N. E. Le Brun, School of Chemistry, University of East Anglia, Norwich NR4 7TJ, UK Fax: +44 1603 592003 Tel: +44 1603 592699 E-mail: firstname.lastname@example.org
CopA, a P-type ATPase transporter involved in copper detoxification in Bacillus subtilis, contains two soluble Atx1-like domains separated by a short linker at its N-terminus, an arrangement that occurs widely in copper transporters from both prokaryotes and eukaryotes. Both domains were previously found to bind Cu(I) with very high affinity. Above a level of 1 Cu(I) per CopAab, dimerization occurred, leading to a highly luminescent multinuclear Cu(I) species [Singleton C & Le Brun NE (2009) Dalton Trans, 688–696]. To try to understand the contributions of each domain to the complex Cu(I)-binding behaviour of this and related proteins, we purified a wild-type form of the first domain (CopAa). In isolation, the domain bound Cu(I) with very high affinity (K = ∼ 1 × 1018m−1) and underwent Cu(I)-mediated protein association, resulting in a mixture of dimer and tetramer species. Addition of further Cu(I) up to 1 Cu(I) per CopAa monomer led to a weakly luminescent species, whereas further additions [2 Cu(I) per CopAa monomer] resulted in protein unfolding. Analysis of the MTCAAC binding motif Cys residue acid–base properties revealed pKa values of 5.7 and 7.3, consistent with the pH dependence of Cu(I) binding, and with the proposal that low proton affinity is associated with high Cu(I) affinity. Finally, Cu(I) exchange between CopAa and the chelator bathocuproine sulfonate revealed rapid exchange in both directions, demonstrating an interaction between the protein and the chelator that catalyses metal ion transfer. Overall, CopAa exhibits similarities to CopAab in terms of affinity and complexity of Cu(I) binding, but the details of Cu(I) binding are distinct.
Transmembrane copper-transporting P-type ATPases play crucial roles in the trafficking of this essential but toxic metal ion in organisms ranging from complex multicellular eukaryotes to simple prokaryotes [1,2]. These proteins typically have eight transmembrane segments, actuator (A-domain), phosphorylation (P-domain) and nucleotide-binding (N-domain) domains, and a number of metal-binding soluble domains usually located at the N-terminus . The human P-type ATPases ATP7A and ATP7B, also known as Menkes and Wilson proteins, respectively, contain six domains, the fruitfly homologue has four, whereas the yeast ATPase, Ccc2, contains two, as do many bacterial transporters. In some cases, bacterial transporters contain only one N-terminal domain [2,4]. These soluble domains are similar in primary sequence and structure to the Atx1 family chaperone with which the ATPase interacts, and structural information is now available describing the Cu(I)-bridged heterocomplex that forms between the N-terminal domains and the chaperone . The function and number of these N-terminal domains remain unclear. Studies of bacterial transporters indicate that they are not essential for the transport function of the protein, but they are important for trafficking between membranes in higher eukaryotes  and were also shown to be important for the transfer of Cu(I) from Atx1 to the transporter domain of Ccc2 . It has also been proposed that they function as regulators of transport activity, through interactions with other domains of the transporter [3,8,9].
The P-type ATPase CopA from Bacillus subtilis plays a major role in the resistance of the cell to copper by exporting the metal across the cytoplasmic membrane . It contains two N-terminal soluble domains and, as such, represents a relatively simple system for studying the Cu(I)-binding properties of, and interaction between, the N-terminal domains of a Cu(I)-transporting ATPases, and subsequently interactions with other domains of the transporter. Initial attempts to structurally characterize the two soluble domains (in isolation from the remainder of the transporter) were hampered by the instability of the first domain, referred to as CopAa (residues 1–72), to unfolding, either alone or as part of a two-domain protein with CopAb (residues 73–147) . A sequence analysis and modelling study indicated that Ser46 in CopAa, which is present instead of a highly conserved Val or Ala, may destabilize the protein’s hydrophobic core. A S46V mutation produced a single, well-folded form in solution, which adopted a βαββαβ fold . Despite the relative instability of domain CopAa, a wild-type version of the two-domain protein CopAab was recently purified and found to be sufficiently stable for structural and Cu(I)-binding studies . The solution structure was similar to that reported for the higher stability S46V variant, with minor differences mostly confined to the Ser46-containing β3 strand of domain CopAa. The two domains are connected by a dipeptide linker and interact with one another principally through hydrogen bonding and hydrophobic interactions. Such a ‘domain–short link–domain’ arrangement is also found in other transporters that contain two soluble N-terminal domains, and also characterizes the last two domains of the eukaryotic copper ATPases that contain multiple domains, suggesting that this may represent a functionally important motif in copper transporters .
Cu(I)-binding studies of CopAab revealed that the two domains, which each contain the Cu(I)-binding motif MTCAAC (located at different ends of the protein molecule), are both able to bind Cu(I) with very high affinity (K = ∼ 4 × 1017m−1) . However, spectroscopic studies showed that Cu(I) binding to CopAab is complex, with the protein undergoing Cu(I)-mediated dimerization above a level of 1 Cu(I) per protein, resulting in a highly luminescent species that likely contains a solvent-shielded Cu(I)-cluster. In fact, CopAab can accommodate up to 4 Cu(I) per protein and remains dimeric at higher Cu(I) loadings [13,15].
We wish to better understand this highly complex Cu(I)-binding behaviour and, in particular, the individual contributions of the two domains to the overall properties of CopAab, and the consequences of the physical link between them. To achieve this requires studies of the two N-terminal domains, first in isolation and then in a 1 : 1 mixture. Here, we report the isolation of wild-type CopAa in a form that is sufficiently stable for spectroscopic and bioanalytical studies of Cu(I) binding, and of the acid–base properties of the Cys residues of the Cu(I)-binding motif. The data reveal complex Cu(I)-binding behaviour that is distinct from that of the CopAab protein.
CopAa is stable to unfolding
To begin to understand how the two N-terminal domains contribute to Cu(I) binding, the first N-terminal domain of CopA alone (CopAa) was overproduced and purified. The far-UV CD spectrum of apo-CopAa was consistent with a folded protein, containing features that are characteristic of both α helices (negative signal at 222 nm, shoulder at 208 nm and positive signal below 200 nm) and β sheets (broad positive feature below 205 nm), see Fig. S1.
To further investigate the stability of the wild-type CopAa domain, 1H–15N heteronuclear single quantum correlation (HSQC) NMR spectra were measured for freshly reduced apo-CopAa and again after 24 and 48 h. Overlaid spectra (Fig. 1) showed that only three resonances at 1H, 15N chemical shifts (ppm) [121, 8.4], [120, 8.08] and [118, 8.14], were lost from the spectrum over the 48-h period, demonstrating good stability of the protein under these conditions. However, the spectra contained 116 peaks, which is many more than the number of residues in the protein (72 residues). This indicates that the CopAa sample is heterogeneous. The possibility that the additional signals might be due to a contaminant can be ruled out because no significant additional protein component was detected by SDS/PAGE and the additional number of resonances would represent a very small protein, and is also not due to a partially metallated sample because the protein was metal-free following purification. Thus, the protein most likely adopts more than one stable conformation in solution, and because the intensity across all the peaks is similar, this suggests that both conformations are present in equal amounts.
CopAa can bind multiple Cu(I) ions
The Cu(I)-binding properties of the domain were investigated under anaerobic conditions by titration with Cu(I), and by following changes using UV-visible absorbance spectroscopy. The two tyrosine residues in CopAa give rise to a characteristic absorption band around 276 nm (Fig. 2A). Additions of Cu(I) resulted in absorbance shoulders at 255 and 292 nm, similar to those previously reported for B. subtilis CopZ  and CopAab [13,15], and are due to CysS–Cu(I) charge transfer (ligand to metal charge transfer; 240–260 nm) and cluster-localized Cu(I) (d–s) transitions (260–360 nm) . Changes at 255 and 292 nm were plotted as a function of Cu(I)/CopAa (Fig. 2A, inset). The plot reveals distinct Cu(I)-binding phases between 0–0.5, 0.5–1.0 and 1.0–1.5 Cu(I)/CopAa. As also observed for CopAab, the distinction between the 0.5–1.0 and 1.0–1.5 phases was less clear than between other phases .
The Cu(I)-binding behaviour of CopAa was also monitored by fluorescence spectroscopy. Like CopAab, CopAa gave rise to an emission signal at 306 nm , because of the presence and absence, respectively, of tryosine and tryptophan residues. Fluorescence intensity was quenched during titration with Cu(I) ions (see Fig. 2B). A plot of the change in fractional fluorescence against Cu(I) stoichiometry (Fig. 2B, inset), shows that Cu(I) did not affect the tyrosine fluorescence significantly up to a stoichiometry of 0.5 Cu(I)/CopAa. Above this level, a significant quenching of intensity was observed. This behaviour contrasts with that of CopAab, for which quenching was only observed above 1 Cu(I)/CopAab . It is noted that, even with an excess of Cu(I), tyrosine fluorescence was not fully quenched, indicating that one tyrosine might be less affected by Cu(I) binding than the other.
Previous studies showed that CopAab gives rise to a luminescence emission band around 600 nm when bound by multiple Cu(I) ions, and it is known that such luminescence is characteristic of Cu(I)–thiolate clusters coordinated in a solvent-shielded environment [13,15]. A Cu(I) titration of CopAa was carried out and luminescence was monitored between 500 and 700 nm, following excitation at 295 nm. Figure 3A shows that addition of Cu(I) to CopAa led to the formation of a luminescent species with emission centred around 600 nm. Intensity at 600 nm increased only marginally between 0 and 0.5 Cu(I)/CopAa and more significantly in the range 0.5–1.0 Cu(I) per protein. Above this level, Cu(I) additions resulted in a slight decrease in intensity (see Fig. 3B). It is noted that the intensity of the observed luminescence is rather weak and is much weaker than that observed for an equivalent concentration of CopAab . Nevertheless, the data suggest that a (partially) solvent-shielded multicopper cluster may form in CopAa, maximizing at 1.0 Cu(I)/CopAa. Intensity decreases observed on further addition of Cu(I) indicated that the binding of further Cu(I) affects the existing copper environment.
Cu(I) binding promotes association of CopAa domains and unfolding at high loadings
Analytical gel-filtration experiments were performed to investigate the association state of CopAa with increasing Cu(I) concentration. CopAa samples containing 0, 0.5, 1.0 and 2.0 Cu(I)/protein were run on a Superdex 75 analytical gel-filtration column. The chromatogram for apo-CopAa, Fig. 4, contained a major peak at an elution volume of 14.6 mL. Our previous studies of Cu(I)-binding domains of this type, for example, CopAab  and CopZ , showed that, in the absence of metal, the proteins are monomeric and so it is very likely that the peak at 14.6 mL represents the monomer form of CopAa. That CopAa is a monomer in solution was also concluded from analytical ultracentrifugation experiments (see below). A number of lower intensity band shoulders were also observed at lower elution volumes, suggesting that a proportion of CopAa was present as higher molecular mass forms, perhaps as disulfide-bonded oligomers. The chromatograms for Cu(I)-bound CopAa revealed very different elution profiles. CopAa containing 0.5 Cu(I)/protein (Fig. 4) gave a main elution peak volume of 11 mL, with shoulders at 13.6 and 14.6 mL, indicating that most CopAa was present as a significantly higher molecular mass form, with smaller proportions present as monomer and perhaps dimer forms. We note that because CopAa-bound Cu(I) gives rise to additional absorbance intensity, it is not straightforward to draw conclusions about the precise distribution of species present in this and higher Cu(I)-loaded samples. CopAa containing 1.0 Cu(I)/protein also gave three major peaks at elution volumes of 14.6, 13.5 and 11 mL, with a shoulder at 11.9 mL. The intensities are quite different from those at 0.5 Cu(I)/protein, generally indicating a shift back towards lower molecular mass forms. At 2.0 Cu(I)/CopAa, the chromatogram contained only one major peak at an elution volume of 13.6 mL, indicating that further significant changes occurred in this range of Cu(I) binding.
To determine more precisely the molecular masses of Cu(I)-bound forms of CopAa, analytical ultracentrifugation experiments were performed at 28 000 and 25 000 rpm on CopAa samples containing 0, 0.5 and 2.0 Cu(I)/protein. Fits of data for apo-CopAa (see Fig. 5A for data at 28 000 rpm) gave an average molecular mass of 8.2 ± 0.2 kDa, slightly higher than the calculated molecular mass of CopAa, (7.6 kDa). Thus, apo-CopAa is a monomer, as concluded from analytical gel-filtration chromatography (see above). At a loading of 0.5 Cu(I)/CopAa, the molecular mass increased significantly to 20.5 ± 0.5 kDa. Clearly, the protein undergoes Cu(I)-mediated association; 20 kDa is closest to a trimeric form of CopAa, but reference to the analytical gel-filtration data, which showed two major species, suggests that, at 0.5 Cu(I)/CopAa, there is a mixture, which may be in slow equilibrium, between likely tetramer and monomer species, with an average molecular mass of 20 kDa. At a loading of 2 Cu(I)/CopAa, fitting of the data gave a molecular mass of 9.8 ± 0.6 kDa. Again, with reference to analytical gel-filtration data, it can be concluded that, at this level of Cu(I), the protein is present principally as an equilibrium mixture of monomer and dimer, with the monomer as the major species. This is somewhat at odds with the gel-filtration chromatography data, which suggested that a likely dimeric species is the major species. We note that the experiments at high Cu(I)-loading were performed at quite different concentrations (much lower in analytical ultracentrifugation experiments) and this may account for the differences in proportion of species observed.
NMR spectroscopy was also used to investigate Cu(I) binding to CopAa. The 1H–15N-HSQC spectrum of CopAa containing 0.5 Cu(I)/protein was measured and is overlaid with that of apo-CopAa in Fig. 6. This shows clearly that many of the resonances due to 0.5 Cu(I)–CopAa were lost or significantly shifted, such that the majority of remaining resonances occur in the crowded central region of the spectrum, which is characteristic of protein association or unfolding. These data indicate that all of the apo-CopAa present at the beginning of the experiment (which was shown to be heterogenous; Fig. 1) participates in Cu(I) binding. An identical loss of intensity was observed upon addition of 1.5 and 2.0 Cu(I)/CopAab . Here, the stoichiometry of 0.5 Cu(I)/CopAa is equivalent to 1.0 Cu(I)/CopAab in terms of Cu(I) per CopA N-terminal domain. However, the spectrum of 1.0 Cu(I)/CopAab was not similar to that of Fig. 6, in that only resonances due to residues in the metal-binding domains were shifted or lost .
Additions of Cu(I) resulted in changes in the far-UV CD spectrum of CopAa (Fig. 7). These were most significant above 1.0 Cu(I)/CopAa, and at 2.5 Cu(I)/CopAa the protein appeared to be largely unfolded. This was presumably driven by the need to accommodate additional Cu(I) ions. We note that similar structural rearrangements were observed in the two-domain CopAab protein at high Cu(I) loading .
Determination of Cu(I)–CopAa binding affinity at low Cu(I)
Given the very high affinities associated with Cu(I) binding to thiolate ligands, affinities cannot be determined directly, but can be obtained through competition binding experiments using a well-characterized high-affinity Cu(I) ligand. We have previously used ligands such as bathocuproine sulfonate (BCS) to determine the initial binding affinities for Cu(I) of CopZ and CopAab [15,18]. In the case of CopAa, the initial binding of Cu(I) is a complex process leading to dimeric and tetramer species involving three linked equilibrium processes (K1–K3) (Data S1). In order to determine the overall equilibrium formation constant, β3, the concentrations of three forms of CopAa, apo-CopAa, Cu(CopAa)2 and Cu2(CopAa)4, would need to be accurately determined in each of the experiments performed. Unfortunately, chelator competition experiments do not provide information on the different Cu(I)-bound forms of CopAa. However, because the Cu(I)-mediated protein association steps have much lower affinities than the Cu(I)-binding step itself, previous binding analyses of copper-trafficking proteins have in some cases ignored them altogether , and a major simplification can be achieved here if one or both of the CopAa association steps are ignored. For CopAa, the bioanalytical data shown above indicated that at low Cu(I), dimer/monomer forms are present in significant amounts, consistent with relatively weak protein association to the tetrameric form. Therefore, competition experiments with the chelator BCS were performed in both directions, as described in Materials and methods. Data were analysed according to Eqns (1) and (2), in which only the tetramer–dimer equilibrium step is ignored, and, alternatively, according to Eqn (3), in which monomer–dimer and dimer–tetramer protein association steps are ignored.
Analysis using Eqns (1) and (2) gave β2(CopAa) = 1.35 × 1022m−2 (Table 1). Analysis using Eqn (3) gave an equilibrium association constant of K = 1.08 × 1018m−1 (Table S1). These results are consistent with the Cu(I)-binding step being by far the highest affinity binding event, and suggest that the Cu(I)-mediated protein association step is weak (K = 104–105m−1).
Table 1. Determination of CopAa Cu(I)-binding affinity through competition with BCS.
a Data from addition of BCS to CopAa. The buffer used was 100 mm Mops, 100 mm NaCl, pH 7.5. b Data from addition of CopAa to BCS. c From Eqns (1) and (2). d Calculation based on β2(BCS) = 6.3 × 1019m−2 for [Cu(BCS)2]3− . e Indicated error is ± SD.
β2(CopAa) (× 1022m−2)d
β2(CopAa) (× 1022m−2) average
1.35 ± 0.46e
CopAa contains the Cu(I)-binding motif MXCXXC, in which the two Cys residues, in their deprotonated (thiolate) form, coordinate the Cu(I). Free cysteine has a pKa of ∼ 8.5, but the acid–base properties of cysteine side chain in proteins can vary significantly from this . Therefore, depending on the pKa of the binding motif Cys residues of CopAa, pH may have a significant effect on the binding affinity for Cu(I). Therefore, the pH dependence of Cu(I) binding to CopAa was investigated, through competition experiments with BCS at pH values of 7.0, 7.5 and 8.0 (Tables 2 and S2). Analysis of the data using Eqn (4) indicated that the Κex values exhibit an approximately sevenfold variation over this pH range. The data fitted well to a single proton dissociation event with a pKa of 7.5 (Table S3), giving an average corrected formation constant β2(CopAa) = ∼ 3.6 × 1022m−2.
Table 2. Correction of β2(CopAa) to take account of the pH dependence of Cu(I) binding.
a See Table S2 for a complete data set. The buffer was 100 mm Mops, 100 mm NaCl at the pH values indicated. b From Eqns (1) and (2). c The pH correction of Kex was obtained using a pKa of 7.5 resulting from fitting with Eqn (4) (see main text and Table S3). d Calculation based on β2(BCS) = 6.3 × 1019m−2 for [Cu(BCS)2]3−  using an average value of pH corrected Kex, obtained from Table S2.
pH corrected Kexc
β2(CopAa), m−2 d
3.71 × 1022
4.54 × 1022
2.57 × 1022
Average β2(CopAa), m−2 d
3.61 (± 0.99) × 1022
The pKa properties of the Cu(I)-binding motif Cys residues
To determine whether the observed pKa is associated with the protonation/deprotonation of a cysteine residue, or perhaps from another residue close to the binding motif, the pKa values of cysteine residues were investigated using the alkylating reagent 6-bromoacetyl-2-dimethylaminonaphthalene (badan) , which forms a thioether bond with the side chain of cysteine thiolates, leading to a significant increase in its fluorescence intensity at 540 nm. Free badan gives a band at ∼ 540 nm; addition of CopAa increased the fluorescence intensity at ∼ 540 nm. Measurement of the rate of alkylation as a function of pH can be used to determine the pKa values of protein cysteine thiol groups . CopAa was added to a badan solution and fluorescence spectra were recorded between 400 and 600 nm at 15 °C, see Fig. 8A,B, in which fluorescence changes in the pH ranges 4.0–6.5 and 7.0–9.0 are plotted separately for clarity. The rate of modification was found to increase over the range pH 4.0–9.0. At pH 6–6.5, the initial increase was followed by a decrease, which is indicative of local unfolding that affects the environmentally sensitive badan intensity . The data were fitted to exponential functions as described in Materials and methods, and the resulting pseudo-first-order rate constants were plotted as a function of pH (Fig. 8C). The data did not fit to a single deprotonation (Eqn 5), but fitted well to Eqn (6), which describes two protonation/deprotonation events, giving pKa1 = 5.7 ± 0.6 and pKa2 = 7.3 ± 0.5 (Fig. 8C).
Kinetic studies of Cu(I) transfer between CopAa and the chelator BCS
Competition experiments with the chelator BCS showed that CopAa has an extremely high affinity for Cu(I). Given that the binding constant is related to the on and off rates for binding, and that the on-rate for binding cannot be greater than that for a diffusion-controlled reaction (∼ 1 × 108m−1·s−1), the off-rate for the dissociation of Cu(I) from CopAa must be extremely low. However, the competition studies carried out with BCS (Table 1) required only minutes to reach equilibrium. From this, we conclude that BCS is not passive in the equilibration reaction, but instead participates in facile metal ion exchange, as we have demonstrated previously for CopZ .
To investigate the concentration dependence of the observed rate of the Cu(I)-exchange reaction, kinetic measurements of exchange were performed in both directions. Prereduced CopAa was loaded with Cu(I) up to 0.5 Cu(I)/CopAa. BCS over the range of 0.2–0.8 mm was added and ΔA483 measured at 25 °C. For the opposite direction, apo-CopAa was added to Cu(BCS)23−. For both directions, A483 was plotted as a function of time (Fig. 9A,B). Each of the data sets fitted well to a single exponential function, and the resulting first-order rate constants were plotted as a function of BCS or CopAa, as appropriate (Fig. 9C). The plots show that the rate constant was independent of the BCS concentration (k = 0.063 ± 0.008 s−1 at 25 °C) and the CopAa concentration (k = 0.050 ± 0.003 s−1 at 25 °C).
The spectroscopic and bioanalytical data reported above provide the basis for understanding Cu(I) binding to CopAa. UV-visible absorbance, fluorescence and luminescence spectra reveal multiphasic binding of Cu(I), with clear phases in the ranges 0–0.5, 0.5–1.0 and > 1.0 Cu(I)/CopAa. Far-UV CD data showed that significant conformational changes in CopAa occur upon binding Cu(I), particularly above 1 : 1 Cu(I)/CopAa, and that at above a ratio of 2:1, the protein appears to be essentially unfolded. Analytical ultracentrifugation and gel filtration showed that CopAa exhibits unusual oligomerization state behaviour. Immediately on binding Cu(I), CopAa undergoes association to a likely tetrameric state, which is in slow equilibrium with dimeric and monomeric forms. Cu(I)-mediated dimerization was previously observed for CopZ, although, in that case, further association to a tetramer did not occur .
Luminescence data showed that intensity due to Cu(I)-bound CopAa reached a maximum at ∼ 1 Cu(I)/CopAa. Luminescence intensity of this type (at around 600 nm) is normally only observed for proteins in which solvent-excluded multinuclear Cu(I) centres are generated, as observed for CopAab [13,15], CopY , ACE1  and AMT1 . Thus, we propose that at 1 : 1 Cu(I):CopAa, the luminescent species is Cu4(CopAa)4. Gel-filtration data indicated that this species is in slow equilibrium with the dimeric species, Cu2(CopAa)2, which may also be luminescent. Addition of Cu(I) up to 2 Cu(I)/CopAa led to further binding and gel filtration indicated that the major species was a lower mass species, presumably because dissociation generates an increase in Cu(I)-binding capacity. The form of CopAa at higher Cu(I) loadings cannot be determined easily from these data; although the elution volume is consistent with it being principally dimeric, far-UV CD data showed that the protein is largely unfolded at this level of Cu(I), and because gel filtration is a hydrodynamic method, elution volumes will vary with the state of folding, such that a volume that corresponds to a dimeric folded state at low Cu(I) is unlikely to correspond to the same association state if the protein is unfolded. Indeed, analytical ultracentrifugation data indicated that CopAa was mainly monomeric at 2 Cu(I)/protein, although it is not immediately clear how a monomeric form of the protein could accommodate more than one Cu(I) ion. Luminescence emission decreased gradually upon addition of Cu(I) above 1 Cu(I)/CopAa, presumably as a result of an increase in the exposure of the Cu(I) cluster to solvent caused by the incorporation of excess copper, as previous reported for CopAab . The Cu(I)-binding behaviour of CopAa is summarized in Fig. 10.
Initial Cu(I) binding by CopAa was analysed through competition experiments. Although the multiple equilibria involved makes analysis problematic, it was recognized that the Cu(I)-binding step is by far the highest affinity event, and by ignoring one or both of the subsequent protein association steps allowed we were able to obtain estimates of the affinity. Ignoring protein association altogether gave an affinity (K = ∼ 1018m−1) comparable with that of initial binding to CopAab (K = ∼1017–1018m−1) , entirely consistent with the conclusion that both domains of CopAab are capable of binding Cu(I) in competition with one another. When protein dimerization was taken into account, the resulting formation constant (β2(CopAa) = ∼ 1.35 × 1022m−2) was similar to that for the initial binding of Cu(I) by CopZ (β2(CopZ) = ∼ 1.11 × 1022m−2) . Overall, the data suggest that there is no strong thermodynamic gradient for transfer of Cu(I) from CopZ to CopAab.
The pH dependence of the Cu(I)-affinity was consistent with a protonatable group at or near the binding motif with a pKa of ∼ 7.5. Examination of the acid–base properties of the binding motif Cys residues revealed pKa values of 5.7 and 7.3 and the latter is likely to correspond to the apparent pKa of 7.5 detected through Cu(I) competition binding experiments. The CopAa pKa values are both significantly lower than the pKa of free cysteine (∼ 8.5) and are consistent with our previous proposal that pKa is correlated with the cysteine-binding site Cu(I) affinity , such that increased acidity of the Cys (lower proton affinity) results in a high Cu(I)-binding affinity. We note that both CopAa pKa values are significantly higher than those of CopZ (pKa1 = < 4 and pKa2 = ∼6.1) . A pH dependence of Cu(I) binding to CopZ was still observed in the pH range 6.5–8.0, and it was suggested that this might arise from a His residue located between the Cys residues at the CopZ-binding motif . For CopAa, the binding motif is MTCAAC, and so the observed pH dependence cannot arise from a binding motif His residue.
Kinetic studies showed that the Cu(I) exchange reaction is rapid and is a first-order process in both directions, with a rate constant that is essentially the same in both directions. This suggests that the rate-determining step in both directions is the same process. Similar observations were made for CopZ, from which it was concluded that the rate-limiting step most likely corresponds to the formation or dissociation of the hetero CopZ/BCS–Cu(I)-coordinated complex . A similar conclusion is reasonable here. We note that the rate constants for CopAa are approximately threefold higher than for CopZ, indicating a more rapid interaction of CopAa with BCS compared with CopZ. Although the interaction with a chelator is not a physiologically relevant reaction, these studies give an indication of how facile Cu(I) transfer between two interacting molecules can overcome the kinetic barriers normally associated with very high-affinity binding. Thus, the interaction of the chelator with domain CopAa, which catalyses release of Cu(I) from the protein, may mimic the reversible transfer of Cu(I) between chaperone and the transporter soluble domains.
The Cu(I)-binding properties of CopAa described here illustrate further the propensity of these protein domains to bind multiple Cu(I) ions, facilitated by protein association. There are clear similarities with Cu(I) binding by CopAab. For example, CopAab undergoes Cu(I)-mediated dimerization and partial unfolding at higher Cu(I) loadings [13,15]. However, the behaviour of CopAa is also distinct: Cu(I)-mediated protein association occurred immediately upon addition of Cu(I) to CopAa, and resulted in a tetrameric species, whereas for CopAab, association was only observed above a level of 1 Cu(I) per protein, and did not go beyond dimerization. CopAa was observed to unfold almost entirely at > 2 Cu(I) per protein, whereas significant partial unfolding of CopAab occurred at a level of 1 Cu(I) per protein.
Determining the Cu(I)-binding properties of the CopAab soluble domains is likely to be important for understanding their biological function, which remains unclear. For example, the ability of the domains to function either in Cu(I) delivery , or in regulating the activity of the transmembrane Cu(I) pump [3,8,9] would likely be affected by the Cu(I)-triggered major conformational changes reported here and previously for CopAab [13,15]. The functional properties of the soluble domains will also be dependent on their number; eukaryotic transporters have more soluble domains than their bacterial counterparts, perhaps because they have evolved additional functions related to protein trafficking within the eukaryotic cell. In bacteria, transporters have either one or two domains, but even here the functional significance of this variation is not yet clear. For CopA, even though the binding motifs of domains (a) and (b) are at opposite ends of the CopAab molecule and cannot participate in the binding of the same Cu(I) ions, the presence of domain (b) is clearly important for modulating the properties of domain (a). This work represents the first steps towards trying to understand the individual contributions of the two domains to the collective properties of CopAab, and similar studies of the isolated domain CopAb are now underway to assess its contribution to the Cu(I)-binding properties of the two-domain protein.
Materials and methods
Construction of an expression vector for the overproduction of CopAa
An overexpression construct for the production of CopAa was generated by PCR amplifying the region of copAa encoding the first 72 residues of the protein using the primers yvgX5: ATCGAATTCAGCTTTTTCTCAAACGACGTG (EcoRI site underlined) and yvgX3: AAGGAGTGCATATGTTGAGTGAACAAAAGG (NdeI site underlined) and B. subtilis 1A1 chromosomal DNA as template. The product was cloned into pET21a, generating pCSNC2 and verified by DNA sequencing of the cloned insert (Eurofins MWG Operon, Ebersberg, Germany).
Purification of CopAa
BL21(DE3)/pCSNC2 cultures were used to inoculate, typically, 5 × 500 mL of Luria–Bertani media in 2-L flasks, followed by incubation at 37 °C 200 rpm until D600 reached 0.6. Isopropyl-β-d-thiogalactoside was added (0.4 mm, final concentration) and the cultures were incubated for a further 3 h. Cells were harvested by centrifuging at 9500 g at 4 °C for 20 min. Cell pellets resuspended in 100 mm Hepes buffer, pH 7.0, were treated with lysosyme (0.1 mg·mL−1 final concentration; Sigma, St Louis, MO, USA) and incubated at 30 °C for 15 min with gentle shaking. RNase (80 μg·mL−1; Sigma) and DNase (6 μg·mL−1; Sigma) were added. Cells were sonicated for 2 × 8 min 20 s using a Status US200 ultrasonicator (Philip Harris Scientific, Lichfield, UK) in pulse mode (0.2 s·s−1) set at 50% power. The lysate was centrifuged at 39 000 g for 20 min at 4 °C and the supernatant heated slowly to 50 °C, stirred continuously for 15 min, and placed on ice for a further 15 min. The suspension was centrifuged at 39 000 g for 20 min at 4 °C. Dithiothreitol (final concentration 5 mm; Melford Laboratories, Ipswich, UK) was added to the supernatant and the solution passed through a 0.45 μm filter (Satorius, Goettingen, Germany) before loading on to a 5 mL Hi Trap Q anion-exchange column. CopAa-containing fractions (as determined by SDS/PAGE analysis) were concentrated to < 5 mL using an ultrafiltration cell fitted with an YM3 membrane (Amicon, Millipore, Billerica, MA, USA) operating at a pressure of 55 psi. The filtered protein solution was applied to a 120 mL Sephacryl S-100 gel-filtration column, previously equilibrated with 250 mL 100 mm Hepes, pH 7.5, 100 mm NaCl, 5 mm dithiothreitol. CopAa-containing fractions were concentrated as above, and buffer exchanged into 100 mm Hepes, pH 7.0. The molecular mass of purified CopAa was determined by MALDI-TOF MS as 7690.78 Da, which is consistent with the predicted molecular mass of CopAa from sequence (7692.84 Da). Analysis of the metal content of CopAa by inductively coupled plasma-atomic emission spectrometry revealed that the protein was isolated in the metal-free (apo) form: copper was present at < 0.02 copper/protein and no other metal ions were detected. CopAa samples were quantified using ε276 = 3560 ± 140 m−1·cm−1, determined by the method of Pace et al. .
Metal ion additions to CopAa
Prior to the addition of metal ion solutions, the protein was reduced using 5 mm dithiothreitol and excess dithiothreitol was subsequently removed using a desalting column (PD-10, GE Healthcare Life Sciences, Uppsala, Sweden). The oxidation state of the cysteines following removal of dithiothreitol was assessed by reaction with Ellman’s Reagent [5′5-dithio-bis (2-nitrobenzoic acid)], which confirmed the presence of approximately two reactive thiols per protein molecule. Additions of Cu(I) to protein solutions were made using a microsyringe (Hamilton, Reno, NV, USA), as described previously [13,18]. Titration experiments were conducted using a single sample cuvette; after each metal ion addition, samples were incubated for 2 min before spectra (see below) were recorded. Data were corrected for dilution effects.
Spectroscopic and analytical methods
UV-visible absorbance spectra were recorded on a Jasco V-550 spectrophotometer. CD spectra in the far-UV range (190–250 nm) were recorded using a Jasco J-810 spectropolarimeter interfaced to a PC. CD intensity is expressed as molar ellipticity ([θ]) in units of deg cm2·dmol−1. Spectra were corrected for intensity due to the buffer. Fluorescence emission spectra were recorded using a Perkin–Elmer LS55 spectrophotometer at 25 °C with excitation at 276 nm and excitation and emission slit widths of 9 nm. For the measurement of emission in the region 500–700 nm, slit widths were set to 10 nm (with excitation at 295 nm) and a 390 nm cut-off band pass filter was employed.
NMR spectra were acquired using a Bruker Avance III 800 MHz spectrometer equipped with a triple resonance, pulsed field gradient probe, operating at 1H frequency of 800.23 MHz and 15N frequencies of 81.09 MHz, using pulse sequences incorporated into the Bruker topspin 2.1 software. 2D 1H–15N HSQC spectra were recorded at 298 K. The 1H carrier frequency was positioned at the resonance of the water during the experiments. The 15N carrier frequency was at 115 ppm. Spectra were processed using nmrpipe . Before Fourier transformation, a cosine-bell window function was applied to each dimension for apodization. The indirect dimensions were first linear-predicted to double the number of data points, and then zero-filled to round up the number of data points to the nearest power of 2.
Protein samples for sedimentation equilibrium experiments were prepared and loaded into in 12-mm charcoal-filled Epon double-sector cells with quartz windows cells under anaerobic conditions. Experiments were run using a Beckman XL-I analytical ultracentrifuge in an AN50Ti rotor at 25 °C. The sample volume was 110 μL and the reference sector of the cell contained identical buffer (100 mm Mops, pH 7.5). Samples were spun at speeds of 28 000 and 25 000 rpm until equilibrium was reached, as judged by cessation of changes in scans collected 4 h apart. Data were analysed using ultrascan, v. 6.1 . The density of the buffer was taken as 1.005 g·mL−1 and the partial specific volume of CopAa was calculated to be 0.7381 mL·g−1 using the program sednterp . Analytical gel filtration of samples of CopAa was performed using a Superdex 75 10/300 GL column (GE Healthcare), with a total column volume (Vt) of 24 mL and a void volume (V0) of 7.6 mL, as determined using blue dextran. The column was equilibrated in thoroughly deoxygenated buffers and operated at a flow rate of 0.8 mL·min−1.
Determination of CopAa–Cu(I) binding affinity
BCS is a popular Cu(I) chelator and was used to act in competition with CopAa for Cu(I). Formation of [Cu(BCS)2]3− gives a strong intensity at 483 nm with ε483 = 13 300 cm−1·m−1 . To determine the extent of binding of Cu(I) by BCS, experiments were carried out in both directions [i.e. BCS was added to Cu-CopAa and apo-CopAa was added to Cu(BCS)23−] under anaerobic conditions . CopAa was always in excess of Cu(I). Following additions of either apo-CopAa or BCS, solutions were left for 10 min to reach equilibrium (at this time point, changes in A483 were complete) and the absorption at 483 nm was recorded using a Jasco V550 spectrophotometer. Data were corrected for dilution effects. A483 values were used to calculate the final concentration of Cu(BCS)23− in each solution. From the partitioning of Cu(I) between CopAa and the chelator, the affinity of CopAa for Cu(I) was determined either through analysis using Eqns (1) and (2) , or using Eqn (3) .
Where Kex is the exchange equilibrium constant for competition between CopAa and BCS and β2 is the overall formation constant for Cu(BCS)2 (logβ2 = 19.8)  and θ is the fraction of CopAa present as the Cu(I)-bound form .
For studies of the pH dependence of Cu(I) binding to CopAa, BCS competition experiments as above were carried out over a range of pH values. Equation (4) was used to model a one proton dependence .
where K is 1/Ka (i.e. the reciprocal of the acid dissociation constant) and [CopAa]f,T is the total concentration of free CopAa (irrespective of the protonation state).
Determination of Cys thiol pKa values
The reaction of Cys thiols with alkylating reagents is well established and occurs only with the ionized thiolate anion  and pKa values of protein thiol groups can be measured through the variation in the rate of alkylation with pH [18,20,21], where the observed rate constant is proportional to the extent of thiol deprotonation at a given pH value . CopAa was prereduced with 5 mm dithiothreitol and excess dithiothreitol was removed via a Sephadex G25 column. CopAa (final concentration 1 μm) was added to a 13 μm (final concentration) solution of badan (Molecular Probes, Invitrogen, CA, USA)  in a mixed buffer system containing potassium acetate, Mes, Mops and Tris (10 mm each) and 200 mm KCl , and incubated for 2 h. Fluorescence spectra were recorded between 400 and 600 nm (excitation wavelength 391 nm) at 15 °C using a Perkin–Elmer LS55 fluorescence spectrophotometer.
Fluorescence data were fitted to a single exponential function to obtain an observed, pseudo-first-order rate constant (k0). Where necessary, a double exponential fit was used and the rate constant for the initial reaction was taken as k0. A pKa value was determined by plotting k0 values as a function of pH and fitting to Eqn (5) or Eqn (6), which describe single and double pKa processes, respectively, and where kSH, kS−, kSHSH, kS−SH and kS−S− are the rate constants for the protein in different states of protonation .
This work was supported by UEA’s School of Chemistry through the award of financial support to LZ and the the UK’s BBSRC (through the award of a PhD studentship to CS). We thank the Wellcome Trust for supporting Biophysical Chemistry at UEA through an award from the Joint Infra-structure Fund for equipment and the Wolfson Foundation for its support of our NMR facility.