Fig 1.
Overview of glutathione-dependent formaldehyde metabolism.
(A) Summary scheme of glutathione-dependent formaldehyde metabolism. Formaldehyde (HCHO) reacts with glutathione (GSH) via its nucleophilic thiol group to form S-hydroxymethylglutathione (HMG), which is a substrate of glutathione-dependent alcohol dehydrogenase (ADH, ADH5 in humans). The product, S-formylglutathione, is then further metabolised by S-formylglutathione hydrolase to give formate and GSH. The reaction of HCHO and GSH, i.e. the first step in GSH-dependent metabolism, occurs spontaneously in aqueous solution; however, the reaction might also be catalysed by GFA (and homologues in other organisms, e.g. CENPV in humans[13]). There is also evidence, at least in vitro, that GSH can react with HCHO to form cyclised adducts[10–12]. (B) Views of X-ray crystal structures of GFA from Paracoccus denitrificans (PDB IDs: 1X6M and 1XA8[14]). The GFA domain contains two zinc binding sites; one zinc ion is coordinated by four cysteinyl thiols (C31, C33, C99 and C102) in a tetrahedral geometry, whereas the other zinc ion is coordinated by three cysteinyl thiols (C52, C54 and C57) in a trigonal planar geometry. Crystallographic studies have proposed that GSH binding induces translocation of the second zinc ion (circles)[14].
Fig 2.
GFA is a GSH-binding protein that induces an increase in EXSY correlation intensities between GSH and HMG, but does not catalyse HMG formation / fragmentation.
(A) 2D EXSY spectra of equilibrium mixtures of GSH (initial concentration 15 mM) and HMG in the absence (left) and presence (right) of GFA (20 μM). Mixing time (τm) = 400 ms. EXSY-correlation intensities between GSH and HMG are increased in the presence of GFA. (B) 1D EXSY spectra of equilibrium mixtures of GSH (initial concentration 15 mM) and HMG in the presence of GFA (20 μM) conducted at different mixing times (τm = 32–300 ms). Irradiation (inversion) of a β-cysteinyl resonance of HMG (δH 2.95 ppm) induced an exchange correlation at δH 2.87 ppm, corresponding to the β-cysteinyl resonance of GSH, which increased in intensity at longer mixing times. (C) Graph showing the intensity of the GSH cross-peak relative to the inverted HMG resonance in the absence and presence of GFA at different mixing times, using either NOESY or ROESY pulse sequences. τm = 4–400 ms. (D) Bar graph showing the intensity of the GSH 1D EXSY-correlation relative to the irradiated HMG resonance in the absence (blue) and presence (green) of GFA (τm = 80 ms). The build-up rates of the 1D EXSY analyses (note: a τm of 80 ms is within the linear range of the EXSY build-up curves, Fig 2C) correlate with the rates of GSH/HMG exchange at equilibrium. Therefore, the observed increase in correlation intensity in the presence of GFA implies an increase in GSH/HMG inter-conversion rate. (E) Non-denaturing MS analyses of GSH binding to GFA. Two new peaks corresponding to the masses of monomeric GFA (with two zinc ions in complex) bound to one and two GSH molecules respectively were observed upon incubation with GSH (4 equivalents, right). (F) Binding curve of GSH binding to GFA obtained using waterLOGSY. Selective irradiation of the solvent H2O 1H resonance results in magnetisation transfer to GSH, resulting in the emergence of GSH 1H resonances with opposite sign to the irradiated H2O resonance. The (negative) intensities of the GSH resonances are linearly dependent on the GSH concentration (blue). Addition of GFA results in a slower net tumbling rate for GSH in solution due to binding with GFA. The slower tumbling rate leads to ‘(more) positive’ GSH resonance intensities as a function of the extent of ligand binding (green). Subtraction of the intensities in the absence (blue) and presence (green) of GFA gives a normalised binding curve (orange, KD value of roughly 500 μM assuming binding of one GSH molecule per GFA subunit). The experiments were carried out at 280 K. τm = 1 s. (G) Graph showing production of HMG from mixtures of GSH (13.3 mM) and HCHO (13.3 mM) in the absence (blue) and presence (green) of GFA (16 μM) in BisTris buffer pH 6.0. GFA does not affect the initial HMG formation rate.
Fig 3.
GFA does not affect HCHO metabolism in E. coli cell lysate.
(A) 1H NMR (top) and 1D-13C-HSQC spectra (bottom) of E. coli BL21 (DE3) cell lysate (0.5 mg/mL in 50 mM Tris buffer pH 7.5) incubated with [13C]-HCHO (6 mM). Resonances in the 1D-HSQC spectra are annotated. [13C]-satellites for the Tris buffer are observed in the 1D-HSQC spectrum due to its high abundance of in the mixture. The doublet resonance at δH 3.25 ppm is assigned to methylamino groups from the lysate. Inset top left and bottom right: 1H NMR spectra of E. coli BL21 (DE3) cell lysate (0.5 mg/mL in 50 mM Tris buffer pH 7.5) incubated with [13C]-HCHO (6 mM) over time. Production of 13C-formate (as indicated by the increase in intensity of the doublet resonance at δH 8.36 ppm, red top left) and [13C]-methanol (as indicated by the increase in intensity of the resonance at δH 3.36 ppm, red bottom right) are clearly observed. Only one half of the expected doublet resonance is observed for [13C]-methanol in the 1H spectra due to overlap with the Tris buffer. Note: the relatively high initial [13C]-methanol level (relative to the level of [13C]-formate) is due to contamination of the commercial source of [13C]-HCHO (S15 Fig). (B) 1H NMR spectra of E. coli BL21 (DE3) cell lysate (0.5 mg/mL in 50 mM Tris buffer pH 7.5) incubated with GSH (4 mM) and [13C]-HCHO (6 mM) after 5 min (blue) and 45 min (red) respectively. [13C]-HMG (as indicated by the resonances at δH 2.96 ppm and δH 3.09 ppm) was most abundant over early time points and decreased during the experiment, which correlated with time-dependent formation of [13C]-formate, [13C]-methanol and GSH. Resonances corresponding to [13C]-HMG were poorly resolved in the time-course experiments without added GSH due to overlap with other resonances (see S16 Fig). (C) GFA does not affect the rate of formate production in E. coli cell lysate. Initial [13C]-formate production rate is dependent on the concentration of added [13C]-HCHO over the tested range; however, addition of recombinant GFA (20 μM) to the lysate before incubation with 1 mM [13C]-HCHO (green) does not affect the rate of [13C]-formate production (relative to the control without added enzyme, blue).