EPR spectroscopic and computational characterization of the hydroxyethylidene-thiamine pyrophosphate radical intermediate of pyruvate:ferredoxin oxidoreductase

Abstract

The radical intermediate of pyruvate:ferredoxin oxidoreductase (PFOR) from Moorella thermoacetica was characterized using electron paramagnetic resonance (EPR) spectroscopy at X-band and D-band microwave frequencies. EPR spectra, obtained with various combinations of isotopically labeled substrate (pyruvate) and coenzyme (thiamine pyrophosphate (TPP)), were analyzed by spectral simulations. Parameters obtained from the simulations were compared with those predicted from electronic structure calculations on various radical structures. The g-values and 14N/15N-hyperfine splittings obtained from the spectra are consistent with a planar, hydroxyethylidene-thiamine pyrophosphate (HE-TPP) pi-radical, in which spin is delocalized onto the thiazolium sulfur and nitrogen atoms. The 1H-hyperfine splittings from the methyl group of pyruvate and the 13C-hyperfine splittings from C2 of both pyruvate and TPP are consistent with a model in which the pyruvate-derived oxygen atom of the HE-TPP radical forms a hydrogen bond. The hyperfine splitting constants and g-values are not compatible with those predicted for a nonplanar, sigma/n-type cation radical.

Figure 1

Stack-plot of EPR spectra obtained from solutions of PFOR incubated with A) ¹⁵N3-TPP and [3″-²H₃]pyruvate, B) unlabeled TPP and [3″-²H₃]pyruvate, C) unlabeled TPP and pyruvate, D) unlabeled TPP and [3-¹³C]pyruvate, E) unlabeled TPP and [2- ¹³C]pyruvate, and F) ¹³C2-TPP and unlabeled pyruvate. The experimental parameters were: temperature 80 K; receiver gain, 5 ×10³; modulation frequency, 100 kHz; modulation amplitude, 0.40 G; center field, 3400 G; sweep width, 100 G; microwave power, 0.4 mW; and number of scans, 32-64.

Figure 2

Comparison of the experimental and simulated high-field EPR spectra of solutions of PFOR incubated with A) [3-¹⁵N]-TPP and [3-²H₃]pyruvate and B) unlabeled TPP and [3-²H₃]pyruvate. Spectral simulations of B) were performed with and without inclusion of a contaminating free radical signal at a spin concentration of 15% that of the HE-TPP radical (upper and lower traces, respectively). The contaminant had a g-value of 2.0032 and a line width 4× greater than that of the HE-TPP radical. The experimental parameters for the D-band two-pulse echo-detected spectra were as follows: EPR frequency 129.9997 GHz; pulse widths, 50 ns; time between pulses, 120 ns; rep rate, 300 Hz; averages per data point, 300; temperature 50 K. The following parameters were used to simulate the spectra: g-tensor (2.0079; 2.0053; 2.0021); A) ¹⁵N hyperfine splitting tensor (|A_⊥| = 2G, | A_‖| = 18G); B) ¹⁴N hyperfine splitting tensor (|A_⊥| = 1G, | A_‖| = 13G). In A) and B), A_⊥ was collinear with the g_x,y and A_‖ was collinear with g_z. A line width of 3.5 G and Gaussian line shape were used.

Figure 3

Comparison of the experimental and simulated X-band EPR spectra of solutions of PFOR prepared with A) ¹⁵N3-TPP and [3-²H₃]pyruvate and B) unlabeled TPP and [3-²H₃]pyruvate. Spectral simulations of B) were performed with and without inclusion of a contaminating free radical signal at a spin concentration of 15% that of the HE-TPP radical (upper and lower traces, respectively). The contaminant had a g-value of 2.0032 and a line width 4× greater than that of the HE-TPP radical. The resolution of the experimental EPR spectra in A) and B) was enhanced using cosine window functions with a line width of 3G and a truncation parameter of 0.5G^-1. In addition to the g tensor and ^14,15N hyperfine splitting tensors (see legend of Figure 2) two isotropic doublet splittings (|A_iso| = 5G, |A_iso| = 3G) were included. A linewidth of 1.3 G and Gaussian line shape were employed in the simulation.

Figure 4

Comparison of the experimental and simulated X-band EPR spectra of solutions of PFOR prepared with unlabeled TPP and A) unlabeled pyruvate and B) [3-¹³C]pyruvate. The resolution of the experimental EPR spectra in A) and B) was enhanced using cosine window functions with a line width of 2G and a truncation parameter of 0.5G^-1. The simulations included the parameters listed in the legend of Figure 3. In A) and B), two isotropic, doublet hyperfine splittings (|A_iso| = 10G, |A_iso| = 5G) were included. In B) an isotropic doublet ¹³C-hyperfine splitting was included (|A_iso| = 5G).

Figure 5

Comparison of the experimental and simulated X-band EPR spectra of solutions of PFOR prepared with A) unlabeled TPP and [2- ¹³C]pyruvate, and B) ¹³C2-TPP and unlabeled pyruvate. The resolution of the experimental EPR spectra in A) and B) was enhanced using cosine window functions with a line width of 3G or 2G, respectively, and a truncation parameter of 0.5G^-1. There is a small amount of contaminating signal evident in the EPR spectrum of B). However, attempts were not made to incorporate the contaminant into the simulations. The simulations included the parameters listed in the legend of Figure 3 and those in the legend of Figure 4A. In addition, ¹³C hyperfine splitting tensors: A) (|A_⊥| = 6 G, |A_‖| = 18 G) ; B) (|A_⊥| = 4 G, |A_‖| = 9 G) were included in the simulations. A_⊥ is collinear with g_xy and A_‖ is collinear with g_z.

Figure 6

Isosurface plots of the spin densities of different truncation models of the HE-TPP radical structure: [1] geometry-optimized protonated HE-TPP π-radical, [2] geometry-optimized unprotonated HE-TPP π-radical, [3] geometry-optimized HE-TPP σ/n-type cation radical, [4] geometry-optimized HE-TPP σ/n-type cation radical C5-C4′ tautomer, [1′] protonated HE-TPP π-radical from X-ray coordinates (PDB code 1KEK), [2′] unprotonated HE-TPP π-radical from X-ray coordinates, [3′] HE-TPP σ/n-type cation radical from X-ray coordinates (11), [4′] HE-TPP σ/n-type cation radical C5-C4′ tautomer from X-ray coordinates (15). Surface plots were generated with the program Molekel 4.3 (38) from the corresponding Gaussian log files using a cut-off of 0.005. Gray clouds represent positive spin density, and blue clouds represent negative spin density.

Figure 7

Overlays of heavy atoms of geometry-optimized models of the HE-TPP radical (red) with X-ray coordinates (blue) from PDB accession code 1KEK (11): A) Protonated π–radical (model 1), RMSD 0.4 Å; B) Neutral π–radical (model 2), RMSD 0.4Å; C) σ/n-type cation radical (model 3) (11), RMSD 1.1Å; D) C5-C4′ tautomer of σ/n-type cation radical (model 4) (15), RMSD 0.8Å.

Scheme 1

Models used in electronic structure calculations.