS-NRP complex. Secondary-structure elements and terminal residues are indicated. Bound ligands are also depicted. (d) Interactions between chains in the C-terminal three-helix bundle. The side chains of the interacting residues are represented as sticks. Hydrogen bonds are marked with dotted lines. The colours for each chain are the same as in (a), but the orientation of the figure has been shifted slightly for clarity.

Figure 3

Sequence alignment of the N- and C-terminal barrels. Secondary-structure elements are indicated. Identical and similar residue pairs are highlighted in dark and light orange, respectively. Residues in direct contact with the ligand molecules, as seen in the complex structures solved in this work, are marked with a star. Gly99 and His101, which belong to the central loop bridging both β-barrels and are not present in the alignment, are also involved in ligand binding.

Figure 4

Superposition of the RS monomers from B. abortus (chain A from RS-APO, green), E. coli (chain A from PDB entry 1i8d, pink) and S. pombe (unique chain from PDB entry 1kzl, yellow). N- and C-termini are indicated, as well as the main structural elements. (a) Front view. (b) Top view.

Figure 5

Sequence alignment of RSs with known three-dimensional structure. Colours and features are similar to those in Fig. 3 ▶. The location of the secondary-structure elements in the picture refers to the B. abortus enzyme.

Figure 6

Superposition of the RS-RBF, RS-ROS and RS-NRP structures (green) with RS-APO (yellow). Only chains B and C are used for the structural alignment in order to stress the displacement observed in chain A from RS-APO (approximately 7°). The relative orientation of chains B and C is fairly similar in the four structures. The calculated r.m.s.d. values with respect to RS-APO for these two chains together (366 superimposed C^α atoms) are only 0.73, 0.92 and 0.70 Å for RS-RBF, RS-ROS and RS-NRP, respectively. Bound ligands are not included for clarity.

Figure 7

Representation of the ligands found at the substrate-binding sites. The left panels highlight the locations of the different molecules observed, together with some of the most representative residues involved in binding (drawn in sticks). Monomers A and B are depicted in green and pink, respectively. 2mF _o − DF _c Fourier maps are represented at the 1.0σ level. The right panels summarize the hydrogen-bond interactions between the individual ligands and the RS enzyme, indicating residue numbers and atoms, distances (in Å, with a maximum value of 3.5) and ligand atom numbers. (a) RS-RBF. (b) RS-ROS (roseoflavin molecule bound at the N-terminal barrel from chain A). (c) RS-ROS (roseoflavin molecule bound at the C-terminal barrel from chain A). (d) RS-NRP [pair of NRP molecules bound at the N-terminal barrel from chain A (i) and at the C-terminal barrel from chain B (ii)]. Absolute contour levels for the electron-density maps (in e Å⁻³) are as follows: 0.196 (a), 0.288 (b), 0.274 (c) and 0.284 (d).

Figure 8

Cartoon representation of the complex RS structures highlighting the different ligand-binding stoichiometries in each case. Each monomer is depicted as a violet drop. N- and C-terminal barrels are identified inside the monomers with the letters N and C. Individual chains are labelled outside the drops. Ligands are represented with their aromatic groups in orange (riboflavin), pink (roseoflavin) and yellow (NRP). Ribityl moieties are represented as tails. (a) RS-RBF. (b) RS-ROS. (c) RS-NRP.

Figure 9

Modelling of 6,7-dimethyl-8-ribityllumazine molecules into the active site of RS-NRP (a) and RS-APO (b). Monomers are depicted in green (chain A) and pink (chain B). Substrate molecules are identified with the letters a and d, which correspond to the acceptor and donor molecules, respectively. Dashed lines represent the distances between the C6α methyl group from the acceptor and the C7 atom from the donor (top) and between the C7α methyl group from the acceptor and the C6 atom from the donor (bottom). Substrate molecules were placed by superposition taking into account the location of the aromatic systems and the ribityl moieties of the ligands in RS-NRP. For (b), the individual subunits from RS-APO were superimposed on the RS-NRP structure first.

Figure 10

Kinetic study of RS. The initial velocities of riboflavin formation (V _i) for different 6,7-dimethyl-8-ribityllumazine concentrations are represented by a solid line. The inset highlights the substrate inhibition observed at high substrate concentrations. For comparison, the curve representing the classic Michaelis–Menten model is drawn as a dashed line.

Figure 11

Kinetic scheme used for the calculations in the substrate-inhibition model. E, enzyme; S, substrate; P, product (riboflavin).