Structural basis and mechanism for metallochaperone-assisted assembly of the CuA center in cytochrome oxidase

See allHide authors and affiliations

Science Advances  31 Jul 2019:
Vol. 5, no. 7, eaaw8478
DOI: 10.1126/sciadv.aaw8478


The mechanisms underlying the biogenesis of the structurally unique, binuclear Cu1.5+•Cu1.5+ redox center (CuA) on subunit II (CoxB) of cytochrome oxidases have been a long-standing mystery. Here, we reconstituted the CoxB•CuA center in vitro from apo-CoxB and the holo-forms of the copper transfer chaperones ScoI and PcuC. A previously unknown, highly stable ScoI•Cu2+•CoxB complex was shown to be rapidly formed as the first intermediate in the pathway. Moreover, our structural data revealed that PcuC has two copper-binding sites, one each for Cu1+ and Cu2+, and that only PcuC•Cu1+•Cu2+ can release CoxB•Cu2+ from the ScoI•Cu2+•CoxB complex. The CoxB•CuA center was then formed quantitatively by transfer of Cu1+ from a second equivalent of PcuC•Cu1+•Cu2+ to CoxB•Cu2+. This metalation pathway is consistent with all available in vivo data and identifies the sources of the Cu ions required for CuA center formation and the order of their delivery to CoxB.


Cytochrome oxidase (Cox; EC is the terminal enzyme of the respiratory chain in mitochondria of eukaryotes and in many aerobic bacteria and archaea (1, 2). Located in the inner mitochondrial membrane and in the cytoplasmic membrane of prokaryotes, it receives electrons from reduced cytochrome c and transfers them to molecular oxygen as the final electron acceptor. Oxygen reduction is coupled to proton translocation (3), and the resulting electrochemical membrane potential drives adenosine 5′-triphosphate (ATP) synthase. The aa3-type Cox is the most thoroughly studied cytochrome oxidase (46). Although its overall subunit composition varies among different organisms, the catalytically active core is highly conserved, consisting of a membrane-integral subunit I (CoxA in bacteria; COX1 in mitochondria) and a membrane-anchored subunit II (CoxB or COX2) with a globular domain protruding into the mitochondrial intermembrane space or the periplasm of Gram-negative bacteria. A single electron from reduced cytochrome c is transferred to the exposed, binuclear CuA center of the globular CoxB domain and passed on to a low-spin heme a site and a high-spin heme a3-CuB center, the latter being the active site for O2 reduction to H2O (7).

In this study, we elucidate the reaction mechanisms underlying the biogenesis of the CuA center of CoxB. The term CuA refers to a mixed-valence, binuclear copper site, with the unpaired electron delocalized between two adjacent Cu ions in the globular CoxB domain (Cu1.5·Cu1.5+, oxidized form) (810). Incidentally, the CuA architecture occurs not only on CoxB but also on a few other metalloenzymes (11, 12). In CoxB, the two copper atoms are typically coordinated by the nitrogen atoms of two histidine imidazoles, the sulfur atoms of one methionine and two bridging cysteine thiols, and a protein backbone carbonyl oxygen (usually from a glutamate residue) present in the highly conserved amino acid sequence motif H-X34-CXEXCX3HX2M (ligands underlined) (6, 8). It seems compelling to assume that the binuclear CuA center originates from one Cu1+ ion and one Cu2+ ion (10), but how CuA assembly is achieved is still not fully understood. Part of the uncertainty is due, on the one hand, to the substantial diversity between organisms under investigation, which precludes the formulation of a unified assembly pathway. On the other hand, the peculiar topology and cellular location of CoxB·CuA might ask for different mechanisms of Cu incorporation, depending on whether assembly takes place in the mitochondrial intermembrane space, in the periplasm of Gram-negative bacteria, or on the extracellular surface of the cytoplasmic membrane in Gram-positive bacteria and in archaea. In all of these environments, however, Cu-binding metallochaperones (13) play a decisive role (see below). In general, these cuproproteins ensure coordinated and controlled Cu ion delivery to cytochrome oxidase, and prevent formation of reactive oxygen species or displacement of other transition metals from essential enzymes by free intracellular Cu ions (14, 15).

At this point, it is pertinent to introduce the organism studied in the authors’ laboratories: Bradyrhizobium diazoefficiens (formerly called Bradyrhizobium japonicum strain USDA 110) (16), a Gram-negative α-proteobacterium and, hence, one of the closest extant relatives of mitochondria. This trait allows interesting comparisons of related processes in a prokaryote versus the eukaryotic organelle. Respiration in this bacterium is highly flexible and diverse, as it has up to eight different oxygen reductases that are probably used for growth under different oxygen regimes (17). In cells growing strictly aerobically, the aa3-type cytochrome oxidase is the most prevalent oxygen reductase.

Regarding CuA formation, B. diazoefficiens was the first organism in which a specific thioredoxin-like reductase (TlpA) was found to satisfy an obligate requirement before metalation, i.e., the need to keep the two Cu-bridging cysteine ligands of the periplasmic CoxB domain in the reduced (dithiol) state (18). Just like CoxB, TlpA is anchored to the cytoplasmic membrane, and its thioredoxin-like domain faces the periplasm. Only reduced CoxB can serve as a target for copper insertion. In addition to TlpA, two different copper chaperones were identified previously in various organisms, including B. diazoefficiens, to be involved in Cu ion trafficking to CoxB, namely, ScoI and PcuC (13, 19, 20). ScoI is phylogenetically the more widespread of the two, occurring both in mitochondria and in many aerobic prokaryotes (alternative names are SCO1, SenC, PrrC, and YpmQ) (21, 22). PcuC, originally discovered in Deinococcus radiodurans and Thermus thermophilus [therein named PCuAC (23, 24)], was recognized as a second important Cu chaperone for CuA center formation in several bacteria, but not in mitochondria, in which COX17 (13) may act as a functional equivalent.

Akin to CoxB and TlpA, all known ScoI proteins are tethered to the membrane and have a periplasmic, thioredoxin-like domain. However, unlike the thioredoxin domain of TlpA with its Cys-Val-Pro-Cys (CVPC) active-site motif, the thioredoxin domain of ScoI carries a conserved CX3CP motif, in which the two cysteines form the copper-binding site together with a more distant histidine (2527). If these two vicinal cysteines are oxidized to a disulfide bridge, they need to be reduced by TlpA in B. diazoefficiens before Cu can bind (18, 28). In organisms that lack a TlpA-like protein, it was argued that ScoI itself or a second Sco homolog [e.g., SCO2 in mitochondria (13, 2931)] might fulfill this reductase function. While B. diazoefficiens ScoI (19) readily associates with Cu2+, the binding of Cu1+ to other ScoI-like proteins was reported (13, 3234), and a redox change in the course of Cu transfer was uncovered for the Bacillus subtilis ScoI-like protein (35). Binding and delivery of both Cu2+ and Cu1+ ions by Sco proteins might be an option in those organisms that lack a second Cu1+-specific chaperone such as PcuC (PCuAC). Where tested (20, 23, 36, 37), the specificity of the latter is the highest for Cu1+. PcuC is unrelated to ScoI in both amino acid sequence and structure. It has a cupredoxin fold carrying a novel type of copper-binding site with the consensus motif HXnMX21HXM, in which the two histidines and two methionines act as Cu1+ ligands (23, 24). Unlike the membrane-anchored ScoI, PcuC is a soluble protein in the bacterial periplasm. Null mutations in scoI- and pcuC-like genes in B. diazoefficiens cause a strongly diminished activity of the aa3-type oxidases (20). Wild-type oxidase activity can be restored by supplementing the growth medium with high Cu2+ concentrations (19, 20, 3739). Therefore, chaperone-independent assembly of the CuA center on cytochrome oxidases may be possible in the laboratory. In their natural habitat, however, bacteria will not encounter high concentrations of free Cu2+.

In recent studies on the cooperation between ScoI- and PcuC-like chaperones in the delivery of Cu to cytochrome oxidase, it was hypothesized that PcuC might function upstream of both ScoI and the Cox targets, but the mechanisms of Cu1+ and/or Cu2+ transfer to CoxB remained unknown (20, 37, 38, 40). In the present report, this question is addressed regarding CuA center formation by using the purified B. diazoefficiens proteins ScoI, PcuC, and CoxB and by applying an arsenal of techniques for the analysis of reaction kinetics, stoichiometry, and structural aspects of protein-metal and protein-protein interactions. We present the complete pathway for chaperone-mediated Cu ion transfer to CoxB, reconstituted in vitro. Specifically, we found that PcuC has distinct binding sites for Cu1+ and Cu2+, that a stable ScoI·Cu2+·CoxB complex is formed as an obligatory reaction intermediate, and that the ScoI·Cu2+·CoxB complex can only be resolved to the mature CoxB·CuA center when it specifically interacts with PcuC·Cu1+·Cu2+.


High-resolution crystal structures of ScoIox, ScoI·Cu2+, and CoxB·CuA

As a prelude to this study, we attempted to solve the x-ray structures of all relevant B. diazoefficiens copper proteins (ScoI, PcuC, and CoxB) to confirm that they showed the same overall fold and Cu-binding site architecture as their closest homologs with known three-dimensional structure. For this purpose, we produced and purified the soluble periplasmic domains of ScoI and CoxB (18.0 and 15.6 kDa, respectively) and soluble, mature PcuC (15.7 kDa). We succeeded in determining the x-ray structures of the periplasmic ScoI domain in its oxidized state [both cysteines in the Cu2+-binding site linked by a disulfide bond, no Cu2+ bound; Protein Data Bank (PDB) ID: 4WBJ] and in the reduced state in complex with Cu2+ (PDB ID: 4WBR) with very high resolutions of 1.3 and 1.4 Å, respectively (Fig. 1A, fig. S1A, and table S1). Likewise, the x-ray structure of the periplasmic CoxB domain containing the binuclear CuA center of cytochrome c oxidase (CoxB·CuA) (PDB ID: 4W9Z) was solved with the highest resolution (1.3 Å) reported thus far for a protein harboring a CuA center (Fig. 1B and table S1). In contrast, our attempts to crystallize apo-PcuC and holo-PcuC failed, which was possibly caused by the flexible, C-terminal extension of PcuC (see below). We therefore used the solved nuclear magnetic resonance (NMR) structure of the folded core of PCuAC from T. thermophilus (PDB ID: 2K6Z) (24) as a model for B. diazoefficiens PcuC, although PCuAC from T. thermophilus lacks the C-terminal extension of B. diazoefficiens PcuC (figs. S1F and S2; see also below).

Fig. 1 X-ray structures of the ScoI·Cu2+ and CoxB·CuA holo-proteins, and formation of the ScoI·Cu2+·CoxB complex.

(A and B) Cartoon representations of the 1.4-Å structure of the periplasmic ScoI domain (18.1 kDa) with bound Cu2+ (A) and the 1.3-Å structure of the periplasmic CoxB domain (15.6 kDa) harboring the CuA center (B). Top, overall folds; bottom, coordination spheres around the bound copper ions (orange). Residues coordinating Cu are shown as stick models, and distances between the liganding atoms and the Cu ions are given. (C) Cw EPR spectra of ScoI·Cu2+ (red) and the ScoI·Cu2+·CoxB complex (dark gray), normalized to identical concentration. (D) Cw EPR spectrum of CoxB·CuA. (E) Analysis of ScoI·Cu2+·CoxB complex (CP) formation at pH 7.0 and 25°C by gel filtration. CoxB (40 μM) was titrated with the indicated amounts of ScoI·Cu2+. (F) Titration of 15 μM CoxB with ScoI·Cu2+ at pH 7.0 and 25°C, recorded via the ScoI·Cu2+·CoxB–specific absorption at 520 nm. The sharp kink at the 1:1 ratio indicates that ScoI·Cu2+·CoxB is a high-affinity complex with a dissociation constant below 10−7 M. (G) Stopped-flow absorbance kinetics of ScoI·Cu2+·CoxB complex formation at 25°C and pH 7.0, recorded at 520 nm. CoxB (25 μM) was mixed with a 1.5-, 3-, or 5-fold excess of ScoI·Cu2+. Absorbance traces (dotted lines) were fitted globally (solid lines) to a mechanism assuming formation of an encounter complex (association: k1 = 1.5 ± 0.014 × 105 M−1 s−1; dissociation: k2 = 1.68 ± 0.04 s−1) followed by intramolecular rearrangement to the final complex (k3 = 1.20 ± 0.01 s−1).

ScoI displays a characteristic thioredoxin-like fold, composed of six α helices and nine β strands with a β-hairpin extension between α helix 4 and β strand 8 (26, 41). The CXXXC…H motif of ScoI with the Cu2+-coordinating cysteine pair C74/C78 is located in the loop segment 71–80 between strand β3 and helix α1. The Cu2+ ion in the ScoI·Cu2+ complex shows a distorted, squared planar coordination sphere. The Cu-SCys74 and Cu-SCys78 distances in the complex are both 2.1 Å, the Cu-Nε2His162 distance is 2.2 Å, and a water molecule is the fourth Cu2+ ligand (Fig. 1A). Superposition of the ScoI·Cu2+ structure with that of oxidized apo-ScoI showed that no major conformational changes occurred upon Cu2+ binding, and only small local structural differences around the Cys74-Cys78 disulfide bond of oxidized apo-ScoI were detected (fig. S1B). These two structures might differ from the structure of reduced apo-ScoI, in which the His-containing loop possibly adopts a different conformation.

Our solved x-ray structure of the periplasmic CoxB·CuA domain is virtually identical to the previously reported structure of holo-subunit II in the aa3-type cytochrome c oxidase from Paracoccus denitrificans (PDB ID: 3HB3) (fig. S1, C and D) (42). The high-resolution structure reported here revealed the environment of the delocalized mixed-valence (Cu1.5+·Cu1.5+) binuclear CoxB·CuA center in exquisite detail. The two copper ions are coordinated by the highly conserved copper-binding motif (HX34CXEXCX3HXXM). The distance between the two copper ions is 2.5 Å, which is in agreement with previously published CuA center structures (42, 43). Copper-ligand distances (Fig. 1B) are as follows: Cu1–Nδ1His237 and Cu2–Nδ1His194, 2.0 Å each; Cu2–SMet240, 2.4 Å; and Cu1 and Cu2 to S of Cys229 and Cys233, 2.3 Å each, giving rise to a highly symmetric CuA center reminiscent of Fe2S2 clusters (11). Comparison of the CoxB·CuA structure with the previously determined apo-CoxB structure (PDB ID: 4TXV) (18) revealed a major structural rearrangement upon CuA center formation, in which the loop segment 228–239 that adopts an open conformation in apo-CoxB closes over the binuclear CuA center in the CoxB·CuA structure (fig. S1E).

Figure 1 (C and D) depicts the characteristic continuous wave electron paramagnetic resonance (cw EPR) spectra of ScoI·Cu2+ and CoxB·CuA. They can be taken as fingerprints for these two metalloproteins that reflect the local electronic structure. ScoI·Cu2+ shows a well-resolved EPR spectrum, with three of four copper hyperfine lines clearly visible on the low-field side. Their center and splitting is described by a gz value of 2.158 and a hyperfine coupling of Az = 523 MHz, indicative of a coordination by two sulfur ligands (44). CoxB·CuA shows a spectrum typical (45) for a CuA site with a small and barely resolved hyperfine structure in the low-field region around a gz value of 2.198.

ScoI·Cu2+ forms a stable 1:1 complex with reduced CoxB

In the initial design of Cu transfer experiments, we started out with the naive idea that mature CoxB·CuA might be formed simply by mixing ScoI·Cu2+ with apo-CoxB, because formation of a disulfide bond in either CoxB or ScoI could potentially provide the reducing equivalents for generation of the Cu1+ ions necessary for CuA center formation (10). However, titration of reduced CoxB (40 μM) with increasing amounts of ScoI·Cu2+ did not yield CoxB·CuA. Instead, analytical gel filtration of the reaction products demonstrated that a previously unknown binary ScoI·Cu2+·CoxB complex (labeled CP in Fig. 1E) was quantitatively formed when ScoI·Cu2+ was mixed with equimolar amounts of CoxB, which eluted at the retention time expected for this 33.6-kDa complex (Fig. 1E). By contrast, mixing apo-ScoI with CoxB or ScoI·Cu2+ with oxidized CoxB did not result in the formation of a complex. The ScoI·Cu2+·CoxB complex was then purified on a preparative scale and further characterized. Its cw EPR Cu2+ spectrum features a copper hyperfine coupling of Az = 465 MHz centered around gz = 2.135 that is consistent with a four-sulfur ligand coordination (44) and clearly distinct from that of ScoI·Cu2+ and CoxB·CuA, showing the unique Cu2+ coordination sphere in the complex (Fig. 1C, gray) (44). In addition, the ScoI·Cu2+·CoxB complex exhibited a specific diagnostic absorbance maximum at 520 nm (manifesting its pinkish color in solution), which allowed us to confirm its 1:1 stoichiometry by absorbance titration (constant CoxB concentration of 15 μM, 0 to 5 molar equivalents (meq) of ScoI·Cu2+; Fig. 1F). The sharp kink in the titration profile indicates that ScoI·Cu2+·CoxB is a high-affinity complex with a dissociation constant (KD) below 10−7 M (Fig. 1F).

Stopped-flow absorbance kinetics of formation of ScoI·Cu2+·CoxB, recorded at 520 nm after mixing of CoxB (25 μM) with 1.5-, 3-, or 5-fold excess of ScoI·Cu2+, revealed rapid quantitative ScoI·Cu2+·CoxB formation within 10 s (Fig. 1G). The reaction mechanism, however, proved to be complex. Among several kinetic models tested, the best global fits were obtained for a mechanism with reversible formation of an encounter complex [ScoI·Cu2+·CoxB]*, followed by intramolecular rearrangement of the encounter complex to the final ScoI·Cu2+·CoxB complex (reaction scheme 1)ScoICu2++CoxBk1k2[ScoICu2+CoxB]*k3[ScoICu2+CoxB](1)

The deduced rate constants k1 to k3 are given in the figure legend (Fig. 1G). Specifically, the fast rate constant with which the encounter complex is formed (k1 = 1.5 ± 0.014 × 105 M−1 s−1) provides a strong hint that the reaction between ScoI·Cu2+ and CoxB is highly specific, as it is clearly above the threshold of 103 M−1 s−1 for rate constants of formation of physiologically relevant protein-protein complexes.

PcuC has a Cu1+-specific binding site and a second Cu2+-specific binding site

Apart from the Cu2+ chaperone ScoI, the periplasmic, Cu1+-specific chaperone PcuC is considered the second candidate protein acting as Cu transfer catalyst for CuA center biogenesis in B. diazoefficiens (20). Its homologs PCuAC from T. thermophilus (ThPCuAC) and Rhodobacter sphaeroides were shown to assist in the metalation of the CuA site of their ba3 and aa3 oxidases, respectively (24, 38). The NMR structure of the ThPCuAC·Cu1+ complex (PDB ID: 2K6Z) revealed a β-sheet fold harboring a conserved H(M)X10MX21HXM motif to which Cu1+ is bound (fig. S1F) (24). Like most other α-proteobacterial members of this chaperone family, B. diazoefficiens PcuC, however, differs from ThPCuAC in that a 23-residue extension is fused to the C terminus of its folded core domain. This extension contains a strikingly high content of typical copper-binding residues (five Met and two His residues). Although these residues are not conserved in the C-terminal extensions of other PcuC members, they share a high Met and His content with B. diazoefficiens PcuC (fig. S2). To test if the C-terminal extension B. diazoefficiens PcuC is also involved in copper binding and to determine its Cu ion specificity, PcuC was loaded with 4 meq of either Cu2+ or Cu1+ under anoxic conditions. After removal of excess copper by gel filtration, the samples were analyzed by electrospray ionization (ESI) mass spectrometry. With an expected mass increase relative to apo-PcuC (15,689.5 Da) of 63.5 Da per bound Cu ion, the measured masses of 15,813 Da for the holo-proteins showed that PcuC has two Cu-binding sites and is able to bind either two Cu1+ ions or two Cu2+ ions (Fig. 2A).

Fig. 2 Copper-binding properties of PcuC.

(A) ESI mass spectra of apo-PcuC (blue), PcuC·Cu1+·Cu1+ (pink), and PcuC·Cu2+·Cu2+ (violet). (B) EPR spectra (continuous wave, 9.5 GHz, 40 K), normalized to identical protein concentration, of PcuC·Cu2+·Cu2+ (violet), PcuC·Cu1+·Cu1+ in the presence of 50 mM Na-dithionite (gray), PcuC·Cu1+·Cu1+ in the absence of Na-dithionite (pink), and PcuC after incubation with 1.5 (green) or 3 meq (blue) of a Cu1+/Cu2+ mixture (1:1). Copper complexes were prepared under anaerobic conditions, and excess Cu ions were removed by buffer exchange with 20% glycerol, 20 mM Mops-NaOH (pH 7.0), and 50 mM NaCl before EPR measurements. (C) EPR determination of distances between the Cu2+ ions in PcuC·Cu2+·Cu2+ by four-pulse ultra-wideband DEER, with uncertainties within 15% root mean square deviation. The DEER distance distribution shows Cu(II)-Cu(II) interspin distances widely distributed with a full width at half maximum (FWHM) ranging from 1.6 to 2.6 nm. The small peak observed in the 3- to 4-nm distance region is at the edge of resolution for the achievable dipolar evolution time of 1.4 μs (fig. S3), is therefore affected by trace length and noise, and was not interpreted.

To determine the specificity of PcuC for Cu1+/Cu2+, we next recorded Cu2+-specific cw EPR spectra of PcuC after incubation with either only Cu1+ or only Cu2+ or 1:1 mixtures of Cu1+ and Cu2+ and desalting under anaerobic conditions. The EPR spectra of PcuC·Cu2+·Cu2+ and PcuC·Cu1+·Cu1+ served as reference spectra (Fig. 2B). PcuC·Cu1+·Cu1+ proved to be highly sensitive to oxidation, and the complete absence of Cu2+ in the EPR spectra could only be established in the presence of the reductant dithionite (Fig. 2B). Notably, identical EPR spectra were obtained for PcuC incubated with a 1.5- or 3-fold molar excess of the Cu1+/Cu2+ mixture, from which a Cu2+ content of ~1.3 Cu2+ per PcuC polypeptide could be deduced after normalization to identical protein concentrations (Fig. 2B). Within experimental error, this result indicated that PcuC has a Cu1+- and a Cu2+-specific binding site and selects one Cu1+ and one Cu2+ from solutions with excess Cu1+ and Cu2+. The presence of a highly specific Cu1+-specific binding site in PcuC that protects the bound Cu1+ from oxidation is supported by the observation that only one of the two Cu1+ ions proved to be oxidation insensitive when PcuC·Cu1+·Cu1+ was stored in the absence of dithionite (Fig. 2B). Together, the results indicate that the PcuC·Cu1+·Cu2+ complex is the thermodynamically preferred metalation state of the chaperone.

As Cu1+ is spectroscopically silent in EPR spectra, we measured the distance between the Cu2+ ions in PcuC·Cu2+·Cu2+ using a four-pulse ultra-wideband double electron-electron resonance (DEER) experiment (46, 47). The results showed that the two Cu2+ ions have a broad distance distribution (up to 4 nm), with most distances in the range between 1.6 and 2.6 nm (Fig. 2C and fig. S3). We assume that the distance between Cu1+ and Cu2+ in the PcuC·Cu1+·Cu2+ complex is in the same range because the Cu1+-specific binding site of PcuC can also be occupied by Cu2+, albeit with lower affinity (fig. S4).

PcuC has a high-affinity Cu1+-binding site in its folded core domain and a Cu2+-specific binding site in its C-terminal 23-residue extension

PcuC contains 10 methionines. Five Met residues are predicted to be located in the structured PcuC core domain, of which at least two are part of the conserved Cu1+-binding motif (see above). The other five Met residues are located in the C-terminal PcuC extension (fig. S2). To identify the Cu1+- and Cu2+-specific binding sites of PcuC, we labeled all 10 Met residues with Met-(methyl-13C) and measured changes in the 13C methyl NMR resonances upon binding of Cu1+ and Cu2+. To assign Met signals from the folded PcuC core and the C-terminal extension, we first compared the Met-(methyl-13C) resonances of full-length apo-PcuC with those of a Met-(methyl-13C)–labeled, truncated variant, apo-PcuC ΔC, lacking the C-terminal 23 residues. Superposition of the 2D [13C,1H]-HMQC NMR spectra of apo-PcuC and apo-PcuC ΔC identified the well-dispersed resonances of the five Met residues in the folded core, while the very low chemical shift dispersion of the five Met residues in the C-terminal extension indicated a lack of defined tertiary structure in this segment of apo-PcuC (Fig. 3A).

Fig. 3 Differential binding of Cu1+ and Cu2+to PcuC studied by NMR.

(A to F) Overlays of 2D-[13C,1H]-HMQC NMR spectra of Met-(methyl-13C)–labeled apo-PcuC and PcuC loaded with Cu1+ and/or Cu2+. Cartoons of PcuC are shown at the bottom of each panel, depicting the core domain (filled oval) and the flexible C-terminal extension (curved line). Cu1+ and Cu2+ are represented by yellow and orange balls, respectively. The colored circles around the cartoons indicate the color of the spectrum of each species. (A) C-terminally truncated PcuC (apo-PcuC ΔC) was used for identifying methionine signals from the flexibly disordered C-terminal segment (“C-term”) and the structured core domain. (B) Upon addition of 1 meq of Cu1+ to PcuC, only signals from the core domain shift, indicating preferential Cu1+ binding to the core domain. (C) Cu2+ binds to the C-terminal extension of PcuC, evidenced by the quenching of the signals from this region due to the paramagnetic properties of Cu2+. (D) In the presence of excess Cu1+, shifts of the methionine peaks in both the core and C-terminal extension indicate binding of Cu1+ ions to both sites. The sample contained 50 mM sodium dithionite to suppress Cu1+ oxidation. (E) In presence of 1.5 meq of both Cu1+ and Cu2+, Cu1+ preferentially binds to the conserved, primary Cu1+-binding site in the PcuC core. (F) Same sample of PcuC·Cu1+·Cu1+ as in (D) but in the absence of sodium dithionite. The Cu1+ bound at the C terminus became oxidized to Cu2+, leading to quenching of the C-terminal methionine peaks. In contrast, the Cu1+ bound to the PcuC core stayed resistant against oxidation.

Next, we prepared NMR samples of PcuC and PcuC ΔC in the presence of Cu1+ and/or Cu2+ and, when applicable, removed excess Cu ions by gel filtration under anoxic conditions before recording the NMR spectra. We expected that resonances of Met residues involved in Cu1+ binding would be shifted, whereas the signals of Met residues contributing to Cu2+ binding would be quenched due to the paramagnetic properties of Cu2+. Figure 3B shows that four of five Met signals from the PcuC core were shifted in the presence of 1 meq of Cu1+, with two signals shifting by more than 4 ppm in the 13C dimension, while the Met peaks from the C-terminal extension were not affected by Cu1+. Figure 3C shows the analogous NMR experiments in the presence of 1 meq of Cu2+ relative to PcuC, where Cu2+ binding only quenched the Met peaks from the C-terminal extension but left the resonances from the core domain unaffected. In addition, a small fraction of the resonances in the C-terminal extension remained visible, which was likely caused by a small excess of PcuC over Cu2+ in the sample. These resonances were significantly broadened compared to apo-PcuC, indicating fast, dynamic exchange of Cu2+ ions bound to the C-terminal extension, which was confirmed by NMR measurements recording Cu2+ binding to a synthetic 23-residue peptide corresponding to the C-terminal PcuC extension (fig. S5). In contrast to dynamic binding of Cu2+ to the C-terminal PcuC segment, binding of Cu1+ to the PcuC core domain exhibited slow exchange kinetics. Together, the results demonstrate that the PcuC core domain harbors the preferred binding site for Cu1+ and that the C-terminal extension is the preferred binding site for Cu2+. This was confirmed by spectra of PcuC after incubation with a 1:1 mixture of excess Cu1+/Cu2+, which showed complete quenching of the Met signals from the C-terminal extension by Cu2+ and the Met resonance shifts in the core domain diagnostic for Cu1+ binding (Fig. 3E). Notably, the spectra recorded for PcuC after incubation with a threefold excess of Cu1+ together with 50 mM dithionite showed peak shifts for the Met resonances from both the core domain and the C-terminal extension, confirming that Cu1+ can also bind to the C-terminal PcuC extension, albeit with lower affinity than to the core domain (Fig. 3D, see also Fig. 2A). In addition, comparison of the PcuC·Cu1+·Cu1+ spectra recorded in the presence and absence of the reductant dithionite showed that Cu1+ bound to the C-terminal PcuC extension was oxidation sensitive (leading to quenching of the C-terminal Met signals), whereas Cu1+ bound to the PcuC core was protected against air oxidation (Fig. 3, D and F), which may have been caused by the marginal air permeability of the top cover of the NMR tube. Together, our NMR experiments on PcuC (Fig. 3) are fully consistent with the mass spectrometry and EPR results shown in Fig. 2 (A and B) and demonstrate that (i) the primary Cu1+-binding site is located in the PcuC core, (ii) the primary Cu2+-binding site is in the C-terminal PcuC extension, and (iii) the C-terminal PcuC extension can also bind Cu1+.

Two molar equivalents of PcuC·Cu1+·Cu2+ are required for quantitative formation of the CoxB·CuA center from the ScoI·Cu2+·CoxB complex

The fast and specific formation of the stable ScoI·Cu2+·CoxB complex from ScoI·Cu2+ and CoxB (Fig. 1, E to G) indicated that ScoI·Cu2+·CoxB is an obligatory intermediate in CoxB·CuA biogenesis. Therefore, we tested whether apo-PcuC or PcuC in its different metalation states would be able to dissolve the ScoI·Cu2+·CoxB complex and/or form CoxB·CuA. For this purpose, the ScoI·Cu2+·CoxB complex was incubated overnight under anaerobic conditions with different amounts (0 to 3 meq) of either apo-PcuC, PcuC·Cu1+, or PcuC·Cu1+·Cu2+, and the reaction products were separated by analytical gel filtration, in which PcuC and ScoI exhibited identical retention times but could be well separated from ScoI·Cu2+·CoxB and apo-CoxB/CoxB·CuA (Fig. 4, A and B).

Fig. 4 Neither apo-PcuC nor PcuC·Cu1+ can release CoxB from the ScoI·Cu2+·CoxB complex.

(A) SDS-polyacrylamide gel (20%) of the purified protein components used in the titration experiments shown here and in Fig. 5. (C to E) Titration of ScoI·Cu2+·CoxB with apo-PcuC or PcuC·Cu1+ at pH 7.0 and 25°C, analyzed by analytical gel filtration. Eluted proteins were detected via their absorbance at 280 nm. The ScoI·Cu2+·CoxB complex (20 μM) was kept constant in all titration experiments and mixed with 0.25 to 3 meq of apo-PcuC or PcuC·Cu1+, and the reaction products were separated by gel filtration. (B) Gel filtration profiles of the individual components and controls used in the titration experiments shown here and in Figs. 1E and 5, demonstrating that (i) PcuC and ScoI exhibit the same retention times but can be well separated from ScoI·Cu2+·CoxB and CoxB and (ii) no complex is formed between apo-ScoI and CoxB and between ScoI·Cu2+ and oxidized CoxB (CoxBox). (C) Titration of ScoI·Cu2+·CoxB with apo-PcuC. Neither dissociation of ScoI·Cu2+·CoxB nor CoxB·CuA formation could be observed. (D) Titration of ScoI·Cu2+·CoxB with PcuC·Cu1+, showing that only a tiny fraction of ScoI·Cu2+·CoxB dissociated at high excess of PcuC·Cu1+. (E) Titration experiment from (D) but with protein detection at the CoxB·CuA–specific absorbance maximum of 813 nm instead of 280 nm, showing that no CoxB·CuA was formed. Purified CoxB·CuA (20 μM) was used as reference for 100% CoxB·CuA formation (magenta peak).

Figure 4C shows that apo-PcuC did not react with the ScoI·Cu2+·CoxB complex. In the presence of excess PcuC·Cu1+ over ScoI·Cu2+·CoxB, only a small fraction of CoxB was released from the ScoI·Cu2+·CoxB complex, but no CoxB·CuA was formed (Fig. 4, D and E). Recording the gel filtration profiles at both 280 and 813 nm (the CoxB·CuA–specific absorbance maximum; see fig. S6) demonstrated the complete absence of CuA in the fraction of released CoxB molecules (Fig. 4E). Therefore, Cu1+ cannot be transferred directly from its primary binding site in the PcuC core domain to CoxB. The inability of PcuC·Cu1+ of transferring the Cu1+ from the core domain was additionally confirmed with titration experiments using the Cu1+ indicator dye bathocuproine disulfonate (BCS), which forms a high-affinity Cu1+(BCS)2 complex with Cu1+, with a characteristic absorbance maximum at 483 nm (fig. S7) (48, 49). BCS failed to extract Cu1+ from PcuC·Cu1+ or PcuC·Cu1+ ΔC and only extracted the more weakly bound Cu1+ at the C-terminal PcuC extension from PcuC·Cu1+·Cu1+. Release of Cu1+ from the PcuC core domain could only be detected in triple variants of PcuC or PcuC ΔC, in which the Cu1+ chelating residues His79, His113, and Met115 of the core domain had been replaced by alanines (fig. S7, A to C). These results are consistent with the observation that a mutant B. diazoefficiens strain harboring only PcuC ΔC shows strongly decreased cytochrome oxidase activity compared with the wild type, although cytochrome oxidase activity was not completely absent like in the strain harboring the PcuC ΔHHM triple variant (fig. S8).

In contrast to mixing ScoI·Cu2+·CoxB with apo-PcuC or PcuC·Cu1+, which did not yield CoxB·CuA, we essentially achieved complete (97% yield) CoxB·CuA formation from the ScoI·Cu2+·CoxB complex when PcuC·Cu1+·Cu2+ was added to the complex (Fig. 5). Unexpectedly, however, titration of ScoI·Cu2+·CoxB with PcuC·Cu1+·Cu2+ revealed that a twofold excess of PcuC·Cu1+·Cu2+ over ScoI·Cu2+·CoxB was required for quantitative CoxB·CuA formation. Figure 5A shows the two-step reaction mechanism deduced from our titration experiments. In the first step (I) of CoxB·CuA formation, the first equivalent of PcuC·Cu1+·Cu2+ dissociates the ScoI·Cu2+·CoxB complex to CoxB·Cu2+, ScoI·Cu2+, and PcuC·Cu1+ (Fig. 5, B, C, and F, and fig. S9A), but CoxB·CuA is not quantitatively formed yet. Thus, only PcuC·Cu1+·Cu2+ can release CoxB·Cu2+ from the ScoI·Cu2+·CoxB complex, confirming that Cu1+ bound to the PcuC core domain cannot be transferred to CoxB. In the second reaction step (II), Cu1+ is then transferred from the second PcuC·Cu1+·Cu2+ equivalent to CoxB·Cu2+, yielding the genuine CoxB·CuA center with its characteristic absorbance maxima at 367, 479, and 813 nm (Fig. 5D and fig. S6). Strikingly, the results indicated that only copper ions bound to the C-terminal PcuC extension can be donated to CoxB and that an intramolecular transfer of a single electron from Cu1+ in the PcuC core to Cu2+ in the C-terminal PcuC extension occurred before the resulting Cu1+ in the C-terminal extension was transferred to CoxB·Cu2+ in step II. Intramolecular electron transfer from Cu1+ to Cu2+ in PcuC·Cu1+·Cu2+ is supported by the following observations: First, the two bound copper ions are sufficiently close (1.6 to 2.6 nm, Fig. 2C) to allow sufficiently fast electron transfer (50, 51). Second, our NMR titration experiments showed that the PcuC core can also bind Cu2+ instead of Cu1+ (fig. S4) and that the C-terminal PcuC extension can also bind Cu1+ instead of Cu2+ (Fig. 3D). Both findings are strict requirements for intramolecular electron transfer and consistent with a dynamic electron transfer equilibrium in which the PcuC state with Cu1+ in the core and Cu2+ in the C-terminal extension is thermodynamically favored. Recent studies on the PcuC homolog PccA from Rhodobacter capsulatus had shown that the core domain of PccA can also bind either Cu1+ or Cu2+, in accordance with our own findings, albeit a second Cu-binding site in its C-terminal extension that is 10 residues shorter than that of PcuC had not been reported (37). Third, as predicted by the mechanism in Fig. 5A, an NMR titration experiment in which Met-(methyl-13C)–labeled PcuC·Cu1+·Cu2+ was mixed with different amounts of ScoI·Cu2+·CoxB confirmed that the intensities of resonances specific to the Cu1+-bound PcuC core domain only started to decrease when more than one equivalent of PcuC·Cu1+·Cu2+ was added to ScoI·Cu2+·CoxB (fig. S9, A to D). This intensity decrease of Cu1+-bound core domain resonances was not accompanied by simultaneous appearance of resonances of the apo-core domain (fig. S9E), in agreement with the appearance of Cu2+ in the core domain as a consequence of intramolecular electron transfer.

Fig. 5 Two equivalents of PcuC·Cu1+·Cu2+ are required for conversion of ScoI·Cu2+·CoxB to CoxB·CuA.

(A) Mechanism of in vitro formation of CoxB·CuA (97% yield), deduced from the experiments shown in (B to H). (B and F) Titration of ScoI·Cu2+·CoxB (20 μM) with PcuC·Cu1+·Cu2+ at pH 7.0 and 25°C, analyzed by gel filtration (protein detection at 280 nm). ScoI·Cu2+·CoxB dissociation was completed after addition of one PcuC·Cu1+·Cu2+ equivalent. (C) Gel filtration runs from (B) but with protein detection at the CoxB·CuA–specific absorbance maximum at 813 nm. CoxB·CuA formation only reached its maximum after addition of two PcuC·Cu1+·Cu2+ equivalents to ScoI·Cu2+·CoxB. (D) Absorption spectra of the isolated CoxB·CuA peaks from (B) and (C), showing that two equivalents of PcuC·Cu1+·Cu2+ added to ScoI·Cu2+·CoxB quantitatively reconstituted the CoxB·CuA center (absorbance maxima at 367, 479, and 813 nm). (E) CoxB·CuA peak areas from (C) plotted against the added equivalents of PcuC·Cu1+·Cu2+. (G) Stopped-flow absorbance kinetics of the dissociation of the ScoI·Cu2+·CoxB complex (25 μM) by 1 equivalent (25 μM) of PcuC·Cu1+·Cu2+, yielding a second-order rate constant of 6.83 ± 0.02 × 102 M−1 s−1 (see Materials and Methods for the details). (H) Detection of cytochrome oxidase activity in B. diazoefficiens wild type (WT; positive control) and deletion mutants deficient in ScoI or PcuC. A coxB deletion mutant served as a negative control. Bacteria were grown aerobically in peptone–salt–yeast extract medium with or without 50 μM CuCl2. Identical amounts of cells were spotted onto a filter paper soaked with the indicator dye TMPD that reacts to indophenol blue when CoxB is active.

Notably, complete dissociation of the ScoI·Cu2+·CoxB complex was already achieved when only a single PcuC·Cu1+·Cu2+ equivalent was added to the complex (Fig. 5F). Stopped-flow absorbance kinetics of this reaction, recorded via the decrease in the absorbance of the complex at 520 nm, showed that the reaction is comparably fast, with a rate constant of ~700 M−1 s−1 (Fig. 5G). Figure 5 (C to E) shows that about 50% of the CoxB·Cu2+ molecules that had been generated after complex dissociation by one equivalent of PcuC·Cu1+·Cu2+ (reaction I) reacted further to CoxB·CuA after overnight incubation. We interpret this result such that CoxB·CuA can also be formed when ScoI·Cu2+·CoxB reacts with only one PcuC·Cu1+·Cu2+ equivalent but only very slowly (half-life above 10 hours). This alternative mechanism would require the regeneration of PcuC·Cu1+·Cu2+, the only species capable of donating Cu1+ to CoxB·Cu2+, by Cu2+ transfer from ScoI·Cu2+ to PcuC·Cu1+ (fig. S10A). However, as ScoI has a higher affinity for Cu2+ than PcuC·Cu1+ (fig. S10B), this reaction is energetically unfavorable and makes CoxB·CuA formation with only one equivalent of PcuC·Cu1+·Cu2+ so slow that this alternative pathway is most likely not physiologically relevant. Last, we also attempted to record formation of CoxB·CuA from CoxB·Cu2+ and PcuC·Cu1+·Cu2+ (reaction II in Fig. 5A) directly. This experiment, however, proved to be unfeasible, because we could not obtain stable preparations of CoxB·Cu2+. Specifically, a mixture of the disulfide bonded form of CoxB (CoxBox disulfide between the Cu-coordinating CoxB cysteines) and CoxB·CuA was formed very rapidly when apo-CoxB was mixed with Cu2+ in vitro according to reaction scheme 2, which previously had also been reported for the CoxB homolog from T. thermophilus (see also Materials and Methods) (10)2apoCoxB+3 Cu2+CoxBCuA+CoxBox+Cu1++2H+(2)

Although this reaction generates CoxB·CuA with a maximum yield of 50%, it is an artificial in vitro reaction, because there is no free Cu2+ in the cell (14, 52, 53). Therefore, we believe that the CoxB·Cu2+ complexes in the reaction scheme of Fig. 5A are only formed transiently in ScoI/PcuC-dependent CoxB·CuA formation and readily react with PcuC·Cu1+·Cu2+ to CoxB·CuA (reaction II in Fig. 5A). This was independently confirmed by cw EPR spectra that also demonstrated nearly complete formation (89%) of CoxB·CuA when ScoI·Cu2+·CoxB was mixed with two equivalents of PcuC·Cu1+·Cu2+ (fig. S11). Moreover, the data obtained from these cw EPR spectra revealed that ScoI retained a Cu2+ ion during CuA site formation (fig. S11), which is in agreement with the proposed reaction mechanism. Cu2+ was, however, slowly released from ScoI with increasing incubation time (fig. S11, B and C), most likely due to disulfide bond formation between the oxidation-prone active-site cysteines of ScoI (fig. S12).

Both chaperones, ScoI and PcuC, are required to form the CuA center of cytochrome oxidase in vivo

Our data presented so far showed that both ScoI·Cu2+ and PcuC·Cu1+·Cu2+ are required for quantitative formation of the CoxB·CuA center in vitro. In addition, apo-CoxB reacted with PCuC·Cu1+·Cu2+ alone (no ScoI present) to CoxB·CuA with only ~50% yield (fig. S13). Thus, our in vitro results predicted that both ScoI and PcuC are required for efficient formation of the CuA center in the CoxB subunit of aa3-type cytochrome oxidase in vivo. To test this, we generated B. diazoefficiens deletion mutants deficient in PcuC (ΔpcuABCDE), ScoI (ΔscoI), or CoxB (ΔcoxB). The strains were grown either under Cu-limiting conditions or in the presence of excess Cu2+ (50 μM CuCl2) in the medium. Cell extracts were tested for aa3-type cytochrome oxidase activity with the specific redox indicator dye tetramethyl-p-phenylenediamine (TMPD). The ΔpcuC and ΔscoI deletion mutants, like the CoxB deletion mutant ΔcoxB (negative control), both lacked cytochrome oxidase activity when grown under Cu-limiting conditions (Fig. 5H). Hence, the absence of one of the two copper chaperones could not be compensated by the other chaperone, and cytochrome oxidase activity in the ΔpcuC and ΔscoI mutants could only be restored with artificially high Cu2+ concentrations in the medium (Fig. 5H).


The electronically coupled di-copper CuA center on subunit II (CoxB) of cytochrome c oxidase is unique because of its highly symmetric architecture and its peculiar extracytoplasmic membrane topology. Its location in an oxidizing compartment (bacterial periplasm or mitochondrial intermembrane space) demands an inherently complex, multifactorial assembly process. Research of different laboratories using disparate model organisms had shown that a disulfide reductase for CoxB and two chaperones for Cu delivery (ScoI- and PcuC-like) are involved in CoxB·CuA biogenesis. Although a variety of scenarios for CuA formation had been proposed [reviewed in (54)], our mechanistic understanding of how the assembly factors cooperate remained incomplete. Previous genetic and physiological investigations in vivo (19, 20, 3840) were limited by the difficulty to assign specific functions to individual CuA biogenesis factors because of their intimate interactions and interdependencies. In the present study, we have taken a rigorously different approach by reconstituting CoxB·CuA assembly in vitro using the purified B. diazoefficiens CoxB protein as target and the purified, copper-loaded metallochaperones ScoI·Cu2+ and PcuC·Cu1+·Cu2+ for copper delivery. A detailed and comprehensive mechanism of CuA assembly has emerged, which is summarized and illustrated in Fig. 6. The entire pathway of copper insertion into CoxB can be dissected in three steps (Fig. 6).

Fig. 6 Model for copper trafficking to the CoxB subunit of aa3-type cytochrome c oxidase in the periplasm of B. diazoefficiens.

CuA center formation in subunit II (CoxB) of aa3 oxidase of B. diazoefficiens requires soluble periplasmic PcuC (violet) and membrane-anchored ScoI (red). The periplasmic thioredoxin-like reductase TlpA maintains the active-site cysteine pairs of ScoI and CoxB in the reduced (dithiol) state that is required for Cu2+ binding. The Cu2+-specific chaperone ScoI readily reacts with apo-CoxB to a stable ScoI·Cu2+·CoxB complex (reaction I). The preferred metalation state of PcuC has a Cu1+ ion bound to the folded PcuC core domain and a Cu2+ ion bound to its flexible, C-terminal extension. PcuC·Cu1+·Cu2+ specifically reacts with the ScoI·Cu2+·CoxB complex and releases CoxB·Cu2+ from the complex (reaction II). A second equivalent of PcuC·Cu1+·Cu2+ then transfers Cu1+ to CoxB·Cu2+, and the CoxB·CuA center is formed (reaction III). As only Cu ions bound to the C-terminal PcuC extension can be transferred to CoxB, donation of Cu1+ to CoxB·Cu2+ requires intramolecular transfer of an electron from Cu1+ in the core domain to Cu2+ at the C-terminal PcuC extension. The overall reaction scheme at the bottom shows that, formally, ScoI·Cu2+ is not consumed in ScoI/PcuC-mediated CoxB·CuA biogenesis, indicating that catalytic amounts of ScoI relative to CoxB might be sufficient for efficient CoxB·CuA center formation in vivo.

In step I, ScoI·Cu2+ forms a 1:1 complex with apo-CoxB (Fig. 6). The ScoI·Cu2+·CoxB complex not only formed very rapidly but proved to be remarkably stable against dissociation and oxidation of the active-site cysteine pairs in ScoI and CoxB. By contrast, we noticed a relatively fast Cu2+ loss from isolated ScoI·Cu2+ (fig. S12), most likely due to Cu2+-mediated oxidation of the ScoI cysteine pair, similar to oxidation of the CoxB thiol pair (10) by Cu2+. In addition, the periplasmic domain of apo-CoxB proved to be only marginally stable and aggregated unspecifically at concentrations above 100 μM. These results strongly support the notion that ScoI·Cu2+·CoxB is an obligatory intermediate in CuA center formation. Besides its stabilizing function for CoxB and ScoI, it may also function as temporary Cu2+ depository preventing Cu-provoked oxidative stress. The stability of ScoI·Cu2+·CoxB is reminiscent of PmoD from methane-oxidizing microbes, a homodimeric copper-binding protein forming a bimolecular CuA-like site at the subunit/subunit interface in which the two Cu-bridging cysteines might be similarly protected from oxidation (12). In the B. diazoefficiens ScoI·Cu2+·CoxB complex, the Cu2+ ion is most likely coordinated by four cysteines (one cysteine pair from each protein), giving rise to a previously unknown, pink metal center with its characteristic absorbance and EPR spectra (Fig. 1C and fig. S6).

Step II in CoxB·CuA biogenesis is the dissociation of the ScoI·Cu2+·CoxB complex by PcuC·Cu1+·Cu2+ and the transfer of Cu2+ from the C-terminal PcuC-extension to ScoI, and step III is the transfer of Cu1+ from a second equivalent of PcuC·Cu1+·Cu2+ to CoxB·Cu2+ (Fig. 6). The following discoveries, based on a combination of mass spectrometry, EPR and NMR spectroscopy, titration experiments, and stopped-flow kinetics, proved to be critical for establishing these reactions.

First, we demonstrated that PcuC binds one Cu1+ and one Cu2+ ion. In addition to the previously described Cu1+-binding site in the PcuC core domain (24), a Cu2+ ion specifically binds to the C-terminal PcuC extension. A Cu2+-binding site is most likely also present in the Met- and His-rich C-terminal extensions of other PcuC-like proteins but may have been missed in previous studies due to the lack of obvious sequence conservation (fig. S2). Our results on B. diazoefficiens PcuC, together with previously reported results on T. thermophilus PCuAC (24), indicate that PCuAC/PcuC-like proteins can be divided into two functionally distinct groups, namely, PCuAC domains with and without His/Met-rich C-terminal extension. Among the 920 representative proteomes in the Pfam database with 15% co-membership threshold ( (55), 674 representative proteomes contain a PCuAC-like protein domain. Of these, only 4% (including PCuAC from T. thermophilus) lack a C-terminal extension, while the other members have a C-terminal extension with median length of 20 residues in which Met and His residues are enriched two- and fourfold, respectively, relative to their average abundance pattern in proteins. It thus appears as if PCuAC/PcuC-like proteins with a Met/His-rich C-terminal extension that might harbor a second Cu-binding site are rather frequent, and that the mechanism of PcuC function in B. diazoefficiens, as characterized in our study, could apply to a number of other PcuC homologs. Some PcuC-like proteins (e.g., that of P. denitrificans) lack His or Met residues in their C-terminal region but have additional Met or His residues near their N terminus to potentially form a second Cu-binding site (UniProtKB: A1BAG4).

Second, we showed that only the Cu ion (Cu2+ or Cu1+) bound to the C-terminal PcuC extension can be transferred to CoxB and that the C-terminal PcuC extension is strictly required for the dissociation of ScoI·Cu2+·CoxB by PcuC·Cu1+·Cu2+. The intrinsic flexibility of the C-terminal PcuC extension (Figs. 2C and 3A) might enable PcuC·Cu1+·Cu2+ to target the ScoI/CoxB interface. This reaction is highly specific: ScoI·Cu2+·CoxB could be dissociated under physiological buffer conditions only with PcuC·Cu1+·Cu2+, which readily explains why PcuC is indispensable for CoxB·CuA formation in vivo under Cu-limiting conditions (Fig. 5H and fig. S8).

Third, our data suggest that a single electron can be transferred from Cu1+ in the PcuC core to the Cu2+ at the C-terminal extension, which gives PcuC the unique ability to transfer either Cu1+ or Cu2+ to its substrates. A putative redox reaction in the course of Cu transfer from either ScoI- or PcuC-like chaperones has been a matter of debate for years (54). Notably, Abriata et al. (24) quantitatively reconstituted the CuA center in cytochrome oxidase subunit II of T. thermophilus by addition of two equivalents of PCuAC·Cu1+, in which aerobic oxidation of one of the two Cu1+ ions to Cu2+ led to the formation of the mixed-valence CuA center. In our B. diazoefficiens system, we could not achieve complete CoxB·CuA center formation by incubating apo-CoxB with excess PcuC·Cu1+·Cu2+ (fig. S13). Together with the observation that both PcuC and ScoI are required for the formation of a functional terminal oxidase in vivo (Fig. 5H), this rather excludes the mechanism of CoxB·CuA formation in T. thermophilus for B. diazoefficiens. Instead, the results indicate that there are distinct mechanisms of CuA center biogenesis in bacteria and that pathways of CuA center formation analogous to those in T. thermophilus or B. diazoefficiens can be distinguished via the presence or absence of a Met/His-rich C-terminal PcuC extension in the respective proteome.

We are fully aware that the three-step mechanism of CoxB·CuA formation in B. diazoefficiens depicted in Fig. 6 is a simplified reaction scheme that does not address the mechanisms of loading of ScoI with Cu2+ and PcuC with Cu1+/Cu2+ in vivo, which certainly involves further upstream Cu uptake and delivery systems [reviewed in (14, 56)]. To which extent the mechanism established here for CoxB·CuA assembly in B. diazoefficiens can be extrapolated to other organisms remains another open question. In any case, our results allow us to make predictions as to whether bacterial organisms having ScoI- and PcuC-like metallochaperones might share the CuA assembly mechanism with that of B. diazoefficiens. The length of the C-terminal extension in the respective PcuC orthologs and the number of Met/His residues therein could serve as conspicuous indicators. The predictions then ought to be validated by in vivo and in vitro experiments analogous to those reported in this study. Even in mitochondria that have Sco1/Sco2 but lack a PcuC homolog, the first step—formation of a complex between ScoI·Cu1+ and COX2 or ScoI·Cu2+ and COX2 analogous to ScoI·Cu2+·CoxB—ought to be considered and examined, because there is ample evidence for physical interactions between Sco and cytochrome oxidase subunit II (COX2) (13, 54). If a Cu-bridged Sco1-COX2 complex were formed in mitochondria, a promising candidate to resolve that complex in a PcuC-like manner would be the mitochondrial Cu chaperone COX17. Cu transfer from COX17 to at least Sco1 has already been demonstrated (57).

Last, we would like to point to another interesting aspect of ScoI- and PcuC-assisted CoxB·CuA center formation in B. diazoefficiens as depicted in Fig. 6: The total reaction (sum of individual steps I to III) shows that ScoI·Cu2+ is not consumed during CoxB·CuA formation and therefore can be considered, at least formally, a catalyst. Consequently, substoichiometric amounts of ScoI relative to CoxB might be sufficient for quantitative CoxB·CuA formation in vivo.


Expression plasmids

Bacterial expression plasmids for the production of the periplasmic domains of CoxB (residues 128 to 265) and ScoI (residues 34 to 196) (18) and the expression plasmid for PcuC (for residues, see fig. S3) (20) were described before. Plasmid variants were constructed using standard molecular cloning techniques. Detailed information on the individual plasmids including the protein-encoding nucleotide sequences is available from the authors on request.

Protein production and purification

For expression of CoxB in inclusion bodies and the cytoplasmic expression of ScoI and PcuC under T7 promoter/lac operator control, Escherichia coli BL21 (DE3) carrying the corresponding expression plasmid was grown at 37°C in 2YT medium containing ampicillin (100 μg/ml) until an OD600 (optical density at 600 nm) of 0.6 had been reached. Expression was induced by the addition of isopropyl-β-d-thiogalactoside (final concentration, 0.1 mM), and cells were further grown at 30°C for 4 hours and harvested by centrifugation. Proteins were purified according to the following protocols:

CoxB: Cells were suspended in 100 mM tris-HCl (pH 8.0) and 1 mM EDTA (3 ml/g wet cells), mixed with deoxyribonuclease I (DNase I) (50 μg/ml final concentration), and lysed with a microfluidizer (M-110L, Microfluidics, Westwood, MA). After the addition of 0.5 volume of 60 mM EDTA-NaOH (pH 7.0), 1.5 M NaCl, and 6% (v/v) Triton X-100, the lysate was stirred at 4°C for 1 hour. The CoxB inclusion bodies were harvested by centrifugation (30 min, 47,850g, 4°C) and washed five times at 4°C with 100 mM tris-HCl (pH 8.0) and 20 mM EDTA to remove Triton X-100. The inclusion bodies were solubilized in 100 mM tris-HCl (pH 8.0), 6 M guanidinium chloride, 1 mM EDTA, and 100 mM dithiothreitol (DTT) (20 ml/g of inclusion bodies) at room temperature under stirring for 2 hours. Insoluble material was removed by centrifugation (30 min, 142,159g, 20°C). CoxB was refolded at room temperature from the supernatant by rapid, 100-fold dilution with refolding buffer [0.5 M arginine, 1 mM EDTA, 10 mM DTT, and 20 mM tris-HCl (pH 8.0)] and stirring for 1 hour. Refolded and reduced CoxB was concentrated at 4°C to ca. 1.5 mg/ml and dialyzed against 2 × 10 volumes of 20 mM tris-HCl (pH 8.5), 1 mM EDTA, 5 mM DTT, and 1 × 10 volumes of 20 mM tris-HCl (pH 8.5). Precipitated protein was removed by centrifugation, and the supernatant was applied at 4°C to a Resource Q column (GE Healthcare) equilibrated with 20 mM tris-HCl (pH 8.5). CoxB was eluted with a linear NaCl gradient (0 to 1.0 M). Fractions containing pure CoxB according to SDS–polyacrylamide gel electrophoresis (PAGE) were combined; supplemented with 5 mM DTT, 1 mM EDTA, and 5% glycerol; frozen in liquid nitrogen; and stored at −20°C until further use. The final yield of purified CoxB was 60 mg/liter of bacterial culture.

ScoI: Cells were suspended at 4°C in 50 mM acetic acid–NaOH, 1 mM EDTA, 2 mM DTT, 1 mM phenylmethylsulfonyl fluoride (PMSF), DNase I (50 μg/ml), and cOmplete protease inhibitor mix (1 tablet/60 ml, Roche Applied Science) (3 ml/g wet cells) and lysed with a microfluidizer (M-110L, Microfluidics). After centrifugation, the supernatant was filtered (0.2-μm pore size) and loaded onto an SP Sepharose column (GE Healthcare) equilibrated with 50 mM acetic acid–NaOH, 1 mM EDTA, and 2 mM DTT. ScoI was eluted with a linear NaCl gradient (0 to 0.5 M). ScoI-containing fractions were pooled, dialyzed at 4°C against 20 mM tris-HCl (pH 9.0), and loaded onto a Source 30Q column (GE Healthcare) equilibrated with the same buffer. ScoI was eluted with a linear NaCl gradient (0 to 0.5 M). ScoI-containing fractions were pooled, concentrated, and loaded onto a Superdex 75 (HiLoad 26/60) column (GE Healthcare) equilibrated with 20 mM tris-Cl (pH 8.5), 0.3 M NaCl, 5 mM DTT, and 0.5 mM EDTA. Fractions containing pure ScoI according to SDS-PAGE were combined, frozen in liquid nitrogen, and stored at −20°C until further use. The final yield of purified ScoI was 30 mg/liter of bacterial culture.

PcuC variants: PcuC and PcuC ΔC were produced as fusion proteins with N-terminal His10 tag, cleavable with tobacco etch virus (TEV) protease. Cells were suspended at 4°C in 50 mM Hepes-NaOH (pH 7.5), 300 mM NaCl, 1 mM PMSF, cOmplete protease inhibitor mix (1 tablet/60 ml), and DNase I (50 μg/ml) (3 ml/g wet cells) and lysed with a microfluidizer (M-110L, Microfluidics). PcuC variants were purified from the soluble fractions of the lysates by application to prepacked Ni2+-NTA (nitrilotriacetic acid) columns (GE Healthcare) equilibrated with 50 mM Hepes-NaOH (pH 7.5), 300 mM NaCl, and 10 mM imidazole. The proteins were eluted with a linear imidazole gradient (10 to 250 mM). PcuC-containing fractions were pooled, and the His10-tags were removed by TEV protease (40 μg/ml; protease/substrate ratio, 1:25) cleavage during overnight dialysis at 4°C against 50 mM Hepes-NaOH (pH 7.5) and 300 mM NaCl. The cleaved His10-tag and TEV protease [His6-tagged variant (58)] were removed by loading the sample onto a Ni2+-NTA column. Flow-through fractions containing PcuC were pooled and incubated with 50 mM EDTA overnight, concentrated, and loaded onto a Superdex 75 (HiLoad 16/60) column (GE Healthcare) equilibrated with 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl. apo-PcuC–containing fractions were pooled, frozen in liquid nitrogen, and stored at −20°C until further use. The final yield of purified apo-PcuC and apo-PcuC ΔC was 26 and 10 mg/liter of bacterial culture, respectively.

Copper loading of ScoI, CoxB, and PcuC

As the active-site cysteine pairs of ScoI and CoxB need to be in the dithiol form for copper binding, both proteins were treated with reducing agents before loading with Cu ions. ScoI was incubated overnight at 4°C with 10 mM DTT, 20 mM tris-Cl (pH 8.5), 0.3 M NaCl, and 0.5 mM EDTA. CoxB (20 mM) was incubated with 5 to 10 mM tris(2-carboxyethyl)phosphine (TCEP) overnight at 4°C in 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl. Reduced proteins were buffer-exchanged against 20 mM Mops-NaOH (pH 6.5), and concentrations were determined via their absorbance at 280 nm. For loading of ScoI with Cu2+, reduced apo-ScoI was mixed with 1.5 meq of CuCl2 and incubated for 15 min at room temperature. For preparation of CoxB·CuA, reduced apo-CoxB (10 to 20 μM) was mixed with 3 meq of CuCl2 and incubated for 5 min at room temperature. The CoxB·CuA–specific pink color formed immediately after mixing. Cu-loaded proteins were desalted and buffer-exchanged against 20 mM Mops-NaOH (pH 7.0) by gel filtration to remove excess, unbound Cu ions and subjected to a final purification step by hydrophobic chromatography. The ScoI·Cu2+ and CoxB·CuA solutions were supplemented with 1 M ammonium sulfate and 1.5 M ammonium sulfate, respectively, and the holo-proteins were separated from the apo-proteins on Butyl Sepharose 4 Fast Flow columns (GE Healthcare). ScoI·Cu2+ and CoxB·CuA were eluted with a linear gradient from 1 or 1.5 M to 0 M ammonium sulfate, respectively. Separation from the apo-protein was monitored using the ScoI·Cu2+–specific absorbance at 360 nm or the CoxB·CuA–specific absorbance at 479 nm. Fractions containing the holo-proteins were pooled, buffer-exchanged against 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl, concentrated, and stored at 4°C until further use within 48 hours.

Samples with PcuC variants were transferred to an anaerobic chamber (Coy Laboratory); loaded with different amounts of Cu1+ [tetrakis(acetonitrile)copper(I) hexafluorophosphate, Sigma-Aldrich, anaerobically dissolved in dimethyl sulfoxide], Cu2+ (CuCl2 in H2O), or a mixture of both; and incubated overnight at 5°C. Cu-loaded PcuC samples were desalted in the anaerobic chamber by gel filtration over PD MidiTrap G-25 columns (GE Healthcare), their concentration was measured, and protein samples were directly used for the experiments.

Purification of the ScoI·Cu2+·CoxB complex

Reduced apo-CoxB was mixed with equimolar amounts of freshly purified ScoI·Cu2+ in 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl (final protein concentrations, 20 to 30 μM). The pink ScoI·Cu2+·CoxB complex formed within seconds after mixing and was loaded onto a Superdex 75 (16/60) column equilibrated with 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl. Fractions containing ScoI·Cu2+·CoxB were pooled, concentrated with ultrafiltration (10 kDa cutoff), and immediately used for further experiments.

Determination of protein concentrations

Protein concentrations were measured via the specific absorbance at 280 nm using the following extinction coefficients: CoxBSS (28,085 M−1 cm−1), apo-CoxB (27,960 M−1 cm−1), CoxB·CuA (26,950 M−1 cm−1), ScoISS (10,555 M−1 cm−1), apo-ScoI (10,430 M−1 cm−1), ScoI·Cu2+ (10,430 M−1 cm−1), ScoI·Cu2+·CoxB complex (38,390 M−1 cm−1), apo- and holo-PcuC, PcuC ΔC (8480 M−1 cm−1), and TEV-protease (32,290 M−1 cm−1).

Crystallization of ScoIox, ScoI·Cu2+, and CoxB·CuA

Crystallization experiments were carried out with the sitting-drop vapor diffusion method. ScoIox [0.15 μl, 54 mg/ml in 10 mM tris-HCl (pH 8.0)] was mixed with 0.15 μl of precipitant [25% (w/v) polyethylene glycol (PEG) 1500, 0.1 M succinic acid/sodium phosphate/glycine (SPG) buffer (pH 4.0)], and crystals (space group P 1 21 1) grew within several weeks at 20°C. ScoI·Cu2+ [0.1 to 0.15 μl, 47 mg/ml in 10 mM tris-HCl (pH 8.0)] was mixed with 0.15 to 0.2 μl of 20% (w/v) PEG 6000 and 0.1 M Na-citrate (pH 5.0) at 20°C. Crystals (space group P 21 21 2) grew within several weeks at 20°C. Crystals of CoxB·CuA were obtained as a side product of our attempts to crystallize the ScoI·Cu2+·CoxB complex. ScoI·Cu2+·CoxB [0.1 to 0.15 μl, 25.9 mg/ml in 5 mM Mops-NaOH (pH 7.0)] was mixed with 0.15 to 0.2 μl of 10% (w/v) PEG 1000, 10% (w/v) PEG 8000, 0.2 M potassium bromide, and 0.1 M tris-HCl (pH 7.5). Crystals of CoxB·CuA (space group P 1 21 1) grew within ~10 days at 4°C.

Crystallographic data collection and structure determination

All crystals were cryopreserved in precipitant solution and supplemented with 15 to 20% glycerol and additional 4 to 5% of the corresponding PEG. Crystals were flash-frozen in liquid nitrogen. All x-ray diffraction data were recorded at the SLS beamline X06sa (Swiss Light Source, Paul Scherrer Institute, Switzerland).

Data were indexed and scaled with XDS (59) and refined with Refmac (60), and the models were built with Coot (61). Ramachandran plot: 4W9Z: Pref. 93.1% Allowed 6.1% Outliers 0.8%; 4WBR: Pref. 97.3% Allowed 2.1% Outliers 0.7%; 4WBJ: Pref. 96.3% Allowed 2.4% Outliers 1.4%.

EPR spectroscopy

For cw EPR measurements, proteins were concentrated to 100 to 400 μM and supplemented with 20% (w/v) glycerol as a cryoprotectant. Samples of 60 μl were filled into 3-mm quartz tubes and flash-frozen in liquid nitrogen. Samples containing Cu1+ were kept anaerobically, and the quartz tubes were sealed under vacuum after freezing.

Cw EPR spectra were recorded in independent duplicates on an ElexSys E500 spectrometer (Bruker BioSpin, Rheinstetten, Germany) at X-band (~9.4 GHz) using an SHQ resonator and an ESR900 helium flow cryostat (Oxford Instruments, Oxfordshire, UK). All spectra were baseline-corrected by a first-order polynomial, the magnetic field offset was corrected using diphenyl-1-picrylhydrazyl (Sigma-Aldrich, Buchs, Switzerland) as a reference, and the magnetic field axis was rescaled to display all spectra at a common microwave frequency of 9.5 GHz. The spectra were recorded at 40 K using microwave powers of 8 to 32 μW (nonsaturating conditions). The magnetic field modulation amplitude was set to 0.3 mT with a frequency of 100 kHz for lock-in amplification with a conversion time and time constant of 81.92 ms. EPR spectra were analyzed using MATLAB and EasySpin (62) and scaled to correct for differences in protein concentration in the samples. EPR spectra of mixtures were decomposed by least-squares fitting of weighting factors of individual experimental spectra such that the weighted sum corresponded to the experimental spectrum of the respective mixture.

Pulsed EPR distance measurements between Cu2+ ions were carried out by ultra-wideband four-pulse DEER (46, 47) at ca. 9.3 GHz and a temperature of 10 K using a homebuilt X/Q band pulse EPR spectrometer (63) equipped with an MS-3 resonator (Bruker BioSpin, Rheinstetten, Germany) and 40-μl sample in 3-mm quartz tubes (as above). Observer pulses were all rectangular 12-ns pulses applied 50 MHz below the spectral maximum with an initial interpulse delay of 200 ns and incremented in eight steps of 8 ns to average out nuclear modulations. The pump pulse was a hyperbolic secant pulse of order 6 with a length of 64 ns and an excitation bandwidth of 450 MHz, was applied with an offset of +125 MHz, and included nonlinear compensation for resonator and microwave amplifier profiles (63). The pump pulse was stepped by 8 ns, and the repetition rate of the pulse sequence was 100 Hz. The data were analyzed using DeerAnalysis 2018 (64).

NMR spectra of PcuC variants

NMR samples consisted of 180 μl of 100 μM PcuC (13C-methyl) methionine in 20 mM Mops (pH 7.0) (25°C), 50 mM NaCl, and 10% D2O in 3-mm NMR tubes (Norell, S-3-HT-7). Labeled protein was produced by adding l-methionine-(methyl-13C) (1 g/liter) (Sigma-Aldrich, 299146) to LB medium at the time point of induction (65). Depending on the sample, 0 to 3 meq of CuCl2 or tetrakis(acetonitrile)copper(I) hexafluorophosphate were added under anaerobic conditions and incubated overnight at 5°C. All copper-containing samples were desalted over a disposable PD-10 column before the measurement to remove unbound copper. Only the PcuC ΔC and the peptide samples used for the titration with CuCl2 were measured without previous desalting step. For obtaining stable preparations of Cu1+ complexes, 50 mM Na-dithionite was added after desalting. None of the substances were deuterated.

For the titration of PcuC·Cu1+·Cu2+ with the ScoI·Cu2+·CoxB, 50 μM of the (13C-methyl) methionine–labeled PcuC·Cu1+·Cu2+ in 20 mM Mops-NaOH (pH 7.0) (25°C), 50 mM NaCl, and 10% D2O was titrated with 0 to 2 meq of the unlabeled ScoI·Cu2+·CoxB complex. The samples were incubated at 10°C overnight before recording NMR spectra.

All NMR measurements were carried out at 25°C (temperature calibrated with methanol-d4, 99.8%) on a Bruker AVNEO 700 MHz spectrometer equipped with z-gradient coils. Gradient-selected 2D-[13C,1H] ALSOFAST-HMQC experiments (66) were used to suppress the intense signals of the protonated buffer substances. To resolve all methionine signals, a maximum t1 evolution time of 121 ms was used, yielding a 13C resolution of 8.3 Hz in the indirect dimension. Each experiment was recorded in 34 min. All NMR measurements were independently repeated at least two times and proved to be fully reproducible.

Kinetics of ScoI·Cu2+·CoxB complex formation

CoxB (constant final concentration of 25 μM) was mixed with different amounts of ScoI Cu2+ (37.5, 75, or 125 μM final concentration) at 25°C in 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl using a SX20 stopped-flow instrument (Applied Photophysics), and complex formation was monitored by the increase in absorbance at 520 nm. Each dataset was recorded six times and averaged. The absorbance traces were then globally fitted with DynaFit 4 according to a mechanism with reversible formation (k1) and dissociation (k2) of an encounter complex that irreversibly reacts to the final complex (k3). Indicated errors of rate constants correspond to errors from the fits performed with DynaFit.

Kinetics of PcuC·Cu1+·Cu2+–mediated ScoI·Cu2+·CoxB complex dissociation

The rate of ScoI·Cu2+·CoxB complex dissociation by PcuC·Cu1+·Cu2+ was measured at 25°C in 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl after stopped-flow mixing. Identical initial concentrations of 25 μM were used for ScoI·Cu2+·CoxB and PcuC·Cu1+·Cu2+, and the reaction was monitored by the decrease in absorbance at 520 nm. Absorbance kinetics were measured in triplicates, averaged, and fitted with DynaFit 4 according to a second-order reaction followed by a monoexponential decay.

Kinetics of Cu2+ transfer from PcuC·Cu1+·Cu2+ to reduced ScoI

The Cu2+ transfer from PcuC·Cu1+·Cu2+ to ScoI (fig. S9B) was measured at 25°C in 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl in a SX20 stopped-flow instrument. ScoI (25 μM) was mixed with 25 μM PcuC·Cu1+·Cu2+, and Cu2+ transfer was monitored by the increase in the ScoI·Cu2+–specific absorbance at 360 nm. Three independent absorbance traces were recorded and averaged.

Titration of reduced CoxB with ScoI·Cu2+

The initial concentration of CoxB (reduced with TCEP and desalted) was kept constant (40 μM) and mixed with 0.5 to 5.0 meq of ScoI·Cu2+ in 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl. After incubation at 10°C (overnight), samples were analyzed by gel filtration on a Superdex 75 10/300 column (GE Healthcare) equilibrated with the same buffer, and eluted proteins were detected via their absorbance at 280 nm. Titrations were repeated independently at least two times and proved to be reproducible within experimental error.

For absorbance titration of CoxB with ScoI·Cu2+, the initial apo-CoxB concentration was kept constant (15 μM) and the ScoI·Cu2+ concentration was varied between 0 and 75 μM. The absorbance signal at 520 nm was plotted against added equivalents of ScoI·Cu2+. Data were fitted to a noncovalent binding equilibrium according to the following equation, where A is the monitored absorbance signal, A0 and A are the absorbance values of 0 and 100% complex formation, KD is the dissociation constant, [C]0 is the total concentration of apo-CoxB, and [S]0 is the total concentration of ScoI·Cu2+A=A0(A0A)·[C0]+[S0]+KD([C0]+[S0]+KD)24·[C0]·[S0]2·[C0]

Fitting the data yielded a KD value of 1.54 ± 0.8 × 10−7 M, showing that the concentration of apo-CoxB was at least two orders of magnitude above the KD value, thus not allowing accurate KD determination. As we could not lower protein concentrations due to the low sensitivity of the absorbance measurements, we confined the analysis to the estimation that the KD value of the ScoI·Cu2+·CoxB complex is below 10−7 M. Absorbance titrations were repeated independently at least two times, each confirming the 1:1 stoichiometry with the sharp kink in the titration profile (Fig. 1F).

Analysis of CuA center formation by analytical gel filtration

All samples were mixed in an anaerobic chamber and sealed in a silanized microvial (8004-HP-H/iV2μ/SZ, Infochroma) with a polypropylene screw cap (G004-HP-CB-FKSKFK10, Infochroma). Samples were incubated at 10°C overnight before gel filtration at 25°C on an AdvancedBio SEC column (PL1580-5350, Agilent Technologies) equilibrated with 20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl. Gel filtration runs were performed with a high-performance liquid chromatography system (Agilent Technologies, 1100 Series) equipped with a diode array detector, allowing online recording of ultraviolet-visible (UV-Vis) spectra of eluting proteins. The initial concentration of ScoI·Cu2+·CoxB was kept constant (20 μM) in all experiments, and PcuC·Cu1+·Cu2+ was varied between 5 and 60 μM. CuA center formation was followed by recording the CoxB·CuA–specific absorbance at 813 nm. The amount of CoxB·CuA formed was quantified by plotting the peak areas at 813 nm against the molar equivalents of PcuC·Cu1+·Cu2+ and comparing them to the peak area of a 20 μM CoxB·CuA reference sample. Peak areas were determined with PeakFit v4.12 (exponentially modified Gaussian model). Furthermore, the absorbance spectra at the peak maxima recorded at 280 nm were used to quantify the CoxB·CuA center by comparing the increase in the CoxB·CuA–specific absorbance peaks at 367, 479, and 813 nm to the absorbance spectra of the CoxB·CuA reference sample. For the titration of apo-CoxB with PcuC·Cu1+·Cu2+, the initial concentration of apo-CoxB was kept constant (15 μM), and PcuC·Cu1+·Cu2+ was again varied between 0.25 and 3 meq. Titrations were repeated independently at least two times and proved to be reproducible within experimental error.

Analysis of Cu1+ binding to PcuC or ScoI with the Cu1+-specific indicator dye BCS

Potential contaminations with Cu2+ ions in the buffer used for BCS titration experiments [20 mM Mops-NaOH (pH 7.0) and 50 mM NaCl] were removed by filtering the buffer through a 5-ml HisTrap agarose column (GE Healthcare) that had been extensively washed with 250 mM EDTA and deionized water. Apo-PcuC was then transferred into this buffer (PD-10 desalting column), and the absence of Cu ions in the apo-PcuC preparation was verified by ESI mass spectrometry. A stock solution of Cu1+(BCS)2 was prepared by mixing tetrakis(acetonitrile)copper(I) hexafluorophosphate with 2 meq BCS acid (B1125, Sigma-Aldrich). The mixture was centrifuged, and the concentration of Cu1+(BCS)2 in the supernatant was determined via its specific absorbance at 483 nm (13,000 M−1 cm−1). Titration experiments were performed under anaerobic conditions at a constant initial Cu1+(BCS)2 concentration of 15 μM, mixed with 0 to 4 meq of apo-PcuC, apo-PcuC ΔC, apo-PcuC ΔHHM, apo-PcuC ΔC ΔHHM, or apo-ScoI. The samples were sealed and incubated for 1 hour or overnight at 25°C. Then, the absorbance at 483 nm of each sample was recorded and plotted against the molar equivalents of the respective protein added to Cu1+(BCS)2.

For the reverse experiments, PcuC variants were loaded with 1 or 2 meq of Cu1+ as described above. Then, full-length PcuC or the PcuC ΔC variants (constant concentration of 15 μM each) were mixed with 0 to 5 meq of BCS stock under anaerobic conditions and incubated overnight at 25°C. The recorded absorbance values at 483 nm were plotted against the molar BCS/protein ratio. Two independent sets of experiments confirmed reproducibility.

Detection of cytochrome c oxidase activity in B. diazoefficiens mutants

B. diazoefficiens strains 110spc4 (wild type) (67), 6611 (ΔpcuABCDE::aphII) (20), 2575 (ΔscoI::aphII) (19), and 3563 (ΔcoxB::aphII) (19) were grown aerobically at 30°C to late exponential growth phase in peptone–salt–yeast extract medium supplemented with 0.1% L(+)-arabinose (68) and, where applicable, 50 μM CuCl2. Spectinomycin (to wild type) or spectinomycin and kanamycin (to all other strains) were added to a final concentration of 100 μg/ml each. Cells were harvested by centrifugation and washed in 0.9% NaCl solution and adjusted to a final OD600 of 10.0. Samples of 50 μl were immediately spotted in duplicate on a filter paper soaked in freshly prepared 1% TMPD, and the reaction product indophenol blue was detected after incubation for 15 min.

Detection of cytochrome c oxidase activity in B. diazoefficiens mutants under copper starvation

Strains 110spc4 (wild type), 3563 (ΔcoxB::aphII) (19), 6611 (ΔpcuABCDE::aphII) (20), 6611-33 (ΔpcuABCDE::aphII + pcuABCDE), 6611-34 (ΔpcuABCDE::aphII + pcuABDE) [both (20)], 6611-74 (ΔpcuABCDE::aphII + pcuAB[C3M]DE), 6611-75 (ΔpcuABCDE::aphII + pcuAB[CΔC-term]′DE), 6611-76 (ΔpcuABCDE::aphII + pcuABC), and 6611-1696 (ΔpcuABCDE::aphII + pcuAB[C3M, ΔC-term]′DE) were constructed according to standard protocols (20) and grown at 30°C in V3S minimal medium (9.9 mM K2HPO4, 10.1 mM NaH2PO4, 10.0 mM NH4NO3, 3.3 mM MgSO4, 340 μM CaCl2, 59 μM MnSO4, 49 μM H3BO4, 7 μM ZnSO4, 1 μM NaMoO4, 5 μM KI, 5 μM NaSeO3, 9 μM FeCl3, 9 μM FeCl2, 160 nM CuSO4, 105 nM CoCl2, and 105 nM NiCl2) supplemented with 25 mM Na2-succinate and appropriate antibiotics to mid-exponential growth phase, harvested, washed twice with sterile 0.9% NaCl, and transferred into fresh V3S without CuSO4 to a starting OD600 of 0.05. Strains were grown in the presence of spectinomycin (100 μg/ml) for 48 hours at 30°C to OD600 of 0.8. Bacteria were harvested, washed with copper-free V3S, centrifuged, and adjusted to OD600 of 7.0. Identical amounts (15 μl) of each suspension were immediately spotted onto a filter paper soaked in 1% TMPD, and indophenol blue formation was recorded after 10 to 15 min.


Supplementary material for this article is available at

Fig. S1. Structural characterization of ScoIox, ScoI·Cu2+, CoxB·CuA, and PcuC.

Fig. S2. Amino acid sequence alignment of PcuC homologs.

Fig. S3. EPR determination of distances between two Cu2+ ions in PcuC·Cu2+·Cu2+.

Fig. S4. NMR experiments showing that the C-terminally truncated PcuC variant PcuC ΔC can also bind Cu2+.

Fig. S5. Investigation of the Cu2+-binding properties of the C-terminal PcuC peptide by NMR.

Fig. S6. Absorbance spectra of ScoI·Cu2+·CoxB, CoxB·CuA, ScoI·Cu2+, and PcuC·Cu1+·Cu2+.

Fig. S7. Equilibrium competition between PcuC or ScoI and BCS for binding of Cu1+ reveals a specific high-affinity Cu1+-binding site in the PcuC core domain and a low-affinity Cu1+-binding site in the C-terminal PcuC extension and demonstrates that ScoI does not bind Cu1+.

Fig. S8. Detection of cytochrome oxidase activity in B. diazoefficiens mutants under copper starvation.

Fig. S9. NMR analysis of Cu transfer from PcuC·Cu1+·Cu2+ to the ScoI·Cu2+·CoxB complex.

Fig. S10. Alternative, but disfavored, mechanism of CoxB·CuA formation from CoxB·Cu2+ and only one equivalent of PcuC·Cu1+·Cu2+.

Fig. S11. Cw EPR spectra of the PcuC·Cu1+·Cu2+ + ScoI·Cu2+·CoxB reaction mixture.

Fig. S12. The ScoI·Cu2+ complex shows limited long-term stability due to Cu2+-induced formation of a disulfide bond between the active-site cysteines of ScoI that causes release of Cu2+.

Fig. S13. CoxB·CuA formation from apo-CoxB and PcuC·Cu1+·Cu2+ in the absence of ScoI.

Table S1. X-ray data collection and refinement statistics (molecular replacement).

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank the staff at the Swiss Light Source (Villigen, Switzerland) for outstanding support in crystallographic data collection and B. Blattmann and his team at the Protein Crystallization Center at the University of Zurich for the initial screening and the imaging of our crystals. We thank C. Field (Institute of Microbiology, ETH Zurich) for help in the bioinformatics analysis of PCuAC-like sequences. We also thank N. Ban (Institute of Molecular Biology and Biophysics, ETH Zurich) for support in x-ray structure determination and the fruitful discussions. Funding: This project was funded by the Swiss Federal Institute of Technology Zurich. Author contributions: F.C., R.G., H.H., and H.-M.F. conceived the study and experimental approach; F.C. and H.K.A. cloned, purified, and crystallized proteins; F.C. and M.M.S. carried out kinetic experiments; N.Q. collected x-ray diffraction data and determined crystal structures; D.K. and G.J. performed the EPR measurements and data evaluation; A.D.G. and F.C. performed NMR experiments; R.L. performed, together with F.C., the oxidase activity test in B. diazoefficiens knockout mutants. S.C. analyzed Cu ion binding to proteins under nondissociating conditions with ESI mass spectrometry. F.C., H.H., and R.G. wrote the paper with input from all authors. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Coordinates for the solved x-ray structures of ScoIox, ScoI·Cu2+, and CoxB·CuA have been deposited at the PDB under the accession numbers 4WBJ, 4WBR, and 4W9Z, respectively. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article