S. aureus enzymes are shown in red, while isoforms in other organisms are depicted in black (Payne et al., 2001). Initiation of fatty acid biosynthesis requires malonyl-ACP, formed from acetyl-CoA by acetyl-CoA carboxylase (Acc) and malonyl-CoA:ACP transacylase (FabD). The β-ketoacyl-ACP synthase FabH (KAS III) performs the first Claisen condensation yielding a β-ketoacyl-ACP from malonyl-ACP and an acyl-CoA primer. The two carbons originating from malonyl-ACP are shown in cyan and the first two primer carbons in red. Subsequently, the β-ketoacyl-ACP is converted to a saturated acyl-ACP by the actions of the NADPH-dependent reductase FabG, the dehydrase FabZ, and the mostly NADH-dependent enzyme FabI. The saturated acyl-ACP is a substrate for additional rounds of elongation in which FabF catalyzes the condensation reaction. The long-chain acyl-ACP products are subsequently partially transformed into membrane lipids (a typical S. aureus phosphatidylglycerol is shown (Parsons et al., 2011)). Most bacteria use acetyl-CoA as the primer which results in straight-chain fatty acids (C₁₄ and C₁₆) (Mendoza et al., 2002), however, S. aureus FabH prefers branched-chain acyl-CoAs which yield branched-chain fatty acids (Qiu et al., 2005). The branched-chain primers isobutyryl-CoA, isovaleryl-CoA and 2-methylbutyryl-CoA are derived from valine, leucine and isoleucine, respectively, through the actions of the branched-chain aminotransferase BAT (ilvE (Madsen et al., 2002)), and the branched-chain α-ketoacid dehydrogenase BKD (lpd, bkdA1, bkdA2 and bkdB (Singh et al., 2008)). The branched-chain acyl-CoAs yield iso-C₁₄ and iso-C₁₆ (isobutyryl), iso-C₁₅ and iso-C₁₇ (isovaleryl), and anteiso-C₁₅ and anteiso-C₁₇ (2-methylbutyryl) fatty acids (Mendoza et al., 2002). Exogenous fatty acids are converted into acyl-ACPs by the acyl-ACP synthetase (AAS) for incorporation into cell membranes. Typical fatty acids of the human blood plasma are palmitic (21.3%) and linoleic acid (23.7%) (Holman et al., 1995). For S. aureus, exogenous oleic acid was only incorporated in the 1-position of phosphatidylglycerol (Parsons et al., 2011). Inhibitors of the NADPH-dependent FabI from S. aureus (saFabI) used in this study include 5-chloro-2-(2,4-dichlorophenoxy)phenol (triclosan), 5-chloro-2-phenoxyphenol (CPP) and 5-ethyl-2-phenoxyphenol (EPP) (Xu et al., 2008). See also Figure S1.

(A) Homo-tetrameric structure of saFabI. Each subunit of the TCL-2 structure is shown in a different color. The three perpendicular 2 fold non-crystallographic symmetry axes P, Q and R are indicated. The interfaces spanned by the axes P and R (Q and R) are referred to as the PR- (QR-) interface. (B) Structure of the saFabI monomer. α-Helices are shown in red, β-strands in yellow, loops in green, cofactor and inhibitor as space-filling models in grey. The chair-like structure consists of the Rossmann-fold (seat, lower part) and the active site (back, upper part). (C) Flexibility of the saFabI active site in its ligand-free form. Multi-rmsd values were calculated for the C_α atoms of the apo-1 (one subunit), apo-2 (all subunits) and TCL-2 (one subunit) structures using LSQMAN (Kleywegt and Jones, 1996). These values varied from 0.2 to 20.5 Å and are represented by color (from blue to red) and width of the TCL-2 model (residues 2–256). Rmsd-values for the missing residues 199–201 were interpolated in a linear fashion based on the values for the two adjacent amino acids 198 and 202. See also Figure S2.

Secondary structure matching (SSM) was performed for the TCL-2 structure using PDBe Fold (Krissinel and Henrick, 2004). The secondary structure of saFabI is shown in the first line according to the program DSSP (Kabsch and Sander, 1983). Secondary structure elements participating in the QR and PR-interfaces are depicted in red and blue, respectively. Amino acids highlighted in red are conserved. The alignment comprises the 7 best SSM hits (Q-scores are indicated behind the sequences) including bsFabL and saFabG. The first two letters of the protein names indicate the bacterial source (sa = S. aureus, ba = Bacillus anthracis, bc = B. cereus, bs = B. subtilis, ft = Francisella tularensis, tt = Thermus thermophilus, ec = Escherichia coli, aa = Aquifex aeolicus). PDB codes and subunits are given in parentheses. Blue, red and cyan stars mark residues interacting with the cofactor, inhibitor or cofactor and inhibitor, respectively. Disordered or rearranged regions in all four available saFabI apo structures (apo-1, apo-2, 3GNS, 3GNT) are marked with a continuous line, regions which are additionally disordered or rearranged for distinct apo structures are represented by a dashed line. This Figure was prepared with ESPript (Gouet et al., 1999).

(A) Dose-response curves of diphenyl ether analogues. The steady-state reaction velocities (v_s) at various inhibitor (CPP, EPP, triclosan) concentrations are plotted as a fraction of the uninhibited reaction velocity (v_u). Each plot is fitted to the standard binding isotherm. R² = 0.88, 0.75 and 0.73 for the best fit curves to the binding isotherm for CPP, EPP and triclosan, respectively. (B) Schematic representation of all residues interacting with triclosan (TCL in red). The interaction pattern was derived from the TCL-2 structure. Directional interactions are depicted explicitly (black residues), amino acids creating the hydrophobic pocket are included in 1-letter code (residues in blue boxes). R = adenosine diphosphate. (C) Fit of triclosan in its binding pocket. Triclosan is shown as space filling model in grey. Ala95, Phe96, Ala97, Leu102, Tyr147, Tyr157, Met160, Ile207 and NADP⁺ are shown in surface representation in blue. Additionally interacting residues Ser197, Ala198, Val201 and Phe204 are located within the substrate-binding loop (SBL). (D) Comparison of the TCL and CPP inhibitor binding modes. The TCL-2 structure is shown in grey and yellow, the CPP structure in blue. The cofactor and some protein regions are omitted for clarity. Arrows indicate differences between the two structures. The B-ring is rotated by ~11 ± 2° upon removal of the two chlorine atoms. Simultaneously, the C_α atoms of Ala97, Leu102, Ser197 and Ala198 move towards the inhibitor by about 0.4 ± 0.1 Å, 0.3 ± 0.1 Å, 0.3 ± 0.1 Å and 0.2 ± 0.1 Å, respectively, and the side chain of Met160 rotates away from the inhibitor by ~13 ± 3°.

Panels A to F show consecutive conformations of saFabI proposed to be adopted during cofactor and inhibitor binding. The three regions undergoing the most extensive conformational changes are highlighted (SBL-2 in cyan, ASL in dark blue and the C-terminally extended SBL in red as in Figure 3). One subunit of TCL-2 is continuously shown as a reference in yellow. (A) Superposition of the apo-1 and TCL-2 structures. Apo-1 is shown in light blue. (B–E) Superposition of the apo-2 and TCL-2 structures. Subunits B, C, A and D of apo-2 are shown in light blue. (F) Final ternary complex conformation. The hydrogen bonded residues Arg194 and Asn205 can act as a hinge for SBL-1 closure and are displayed in a larger scale in the black box (Maity et al., 2011). (G) Peptide flip mechanism and SBL-2 closure observed upon superposition of TCL-2 (yellow) with apo-2 (light blue, subunit A). The carbonyl oxygen of Ala95 flips between Ser121 in the apo structure and a water molecule bound to Ser197 in the ternary complex (a rotated view is displayed in the insert). See also Figure S4 and Movie S1.

From left to right the QR-interface in the saFabI crystal structures 3GNT, 3GNS (Priyadarshi et al., 2010), apo-1, apo-2 and TCL-2 is shown to illustrate tetramer association and dissociation (view as in Figure 2A). The two associating dimers are shown in grey and yellow with the respective QR-interface residues (according to the PDBe PISA server (Krissinel, 2010)) in cyan and red. (A) Front view of the dimer - tetramer transition. (B) Side view of the dimer - tetramer transition. (C) Close up view of the upper QR-interface half. Buried surface areas of the monomeric QR-interface (BSA_QR) as well as the numbers of the corresponding hydrogen bonds (N_HB) and salt bridges (N_SB) are given below according to PISA. See also Figures S3 and S5 as well as Movie S2.

(A) The 2′-phosphate binding motif. NADP⁺ is shown in grey, saFabI (CPP structure) in yellow. Additional parts of the protein are omitted for clarity. (B) Superposition of saFabI and ecFabI. SaFabI is shown in yellow, ecFabI (1QSG:A) in green. Asn41 is shifted 2.8 ± 0.3 Å towards the solvent compared to Lys41 preventing 2′-phosphate stabilization by an Asn41 backbone interaction.