Functional Diversity of the Oxidative Stress Sensor and Transcription Factor SoxR: Mechanism of [2Fe-2S] Cluster Oxidation

Abstract

The [2Fe-2S] transcription activator SoxR, a member of the MerR family, functions as a bacterial stress response sensor. The response governed by SoxR is activated by the oxidation of the [2Fe-2S]. In this review, I describe functional differences between Escherichia coli SoxR (EcSoxR) and Pseudomonas aeruginosa SoxR (PaSoxR). Pulse radiolysis demonstrated that the reduced form of EcSoxR reacts directly with O₂⁻ with a second-order rate constant of 5.0 × 10⁸ M⁻¹s⁻¹. PaSoxR was found to undergo a similar reaction, although with a 10-fold smaller rate constant (4.0 × 10⁷ M⁻¹s⁻¹). This difference in rate constants may reflect distinct regulatory features of EcSoxR and PaSoxR. Specifically, mutagenesis studies have shown that Lysine residues―which are located close to [2Fe-2S] clusters, in EcSoxR, but are not conserved in PaSoxR―are essential for EcSoxR activation. In contrast, both EcSoxR and PaSoxR were found to react with various redox-active compounds (RACs), including viologens, phenazines, and quinones, with no apparent differences in the kinetic behavior or specificity of the two proteins. Importantly, both O₂⁻ and RACs oxidize SoxR with the same rate constants. soxR regulon may be induced through multiple pathways, and the activation may depend on the cellular concentration of O₂⁻ and RACs.

1. Introduction

Iron–sulfur clusters are universal cofactors that play many important roles in biological processes [1,2,3,4,5]. Iron–sulfur clusters in signal-transducing sensor proteins are functionally distinct from most iron–sulfur clusters, which act as redox centers for electron transfer reactions. In contrast, iron–sulfur clusters in sensor proteins respond to small signaling molecules in the environment [4,6,7,8], resulting in structural changes such as oxidation state and cluster type. These changes may alter gene expression in cells. The SoxR, one example of such iron–cluster-containing sensor proteins, proceeds reversible oxidation and reduction between SoxR_ox and SoxR_red, according to the following equation.

$${\mathrm{SoxR}}_{\mathrm{ox}}\stackrel{{\mathrm{e}}^{-}}{\rightleftharpoons }{\mathrm{SoxR}}_{\mathrm{red}}$$

SoxR is a 34 kDa homodimer containing one essential [2Fe-2S] cluster per polypeptide chain. Because SoxR_red can be observed by EPR spectra, its redox changes can be monitored quantitatively by EPR spectroscopy. In vivo EPR spectra show that SoxR in E. coli (EcSoxR) is maintained in a reduced form during normal aerobic growth [8]. When O₂⁻ generating compounds such as methyl viologen (MV²⁺) were added to E. coli cells, EPR signals disappeared rapidly, indicating oxidation of the EcSoxR cluster. Thus, SoxR senses oxidative stress. The resulting activated redox state is propagated to the SoxR DNA-binding domain, resulting in the activation of target gene promoters [9,10,11,12,13].

The target of activated SoxR is a single adjacent soxS gene. Increased SoxS levels induce expression of various antioxidant proteins and repair proteins [14]. This two-step soxRS-based regulatory mechanism controls the expression of genes encoding antioxidant defense proteins such as Mn-containing superoxide dismutase (SOD), glucose-6-phosphate dehydrogenase, endonuclease IV, and outer membrane drug efflux pump, as shown in Figure 1.

A number of SoxR homologs have been identified in various bacteria [15], all of which contain a SoxR-specific cysteine motif and DNA binding sites. In contrast to those in E. coli and related enteric bacteria, the majority of SoxR regulons in non-enteric bacteria such as P. aeruginosa and Streptomyces coelicolor lack genes involved in oxidative stress and detoxification [15]. Some of these bacteria produce redox-active compounds (RACs) independent of oxidative stress [15,16]. PaSoxR is activated by endogenous RAC antibiotic compounds, such as pyocyanin (Pyo) and actinorhodin (Act) [15,16,17,18], which are the physiological signals for the regulation of quorum-sensing-regulated genes. These SoxR target genes regulate genes involved in export (via the mexGHI-opmD RND efflux pump [19]) and the metabolism of small molecules [19], including RAC antibiotics and endogenous pigments (Figure 2). In these species, SoxR appears to function as a defense against the toxicity of RACs.

2. Direct Oxidation of [2Fe-2S] Cluster in SoxR by O₂

An important question relates to the chemical nature of the signal sensed by SoxR. The first proposal SoxR-activating signal identified was O₂⁻ [20], but the mechanism by which O₂⁻ activates SoxR remains unclear [21]. Gu and Imaly [22] reported that EcSoxR is not directly activated by O₂⁻, but RACs. To investigate the oxidation of EcSoxR by O₂⁻, I followed the reactions that occurred following pulse radiolysis of oxygen-saturated aqueous solutions in the presence of EcSoxR_ox, which were previously determined [13]. Under these experimental conditions, most of the hydrated electrons (e_aq⁻) are converted to O₂⁻, while a part of the e_aq⁻ population rapidly reduced the [2Fe-2S] cluster. The initial rapid absorption changes due to reduction in the [2Fe-2S] cluster partially recovered on a millisecond time scale. It is noted that the recovery was inhibited by the addition of Cu/Zn-SOD. From pulse radiolysis results, the reaction schemes can be summarized by reactions (1) and (2):

$${{\mathrm{e}}_{\mathrm{aq}}}^{-}\hspace{1em}+\hspace{1em}{[2\mathrm{Fe}\text{-}2\mathrm{S}]}^{2+}\to {[2\mathrm{Fe}\text{-}2\mathrm{S}]}^{1+}$$

(1)

$${{\mathrm{O}}_2}^{-}+{[2\mathrm{Fe}\text{-}2\mathrm{S}]}^{1+}+2{\mathrm{H}}^{+}\to {[2\mathrm{Fe}\text{-}2\mathrm{S}]}^{2+}+{\mathrm{H}}_2{\mathrm{O}}_2$$

(2)

where [2Fe-2S]²⁺ and [2Fe-2S]¹⁺ are oxidized and reduced forms of SoxR, respectively.

Reaction (2), observed in SoxR, represents the first reported example of O₂⁻ oxidation, a reduced form of iron–sulfur protein. Whether other [2Fe-2S] proteins show similar reactivity was tested by examining their reactions with O₂⁻ using spinach ferredoxin and adrenodoxin as examples. However, unlike the case of SoxR, after rapid reduction in these [2Fe-2S] proteins by e_aq⁻, the recovery of absorption changes was not observed. On the other hand, in the reaction of the oxidized form of iron–sulfur proteins, reactions between iron–sulfur proteins and O₂⁻ with second-order rate constants ranging from 10⁶ to 10⁷ M⁻¹ s⁻¹ [23,24,25] were reported. In these cases, the reactions were irreversible and were accompanied by the degradation of the iron–sulfur cluster. By contrast, reaction (2) in SoxR is reversible. Therefore, it can be said that reaction (2) of SoxR is a specific characteristic of SoxR protein.

Pulse radiolysis revealed that a similar process was at work in PaSoxR. To compare the reactions of EcSoxR and PaSoxR, we studied the inhibition by SOD oxidation of SoxR_red (Figure 3). The oxidation of PaSoxR was completely inhibited by 0.5 μM SOD (Figure 3B), whereas the same concentration of SOD inhibited EcSoxR by only 20% (Figure 3A). This suggests that the reactivity of O₂⁻ toward EcSoxR is much greater than toward PaSoxR. The SOD dependence of absorption changes at 420 nm for EcSoxR is presented in Figure 3C. Based on the concentration of SOD with the half-maximal inhibition of SoxR_red and the rate constant for the reaction of O₂⁻ with SOD, we obtained a second-order rate constant of 5.0 × 10⁸ M⁻¹s⁻¹ for the reaction of O₂⁻ with EcSoxR_red. In contrast, the rate constant for PaSoxR_red with O₂⁻ was calculated to be 4.0 × 10⁷ M⁻¹s⁻¹. The difference in rate constant reflects the distinct regulatory roles for EcSoxR and PaSoxR in response to their physiological activation by O₂⁻. The mechanism underlying the different sensitivities of various SoxRs to O₂⁻ remains an important question.

These data clearly suggest that slight alteration in structure between highly homologous SoxR proteins can lead to functional differences. Figure 4A compares sequence alignments for SoxR proteins from different enteric and non-enteric bacteria. SoxR homologs from E. coli, an enteric species, contain two lysine residues, which are substituted with alanine residues in the SoxR of most non-enteric species [26]. In the crystal structure of EcSoxR [27], Lys89 and Lys92, located immediately upstream of dimerization helix 5, are close to the [2Fe-2S] cluster (Figure 4B). In addition, SoxRs from most species of enteric bacteria contain a conserved three-residue hydrophilic motif (R127S128D129) in the proximity of [2Fe-2S] clusters that affect SoxR sensitivity to RACs [26] and it is not conserved in SoxR from other non-enteric bacteria [26].

Mutagenesis experiments have investigated the sensitivity to O₂⁻, focusing on several amino acids that are not conserved among SoxR homologues. Specifically, the EcSoxR mutants (K89A, K92A, K89AK92A, D129A, and R127LS128QD129A) and PaSoxR mutants (A87K, A90K, and L125RQ126SA127D) were prepared, and their O₂⁻ sensitivity was determined by examining the effects of concentration on SOD inhibition [26].

Table 1. Second-order rate constants in O₂⁻ reactions with the reduced forms of wild-type and mutant SoxR proteins.

	k × 108 (M−1 s−1)
E. coli
WT	5.0
R127LS128QD129A	4.8
D129A	5.0
K89A	3.8
K92A	2.2
K89AK92A	0.33
K89RK92R	4.7
K89EK92E	0.31
P. aeruginosa
WT	0.4
A87K	2.1
A90K	5.4
L125RQ126SA127D	0.4

shows the rate constants for EcSoxR and PaSoxR mutants. Although the rate constants for the K89A (3.8 × 10⁸ M⁻¹ s⁻¹) and K92A (2.2 × 10⁸ M⁻¹s⁻¹) mutants were only slightly affected by the indicated substitutions, the rate constant for the double-Lys K89AK92A mutant (3.3 × 10⁷ M⁻¹s⁻¹) was dramatically affected. The rate constant of 3.3 × 10⁷ M⁻¹s⁻¹ was 10 times smaller than that of wild-type SoxR. Conversely, the corresponding substitution of K to A in PaSoxR at residues (A87K) or (A90K) increased rate constants from approximately 10-fold to 2.1 × 10⁸ and 5.4 × 10⁸ M⁻¹s⁻¹, respectively (Table 1). Therefore, substituting an alanine residue for a lysine residue in the [2Fe-2S] region of EcSoxR is sufficient to transform the O₂⁻-sensitivity of EcSoxR such that it mimics that of PaSoxR. The effects of the electrostatic charge of amino acid residues were further analyzed by replacing both positive charged both Lys89 and Lys92 residues in EcSoxR with negatively charged glutamate (K89EK92E). The resulting charge reversal from positive to negative decreased the rate constant; in contrast, substituting arginine, another positively charged amino acid, for these Lysine (K89RK92R) did not change rate constants (Table 1). Thus, these Lysine residues are responsible for the high diffusion-limited rate constant. Collectively, these results demonstrate that these electrostatically positive Lysine residues are critical for the reaction of O₂⁻ with the [2Fe-2S] cluster in EcSoxR and suggest that they guide O₂⁻ into the active site on the surface of SoxR. These findings are consistent with the previously reported involvement of Lysine residues in enzymatic reactions with O₂⁻ [28,29,30].

SoxR from most species of enteric bacteria contains a three-residue hydrophilic motif (R127S128D129) in the vicinity of [2Fe-2S] clusters. These residues are not conserved in the SoxR of other bacteria. It has been proposed that the presence of this conserved motif affects the sensitivity of SoxR to RACs [18,31] and is responsible for the observed species-specific differences in RAC sensitivity. However, unlike the above Lys substitutions, the RSD→LQA substitution in EcSoxR and LQA→RSD substitution in PaSoxR did not affect O₂⁻ reaction rates (Table 1).

3. Response of SoxR [2Fe-2S] to RACs

It has been shown that EcSoxR responds to endogenous, as well as various synthetic RACs [17,18,31,32], suggesting that EcSoxR_red is oxidized directly, but nonspecifically, by RACs. In contrast, PaSoxR is only activated by endogenous molecule RACs and appears to respond to more restricted molecules. The sensitivity and specificity of various RACs were examined by investigating the kinetics of the oxidation of the [2Fe-2S] cluster in SoxR using pulse radiolysis [33]. Pulse radiolysis enables reduction of [2Fe-2S]²⁺ in SoxR_ox with a hydrated electron (e_aq⁻) within 1 μs (reaction 1), which permits subsequent oxidation of [2Fe-2S]_red by RACs such as MV²⁺ (reaction 3).

(3)

Figure 5 shows the absorption changes during pulse radiolysis of PaSoxR in the absence and presence of MV²⁺. After the initial decreasing absorbance, a subsequent partially reverse process was observed in the presence of MV²⁺ on a millisecond time scale. From these findings, it can be inferred that the [2Fe-2S] cluster of SoxR_ox is initially reduced by e_aq⁻, and subsequently becomes partially reoxidized, implying that the recovery process reflects the oxidation of SoxR_red by MV²⁺ (reaction 3). Similar absorption changes were observed in EcSoxR.

To determine the second-order rate constants in the reaction of MV²⁺ with PaSoxR_red, we used conditions in which the concentrations of the RAC (10–25 μM) was much higher than that of SoxR_red (0.5–1 μM). Under these conditions, the oxidation of SoxR_red obeyed pseudo-first-order kinetics, with the observed rate constant increasing linearly with RAC concentration. The rate constant of the reaction, obtained from the slope in Figure 5B was determined to be 3.0 × 10⁸ M⁻¹s⁻¹. Similar absorption changes were observed in the presence of other RACs, including pyocyanin (Pyo), phenazine methosulfate (PMS), duroquinone (Dur), and diquat (Dq). The second-order rate constants increased in the order Dur <MV²⁺ < Dq < Pyo < PMS (Table 2) and were found to correlate with the electrode potential of the RAC. In all cases, a rapid decrease phase was followed by a slower recovery phase. Notably, absorbance changes during the slower reoxidation phase varied depending on the specific RACs. The incomplete reoxidation of the [2Fe-2S] cluster may be explained by a rapid equilibrium for electron transfer between the RAC and the [2Fe-2S] cluster (reaction 3).

Table 2. Rate constants of [2Fe-2S] cluster of EcSoxR, and PaSoxR, with RACs and the electrode potentials E₀ (vs NHE) at pH 7.0.

RACs	E0 (vs. NHE) (mv)	EcSoxR k (× 108 M−1s−1)	PaSoxR k (× 108 M−1s−1)
MV2+	−440 [34]	3.0	3.0
Dq	−358 [35]	5.7	5.5
Dur	−260 [35]	2.1	2.6
Pyo	−34 [36]	7.1	6.8
PMS	+80 [36]	16.0	14.0

Table 2. Rate constants of [2Fe-2S] cluster of EcSoxR, and PaSoxR, with RACs and the electrode potentials E₀ (vs NHE) at pH 7.0.

RACs	E0 (vs. NHE) (mv)	EcSoxR k (× 108 M−1s−1)	PaSoxR k (× 108 M−1s−1)
MV2+	−440 [34]	3.0	3.0
Dq	−358 [35]	5.7	5.5
Dur	−260 [35]	2.1	2.6
Pyo	−34 [36]	7.1	6.8
PMS	+80 [36]	16.0	14.0

The pulse radiolysis results presented here show that the [2Fe-2S]_red cluster of EcSoxR and PaSoxR react with RACs with electrode potentials in the range of −440 mV (MV²⁺) [34] to +80 mV (PMS) [36]. Importantly, these results are consistent with the responsiveness of SoxR to RACs observed in vivo [18,22,32]. In comparison of the sensitivity profiles of SoxR towards RACs, Singh et al. showed that both EcSoxR and PaSoxR responded to all RACs, except Act [18].

SoxR homologs exhibit different selectivity toward RACs with each displaying distinctive effective concentrations and cellular response times [31]. However, our study found that these differences are not reflected in the kinetics of the reactions. Specifically, pulse radiolysis revealed no difference in the reactivity of the [2Fe-2S] cluster between EcSoxR and PaSoxR, despite their differing cellular response times to MV²⁺ [22].

4. Physiological Significance of SoxR Response to O₂⁻ and RACs

In this review, from the in vitro data, O₂⁻ directly oxidizes purified SoxR [11,13,26], suggesting that O₂⁻ itself is the signal that activates SoxR. However, it is not known whether O₂⁻ can interact with SoxR under high levels of SOD and a low level of SoxR constitutively under normal cellular conditions [25,37]. In addition, the following in vivo experiments have shown that O₂⁻ is not the signal for SoxR. Firstly, endogenous O₂⁻ in SOD- mutants does not activate SoxR in SOD-deficient cells [22,38]. Secondly RACs such as MV²⁺ can activate SoxR in anoxic conditions, in the presence of alternative respiratory acceptors [22]. The soxR regulon may be induced through multiple pathways. Since both O₂⁻ and RACs oxidize SoxR with the same rate constants, the activation depends on the cellular concentration of O₂⁻ and RACs.

The data presented here provide new insight into electron transfer between RACs and the [2Fe-2S] cluster of SoxR homologs. The observed oxidation of the [2Fe-2S] in SoxR proceeds via direct electron transfer from the [2Fe-2S] cluster to the RAC. It is likely that structurally distinct RACs can gain access through the solvent-exposed surface area allowing rapid electron transfer between the [2Fe-2S] cluster and RACs. An X-ray analysis of DNA-free EcSoxR [24] showed that the upper sulfur atom (S2) and one of the Fe atoms are completely exposed to the solvent. A similar mechanism may take place at the SoxR active site in PaSoxR. The oxidizability of SoxR_red is assisted by the low potential of its [2Fe-2S] cluster in EcSoxR (−285 mV) [39] and PaSoxR (−290 mV) [39], which provide a strong driving force for electron transfer to even the moderate oxidant (−440 mV), MV²⁺ [33].

5. Conclusions and Future Perspectives

In this review, I followed reactions of the [2Fe-2S] cluster of SoxR with O₂⁻ and RACs directly. Both rate constants were almost a diffusion-controlled rate and characteristic of the individual SoxRs. The next major goal now is to investigate the factors that control the iron–sulfur cluster. To understand how iron–sulfur clusters respond to the molecules in the environment, future progress at the boundary of physical chemistry, biological chemistry, and inorganic chemistry will be crucial.