An α-ketoglutarate conformational switch controls iron accessibility, activation, and substrate selection of the human FTO protein

Abstract

The fat mass and obesity-associated fatso (FTO) protein is a member of the Alkb family of dioxygenases and catalyzes oxidative demethylation of N⁶-methyladenosine (m⁶A), N¹-methyladenosine (m¹A), 3-methylthymine (m³T), and 3-methyluracil (m³U) in single-stranded nucleic acids. It is well established that the catalytic activity of FTO proceeds via two coupled reactions. The first reaction involves decarboxylation of alpha-ketoglutarate (αKG) and formation of an oxyferryl species. In the second reaction, the oxyferryl intermediate oxidizes the methylated nucleic acid to reestablish Fe(II) and the canonical base. However, it remains unclear how binding of the nucleic acid activates the αKG decarboxylation reaction and why FTO demethylates different methyl modifications at different rates. Here, we investigate the interaction of FTO with 5-mer DNA oligos incorporating the m⁶A, m¹A, or m³T modifications using solution NMR, molecular dynamics (MD) simulations, and enzymatic assays. We show that binding of the nucleic acid to FTO activates a two-state conformational equilibrium in the αKG cosubstrate that modulates the O₂ accessibility of the Fe(II) catalyst. Notably, the substrates that provide better stabilization to the αKG conformation in which Fe(II) is exposed to O₂ are demethylated more efficiently by FTO. These results indicate that i) binding of the methylated nucleic acid is required to expose the catalytic metal to O₂ and activate the αKG decarboxylation reaction, and ii) the measured turnover of the demethylation reaction (which is an ensemble average over the entire sample) depends on the ability of the methylated base to favor the Fe(II) state accessible to O₂.

Keywords: Alkb dioxygenases; N6-methyladenosine; NMR; liquid chromatography–mass spectrometry; molecular dynamics simulations.

Fig. 1.

FTO catalytic mechanism. (A) FTO catalyzes the oxidative demethylation of nucleic acids via two coupled reactions. The Fe(IV)═O species formed by the secondary reaction oxidizes the methylated base. (B) Methyl modifications reported to be substrates of FTO. Methyl modifications are highlighted in red.

Fig. 2.

FTO enzymatic activity. (A) LC dimension of the LC–MS data acquired on 10 µM samples of GG(m⁶A)CT (Top), GG(m¹A)CT (Middle), and GG(m³T)CT (Bottom) after a 4-h incubation with 2 µM FTO. The elution profiles for the methylated substrate and demethylated product are shown as dashed and solid lines, respectively. Peak shoulders that appear at high elution times are an artifact of high pH and can be eliminated by a pH neutralization step preceding the LC/MS experiment. (B) Michaelis–Menten kinetics for the DNA demethylation reaction catalyzed by FTO. Experimental data are shown as circles. Modeled curves are shown as solid lines. Data for the m⁶A, m¹A, and m³T modifications are colored red, orange, and green, respectively.

Fig. 3.

Solution NMR analysis of FTO-DNA binding. Close-up views of the (A) Ile and (B) Met regions of the methyl-TROSY NMR spectrum showing the effect of the secondary and primary substrate on the NMR resonances of Ile 269, Ile 276, and Met 297. Signals from apo FTO are colored black. Signals from the FTO/αKG complex are colored blue. Signals from the FTO complexes with αKG and the DNA oligos containing m⁶A, m¹A, and m³T are colored red, orange, and green, respectively. (C) Chemical shift perturbation (CSP) versus residue index induced upon the addition of αKG to apo FTO (Top; blue bars), or upon the addition of methylated DNA oligos to FTO/αKG (Bottom; red, orange, and green bars correspond to the CSP induced by m⁶A, m¹A, and m³T, respectively). (D) Model of FTO bound to αKG (cyan and red sticks), Fe(II) (orange sphere), and the m⁶A containing 5-mer DNA oligo (orange, green, and blue cartoon). FTO is shown as cartoon (white for the N-ter domain and light blue for the C-ter domain). The side chains of Trp 270 and Arg 316 are shown as blue sticks. Ile 269 and 276 are shown as salmon sticks. The β-strand connecting Trp 270 to Ile 276 is colored salmon. (E) Scatter plot showing the positive correlation between k_cat and the CSP measured for Ile 269 (Top) and Ile 276 (Bottom). The dashed line shows the linear regression of the experimental data assuming that k_cat = 0 s⁻¹ at CSP = 0 ppm.

Fig. 4.

The off-line to in-line transition activates FTO. (A) Postulated path for formation of the catalytically competent oxyferryl species. (B) Close-up view of the active site from the crystal structure of FTO in complex with Mn(II) (orange sphere) and m⁶A (red and blue sticks). The off-line (cyan) and in-line (purple) αKG molecules were modeled according to crystal structures of Alkbh5 (PDB code: 7WKV) and Alkbh2 (PDB code: 3RZJ), respectively, in complex with the primary and secondary substrates. The side chains of His 231, Asp 233, and His 307 are shown as gray, red, and blue sticks. (C) The m⁶A (Left; red sticks), m¹A (Center; orange sticks), and m³T (Right; green sticks) bases are displayed in the FTO active site. Off-line and in-line αKG molecules are colored cyan and purple, respectively. The side chains of Trp 270 and Arg 316 are shown as blue sticks. Ile 269 and 276 are shown as salmon sticks. The β-strand connecting Trp 270 to Ile 276 is colored salmon. The position of the methyl modification is highlighted with a sphere in panels (B) and (C).

Fig. 5.

The off-line/in-line equilibrium is modulated by the primary substrate. (A) Definition of the pseudo-dihedral angles used to describe the αKG dynamics in the FTO active site. 2D probability plots showing the distribution of the (B) H307-αKG and H231-αKG angles, (C) bb1 and bb2 angles, and (D) bb2 and R316-αKG angles during the 2 µs MD simulations ran on the complexes formed by FTO with primary and secondary substrates. (E) Distribution of the methyl modification (C atom) to Fe(II) distance in the 2 µs MD simulations ran on the complexes formed by FTO with primary and secondary substrates. The positive charge on m¹A causes the base to adopt larger distances from the Fe center and to rotate away from the catalytic site for a portion of simulation 5. (F) Distribution of the O₂ exposed surface area (%ESA) of the Fe(II) atom in the 2 µs MD simulations ran on the complexes formed by FTO with primary and secondary substrates. Simulations 1 to 3 (first three panels Left to Right in B–E) started from the models of FTO in complex with m⁶A, m¹A, and m³T, respectively, and αKG in the off-line configuration. Simulations 4 to 6 (last three panels Left to Right in B–E) started from the model of FTO in complex with m⁶A, m¹A, and m³T, respectively, and αKG in the in-line configuration.

Fig. 6.

NMR evidence for an off-line/in-line equilibrium. Close-up views of (A) 1D ¹³C NMR spectra and (B) 2D ¹H-¹³C HSQC spectra of αKG free (black), bound to FTO (blue), and bound to the FTO complexes with DNA oligos containing m³T (green), m¹A (orange), and m⁶A (red). Labels on each peak specify the resonance assignment. The signal in the ¹³C spectrum of αKG bound to the FTO/m⁶A complex was assigned to a contaminant because too sharp to belong to the bound small molecule or FTO. Full spectra are shown in SI Appendix, Fig. S6. (C) Exchange contribution to the ¹³C_methyl relaxation rate (R_ex) measured for FTO via RD cpmg experiments. Data measured for apo FTO are colored black. Data measured for the FTO/αKG complex are colored blue. Data measured for the FTO complexes with αKG and the DNA oligos containing m⁶A, m¹A, and m³T are colored red, orange, and green, respectively. The blue, cyan, and red boxes highlight different groups of residues experiencing dynamics on the µs-ms timescale. Full relaxation dispersion curves are shown in SI Appendix, Fig. S7. (D) Model of FTO bound to αKG (green sticks), Fe(II) (orange sphere), and the m⁶A containing 5-mer DNA oligo (orange cartoon) showing the residues experiencing µs-ms timescale as blue, cyan, or red sticks. FTO is shown as cartoon (white for the N-ter domain and light blue for the C-ter domain).