Bacterial Enoyl-Reductases: The Ever-Growing List of Fabs, Their Mechanisms and Inhibition

Abstract

Enoyl-ACP reductases (ENRs) are enzymes that catalyze the last step of the elongation cycle during fatty acid synthesis. In recent years, new bacterial ENR types were discovered, some of them with structures and mechanisms that differ from the canonical bacterial FabI enzymes. Here, we briefly review the diversity of structural and catalytic properties of the canonical FabI and the new FabK, FabV, FabL, and novel ENRs identified in a soil metagenome study. We also highlight recent efforts to use the newly discovered Fabs as targets for drug development and consider the complex evolutionary history of this diverse set of bacterial ENRs.

Keywords: FabI; FabK; FabL2; FabV; Fabl; drug targets; fatty acid synthesis; kinetic mechanisms.

Figure 1

Bacterial fatty acid biosynthesis. There are two types of fatty acid (FA) biosynthesis in nature, FAS-I and FAS-II. In both pathways, fatty acid chains are elongated through iterative cycles in which two-carbon units are added after each cycle. Most bacteria rely exclusively on FAS-II pathway for FA synthesis, some members of the Corynebacteriales order, on the other hand, rely exclusively on FAS-I, while mycobacteria and most corynebacteria (mycolic acid-producing organisms) possess FAS-I and FAS-II pathways. In both systems, there are four sequential steps in each elongation cycle: (1) condensation of acyl and malonyl groups, (2) β-ketoacyl-ACP reduction, (3) β-hydroxyacyl-ACP dehydration, and (4) trans-2-enoyl-ACP reduction, leading to an extended acyl chain containing 2 additional carbon units. In FAS-I, a single megasynthase performs the four sequential steps of chain elongation, while in FAS-II different enzyme components perform each step. The first committed step in FA synthesis is the carboxylation of acetyl-CoA by the enzyme acetyl-CoA carboxylase (AccABCD), leading to the production of malonyl-CoA from acetyl-CoA. During the elongation steps of preformed acyl chains, malonyl-CoA units are the malonyl donor groups in the FAS-I condensation step (ketoacyl synthase, KS), while in FAS-II it is first converted to malonyl-ACP by FabD, which is then used by FabB (or FabF) as the malonyl donor group in the condensation step of FAS-II elongation. In FAS-I, chain initiation requires the condensation of acetyl-CoA and malonyl-CoA by the KS activity of the megasynthase. For bacteria relying exclusively on FAS-II for FA synthesis, the initiation step requires the production of acetoacetyl-ACP (a β-ketoacyl-ACP species) from the condensation of malonyl-ACP and acetyl-CoA performed by FabH. When both FAS-I and FAS-II systems are available, as in mycobacteria, preformed acyl-CoA chains containing from C16-C24 carbon units derived from the FAS-I system are shuttled into FAS-II by an initial condensation step with malonyl-ACP catalyzed by mFabH. Bacterial enoyl-ACP reductases (ENRs) catalyze the last step of trans-2-enoyl-ACP reduction of the FAS-II cycle. ACP, Acyl Carrier Protein; KR, β-ketoacyl-ACP reductase activity; DH, β-hydroxyacyl-ACP dehydrogenase activity; ER, trans-2-enoyl-ACP reductase activity. Created with BioRender.com.

Figure 2

The dinucleotide-binding Rossmann fold. (A) A representative βαβ-motif and its topological representation. An α-helix (purple) connects two parallel β-strands (yellow). (B) Topological representation of the Rossmann fold. The βαβ-motifs form two sets of βαβαβ units that are connected by a long α-helix that function as a crossroad (in blue). (C) Representative parallel β-sheet of a Rossmann fold. The α-helix connecting the two sets of βαβαβ units through strands 3 and 4 is represented in blue. The NAD cofactor is also represented to indicate that the cofactor-binding site lies above this central parallel β-sheet. (D) Structure of InhA (PDB: 1ENY). The central parallel β-sheet is flanked by two sets of α-helices on each side. (E) InhA (PDB: 1ENY) with the central parallel β-sheet in lateral view. The active site motif Yx₂(x)Mx₃K and the cofactor-binding glycine-rich motif (Gx₅SxA) are shown in green and blue, respectively. (F) Detailed view of NAD, the active site and Glycine-rich motifs. The NAD cofactor, the first glycine from the Gx₅SxA motif (G14), catalytic tyrosine (Y158) and lysine (K165) from the signature motif Yx₂(x)Mx₃K are represented as Ball & Stick atomic models. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 3

Multiple alignment of multiple ENRs from the SDR superfamily. Amino acids are colored based on their Turn Propensity, where darker shades of red correspond to a higher propensity. E. coli: Escherichia coli, P. aeruginosa: Pseudomonas aeruginosa, B. subtilis: Bacillus subtilis, Serratia rubidaea: S. rubidaea, Rhizobium meliloti: R. meliloti, Vibrio cholerae: V. cholerae, S. lithotrophicus: Sideroxydans lithotrophicus, F. amnicola: Ferriphaselus amnicola, Mycobacterium tuberculosis: M. tuberculosis. Each sequence was retrieved from the SwissProt database, except for the FabMG, which were retrieved from Kim et al. (2020). Multiple alignment was performed using Clustal Omega, and manually adjusted on JalView (Waterhouse et al., 2009).

Figure 4

FabI reaction mechanism. (A) Reduction of trans-2-acyl-ACP (enoyl-ACP) to acyl-ACP by FabI enzymes. A reduced form of the dinucleotide cofactor (NADH or NADPH, depending on the enzyme) serves as the reductant species. In the first half-reaction, bacterial FabI enzymes catalyze the hydride transfer from the 4S hydrogen position of the nicotinamide ring (represented in the figure) of the cofactor to the C3 position (Cβ) of the α,β-unsaturated thioester of the enoyl-ACP substrate. An enolate intermediate is formed with the concomitant oxidation of NADH to NAD⁺. In the second half-reaction, the enolate intermediate is protonated, leading to the formation of the acyl-ACP product. (B) Covalent ene adduct intermediate between NADH and enoyl-ACP substrate. In this covalent intermediate species, there is a covalent bond between the C2 atom of the NADH nicotinamide ring and the C2 atom of the acyl substrate. Two alternate possibilities for hydride transfer are depicted. The hydride transfer could be the result of a pericyclic ene reaction from a transition state that leads to the C2-ene adduct or could be the result of a direct hydride transfer. In the latter, C2-ene adduct would be formed by Michael addition from the intermediate enolate and NAD⁺. Created with BioRender.com.

Figure 5

InhA bound to NAD and fatty acyl substrate. (A) Structure of the ternary complex of Mycobacterium tuberculosis InhA bound to NAD⁺ and a C16 fatty acyl substrate (PDB: 1BVR). Both cofactor-binding site and fatty acyl binding site lie within the same structural substrate binding pocket (SBP; beige surface). The substrate-binding loop (SBL) of mycobacterial FabIs like M. tuberculosis InhA (residues 196–219, in blue) are longer than their bacterial orthologues, presumably to accommodate longer fatty acyl substrates. The fatty acyl substrate can be viewed in its U-shaped conformation. The Rossmann fold is also represented. (B) View of the substrate binding pocket from the major portal. The U-shaped fatty acyl chain lies just ahead of the cofactor. (C) Detailed view of the catalytic triad (F149, Y158 and K165), together with T196, fatty acyl substrate and cofactor. (D) Hydrogen bond interactions among the catalytic Y158 and K165, the fatty acyl substrate and the NAD cofactor. (E) Lateral view of the substrate binding pocket of InhA displaying a network of water molecules inside a water channel. The residue F149 presumably gates the access of this water channel to the active site of InhA. This network of water molecules was proposed to be implicated in the protonation of the enolate intermediate. The minor portal is also indicated. (F) Surface view of InhA with the external part of the hydrogen-bonded network of water molecules inside the water channel. Binding pocket was identified using CASTp 3.0 (Tian et al., 2018). Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 6

Cofactor-binding pockets of ENR FabIs. (A) Tetrameric structure of E. coli FabI (EcFabI, PDB: 1DFI) as a prototypical tetrameric ENR FabI. Three protomers are represented in surface view and the fourth in cartoon representation. (B) Cofactor-binding site of Staphylococcus aureus FabI (SaFabI) bound to NADP (PDB: 3GR6). The positively charged R40 and K41 residues (in red) that interact with the phosphate bound to the adenosine ribose of NADP are highlighted. (C) Cofactor-binding site of EcFabI bound to NAD⁺ (PDB: 1DFI). The polar residue Q40 is found in the place of the positively charged residues required to accommodate the additional phosphate of NADP. (D) Cofactor-binding site of Mycobacterium tuberculosis InhA bound to NADH (PDB: 1ENY). The nonpolar F41 is found in the same approximate position as Q40 in EcFabI and K41 and R40 in SaFabI. (B–D) The Glycine-rich loop is highlighted in dark blue. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 7

Competitive inhibitors of FabI—Triclosan and PT70. (A) Structure of 5-chloro-2-(2,4-dichlorophenoxy)phenol (Triclosan—TCL) and the TCL derivative 2-(o-Tolyloxy)-5-hexyphenol (PT70). (B) Ternary complex of EcFabI bound to TCL and NAD (PDB:1QSG). Hydrogen bonds between the ether oxygen and the chlorophenol ring hydroxyl oxygen with the 2′-oxygen from the nicotinamide ribose of NAD are highlighted. (C) Ternary complex of InhA bound to NAD⁺ and PT70 (PDB: 2×23). Hydrogen bonds between the inhibitory compound and the catalytic Y158 and K165 residues are highlighted. This compound also makes hydrogen bonds with the 2′-nicotinamide ribose oxygen of NAD⁺. (D) Superposition of two InhA structures highlighting to possible conformations for F149 and Y158. In both cases, the in conformation is the one in which the residue points toward the active site, as opposed to the out conformation. In one structure (PDB: 4D0S) the F149 is in the in conformation, while Y158 is in out conformation. The second structure (PDB: 3FNH) the situation is the opposite (F149 out, Y158 in). Note that the configuration (F149 in, Y158 in) is not allowed due to steric hindrance. (E) Structural flexibility of the substrate-binding loop (SBL) helix 6 (residues 196–206) from InhA. This region is found in variable conformations in different structures, but they can be grouped in four types, each of them represented by one structure in the picture: disordered (4TRM), widely opened (1P44), open (2AQ8) and closed (3FNH). Potent slow-onset inhibitors usually induce the ordering of the SBL H6 helix into the closed conformation. The superposition of structures was performed using the Matchmaker tool from UCSF ChimeraX. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 8

Adduct forming compounds and a bisubstrate inhibitor. (A) Structure of a diazaborine derivative (thienodiazaborine—TDB). (B) Ternary complex of EcFabI bound to NAD⁺ and TDB (PDB: 1DFH). The TDB propyl moiety turns back in a conformation reminiscent to a scorpion’s tail (Baldock et al., 1996). The covalent bond between TDB and NAD⁺ is indicated. (C) Structure of the INH-NADH adduct. Only the nicotinamide portion of NADH is represented. (D) Comparison of the complex of InhA bound to the INH-NAD adduct (PDB: 1ZID) with the binary complex of InhA with NAD⁺ (PDB: 2AQ8). Both structures almost completely overlap, except for the conformation of F149, which rotates from the out conformation in the binary complex InhA-NAD⁺ to the in conformation in the enzyme complex with INH-NADH. This rotation is required to accommodate the isonicotinoyl (INH) moiety of the INH-NADH adduct. (E) Structure of the pyridomycin bisubstrate inhibitor. The portions that occupies the fatty acyl chain binding site and the NAD⁺ binding site are indicated. (F) Binary complex of InhA bound to pyridomycin (PDB: 4BII). The superposition of structures was performed using the Matchmaker tool from UCSF ChimeraX. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 9

FabV structure. (A) Structure of Yersinia pestis FabV (YpFabV—PDB: 3ZU3). The bound cofactor NADH and the eight-stranded Rossmann fold shared by FabVs are indicated. (B) Superposition of available FabV structures: Y. pestis YpFabV (PDB: 3ZU3), Xanthomonas oryzae XoFabV (PDB: 3S8M), Burkholderia pseudomallei BpFabV (PDB:4BKO) and Vibrio fischeri VfFabV (PDB:5XI0). (C) Superposition of YpFabV (PDB: 3ZU3) with EcFabI (PDB: 1C14). Structure of EcFabI is depicted in purple while structure of YpFabV is in yellow. Some structural elements of FabV ENRs not found in FabIs are indicated. (D) Structural comparison between the FabV extended active site signature motif Yx₈K (from YpFabV—PDB: 3ZU3) and the Yx₆K (Yx₂Mx₃K) motif of FabIs (from EcFabI—PDB: 1C14) reveals that the position of the catalytic lysine is almost identical, while the position of the catalytic tyrosine is very similar, in particular its hydroxyl oxygen atom. The superposition of structures was performed using the Matchmaker tool from UCSF ChimeraX. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 10

FabL structure: general view and interaction with triclosan. (A) Structure of Bacillus subtilis FabL (BsFabL—PDB: 3OID) bound with NADP. (B) Superposition of EcFabI (PDB: 1C14) with BsFabL (PDB: 3OID). (C) Superposition of the ternary complex of BsFabL bound to NADP and triclosan (TCL; PDB: 3OID) with the apo BsFabL (PDB: 3OIC). (D) and (E) The position of both catalytic tyrosine and lysine reveals extensive conformational changes in the active site upon complex formation. (D) Position of the catalytic tyrosine (Y151 in BsFabL, in red) in the apo enzyme (PDB: 3OIC) and bound to TCL and NADP (PDB: 3OID). (E) Position of the catalytic lysine (K158 in BsFabL, in red) in the apo enzyme (PDB: 3OIC) and bound to TCL and NADP (PDB: 3OID). (F) Hydrogen bonds between NADP, TCL and the catalytic residues (Y151 and K158, PDB: 3OID). The superposition of structures was performed using the Matchmaker tool from UCSF ChimeraX. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 11

FabMG structure. (A) Structure of FabMG identified from a TCL-resistant clone in a soil metagenomics study (PDB: 6KIA). A bound NADH cofactor and the Rossmann fold are indicated. (B) Superposition of FabMG (represented in yellow, PDB: 6KIA) with EcFabI (in purple, PDB: 1C14). The superposition of structures was performed using the Matchmaker tool from UCSF ChimeraX. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 12

FabK structure. (A) Structure of Streptococcus pneumoniae FabK (SpFabK, PDB: 2Z6I). The LID and the TIM barrel fold are indicated. The latter, together with a bound flavin mononucleotide (FMN) cofactor, are represented in more detail. (B) Superposition of Porphyromonas gingivalis FabK (PgFabK, PDB: 4IQL, represented in pink) with Thermotoga maritima FabK (TmFabK, PDB: 5GVH, represented in cyan) and S. pneumoniae FabK (SpFabK, PDB: 2Z6I, represented in purple). (C) Superposition of SpFabK bound to FMN cofactor (PDB: 2Z6I) with the ternary structure of SpFabK bound to FMN and a phenylimidazole derivative inhibitor (compound 1; PDB: 2Z6J). The orientation of a putative catalytic histidine residue (H144) rotates upon inhibitor binding. The superposition of structures was performed using the Matchmaker tool from UCSF ChimeraX. Created with UCSF ChimeraX (Pettersen et al., 2021) and BioRender.com.

Figure 13

Unrooted phylogenetic tree inferred from the sequences of ENRs from the classes FabK, FabL, FabL2, FabI, and FabV. Branch lengths denote amino acid substitutions for that node. The letters behind a “|” symbol refer to their respective ENR class. The Uniprot identifiers for each one of the species present in the tree are listed on Supplementary Table 1. Sequences were downloaded from Uniprot and aligned using the multiple alignment tool ClustalW (Larkin et al., 2007). The resulting phylip alignment file was used as input for RAxML (Stamatakis, 2014) and the best substitution model was chosen automatically. Trees were visualized and customized in the webserver of iTOL (Letunic and Bork, 2021).