A dual role of cysteine residues in the maturation of prokaryotic Cu/Zn-superoxide dismutase

Abstract

Bacterial Cu/Zn-superoxide dismutase (SodC) is an enzyme catalyzing the disproportionation of superoxide radicals, to which the binding of copper and zinc ions and the formation of an intramolecular disulfide bond are essential. We previously showed that Escherichia coli SodC (SodC) was prone to spontaneous degradation in vivo in an immature form prior to the introduction of the disulfide bond. The post-translational maintenance involving the metal binding and the disulfide formation would thus control the stability as well as the enzymatic function of SodC; however, a mechanism of the SodC maturation remains obscure. Here, we show that the disulfide-reduced SodC can secure a copper ion as well as a zinc ion through the thiolate groups. Furthermore, the disulfide-reduced SodC was found to bind cuprous and cupric ions more tightly than SodC with the disulfide bond. The thiolate groups ligating the copper ion were then autooxidized to form the intramolecular disulfide bond, leading to the production of enzymatically active SodC. Based upon the experiments in vitro, therefore, we propose a mechanism for the activation of SodC, in which the conserved Cys residues play a dual role: the acquisition of a copper ion for the enzymatic activity and the formation of the disulfide bond for the structural stabilization.

Graphical Abstract

Cu/Zn-superoxide dismutase utilizes its conserved Cys residues for metal acquisition as well as conformational stabilization via disulfide formation.

Open in new tabDownload slide

Introduction

Cu/Zn-superoxide dismutase (SodC) is an enzyme that can erase superoxide radicals by facilitating their disproportionation into hydrogen peroxide and molecular oxygen.¹ This enzyme binds a copper ion as an active site for the disproportionation, while the binding of a zinc ion and the formation of an intramolecular disulfide bond are required for maintaining the catalytically competent conformation. Almost all aerobes are equipped with SodC,² and its loss is known to cause deleterious phenotypes such as shortened lifespan in fruit fly³ and progressive loss of motor abilities in dog and human.^4–6 Physiological roles of bacterial SodC are yet to be established, but its enzymatic activity is proposed to be required for resistance to reactive oxygen species generated by endogenous metabolism during the early stationary phase.⁷ Also, deletion of the sodC gene has been reported to result in losing the ability to form biofilms.⁸ Accordingly, maturation of SodC regulates a variety of physiological processes from bacteria to human.

It has been shown that simple addition of copper and zinc ions to eukaryotic SodC (SOD1) polypeptide in vitro can make SOD1 mature as an active enzyme.⁹ In a cell, however, no free copper and zinc ions are considered to be available because of their tight binding,^10,11 and reducing environment of the cytosol is also unfavorable for formation of the disulfide bond.¹² Even under such intracellular environment, a copper chaperone CCS is known to supply the copper ion and the disulfide bond into SOD1.^13,14 Some of eukaryotes do not have the gene coding CCS but exhibit SOD1 activity, suggesting that SOD1 can be activated also via a CCS-independent pathway.¹⁵ In contrast to such extensive insights into the activation of SOD1, few studies have examined how the metal ions and the disulfide bond are introduced into SodC.

A folding pattern of the native state between SOD1 and SodC is well conserved¹⁶; however, unlike homodimeric SOD1, chromosomally encoded SodC in Escherichia coli functions as a monomer with the metal ions and the intramolecular disulfide bond and tends to be unfolded when lacking those metal ions and the disulfide bond.¹⁷ Because SodC of Gram-negative bacteria is localized at their periplasmic space,¹⁸ its unfolded nature before maturation would be required for the effective translocation from the cytosol to the periplasmic space. Due to such unfolded nature, the metal-deficient, disulfide-reduced SodC is rapidly proteolyzed; therefore, the metal binding and/or the disulfide formation in SodC will be necessary for its survival in the periplasmic space. While a disulfide-null variant (C74A/C169A) of SodC in vitro appeared not to bind copper and zinc ions, the metal-binding ability of SodC with the disulfide bond (SodC^S-S) implies the formation of the disulfide bond prior to the metal binding.¹⁷ A protein DsbA, which in general catalyzes the introduction of disulfide bonds in periplasmic proteins,¹⁹ was shown to introduce the disulfide bond in SodC²⁰ but was not essential to the maturation of SodC in vivo.¹⁷ Furthermore, no CCS homologs have been identified in prokaryotes, while a potential copper chaperone for SodC was proposed in Salmonella enterica albeit controversial.^21,22 It hence remains obscure how SodC becomes matured as the active enzyme.

Here, we show that in the disulfide-reduced form of E. coli SodC (SodC^SH), metal ions (Cu²⁺, Cu⁺, Zn²⁺, and Co²⁺) are ligated preferentially by thiolate groups of the Cys residues that form the intramolecular disulfide bond in the matured state. Both Cu²⁺ and Cu⁺ ions were bound more tightly in SodC^SH than in SodC^S-S, and the copper-thiolate coordination in SodC^SH was susceptible to oxidation to form the intramolecular disulfide bond, leading to successful maturation as the enzyme. Based upon our experiments using purified proteins, we thus propose a dual role of Cys residues in maturation of copper proteins; namely, SodC secures its catalytic copper ion through the Cys residues, which are also essential to folding into the native conformation by forming the disulfide bond.

Methods

Protein preparation

A chromosomally encoded E. coli SodC is transported to the periplasmic space using an N-terminal periplasmic signal sequence (Met1—Ala19), which is cleaved to become a mature SodC protein (Ala20—Lys173). To prepare mature SodC proteins, the fragment containing a 6x His tag and an HRV3C recognition sequence was introduced between the signal sequence and the mature SodC. The order in the resultant polypeptide is as follows; the signal sequence—6x His tag—HRV3C recognition sequence—mature SodC. Then, a cDNA corresponding to the polypeptide was substituted for a glutathione-S-transferase gene in a pGEX4T-1 plasmid. The SodC proteins were overexpressed in E. coli BW25113 ΔsodC transformed with the plasmid. The E. coli cells were shaken-cultured in an LB medium with ampicillin at 37°C until the optical density at 600 nm reached around 0.6. Expression of SodC proteins was then induced with 0.5 mM isopropyl-1-thio-β-D-galactopyranoside at 20°C overnight.

The SodC proteins were isolated from periplasmic fractions of the E. coli cells, which were prepared by cold osmotic shock as follows. The E. coli cell pellets were resuspended in a pre-chilled OSI buffer (20 mM Tris/2.5 mM EDTA/2 mM CaCl₂/20% sucrose, pH 8.0) and left on ice for 10 min. The cells were collected by centrifugation at 20 000 × g for 20 min and then resuspended in a pre-chilled solution of 0.5 mM MgSO₄. After 10 min on ice, the solution was centrifuged at 20 000 × g for 20 min, and the supernatant was collected as a periplasmic fraction.

The periplasmic fraction was loaded on a cOmplete^TM His-Tag Purification Column (1 mL, Roche). After washed with a buffer containing 50 mM Na-Pi and 500 mM NaCl at pH 7.0, SodC proteins were eluted from the column with a buffer containing 50 mM Na-Pi, 100 mM NaCl, and 100 mM imidazole at pH 7.0. An N-terminal His-tag of the eluted SodC protein was cleaved by incubation with His-tagged HRV3C protease at 4°C overnight. The cleaved His-tag and the His-tagged HRV3C protease were removed by a cOmplete^TM His-Tag Purification Column, and a tag-free mature SodC was further purified by size-exclusion chromatography using a gel filtration column (Cosmosil 5Diol-300-II, nacalai tesque) equilibrated with a buffer containing 100 mM Na-Pi and 100 mM NaCl at pH 7.0 (NN buffer).

Mutant SodC lacking the disulfide bond (i.e. C74A, C169A, noCys) is known to be proteolyzed in the periplasmic space; therefore, those SodC proteins were prepared as described in our previous study.¹⁷

Preparation of apo/disulfide-reduced forms of SodC

SodC purified as above was equipped with the intramolecular disulfide bond. To prepare apo-SodC^SH, the disulfide bond was reduced by incubating SodC proteins in the NN buffer (pH 7.0) with 50 mM DTT and 1 mM EDTA at 37°C for an hour. DTT/EDTA and any metal ions bound in SodC were then removed by the trichloroacetic acid (TCA) precipitation, in which the proteins were precipitated in 20% TCA, washed with cold acetone, and dried by using a SpeedVac. Apo-SodC^SH can thus be prepared by redissolving the dried protein pellets in an appropriate buffer (pH 7.0) that was demetallated with a Chelex 100 Chelating Resin (Bio-rad). Apo-SodC^S-S was prepared by the TCA precipitation without the disulfide reduction step. Absence of copper and zinc ions in the SodC samples treated as above was confirmed by a graphite-furnace atomic absorption spectrophotometer (Shimadzu, AA-7000).

Assays of metal binding in SodC

To examine the binding of Zn²⁺ ion in SodC proteins, apo-SodC proteins in the Chelex-treated NN buffer were incubated with an equimolar amount of ZnSO₄ at room temperature for 30 min. In order to remove free Zn²⁺ ions, the samples were buffer-exchanged with the Chelex-treated NN buffer using an Amicon Ultra-0.5 Centrifugal Filter Unit 10 K (Millipore), by which free Zn²⁺ ions were expected to be diluted 10 000-fold. The concentration of SodC proteins was spectroscopically estimated from the absorbance at 215 and 225 nm,²³ and the zinc in the sample was quantitated by a graphite-furnace atomic absorption spectrophotometer (Shimadzu, AA-7000).

For the metal-binding assay with PAR, SodC proteins (20 µM, 90 µL) in 50 mM MOPS/100 mM NaCl at pH 7.0 (MN buffer) were mixed with a solution (90 µL) of 50 µM PAR in the MN buffer containing either 20 µM CuSO₄ or 20 µM ZnSO₄. Soon after the manual mixing of those solutions, the sample was set in a cell (Ultra-micro rect cell, #5062-2496, Agilent Technologies), and a UV-vis spectrum was repeatedly measured by a Cary 8454 UV-Vis spectrophotometer (Agilent) at intervals of 1 min.

Binding of a cuprous ion to SodC proteins was examined in a glove box (UN-650L, UNICO) filled with N₂ gas. In the glove box, Cu⁺(BCS)₂ was first prepared by mixing 20 µM [Cu⁺(CH₃CN)₄](PF₆^–) with 50 µM BCS in the MN buffer. The metal-binding reaction was then started by manual mixing of the SodC solution (20 µM in the MN buffer) with either the Cu⁺(BCS)₂ solution or 20 µM [Cu⁺(CH₃CN)₄](PF₆^–) in 1:1, and a UV-vis spectrum was monitored by a spectrophotometer (SEC2000-UV/VIS, BAS Inc.) at intervals of 1 min.

For evaluation of the Co²⁺ binding in SodC proteins, apo-SodC proteins (200 µM, 100 µL) were prepared in the MN buffer pre-treated with the Chelex resins and then manually mixed with 1 µL of 20 mM CoCl₂ in H₂O. UV-vis spectra were measured using a Cary 8454 UV-Vis spectrophotometer (Agilent).

Electrophoretic analysis of the thiol-disulfide status in SodC

We previously reported that SodC^SH and SodC^S-S exhibit distinct electrophoretic mobilities in polyacrylamide gels of SDS-PAGE; more precisely, reduction of the disulfide bond retarded the electrophoretic mobility of SodC.¹⁷ To examine the thiol-disulfide status in SodC, the proteins in the samples were first precipitated with 20% TCA, by which the thiol-disulfide status in the proteins can be quenched. The precipitates were washed with cold acetone to remove any residual TCA and then redissolved in the Laemmli sample buffer containing 100 µM iodoacetamide (IA). IA is a thiol-specific modifier that can protect thiols from being oxidized/modified during the electrophoresis. After incubation at 37°C for an hour, the samples were electrophoresed in a 15% polyacrylamide gel and then stained with Coomassie Brilliant Blue.

Superoxide dismutase activity assay

The assay was performed as described previously.²⁴ Briefly, an assay solution was prepared that was composed of 100 mM sodium phosphate (pH 8.0), 0.1 mM diethylenetriamine pentaacetic acid, 0.1 mM hypoxanthine, 50 µM WST-1, and 10 µg/mL catalase. The assay solution (200 µL) was first aliquoted into each well of a 96-well plate, to which SodC solutions (5 µL) serially diluted from 10 µM with 100 mM sodium phosphate (pH 8.0) containing 0.1 mM diethylenetriamine pentaacetic acid were added and mixed homogenously. Then, generation of superoxide was started by adding 5 µL of xanthine oxidase (Sigma, #X4500) that was pre-diluted 300-fold with H₂O, and reduction of WST-1 with superoxide was examined by monitoring the absorbance changes at 450 nm with a microplate reader (Epoch, BioTek). In our experimental conditions, the absorbance changes exhibited linear kinetics but not a saturation behavior until 10 min after starting the measurement. Using the absorbance changes at 5 min, therefore, the inhibition efficiency of the WST-1 reduction (%) was calculated and then plotted against the amount (ng) of SodC in the sample. The plots were fitted with an exponential curve, which can be used to calculate the amount of SodC that gives 50% inhibition (IC₅₀). In this study, IC₅₀ was regarded as a parameter representing a superoxide dismutase activity.

Results

SodC can bind a Zn²⁺ ion through Cys residues when the disulfide bond is reduced

To examine the ability of SodC to bind a Zn²⁺ ion, apo-SodC was first prepared by precipitation with TCA and then reacted with an equimolar amount of ZnSO₄ (100 µM). Free Zn²⁺ ions in the sample solution were removed by exchanging the buffer to a metal-free buffer with a centrifugal ultrafiltration device (10 000-fold dilution of the original buffer), and residual zinc and SodC in the sample were quantitated with atomic absorption spectrometry and UV spectrometry, respectively. As summarized in Fig. 1A, SodC^S-S proteins were found to retain most (∼90%) of the added Zn²⁺ ions even after the buffer-exchanges, but mutant SodC lacking all Cys residues (C74A/C169A, called SodC^noCys hereafter) retained only 10% of the added Zn²⁺ ions. These results reproduce our previous finding that the Cys74—Cys169 disulfide bond is required for binding a Zn²⁺ ion in SodC.¹⁷ We had thus initially expected that immature SodC in a disulfide-reduced state (i.e. SodC^SH) was incapable of binding a Zn²⁺ ion; however, it was not the case. Notably, SodC^SH was found to retain a significant amount (∼40%) of the added Zn²⁺ ions after the buffer-exchanging procedure (Fig. 1A).

We further examined the ability of SodC to bind a Zn²⁺ ion by a spectrophotometric assay with 4-(2-pyridylazo)resorcinol (PAR).²⁵ PAR binds a Zn²⁺ ion to form Zn²⁺(PAR)₂, which exhibits a distinct absorption peak at 492 nm (black curves, Fig. 1B). When 10 µM Zn²⁺(PAR)₂ (prepared by mixing 10 µM ZnSO₄ with 25 µM PAR) was mixed with 10 µM apo-SodC^S-S, a peak at 492 nm disappeared (a red curve, Fig. 1B), suggesting the transfer of Zn²⁺ from Zn²⁺(PAR)₂ to apo-SodC^S-S. In contrast, addition of apo-SodC^noCys to the Zn²⁺(PAR)₂ solution led to a slight decrease in the peak intensity (a green curve, Fig. 1B), consistent with the weakened Zn²⁺ affinity of SodC^noCys. As in SodC^noCys, the disulfide bond is absent in SodC^SH; however, addition of apo-SodC^SH was found to significantly decrease the absorption peak at 492nm (a blue curve, Fig. 1B). In a Fourier-transformed infrared (FT-IR) spectrum, the spectral shape of the amide I band in apo-SodC^noCys was almost the same with that of apo-SodC^SH (Fig. S1), suggesting little effects of the Cys-to-Ala substitutions on SodC at least in a secondary structural level. These results thus suggest that the thiol/thiolate groups of the Cys residues in SodC play important roles in the binding of a Zn²⁺ ion.

In the native form of SodC, a catalytic copper ion is bound at the ‘canonical’ copper-binding site consisting of His67, His69, His92, and His147 (Fig. 1C). Also, a structural zinc ion is bound at the ‘canonical’ zinc-binding site consisting of His92, His101, His109, and Asp112, among which His92 functions as a bridging ligand binding both Cu²⁺ and Zn²⁺ ions (Fig. 1C). It has been reported in several SOD1 variants that a zinc ion can be bound at the copper-binding site as well as the zinc-binding site.²⁶ Indeed, the absorption peak of Zn²⁺(PAR)₂ at 492 nm was decreased when it was incubated with an apo-form of mutant SodC^S-S lacking either the canonical copper-binding site (SodC with H67A/H69A/H147A triple mutations, SodC(noCu)) or the canonical zinc-binding site (SodC with H101A/H109A/D112A triple mutations, SodC(noZn)) (Fig. 1D). Those results showed the zinc binding in SodC(noCu)^S-S and SodC(noZn)^S-S, while the affinity of the mutant SodC proteins to a zinc ion appears to be decreased compared to SodC^S-S. Given that the copper-binding and the zinc-binding sites are connected with each other through the bridging ligand His92 and also a secondary bridge Asp151,²⁷ the Ala substitution of the ligands in one metal-binding site could weaken the zinc affinity at the other binding site.

When the disulfide bond in SodC was reduced, SodC(noZn)^SH still decreased the absorption peak of Zn²⁺(PAR)₂ at 492 nm, implying the binding of the zinc ion at the copper-binding site (red curve, Fig. 1E). In contrast, the equilibration of Zn²⁺ from the PAR complex to SodC(noCu)^SH was not significant (blue curve, Fig. 1E). These results suggest that, in the disulfide-reduced state, SodC binds a zinc ion predominantly through the ligands of the canonical copper-binding site. We should also note that those ligands are intact both in SodC^SH and SodC^noCys but tightly bind a zinc ion only in SodC^SH and not in SodC^noCys (Fig. 1B). Taken together, therefore, prior to the formation of the disulfide bond, SodC could bind a Zn²⁺ ion through the Cys residues together with the His residues constituting the canonical copper-binding site. Indeed, in the crystal structure of the native SodC protein (Fig. 1C), the disulfide bond between Cys74 and Cys169 appears to be in proximity to the His residues of the canonical copper-binding site.

SodC^SH binds a Cu²⁺ ion through the Cys residues with canonical copper-binding His residues

We also examined the ability of SodC to bind a Cu²⁺ ion by a spectrophotometric assay with PAR. PAR is known to form a 1:1 complex with Cu²⁺, producing a characteristic absorption peak at 508 nm.²⁵ When an equimolar amount of Cu²⁺PAR (10 µM; prepared by mixing 10 µM CuSO₄ with 25 µM PAR) was added to 10 µM apo-SodC^SH, the absorption peak at 508 nm was gradually decreased in its intensity within 10 min (Fig. 2A). While the spectral shapes of PAR and Cu²⁺PAR were slightly affected by the presence of the SodC proteins (possibly due to the interaction of PAR with the protein), the transfer of Cu²⁺ from PAR to SodC^SH was considered. Unlike Zn²⁺, Cu²⁺ is redox-active and can oxidize a thiolate group of a Cys residue to form a disulfide bond. Indeed, the intramolecular disulfide bond was introduced in significant fractions of SodC after the reaction with Cu²⁺PAR for 10 min (inset of Fig. 2A). Cu²⁺-mediated formation of the disulfide bond might precede the binding of a Cu²⁺ ion in SodC; however, an amount of Cu²⁺ ions transferred to apo-SodC^S-S was significantly less than that to SodC^SH (Fig. 2A and B). Those results might suggest that the thiol group of the Cys residue(s) in apo-SodC is required to extract the copper ion from Cu²⁺PAR. Nonetheless, this is not the case, because the absorption at 508 nm representing Cu²⁺PAR was further decreased when the concentration of apo-SodC^S-S was increased from 10 µM to 75 µM (Fig. S2). Instead, our results are consistent with the idea that SodC^S-S is characterized by a weaker affinity to Cu²⁺ than that of SodC^SH. It is also notable that the transfer of Cu²⁺ ions to SodC^noCys was almost negligible (Fig. 2C). These results thus suggest that a Cu²⁺ ion is first bound through the Cys residues of SodC^SH and could then facilitate autooxidation of the Cys residues to form the disulfide bond.

As proposed in the binding of a Zn²⁺ ion, the canonical zinc-binding site in SodC appears to have a lesser role in the binding of a Cu²⁺ ion when the disulfide bond is reduced. Removal of the canonical zinc-binding site from SodC did not significantly affect the transfer of Cu²⁺ from PAR to SodC(noZn)^SH and the disulfide formation (Fig. 2D); in contrast, both of the Cu²⁺ binding and the disulfide formation were suppressed by removal of the canonical copper-binding site (i.e. in SodC(noCu)^SH) (Fig. 2E). Taken together, SodC^SH was found to be equipped with the non-canonical, alternative metal binding site in which the Cys residues and the copper-binding His residues function as ligands for binding a Zn²⁺/Cu²⁺ ion.

Spectroscopic evidence of the Cys-mediated metal binding in SodC^SH using Co²⁺

We further characterized the alternative metal-binding site of SodC^SH by using a Co²⁺ ion as a model of the divalent metal ion, Zn²⁺/Cu²⁺. A Zn²⁺ ion with a d¹⁰ electron configuration is perceived to be spectroscopically silent, and a Cu²⁺ ion in proteins shows a d-d absorption band in the visible region but often with less distinctive features. In contrast, Co²⁺ is similar to Zn²⁺/Cu²⁺ in its ionic radius and exhibits a characteristic d-d absorption spectrum depending upon the types of ligands in a tetrahedral coordination.²⁸ Addition of an equimolar amount of CoCl₂ to 200 µM apo-SodC^S-S resulted in the appearance of an absorption peak at 562 nm due to a d-d transition (a black curve, Fig. 3A), indicating the binding of a Co²⁺ ion at the canonical zinc-binding site.^17,29 Apo-SodC^SH also produced the visible absorption upon the addition of Co²⁺, but the absorption peak of a d-d transition was characterized by a significant shoulder at 630 nm (a red curve, Fig. 3A) and was found to resemble that of Co²⁺-substituted peptides in which Co²⁺ was tetrahedrally coordinated with thiolates (Cys) and imidazoles (His).²⁸ Furthermore, the addition of Co²⁺ to SodC^SH increased the absorption from 400 to 300 nm (a red curve, Fig. 3A), which was considered to be due to the sulfur-to-cobalt charge transfer resulting from a thiolate coordination.³⁰ These results hence support our idea that SodC^SH can bind the divalent metal ions (Zn²⁺, Cu²⁺, and Co²⁺) at the alternative binding site consisting of the disulfide-bonding Cys residues.

Indeed, addition of Co²⁺ to SodC^noCys did not produce a visible absorption peak (a black curve, Fig. 3B), which is consistent with compromised binding of Zn²⁺/Cu²⁺ to SodC^noCys (Figs. 1B and 2C). To evaluate the contribution of Cys74 and Cys169 to the metal binding in SodC^SH, a visible spectrum of SodC with either C74A or C169A mutation was further examined in the presence of Co²⁺. As shown in Fig. 3B, the d-d absorption at 630 nm was observed in SodC with C169A but not with C74A upon addition of Co²⁺. Based upon the examination with the FT-IR spectroscopy, C74A and C169A mutant SodC proteins in the apo form were structurally similar to each other at least in a secondary structural level (Fig. S1). These results thus suggest that compared to Cys169, Cys74 plays more critical roles in the metal binding by SodC^SH.

Also notably, even in the absence of the canonical zinc-binding site, SodC(noZn)^SH with Co²⁺ showed the d-d absorption at 630 nm with the charge-transfer band from 400 to 300 nm; in contrast, SodC(noCu)^SH did not produce an absorption in the visible region (Fig. 3C). These results are consistent with our proposal that the canonical copper-binding site is involved in the Cu²⁺/Zn²⁺ binding of SodC^SH. To further examine which of the copper-binding ligands are involved in the binding of the divalent metal ions in SodC^SH, mutant SodC proteins were prepared in which alanine was substituted for each of the canonical copper-binding ligands, His67, His69, His92, and His147. As shown in Fig. 3D, the addition of Co²⁺ to H92A and H147A mutant SodC^SH proteins resulted in the appearance of the d-d absorption at 630 nm with the charge-transfer band (300–400 nm): in contrast, no obvious absorption in the visible region was observed in H67A and H69A mutant SodC^SH proteins. Collectively, therefore, it is considered that the alternative metal-binding site in SodC^SH is composed at least of His67, His69, and Cys74 (also see Fig. 1C).

SodC exhibits high affinity to Cu⁺ upon reduction of the disulfide bond

As described in the Introduction, SodC exclusively localizes at the periplasmic space of bacterial cells.¹⁸ Given that another periplasmic protein CueO is known to minimize potential toxicities of Cu⁺ by oxidizing it to Cu²⁺,³¹ SodC would interact with Cu²⁺ in the periplasmic space. Nonetheless, it is still obscure whether periplasmic copper ions exist as a cupric or cuprous state, leading us to further examine the binding of Cu⁺ with SodC proteins. Like Zn²⁺, Cu⁺ is a d¹⁰ electron configuration and exhibits no d-d absorption in the visible region. We thus utilized a bathocuproine disulfonate (BCS), which can specifically and tightly chelate Cu⁺ to form Cu⁺(BCS)₂ (β = 10^19.8 M^–2).³² Because Cu⁺(BCS)₂ but not BCS exhibits a distinct absorption peak at 492  nm, we can evaluate the binding of Cu⁺ with SodC by adding Cu⁺(BCS)₂ to SodC proteins and then monitoring decrease in the absorption at 492 nm. Also, Cu⁺ is expected to be easily air-oxidized to Cu²⁺; therefore, the reaction of SodC with Cu⁺(BCS)₂ was performed in a glove box filled with N₂.

As shown in Fig. 4A, Cu⁺(BCS)₂ exhibited the absorption peak at 492 nm (a black curve), and it was not affected by adding an equimolar amount of apo-SodC^S-S (a blue curve). In contrast, the addition of apo-SodC^SH led to significant decrease in the intensity of the absorption peak (a red curve, Fig. 4A). These results thus suggest that a Cu⁺ ion was transferred from Cu⁺(BCS)₂ to SodC^SH but not to SodC^S-S. Almost no changes in the absorption peak were observed when apo-SodC^noCys was mixed with Cu⁺(BCS)₂ (Fig. 4B); therefore, SodC is considered to acquire a Cu⁺ ion by utilizing the thiol/thiolate groups of the Cys residues. Furthermore, a Cu⁺ ion with a thiolate ligand is known to exhibit an absorption around 260 nm due to a charge-transfer from Cu⁺ to a thiolate ligand³³; indeed, the charge-transfer band was observed when a Cu⁺ ion ([Cu⁺(CH₃CN)₄](PF₆^–)) was added to apo-SodC^SH but not in apo-SodC^S-S and apo-SodC^noCys (Fig. 4C).

To test if the canonical copper and zinc-binding sites are involved in the binding of Cu⁺ with SodC^SH, the apo-forms of SodC(noCu) and SodC(noZn) were added to the Cu⁺(BCS)₂ solution. As shown in Fig. 4D, both of the mutant SodC proteins were found to decrease the intensity of the Cu⁺(BCS)₂ absorption peak when the disulfide bond was reduced. These results indicate little roles of those canonical metal-binding sites in the binding of Cu⁺ in SodC^SH, which is in sharp contrast to the involvement of the canonical copper-binding site in the binding of the divalent metal ions (Cu²⁺/Zn²⁺/Co²⁺) in SodC^SH (Figs. 1E, 2D, 2E, and 3C). Nonetheless, our results on the Cys-mediated binding of Cu⁺ to SodC^SH are consistent with many precedents in which Cu⁺ can be chelated by two Cys residues to form a two coordinate structure (e.g.³⁴). Taken together, reduction of the disulfide bond in SodC was found to significantly increase the binding affinity to Cu⁺, which is realized by the chelation of Cu⁺ with the disulfide-bonding Cys residues.

Effects of the thiol-disulfide status upon copper binding in Zn²⁺-bound SodC

SodC is an enzyme catalyzing the disproportionation of superoxide radicals, for which the disulfide formation and the binding of copper and zinc ions are required.¹⁷ While it remains obscure which of those processes occurs first in the maturation of SodC proteins in vivo, the binding of a zinc ion is considered to precede the binding of a copper ion and the disulfide formation in SOD1 proteins,¹³ which led us to test effects of the thiol-disulfide status on the copper binding of Zn²⁺-bound SodC.

As shown in Fig. 5A, addition of 10 µM Cu⁺(BCS)₂ to 10 µM Zn²⁺-bound SodC^SH was found to result in almost instantaneous disappearance of the absorption peak at 492 nm. As described in the Methods, the enzymatic activity can be represented by the amount of SodC that gives 50% inhibition of the water-soluble tetrazolium (WST-1) reduction by superoxide (IC₅₀). Despite the rapid transfer of Cu⁺ from BCS to Zn²⁺-bound SodC^SH, however, activation of SodC as the enzyme was found to be inefficient to the extent that IC₅₀ was 160 ± 12 ng compared to that of holo-SodC (5.4 ± 1.1 ng). This is probably because the disulfide bond was introduced into only a subset of the proteins (8.5 ± 2.1%) (inset of Fig. 5A). Indeed, when the solution (i.e. 10 µM Zn²⁺-bound SodC^SH with 10 µM Cu⁺(BCS)₂) was treated with 100 µM H₂O₂ in the glove box, the disulfide bond was introduced into significant fractions of SodC (56 ± 5.3%) (inset of Fig. 5A), and the enzymatic activity with 20 ± 5.2 ng of IC₅₀ was observed.

Addition of 10 µM Cu⁺(BCS)₂ to 10 µM Zn²⁺-bound SodC^S-S also decreased the intensity of the absorption peak at 492 nm (Fig. 5B), but it was significantly more sluggish than the reaction with SodC^SH. Notably, the apo-form of SodC^S-S was not able to receive Cu⁺ from BCS (Fig. 4A); SodC^S-S would increase the binding affinity to Cu⁺ in the presence of the bound Zn²⁺. Indeed, when Cu⁺(BCS)₂ was added to Zn²⁺-containing SodC(noZn)^S-S, the absorption peak at 492 nm did not change its intensity (Fig. S3). The binding of Zn²⁺ at the canonical zinc site is thus considered to facilitate the binding of Cu⁺ in SodC^S-S, while its mechanism needs to be investigated in more detail.

Given that the binding of a copper ion in Zn²⁺-bound SodC^S-S is expected to result in the activation, we suppose that limited amounts of SodC^S-S would be activated in an early phase (e.g. 1 min) of the reaction between Zn²⁺-bound SodC^S-S and Cu⁺(BCS)₂. Nonetheless, a significant level of the activity (IC₅₀ = 20 ± 4.3 ng) was observed; this is probably because it usually takes at least 30 minutes for the activity assay including the sample preparation. During such procedures for the activity assay, SodC is expected to bind a copper ion and thereby exhibit the activity. In environment maintaining a copper ion in the cuprous state, therefore, successful activation as the enzyme would be achieved with SodC^S-S albeit possibly sluggish, and also with SodC^SH albeit with the need of oxidizing equivalents.

Given that SodC proteins can also bind a cupric ion, we then tested the binding of Cu²⁺ to Zn²⁺-bound SodC^SH; namely, 10 µM Zn²⁺-bound SodC^SH was mixed with an equimolar amount of the Cu²⁺PAR complex prepared by mixing 10 µM CuSO₄ with 25 µM PAR. As shown in Fig. 5C, the absorption peak around 500 nm was unexpectedly increased compared to that of the Cu²⁺PAR complex alone (i.e. 10 µM CuSO₄ with 25 µM PAR, a black curve) and then gradually decreased. A careful inspection of the spectra revealed that the absorption was peaked at 492 nm but not at 508 nm, suggesting the dissociation of Zn²⁺ ion from SodC^SH and formation of the Zn²⁺(PAR)₂ complex upon addition of the Cu²⁺PAR solution. Given that addition of 25 µM PAR to 10 µM Zn²⁺-bound SodC^SH did not dissociate the zinc ion from SodC^SH (data not shown, but a spectrum similar to that shown in Fig. 1B was obtained), it is likely that Cu²⁺ replaced the Zn²⁺ ligated with the Cys residues in SodC^SH. As indicated in the inset of Fig. 5C, the disulfide bond was introduced into almost half as much as SodC proteins (43 ± 6.2% based upon densitometric analysis) after 10 min of the reaction, which is considered to be due to the copper-mediated autooxidation of the thiolate groups. Also, the decrease in the intensity of the absorption peak around 500 nm is consistent with the binding of the once-dissociated Zn²⁺ ion at the canonical zinc-binding site, resulting in the maturation of SodC. This is supported by significant enzymatic activity in the Zn²⁺-bound SodC^SH proteins reacted with Cu²⁺PAR for 10 min (IC₅₀ = 17 ± 3.2 ng vs. 5.4 ± 1.1 ng of holo SodC^S-S). Furthermore, when the Cu²⁺PAR solution was added to zinc-bound SodC(noZn)^SH, the absorption peak at 492 nm characteristic to Zn²⁺(PAR)₂ was observed, again suggesting the replacement of the protein-bound zinc ion with the copper ion (Fig. S4); unlike in SodC^SH, the intensity of the absorption around 500 nm remained almost unchanged during the measurement. This would reflect the lack of the canonical zinc-binding site in SodC(noZn); the once-dissociated Zn²⁺ ion could not be re-bound in the protein.

In contrast, addition of Cu²⁺PAR to Zn²⁺-bound SodC^S-S led to almost no changes in the spectrum from that of the Cu²⁺PAR complex alone (a black curve), and the decrease of the absorption intensity around 500 nm was sluggish (Fig. 5D). Also, the disulfide bond was maintained during the reaction (93 ± 3.3% based upon the densitometric analysis, inset of Fig. 5D); the reaction mixture after the incubation for 10 min showed weak enzymatic activity (IC₅₀ = 74 ± 19 ng), which was almost 1/10 of that of holo SodC^S-S (IC₅₀ = 5.4 ± 1.1 ng). While Cu²⁺ is generally known to exhibit high affinity to proteins and other biomolecules, our results suggest that SodC would be highly competitive for the copper ion when the thiol/thiolate groups are available prior to the formation of the disulfide bond.

Discussion

Roles of SodC in bacterial cells are yet to be established, but this protein is known to be well conserved, implying its importance in a certain physiological process(s) of bacteria. Indeed, previous studies have shown that the SodC activity in cells is significantly changed in response to several factors; during bacterial growth, for example, SodC can be detected only in the stationary phase, which is due to the transcriptional regulation by the stationary phase/stress sigma factor RpoS.⁷ Limiting the availability of copper and zinc ions in culture media by adding metal chelators is known to significantly reduce the sodC transcription in bacterial cells.^7,35 Furthermore, SodC is undetectable in anaerobically cultured E. coli cells, and Fnr has been considered as the primary effector of anaerobic sodC repression.^7,36 Regulation of the SodC activity appears to be largely conducted at the transcriptional level; however, it remains unknown how SodC matures as the active enzyme after the translation. Based upon the experimental results in vitro, we propose a mechanism for the post-translational control of the SodC maturation.

A proposed mechanism of SodC activation

As summarized in Fig. 6, we suggest that apo-SodC^SH, the most immature state, interacts with either copper or zinc ion by using His67, His69, and Cys74 as ligands. Given that Zn²⁺ and Cu²⁺ generally favor tetrahedral and trigonal bipyramidal coordination, respectively, other amino acid residues and/or water molecules would contribute to the metal binding. Also, preferential metal binding through the canonical copper ligands over the canonical zinc ligands is reminiscent of the metal-catalyzed SOD1 folding, where the canonical copper-binding site is contained within the folding nucleus and is thus able to coordinate a metal ion early in the folding process.³⁷ In our preliminary results with FT-IR spectrometry, an absorption peak at 1 641 cm^–1 representing a random coil structure was evident in apo-SodC^SH but not in holo-SodC^S-S, where a peak at 1 637 cm^–1 corresponding to β-sheet was observed instead (Fig. S5). Notably, the absorption peak at 1 637 cm^–1 (β-sheet) was also emerged by addition of a zinc ion to apo-SodC^SH, implying that the metal binding at the alternative site would facilitate the formation of secondary structures in SodC. While the overall structure of a metal-bound, disulfide-reduced SodC remains unknown, the canonical copper-binding site including His67 and His69 is a part of the SodC β-barrel scaffold; in contrast, the canonical zinc-binding site is placed in a loop region (Fig. 1C). Such distinct tendency to form the secondary structure might also contribute to preferential binding of metal ions at the alternative site (His67, His69, Cys74).

In our model, when Zn²⁺ is first bound at the alternative metal binding site, the following Cu²⁺ could replace the bound Zn²⁺ ion. While it remains unclear where the Zn²⁺ ion goes, at least two possibilities can be considered; one is an intramolecular shuttling of the Zn²⁺ from the alternative site to the canonical zinc-binding site. This shuttling would be difficult to directly prove but was previously discussed in human SOD1, where a Zn²⁺ ion is bound first at the canonical copper-binding site early in the folding and is then transferred to the thermodynamically more stable, canonical zinc-binding site after the protein passes the folding transition state.³⁷ The other possibility is that the Zn²⁺ ion dissociates from SodC, which would be supported by the appearance of the Zn²⁺(PAR)₂ absorption after addition of Cu²⁺PAR to the Zn²⁺-bound SodC^SH (Fig. 5C). Although we do not have concrete evidence to distinguish between those two possibilities, we have assumed for the present that SodC adopts the copper-bound state shown as A in Fig. 6 when Cu²⁺ is available, and the following introduction of a zinc ion into the canonical zinc-binding site would complete the maturation of SodC as the active enzyme (the last step in Fig. 6).

Regarding the copper-bound state shown as A in Fig. 6, the thiolate is known to be oxidized by the bound copper ion, and in aerobic environment, molecular oxygen would take one more electron to form a disulfide bond.³⁸ We thus suppose that the disulfide formation in the copper-bound state A could transfer the copper ion from the alternative to the canonical copper-binding site (B in Fig. 6). This is reminiscent of a proposed mechanism in which SOD1 first receives a cuprous ion from a copper chaperone CCS by using an ‘entry site’ and then intramolecularly transfers the copper ion to the canonical copper-binding site.^39,40 Similar to our proposing alternative site in SodC, the entry site is composed of Cys and His residues from SOD1 and CCS, and either the sulfenylation or the disulfide formation of the Cys residues in SOD1 has been suggested as a driving force for the copper transfer.^39,40 Also, a copper transfer from copper chaperones such as Atx1 to their respective partners occurs in a ligand-exchanging mechanism involving two pairs of Cys residues,⁴¹ and the conserved Cys residues of SOD1 might hence function as the ligands exchanging a copper ion with Cys residues from CCS.⁴² While the copper chaperone CCS is not conserved in bacterial cells, the Cys residues conserved in SOD1 and SodC would play a universal role in acquiring a copper ion from its intracellular sources.

The mechanism we proposed in Fig. 6 is not the only pathway for the activation of SodC. As studied in SOD1, it is well expected that several pathways operate to mature SodC as the active enzyme.^15,43 Consistent with the previous study,⁴⁴ indeed, SodC proteins were purified from the periplasmic fraction of E. coli cells cultured in the LB media (see Methods) and obtained mostly as the apo form (Zn/SodC ∼ 0.2; Cu/SodC ∼ 0) with the disulfide bond. Addition of copper and zinc salts to the periplasmic fraction was also found to activate SodC (data not shown). While the metal ions might be lost from SodC during the purification steps, our observation suggests an alternative maturation pathway where the copper/zinc binding occurs after the disulfide formation in SodC. In other words, SodC can be matured via a pathway that does not require the metal binding for the folding of SodC but likely involves DsbA for the folding by introducing the disulfide bond.²⁰ Also, we cannot exclude a possibility that a certain protein(s) functions as a metallochaperone for SodC and that the Cys residues might be important for receiving a copper ion from the potential metallochaperone.

Other than roles of a copper ion in the disulfide formation, furthermore, it is also interesting to note that the ligands of the alternative metal-binding site are close to each other in the primary sequence of SodC: namely, His67, His69, and Cys74 (Fig. 6). During the periplasmic transport, SodC shows up in the periplasmic space from its N-terminus. Once His67, His69, and Cys74 consecutively appear in the periplasmic space, they might immediately bind either copper or zinc ion, which could then function as a nucleus for the protein folding. When overexpressed in E. coli as recombinant proteins, the yield of SodC(noCu)^S-S was almost half as much as that of SodC(noZn)^S-S, which was similar to that of wild-type SodC^S-S. This is consistent with our idea that SodC becomes folded and protected from proteolytic degradation by binding metal ions during the periplasmic transport. In any cases, we found that the conserved Cys residues in SodC behave as ligands for metal ions, but further investigation will be definitely required for proving the significance in vivo of our proposed mechanism.

Conclusions

The Cys residues forming the intramolecular disulfide bond are conserved in SodCs among almost all species from bacteria to mammals. We have shown before that the disulfide formation can endow SOD1 and SodC with significant structural/thermal stability; indeed, losing the disulfide bond led to degradation and aggregation/misfolding of SodC and SOD1, respectively.^17,45 As summarized in Fig. 6, we have proposed here that when the disulfide bond is reduced, the Cys residues of SodC can bind and secure copper/zinc ions even in highly competitive environment for the metal ions. We thus suggest that the Cys residues play a dual role in SodC: the structural stabilization by forming the disulfide bond and the metal acquisition by acting as a ligand.

Acknowledgements

We thank Ms Kaori Sue for the preparation of SodC proteins and the metal analysis. We also thank Dr Yasuyuki Sakurai for his preliminary experiments and fruitful discussions.

Author Contributions

YF conceived and designed the work and wrote the manuscript; YF, AS, TK contributed to acquisition and analysis of the data.

Funding

This work was supported by Grants-in-Aid 19H05765 for Scientific Research on Innovative Areas (to YF) from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and also supported by Grants-in-Aid for Programs for the Advancement of Research in Core Projects under Keio University's Longevity Initiative and for the Advancement of Next Generation Research Projects from Keio University (to YF).

Conflicts of Interest

There are no conflicts of interest to declare.

Data Availability

Data available on request.

References