Impact of subunit and oligomeric structure on the thermal and conformational stability of chlorite dismutases

Abstract

Chlorite dismutases (Cld) are unique heme b containing oxidoreductases that convert chlorite to chloride and dioxygen. Recent phylogenetic and structural analyses demonstrated that these metalloproteins significantly differ in oligomeric and subunit structure. Here we have analyzed two representatives of two phylogenetically separated lineages, namely pentameric Cld from Candidatus “Nitrospira defluvii” and dimeric Cld from Nitrobacter winogradskyi having a similar enzymatic activity at room temperature. By application of a broad set of techniques including differential scanning calorimetry, electronic circular dichroism, UV–vis and fluorescence spectroscopy the temperature-mediated and chemical unfolding of both recombinant proteins were analyzed. Significant differences in thermal and conformational stability are reported. The pentameric enzyme is very stable between pH 3 and 10 (T_m = 92 °C at pH 7.0) and active at high temperatures thus being an interesting candidate for bioremediation of chlorite. By contrast the dimeric protein starts to unfold already at 53 °C. The observed unfolding pathways are discussed with respect to the known subunit structure and subunit interaction.

Highlights

► Comparison of two chlorite dismutases of different subunit and oligomeric structure but similar enzymatic activity. ► Significant differences in conformational and thermal stability due to different subunit interactions. ► Pentameric (canonical) chlorite dismutase exhibits a high thermal stability and enzymatic activity at high temperatures. ► Presented thermodynamic data are representative for chlorite dismutases of two distinct phylogenetic lineages.

1. Introduction

Chlorite dismutases (Clds) are oligomeric heme b-containing oxidoreductases found in prokaryotic organisms. These oxidoreductases (EC 1.13.11.49) are able to catalyze the conversion of toxic chlorite (ClO₂⁻) to chloride and dioxygen. In the last years several X-ray structures of multimeric (di-, penta-, hexa-) Clds and Cld-like proteins from archaea and bacteria have been published [1–5] that helped to critically analyze and complement mechanistic aspects of catalysis [6–10]. The proposed enzyme mechanism includes oxidation of the ferric heme protein to an oxoiron(IV) porphyryl radical intermediate (Compound I) by chlorite which is reduced to hypochlorite. The latter must be captured at the active site since it is rapidly oxidized by Compound I forming a new O―O bond, thereby releasing dioxygen (O₂), chloride (Cl^-) and the enzyme in its native ferric state [8–10]. Both atoms in the oxygen gas product originate entirely from the chlorite substrate [8–10]. The catalytic efficiency (k_cat/K_M) for this reaction varies from 6.0 × 10⁵ M^− 1 s^− 1 to 3.5 × 10⁷ M^− 1 s^− 1 [5].

The physiological role of Clds in prokaryotes is not fully understood, but it has been shown that some microorganisms can use perchlorate or chlorate as terminal electron acceptors for anaerobic respiration thereby producing chlorite that must be detoxified [11]. However, recent phylogenetic analyses showed that many other bacterial and archaeal genomes encode Cld-like proteins, although most of the respective organisms have never been observed to use (per)chlorate or convert chlorite [5,12]. Since the transformation of chlorite to harmless O₂ and Cl⁻ is mediated by Clds with high efficiency, there is increasing interest for using these catalysts for bioremediation [11,12]. Due to its oxidative nature, chlorite reacts with organic material and thus has toxic effects on living cells [13]. Perchlorate, chlorate and chlorite are brought to nature by anthropogenic activities (munition manufacturing, rocket fuel, fertilizer, bleaching agents, disinfectants, pesticides, etc.). They are a serious environmental concern since rising concentrations of these harmful compounds have been detected in groundwater, surface waters, and soils [11]. In order to use Clds for bioremediation it is important to know the conformational and thermal stability of these proteins as well as the temperature and pH-dependence of the enzymatic reaction. Since in addition to pentameric Clds [1–4] also active dimeric forms [5] are found, it is also important to know the correlation between the oligomeric structure and the stability of Clds in order to select the proper candidates for future application and molecular engineering.

In this work we have analyzed the conformational and thermal stability of recombinant dimeric Cld from Nitrobacter winogradskyi (NwCld) and of pentameric Cld from Candidatus “Nitrospira defluvii” (NdCld). Both enzymes have a comparable catalytic efficiency but differ significantly in oligomeric structure and subunit size and belong to different phylogenetic lineages (Fig. 1) [4,5]. The presented comprehensive biophysical analyses [differential scanning calorimetry (DSC), temperature-dependent electronic UV–vis and electronic circular dichroism spectrometry (ECD), denaturant-dependent fluorescence spectroscopy] show significant differences that are discussed with respect to the recently solved X-ray structures of both enzymes [4,5] and the temperature and pH-dependence of chlorite degradation.

Fig. 1.

Structures of dimeric chlorite dismutase from Nitrobacter winogradskyi (NwCld, PDB: 3QPI) [5] and pentameric chlorite dismutase from Candidatus “Nitrospira defluvii” (NdCld, PDB: 3NN1) [4]. (A) Ribbon representation of the NwCld dimer viewed perpendicular to the vertical 2-fold symmetry axis. Subunits are shown in different colors. Heme groups are depicted in orange. (B) Ribbon representation of the NdCld pentamer structure. Monomers are shown in different colors. (C) Ribbon representation of a superposition of subunits of NwCld (green) and NdCld (orange), respectively. (D) Superposition of heme cavity residues of NwCld (green) and NdCld (orange). Amino acid numbering according to NwCld and NdCld (brackets). NdCld shows an imidazole bound to the distal site of the heme. Figures were generated using PyMOL (http://www.pymol.org/).

2. Materials and methods

2.1. Cloning, heterologous expression and purification

Cloning, heterologous expression and purification of chlorite dismutase from Nitrobacter winogradskyi (accession no. YP_319047) was performed as described by Mlynek et al. using the expression vector pET-21b (+) (Merck/Novagen, Darmstadt, Germany) for subsequent production of a C-terminally His-tagged fusion protein [5]. Recombinant NwCld was expressed in E. coli Tuner (DE3) cells (Merck/Novagen) and purified using a HisTrap FF crude column (GE Healthcare) and a HiLoad 26/60 Superdex 200 pg column (GE Healthcare).

A DNA fragment containing the full length coding region of Cld from Candidatus “Nitrospira defluvii” (without the N-terminal signal peptide) was amplified by PCR using primers designed and described by Maixner et al. [14] and cloned into a modified pET-21b (+) vector for the production of a N-terminal TEV-cleavable Strep-II tagged fusion protein. Recombinant NdCld was expressed in E. coli Tuner (DE3) cells (Merck/Novagen, Darmstadt, Germany) and grown in hemin-enriched Luria–Bertani (LB) medium. The LB-Medium was supplemented with ampicillin (100 μg mL^− 1) and was inoculated with a freshly prepared overnight culture (at a dilution 1:100). The culture was grown at 37 °C and 220 rpm agitation until the early stationary phase was reached (OD₆₀₀ = 0.8). In order to induce NdCld expression, isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM. Also hemin (50 μg mL^− 1) was added to the culture. The temperature was lowered to 24 °C. After 4 h the culture was centrifuged (4000 rpm, 10 min, 4 °C) the resulting cell pellet was either processed immediately or stored at − 80 °C until further use. When needed cell pellets were resuspended in 50 mM HEPES, pH 7.4, 5% glycerol, 0.5% Triton X-100, 0.5 mM EDTA, 1 mM phenylmethylsulfonylfluoride (PMSF). The cell suspension was lysed under sonication and clarified by centrifugation (18000 rpm. 25 min, 4 °C). Subsequently the supernatant was loaded to a StrepTrap HP 5 ml (GE Healthcare) column equilibrated with 20 mM HEPES, pH 7.4, 2% glycerol. The protein was eluted using 20 mM HEPES, pH 7.4, 2% glycerol, 2.5 mM desthiobiotin. Strep-II-tag was fully cleaved off using TEV-protease in 20 mM Hepes, pH 7.4, 2% glycerol, 0.5 mM DTT. Finally, 100 μM hemin was added to the cleaved protein and incubated for several hours at 4 °C. Precipitated hemin was removed by centrifugation (18000 rpm, 25 min, 4 °C). Resulting proteins were screened by SDS-PAGE and fractions containing NdCld were pooled and applied on a HiLoad 16/60 Superdex 200 pg (GE Healthcare) column equilibrated with 20 mM HEPES, pH 7.4, 2% glycerol. TEV protease (27 kDa) was separated from NdCld (130 kDa) in this step. Aliquots of purified protein were concentrated to 5 mg ml^− 1, frozen in liquid nitrogen and stored until further use.

Expression and purification of the apo-form of NdCld were performed as described above with the exception that no hemin was added. The absence of any absorbance in the Soret region showed the presence of the pure apoform.

2.2. Steady-state kinetics

The temperature dependence of chlorite degradation was followed as absorbance decrease at 260 nm (Hitachi U-3900 UV–vis) using ε_{260 nm} = 155 M^− 1 cm^− 1 [15]. The temperature was controlled with a water bath connected to the cuvette-holder. The enzymatic activity of NdCld and NwCld were measured between 30 °C–90 °C and 30 °C–55 °C, respectively. Reactions were carried out in 50 mM phosphate buffer, pH 7.0, and using a non-inhibiting NaClO₂ concentration of 40 μM (total reaction volume: 1 mL). Five measurements were performed and the corresponding arithmetic mean is presented. Reactions were started by addition of 32 nM NdCld or 24 nM NwCld. At room temperature the reactions were also followed polarographically using a Clark-type electrode (Oxygraph plus, Hansatech Instruments, UK) [4,5]. These data showed a stoichiometry of ClO₂⁻:O₂ of 1:1 as was reported in the literature [7–10]. However, at temperatures > 45 °C the Clark-type electrode could not be used according to the instructions of the manufacturer.

2.3. Differential scanning calorimetry

Differential calorimetric (DSC) measurements were performed using a VP-capillary DSC microcalorimeter from Microcal (cell volume: 137 μL), controlled by the VP-viewer program and equipped with an autosampler for 96 well plates. Samples were analyzed using a programmed heating scan rate of 60 °C h^− 1 over a temperature range of 20 °C to 110 °C and approximately 60 psi (4.136 bar) cell pressure. Maximum temperature inside the cuvette was 95 °C. Collected DSC data were corrected for buffer baseline and normalized for protein concentration. Conditions: 14.3 μM NdCld and 12.5 μM NwCld in 50 mM phosphate buffer, pH 7, as well as 12 μM apo-NdCld in 20 mM HEPES, pH 7.4, 2% glycerol. For data analysis and conversion the Microcal origin software was used. Heat capacity (C_p) was expressed in kcal mol^− 1 K^− 1 (1 cal = 4.184 J). Data points were fitted to non-two-state equilibrium-unfolding models by the Lavenberg/Marquardt (LM) non-linear least square method.

For measurements of the pH dependence of thermal unfolding 50 mM acetate buffer (pH 3–6) and 50 mM phosphate buffer (pH 6–11) were used. Protein concentrations were 14.3 μM (NdCld) and 12.5 μM (NwCld), respectively.

2.4. Temperature-mediated unfolding followed by electronic circular dichroism spectroscopy

Besides DSC, thermal unfolding was followed by electronic circular dichroism (ECD) spectroscopy (Chirascan, Applied Photophysics, Leatherhead, UK). The instrument was flushed with nitrogen at a flow rate of 5 L min^− 1 and allowed simultaneous UV–vis and ECD monitoring. The instrument was equipped with a Peltier element for temperature control and temperature-mediated denaturation was monitored between 20 °C and 95 °C. Temperature was increased stepwise with 1.0 °C min^− 1.

Single wavelength scans were performed with instrumental parameters set as follows. Visible ECD at Soret maximum was performed with 10 μM NdCld and 10 μM NwCld in 5 mM phosphate buffer, pH 7.0, containing 0.5 M GdnHCl (in order to avoid aggregation at higher temperatures). The pathlength was 10 mm, spectral bandwidth 1 nm and scan time per point was set at 10 s. Far-UV ECD at 208 nm was performed with 10 μM NdCld, NwCld or apo-NdCld in 5 mM phosphate buffer, pH 7.0, containing 0.5 M GdnHCl. The pathlength was at 1 mm, spectral bandwidth 3 nm and scan time per point was set at 10 s. At room temperature ECD spectra in the far-UV region were recorded for 5 μM NdCld and 5 μM NwCld in the presence and absence of 0.5 M GdnHCl in order to probe the effect of 0.5 M GdnHCl on the overall secondary structure.

The fraction α of unfolded protein was calculated according to α = (θ_N − θ)/(θ_N − θ_U) with θ_N being the ellipticity (in mdeg) at 208 or 417 nm/416 nm of the protein in the native folded state, θ the ellipticity at defined temperature (T), and θ_U being the ellipticity at 208 or 417 nm/416 nm of the completely unfolded state.

2.5. Chemical denaturation followed by fluorescence and UV–vis spectroscopy

Guanidinium hydrochloride (GdnHCl) was used to probe the chemical denaturation of both chlorite dismutases. Unfolding was monitored by following the changes in the emission of intrinsic tryptophan fluorescence as well as changes in the Soret region.

The fluorescence spectrophotometer (Hitachi F-7000 Fluorescence) was equipped with a thermostatic cell holder for quartz cuvettes of 10 mm path length. Instrumental parameters were set as follows. Excitation wavelength was at 295 nm, excitation and emission bandwidth at 5 nm and PMT voltage was set at 700 V for NwCld and 400 V for NdCld (scan speed: 60 nm min^− 1). In detail, 500 nM NdCld or NwCld in 5 mM phosphate buffer, pH 7.0, were incubated with increasing concentrations of GdnHCl (0–8 M) over night at room temperature. For each GdnHCl concentration fraction α of unfolded protein was calculated according to α = (F_N − F)/(F_N − F_U), with F_N representing either the relative fluorescence intensity (F/F₀) at 350 nm or the fluorescence emission maximum of the native folded state, F being the relative fluorescence intensity (F/F₀) at 350 nm or the fluorescence emission maximum at defined GdnHCl concentrations and F_U the relative fluorescence intensity (F/F₀) at 350 nm or the fluorescence maximum of the completely unfolded state. The same samples were also measured for elucidation of changes in the Soret region. The UV–vis spectrophotometer (Hitachi U-3900) was equipped with a thermostatic cell holder for quartz cuvettes of 10 mm path length. Scan speed was 120 nm min^− 1. For each GdnHCl concentration the fraction α of unfolded protein was calculated from the shift of the Soret maximum according to α = (A_N − A)/(A_N − A_U), with A representing the Soret band maximum at defined GdnHCl concentrations, A_N the Soret maximum of the native state and A_U the Soret maximum of the completely unfolded state.

3. Results

Recent progress in genome and metagenome sequencing has enormously increased the size and phylogenetic complexity of the Cld-like protein superfamily [12]. At the same time the practical relevance of Clds for the bioremediation of the anthropogenic pollutant chlorite increased. This prompted us to probe the conformational and thermal stability of two representatives of two major lineages that differ significantly in structure but have a comparable efficiency in chlorite degradation [4,5]. The two heme proteins were selected because we have been able to solve their X-ray structure recently and determine basic enzymatic parameters. Chlorite dismutase from Candidatus “Nitrospira defluvii” (NdCld) is a pentameric enzyme with each monomer consisting of two topologically equivalent four-stranded antiparallel β-sheets forming a β-barrel (ferredoxin-like fold) flanked on both sides by six α-helices (Fig. 1B and C) [4]. By contrast, Cld from Nitrobacter winogradskyi (NwCld) is a dimeric protein and each subunit lacks all helices in the N-terminal domain (Fig. 1A and C) [5]. These significant structural differences prompted us to probe the thermal stability of NdCld and NwCld (Fig. 2A–D) as well as the impact of temperature on the enzymatic activity (Fig. 2E).

Fig. 2.

Thermal stability of chlorite dismutases from Nitrobacter winogradskyi (NwCld) and Candidatus “Nitrospira defluvii” (NdCld) measured by differential scanning calorimetry. Black line: original data; red line: fit of endotherm to a non-two-state transition model. (A) Normalized thermograms of NwCld, NdCld and apo-NdCld in 50 mM phosphate buffer, pH 7. (B, C) pH-dependence of thermal stability of NwCld (B) and NdCld (C). Protein concentrations: 12.5 μM NwCld, pH 5 (bottom) to pH 11 (top); 14.3 μM NdCld, pH 3 (bottom) to pH 11 (top). (D) Plot of T_mversus pH. NwCld, empty circles; NdCld, filled circles. Note that for NwCld only the T_m values corresponding to the second transition are depicted. (E) Impact of temperature on chlorite degradation by chlorite dismutase from NdCld (filled circles) and NwCld (empty circles) measured spectrophotometrically at pH 7.0 as described in Material and methods.

3.1. Enzymatic activity and the impact of temperature

The overall chlorite dismutase activities (polarographic measurement of the initial rate of O₂ release at 25 °C and pH 7.0) of NdCld and NwCld are similar. For NdCld K_M, k_cat and k_cat/K_M were determined to be 58 ± 9 μM, 35 ± 5 s^− 1 and 6.0 × 10⁵ M^− 1 s^− 1, whereas for NwCld the corresponding values were 90 ± 12 μM, 190 ± 14 s^− 1 and 2.1 × 10⁶ M^− 1 s^− 1, respectively. The pH optima for transformation of chlorite to chloride and dioxygen are 5.5 (NdCld) and 6.0 (NwCld). The similar enzymatic activity is well reflected by the almost identical active site architecture (see overlay of NdCld and NwCld in Fig. 1D).

However, the impact of temperature on the chlorite degradation reaction was significantly different. Since polarographic measurement of oxygen release could not be used at temperatures higher than 45 °C, chlorite degradation was followed photometrically at 260 nm (see Materials and methods). Up to 40 °C and pH 7.0 both heme proteins exhibited an almost unchanged enzymatic activity (Fig. 2E), but with NwCld the degradation rate of chlorite dramatically decreased with only 8% residual activity at 50 °C and complete inactivation at 55 °C. By contrast, chlorite dismutase from Candidatus “Nitrospira defluvii” showed still 80% and 36% of activity at 50 °C and 70 °C, respectively, and was completely inhibited at around 90 °C.

3.2. Thermal stability evaluated by differential scanning calorimetry and circular dichroism spectroscopy

In order to understand the impact of temperature on enzymatic inactivation we probed the thermal stability of both enzymes by differential scanning calorimetry (DSC). Fig. 2A compares the thermograms of NwCld, NdCld and heme-free NdCld (apo-NdCld) showing huge differences in the thermal stability of the two selected Clds. The pentameric enzyme exhibited a high thermal stability and the thermogram suggests a cooperative two-state transition at 92 °C (T_m) at pH 7.0. In the absence of the prosthetic group the enzyme was dramatically destabilized and followed a broad non-two-state transition with calculated T_m-values at 56 and 67 °C. In contrast to NdCld melting of the dimeric heme protein (NwCld) already occurred at around 58 °C. The asymmetric endotherm (Fig. 2A) suggested a non-two state transition and by fitting to a corresponding model T_m values were found to be at 53 and 58 °C.

Furthermore we probed the impact of pH of the thermal stability of both enzymes (Fig. 2B–C). With both enzymes the highest T_m values were obtained at pH 6 (97 °C and 61 °C for NdCld and NwCld, respectively). NdCld showed a broad range of stability within pH 3–10 (Fig. 2C and D), whereas NwCld precipitated below pH 5 upon heating and showed broad non-two-state transitions in the alkaline region (pH > 7) (Fig. 2B and D). In Fig. 2D the calculated T_m values are plotted versus pH.

Next we followed thermal unfolding by electronic circular dichroism spectroscopy (ECD). Fig. 3A depicts the ECD spectra in the far-UV region for NdCld, apo-NdCld and NwCld. The spectra reflect the overall structure of chlorite dismutases that are composed of two similar domains with a ferredoxin-like fold (β-sheet flanked by varying content of α-helices). The ellipticity of α-helices typically shows two minima at 208 nm and 222 nm, whereas the contribution of β-sheets gives rise to negative ellipticity around 212–214 nm, but with Δε values that are 5-times smaller than those of α-helices [16]. The dimeric enzyme has been shown to miss all N-terminal α-helices present in NdCld [5] thus having a higher portion of β-strands. This is nicely reflected by the differences in the far-UV ECD spectra (Fig. 3A) of NwCld and NdCld. It is also demonstrated that in the absence of heme b the far-UV ECD spectrum was slightly different to that of the holoenzyme (Fig. 3A) suggesting an impact of the prosthetic group on the structural integrity of NdCld.

Fig. 3.

Temperature-mediated unfolding of chlorite dismutases from Nitrobacter winogradskyi (NwCld) and Candidatus “Nitrospira defluvii” (NdCld) followed by electronic circular dichroism in the far-UV and visible region. Conditions: 5 mM phosphate buffer, pH 7.0. (A, B) Comparison of far-UV and visible ECD spectra of NdCld (red), apo-NdCld (green) and NwCld (blue) at 25 °C. For clarity spectra have been shifted. In addition the residual activity at 95 °C is shown (black), which is very similar for NwCld and NdCld. (C, D) Thermal unfolding of NwCld followed at 208 nm and 416 nm. (E, F) Thermal unfolding of NdCld followed at 208 nm and 417 nm. (G), Thermal unfolding of apo-NdCld. The insets show the corresponding van't Hoff plots. Note that the van't Hoff plot in (D) represents is calculated from equilibrium constants deriving only from the first transition.

The near-UV and visible ECD spectra demonstrate a positive ellipticity in the Soret band region at 416 nm (NwCld) and 417 nm (NdCld) as well as differences in the circular dichroism of tyrosine and tryptophan residues. Upon completely unfolding of both proteins at T = 95 °C heme ECD signals were completely lost, whereas in the UV-region some residual ellipticity remained (Fig. 3A and B). Finally, based on these findings we followed temperature-mediated unfolding at 208 nm (reflecting melting of secondary structure, mainly α-helices) and 416 or 417 nm (reflecting release of the heme from the active site). The corresponding melting curves are depicted in Fig. 3C–G. In NwCld α-helices melted within 40 and 50 °C (calculated T_m = 46 °C) in a simple two-state transition. There was a clear linear relationship between the equilibrium constants and the reciprocal temperature allowing the calculation of the vant'Hoff enthalpy for this transition being (261 ± 7) kJ mol^− 1 (inset to Fig. 3C). The loss of ellipticity at 208 nm was accompanied by the loss of the prosthetic group with ΔH_m being (134 ± 7) kJ mol^− 1 (Fig. 3C and D). A further conformational change must occur between 50 and 60 °C as is evident by the change of ellipticity at 416 °C (Fig. 3D). This reflects the non-two-state transition of NwCld seen in the DSC experiments.

The higher thermostability of the pentameric protein could be confirmed by ECD measurements. In Fig. 3E and F the melting of the secondary structures and the release of the heme group followed a two-state transition and allowed the calculation of the van't Hoff enthalpy [(413 ± 8) kJ mol^− 1 and (372 ± 30) kJ mol^− 1, respectively]. Furthermore, Fig. 3G support the finding that heme b is important in stabilizing the subunit structure of chlorite dismutase. In contrast to the holoform the apoform of Cld had a significantly decreased thermal stability. Loss of ellipticity at 208 nm occurred within a broad temperature range (50–70 °C) (Fig. 3G).

3.3. Conformational stability evaluated by UV–vis and fluorescence spectroscopy

Finally, we evaluated the conformational stability of both chlorite dismutases by chemical denaturation with guanidinium hydrochloride (GdnHCl). Firstly, we focused on the stability of the heme cavity by monitoring the release of the prosthetic group upon chemical denaturation. With increasing GdnHCl concentration the Soret absorbance was diminished (Fig. 4A and D) and the decrease of absorbance at the Soret maximum roughly followed a two-state transition (Fig. 4B and E) allowing the calculation of the conformational stability of the active site (Fig. 4C and F). The conformational stability (ΔG°_H2O) of NdCld and NwCld were determined to be (12.7 ± 1.7 kJ mol^− 1) and (4.3 ± 0.8) kJ mol^− 1, respectively. This is reflected by calculated c_m values {corresponding to [GdnHCl] were K = [D]/[N] = 1} of 3.5 and 1.85 M, respectively. From the slope of the linear curves {ΔG°_[GdnHCl] = ΔG°_H2O −  (m × [GdnHCl])} m values were calculated to be 3.7 and 1.5 kJ mol^− 1 M^− 1. The latter reflect the efficacy of the denaturant in unfolding and is proportional to the number of groups in the protein, i.e. large proteins are more sensitive to solvent denaturation than small ones [17].

Fig. 4.

Chemical denaturation of chlorite dismutases from Candidatus “Nitrospira defluvii” (NdCld, A–C) and Nitrobacter winogradskyi (NwCld, D–F) followed by UV–vis spectroscopy. NdCld (0.5 μM) and NwCld (0.5 μM) were incubated for 18 h with various concentrations of GdnHCl (5 mM phosphate buffer, pH 7.0). (A, D) Loss of Soret absorbance upon increasing the concentration of guanidinium hydrochloride. (B, E) Plots of change in Soret maximum absorbance versus guanidinium hydrochloride concentration. (C, F) Plot of change in free enthalphy at various guanidinium hydrochloride concentrations following the linear equation ΔG°_[GdnHCl] = ΔG°_H2O − (m × [GdnHCl]). Arrows indicate changes in absorbance.

Secondly, the overall unfolding was followed by monitoring the change in the intrinsic tryptophan fluorescence during unfolding. Fig. 5A and D show the increase of fluorescence intensity and red-shift of emission maxima of NdCld and NwCld mediated by increasing concentrations of GdnHCl. From the secondary plots (Fig. 5B and E) ΔG_H2O values of (16.3 ± 0.3) kJ mol^− 1 and (5.3 ± 0.7) kJ mol^− 1 kJ mol^− 1 were calculated (c_m-values: 3.8 and 2.5 M; m = 4.31 and 2.1 kJ mol^− 1 M^− 1, respectively). These findings unequivocally underline the higher conformational stability of the pentameric enzyme compared to NwCld. The observed differences in the stability of the heme cavity compared with that of the overall protein indicate that the prosthetic group is released before complete unfolding of the protein is accomplished.

Fig. 5.

Chemical denaturation of chlorite dismutases from Candidatus “Nitrospira defluvii” (NdCld, A–C) and Nitrobacter winogradskyi (NwCld, D–F) followed by fluorescence spectroscopy (excitation at 295 nm) . NdCld (0.5 μM) and NwCld (0.5 μM) were incubated for 18 h with various concentrations of GdnHCl (5 mM phosphate buffer, pH 7.0). (A, D) Change in intrinsic tryptophan emission upon increasing the concentration of guanidinium hydrochloride. (B, E) Plots of change in emission at 350 nm versus guanidinium hydrochloride concentration. (C, F) Plot of change in free enthalphy at various guanidinium hydrochloride concentrations following the linear equation ΔG°_[GdnHCl] = ΔG°_H2O − (m × [GdnHCl]).

4. Discussion

So far chlorite dismutase-like proteins are found in 15 bacterial and archaeal phyla, suggesting ancient roots [4,14]. For several members of the Cld family it has been demonstrated that they are able to convert chlorite to chloride and dioxygen although the physiological relevance of this activity is not fully understood. Other (so far unknown) catalytic properties are possible and likely. Moreover, it has been demonstrated that the family of chlorite dismutases is also structurally related to other ancient and functionally mysterious protein families including dye-decolorizing peroxidases [12].

The pentameric protein from Candidatus “Nitrospira defluvii” is a representative of Lineage I of the Cld family that comprises the so-called canonical Clds that all have a very similar subunit structure and oligomerization [2–4]. All members of this subfamily have been shown to be able to degrade chlorite and occur in different subclasses of Proteobacteria but also some nonproteobacterial organisms (e.g. NdCld). Recently, the Cld from Nitrobacter winigradskyi was detected and found to belong to another Cld subfamily. Actually, it is the first representative of Lineage II of the chlorite dismutase family with known crystal structure [6] as well as known capacity to degrade chlorite. Sequence alignment has demonstrated that NwCld can be considered as a model for the respective subfamily [6,12].

Both heme enzymes exhibit a comparable activity regarding chlorite decomposition between 20 and 40 °C which is in agreement with the active site architecture of NdCld [4] and NwCld [5] showing the presence of identical and superimposable both proximal and distal residues (Fig. 1D). By contrast, significant differences in conformational and thermal stability of Lineage I and II Clds are evident from the presented findings.

The pentameric enzyme exhibits a very high thermal stability within a broad pH-range. Its unfolding pathway follows a simple two-state transition suggesting a cooperative process. At pH 7.0 unfolding starts around 80 °C and is completed at 95 °C. Subunit melting, release of the prosthetic group as well as separation of the subunits occur simultaneously as demonstrated by DSC and ECD measurements. This was also reflected by the fact that NdCld could degrade chlorite even at 80 °C. A closer inspection of the subunit structure and subunit interaction in NdCld [4] demonstrates that there is a pronounced interface between neighboring subunits (around 1400 Å²). Since each subunit is in contact with two other molecules (Fig. 1B), the total area buried in the interface is about 23% of the total area of a subunit. Each monomer of NdCl has two topologically equivalent four-stranded antiparallel β-sheets forming a β-barrel (Fig. 1C) flanked on both sides by six α-helices (ferredoxin-like fold). Comparison with other Cld-like enzymes from the same Cld subfamily (PDB: 1TOT, 3DTZ, 1VDH, 2VXH) reveals high structural conservation thus suggesting comparable stability and unfolding pathway. The interface between two subunits consists mainly of residues from the N-terminal helices α4 and strand β4 that interact with residues in the loop between strands β2 and β3. The compact pentameric structure is dictated by a combination of hydrophobic, ionic and hydrogen-bonding interactions [4] and it is reasonable to assume that unfolding of the individual subunits starts with simultaneous disruption of these close interactions. The heme b is buried in the C-terminal domain of a monomer and is embedded within a defined hydrophobic environment via many interactions. Among others two prominent examples are the fifth coordinating ligand (i.e. proximal His) which is part of the helix α3’ and the heme propionates that form hydrogen bonds to helix α2' and to the β1' strand as well as to the loop between β4 and α1' [4]. This close interaction may explain (i) the simultaneous unfolding of the five subunits and release of the prosthetic groups as well as (ii) the stabilizing effect of the heme b group on the stability of chlorite dismutases. The latter was demonstrated by analysis of the apoform of NdCld.

The subunit structure and interaction of Lineage II Clds is completely different and allows explanation of the low thermal stability of the dimeric protein which has a T_m that is more that 35 °C below that of the pentameric protein. Firstly the primary sequence of NwCld is about 30% shorter than that of NdCld, with significant deletion in the N-terminal region (Fig. 1A). In particular, all terminal α-helices are missing, and the central β-barrel no longer consists of two similar four stranded β-sheets but of one three-stranded and one five-stranded β-sheet (Fig. 1A). But most importantly, the interface between the two monomers in NwCld is entirely different since it lacks all N-terminal helices and has longer loops between β-strands that adopt different conformations. In NwCld the dimer interface is formed from the loop between β4 and α1' that interacts with the loop between helix α2’ and helix α3’ from the other monomer [5]. The surface buried in this interface is 980 Å² which corresponds to 11.5% of the subunit surface (compared to 23% in NdCld). The interactions in NwCld dimer formation are mainly electrostatic and to a lesser extent hydrophobic compared to NdCld [5] which might explain the stronger impact of pH on its stability. In contrast to NdCld the thermograms of NwCld were asymmetric and could not be fitted by a simple two-state model. These phenomenon was more pronounced with increasing (basic) pH. A non-two-state unfolding pathway was also suggested by ECD measurements that focused on the loss of ellipticity from the heme.

Discrepancies in thermal stability between Lineage I and II Clds were also seen in unfolding experiments mediated by chemical denaturation. The conformational stability of both the overall structure as well as of the heme cavity of pentameric NdCld was calculated to be about 3 times higher than that of NwCld. The dimeric protein has a low conformational stability (5.3 kJ mol^− 1) and already at a GdnHCl concentration of 1.85 M 50% of the molecules are in the denatured state.

Summing up, this is the first comprehensive biophysical investigation of the impact of temperature and chaotropic agents on the structural integrity of chlorite dismutases. Our observations clearly suggest that—from a biotechnological point of view—pentameric enzymes from Lineage I (e.g. NdCld) are the more interesting candidates for bioremediation, i.e. degradation of chlorite contaminations. Although Candidatus “Nitrospira defluvii” is a slow growing bacterium and extremely difficult to culture under laboratory conditions [18] large amount of active NdCld can easily be obtained by expression in E. coli as heterologous host [14]. Such recombinant chlorite dismutase from Candidatus “Nitrospira defluvii” can be applied between pH 3 and pH 10 and retains its integrity and chlorite degradation activity even at high temperatures up to 70 °C. Thus, NdCld provides an ideal starting scaffold for further stability engineering by both rational mutagenesis or directed evolution strategies.