From chlorite dismutase towards HemQ - the role of the proximal H-bonding network in haeme binding

Abstract

Chlorite dismutase (Cld) and HemQ are structurally and phylogenetically closely related haeme enzymes differing fundamentally in their enzymatic properties. Clds are able to convert chlorite into chloride and dioxygen, whereas HemQ is proposed to be involved in the haeme b synthesis of Gram-positive bacteria. A striking difference between these protein families concerns the proximal haeme cavity architecture. The pronounced H-bonding network in Cld, which includes the proximal ligand histidine and fully conserved glutamate and lysine residues, is missing in HemQ. In order to understand the functional consequences of this clearly evident difference, specific hydrogen bonds in Cld from 'Candidatus Nitrospira defluvii' (NdCld) were disrupted by mutagenesis. The resulting variants (E210A and K141E) were analysed by a broad set of spectroscopic (UV-vis, EPR and resonance Raman), calorimetric and kinetic methods. It is demonstrated that the haeme cavity architecture in these protein families is very susceptible to modification at the proximal site. The observed consequences of such structural variations include a significant decrease in thermal stability and also affinity between haeme b and the protein, a partial collapse of the distal cavity accompanied by an increased percentage of low-spin state for the E210A variant, lowered enzymatic activity concomitant with higher susceptibility to self-inactivation. The high-spin (HS) ligand fluoride is shown to exhibit a stabilizing effect and partially restore wild-type Cld structure and function. The data are discussed with respect to known structure-function relationships of Clds and the proposed function of HemQ as a coprohaeme decarboxylase in the last step of haeme biosynthesis in Firmicutes and Actinobacteria.

Keywords: H-bonding network; HemQ; chlorite dismutase; electron paramagnetic resonance spectroscopy; haeme binding; resonance Raman spectroscopy.

Figure 1. Representative X-ray structures showing the distal (A) and proximal (B) haeme side in Clds (green with red haeme) and Cld-like proteins (yellow)

Since Cld-like structures were solved without the co-factor, haemes (in grey) from Clds were inserted for orientation. Hydrogen bonds are presented as black dotted lines, other distances as grey dotted lines. Structures are depicted of Clds from D. aromatica (pdb-code: 3Q08), ‘Candidatus Nitrospira defluvii’ (pdb-code: 3NN1) and N. winogradskyi (pdb-code: 3QPI), as well as Cld-like structures from L. monocytogenes (pdb-code: 4WWS), T. thermophilus (pdb-code: 1VDH) and Thermoplasma acidophilum (pdb-code: 3DTZ).

Figure 2. DSC of wild-type NdCld and the variants E210A and K141E

Ligand free thermograms are compared with those in the presence of 10 mM thiocyanate (SCN⁻) or 200 mM fluoride (F⁻). Thermal transitions were fitted to non-two state equilibrium-unfolding models by the Lavenberg/Marquardt (LM) non-linear least squares method and fits are depicted as grey lines.

Figure 3. Spectroscopic signatures of ferric wild-type NdCld and the variants E210A and K141E

(A) UV–visible absorption, (B) EPR and (C) RR spectra (λ_excitation=413.1 nm, laser power at the sample 5 mW, average of 2 spectra with 300 s integration time). Conditions: 50 mM MES, pH 5.5.

Figure 4. UV–vis, EPR and RR spectra of ferric wild-type NdCld and the variants E210A and K141E bound to fluoride

(A) UV–visible absorption spectra. (B) Low-temperature EPR spectra. (C) RR spectra (λ_excitation=413.1 nm, laser power at the sample 1.5 mW, average of 2 spectra with 40 s integration time). Conditions: 50 mM MES, pH 5.5.

Figure 5. UV–vis and EPR spectra of ferric wild-type NdCld and the variants E210A and K141E bound to thiocyanate

(A) UV–visible absorption spectra. (B) Low-temperature EPR spectra. Conditions: 50 mM MES, pH 5.5. The EPR spectra in the region 200–500 mT have been expanded 5-fold.

Figure 6. Resonance Raman spectra of ferrous NdCld and variants

(A) Low frequency resonance Raman spectra of ferrous wild-type NdCld and the variants E210A and K141E. Excitation wavelength: 441.6 nm, laser power at the sample 11 mW, average of three spectra (wild-type) and two spectra (variants) with 600 s integration time. (B) Low and high frequency resonance Raman spectra of the wild-type NdCld and the variant E210A complexes with CO and ¹³CO. Difference spectra are depicted in red. Excitation wavelength: 413.1 nm, laser power at the sample 1 mW, spectra in the low frequency region average of 3 spectra with 600 s integration time, in the high frequency region average of 6 spectra (wild-type, CO), and 12 spectra (wild-type, ¹³CO) with 240 s integration time, 6 spectra (NdCld E210A, CO) and 5 spectra (NdCld E210A, ¹³CO) with 600 s integration time.

Figure 7. Steady-state kinetics of chlorite degradation by wild-type NdCld (black) and the variants E210A (white) and K141E (striped)

(A) Total concentration of released dioxygen. (B) Initial velocities of O₂ generation. Measurements were performed in 50 mM MES buffer, pH 5.5 using 50 nM of enzyme and 175 μM chlorite as substrate.

Figure 8. Transient-state kinetics of chlorite degradation by wild-type NdCld and the variants E210A and K141E

500 nM of enzyme were mixed with 200 μM chlorite (final concentration) in the absence or presence of 10 mM thiocyanate or 200 mM fluoride. Spectra from the following time points are shown: 0 (black), 1 (cyan), 51, 101, 221, 446, 866, 1646, 3099, 5803, 10838 (grey) and 20 s (red).

Figure 9. Haeme transfer from wild-type NdCld and variants to apo-myoglobin

HPLC chromatograms (from bottom to top) of standard proteins, holo-myoglobin, apo-myoglobin and NdCld samples (wild-type, K141E, E210A, R173Q/E210A) in the absence and presence of apo-myoglobin. Elution profiles followed at 280 and 409 nm (Soret absorbance) are depicted in black and grey respectively.