Mycobacterium tuberculosis FasR senses long fatty acyl-CoA through a tunnel and a hydrophobic transmission spine

Abstract

Mycobacterium tuberculosis is a pathogen with a unique cell envelope including very long fatty acids, implicated in bacterial resistance and host immune modulation. FasR is a TetR-like transcriptional activator that plays a central role in sensing mycobacterial long-chain fatty acids and regulating lipid biosynthesis. Here we disclose crystal structures of M. tuberculosis FasR in complex with acyl effector ligands and with DNA, uncovering its molecular sensory and switching mechanisms. A long tunnel traverses the entire effector-binding domain, enabling long fatty acyl effectors to bind. Only when the tunnel is entirely occupied, the protein dimer adopts a rigid configuration with its DNA-binding domains in an open state, leading to DNA dissociation. The protein-folding hydrophobic core connects the two domains, and is completed into a continuous spine when the effector binds. Such a transmission spine is conserved in a large number of TetR-like regulators, offering insight into effector-triggered allosteric functional control.

Fig. 1. Crystal structure of FasR_Δ33-C₂₀-CoA.

a FasR dimer in cartoon representation, the two protomers are distinguished in light blue and pale green. The dimer is strictly symmetric, with one protomer in the asymmetric unit, the crystallographic twofold is depicted as a vertical axis in the plane of the figure. Secondary structure elements are labelled; a prime symbol denotes equivalent elements in the other protomer. The two domains are indicated, an N-terminal DNA-binding domain with a helix–turn-helix (HTH) motif, and a C-terminal domain with the co-crystallised effector molecule C₂₀-CoA bound (in spheres coloured by element). b Details of the effector-binding tunnel. Three orthogonal views illustrate the molecular surface of FasR_Δ33.The leftmost cuts along the middle of a protomer, revealing the top tunnel mouth on the right and a large segment of the tunnel itself with the bound acyl moiety (in sticks). A few amino acid side chains that define the tunnel walls are labelled. The rightmost panel is an open-book perspective, with the footprint of protomer A visible on protomer B’s surface (in pale green), the two openings of the tunnel are visible, showing solvent exposure of both tips of the C₂₀-CoA molecule.

Fig. 2. FasR_Δ33-C₁₄ exhibits larger protein flexibility including dimer asymmetry.

a SigmaA-weighted mF_obs-DF_calc difference Fourier map contoured at 3.5σ (green mesh) calculated with no ligand bound in the effector-binding tunnel during FasR_Δ33-C₁₄ refinement. The FasR_Δ33-C₁₄ model is shown with grey cartoons, and residues at ≤4 Å from the ligand are depicted with thin sticks. The electron density allowed to model a myristic acid, overlaid within the difference map in thick sticks coloured by atom. Note polar and charged residues on the top opening, close to the carboxyl head of the ligand. Hydrophobic residues line up the rest of the tunnel’s walls. b FasR_Δ33-C₁₄ protomers A and B are coloured as in Fig. 1. The dimer is asymmetric, quantitated by superimposing protomer B onto A (red arrow in the top panel), after what the effector-binding domains fit quite well while the HTH domains show significant rotation between one another (bottom panel), with max root-mean-square deviation (rmsd) on helices α2 and α3. c The FasR_Δ33-C₂₀-CoA and FasR_Δ33-C₁₄ structures are superposed, resulting in maximal fit on the central 4-helix bundle that mediates dimerisation (helices α8–α9 from both protomers). The upper portion of the effector-binding domains end up well aligned, but significant shifts affect the lower part together and the HTH domains (dashed arrows). The distance between the centres of mass of Tyr₇₇ on helix α3, differs by more than 10 Å between both structures. The HTH domains move consolidated with the bottom portion of juxtaposed helices α4 and α7 from the effector-binding domains.

Fig. 3. Structure-guided mutations in FasR leading to permanent DNA-binding.

a Selected mutations to occlude the effector-binding tunnel are illustrated. The crystal structure of FasR_Δ33-C₂₀-CoA (left panel) is compared to a model of the triple mutant FasR_LVL (right panel). Val₁₆₃ is actually behind the tunnel (red arrow). The substitutions by bulky phenylalanines anticipate steric hindrance of the acyl chain (shown as sticks coloured by element). b Electrophoretic mobility shift assay was performed by incubating the ³²P-labelled 398 bp Pfas promoter region with wild-type FasR and mutant FasR proteins (FasR_LVL and FasR_L106F) in the absence or in the presence of C₂₀-CoA and C₁₆-CoA. Protein–DNA complexes were separated by electrophoresis on a 6% polyacrylamide gel. The assay was performed in duplicate, producing similar results. Source data are provided as a Source Data file. c Selected mutations to uncouple allosteric signalling from the ligand-binding pocket to the DNA-binding domain. The bottom mouth of the effector-binding tunnel is located towards the centre of the molecule, revealing the tip of C₂₀ (in ball and stick representation, coloured by atom). A few side chains are labelled and shown as orange sticks, chosen among residues bordering the tunnel opening. Two bulky residues (Leu₉₈ on helix α4 and Phe₁₂₃ on α5) that were substituted by alanines, are labelled with bold fonts. d Gel shift assays of the wild-type and selected FasR mutants. The DNA probe that corresponds to the 398 bp Pfas promoter region (Pfas_MT) was ³²P-labelled and incubated either with FasR wild-type, FasR_L98A or FasR_F123A, in the absence or in the presence of the indicated concentrations of C₁₆-CoA and C₂₀-CoA. The assays were performed in duplicate, producing similar results. Source Data are provided as a Source data file.

Fig. 4. FasR can accommodate longer than C₂₀ acyl chains.

A representative 10 ns trajectory of all-atom molecular dynamics is illustrated by showing in cartoon representation the simulated model of FasR_Δ33 in complex with C₂₆-CoA every 250 ps (see Supplementary Movie 1). Shown as superposed ensemble, the secondary structure elements of both domains are preserved, and the cerotic acyl chain (in sticks) also remains stable within the tunnel. The inset shows a projection of a cut section through protomer’s B tunnel, to display the dynamic stability of the dimeric interface and the acyl chain (as opposed to the large flexibility of the nucleotide portion of coenzyme-A). Free volume is available at the interprotomer space, predicting that even longer acyl moieties should be able to accommodate. See Supplementary Fig. 9 for rmsd values within and between domains, comparing also the simulated behaviour of apo-FasR.

Fig. 5. Crystal structure of full-length FasR in complex with DNA.

Electron density map (sigmaA-weighted 2mF_obs-DF_calc Fourier) is shown (magenta mesh), overlaid on the final refined model, showing the protein dimer (protomers in light blue and pale green) and the DNA double helix (orange) in cartoon representation. The map was carved around the atomic model with a border of 3 Å to improve the clarity. Helices α3 are seen fitting within the DNA major groove as expected.

Fig. 6. A hydrophobic spine connects input effector-sensing to output DNA-binding.

a Hydrophobic-spine amino acids in FasR. The colour code depicts residues of the DNA-binding domain in orange, the effector-binding domain in grey and the α6–α7 loop in cyan. b–d distinct TFRs (FasR, RutR, RamR) illustrate the conservation of the hydrophobic spine (residues shown as molecular surface), connecting effector- to DNA-binding domains. The colour code is identical to panel a. The effector molecules in atom-coloured sticks are labelled, with transparent spheres overlaid (insets show their markedly disparate structures). See Supplementary Fig. 13 for additional examples.

Fig. 7. FasR-mediated long acyl-CoA sensing and response mechanism.

The model combines information from the three crystal structures presented in this report (FasR_Δ33-C₁₄, FasR_Δ33-C₂₀-CoA and FasR–DNA complexes). A free structure of FasR with no bound effectors nor DNA has not been determined experimentally, and is not currently known whether it builds up to detectable concentrations within the living cell. The dotted arrows reflect plausible equilibria. See Supplementary Movie 3 to better grasp the anticipated dynamics.