Important Structural Features of Thiolate-Rich Four-Helix Bundles for Cu(I) Uptake and Removal

A family of bacterial copper storage proteins (the Csps) possess thiolate-lined four-helix bundles whose cores can be filled with Cu(I) ions. The majority of Csps are cytosolic (Csp3s), and in vitro studies carried out to date indicate that the Csp3s from Methylosinus trichosporium OB3b (MtCsp3), Bacillus subtilis (BsCsp3), and Streptomyces lividans (SlCsp3) are alike. Bioinformatics have highlighted homologues with potentially different Cu(I)-binding properties from these characterized “classical” Csp3s. Determination herein of the crystal structure of the protein (RkCsp3) from the methanotroph Methylocystis sp. strain Rockwell with Cu(I) bound identifies this as the first studied example of a new subgroup of Csp3s. The most significant structural difference from classical Csp3s is the presence of only two Cu(I) sites at the mouth of the bundle via which Cu(I) ions enter and leave. This is due to the absence of three Cys residues and a His-containing motif, which allow classical Csp3s to bind five to six Cu(I) ions in this region. Regardless, RkCsp3 exhibits rapid Cu(I) binding and the fastest measured Cu(I) removal rate for a Csp3 when using high-affinity ligands as surrogate partners. New experiments on classical Csp3s demonstrate that their His-containing motif is not essential for fast Cu(I) uptake and removal. Other structural features that could be important for these functionally relevant in vitro properties are discussed.


■ INTRODUCTION
Copper (Cu) is required by both prokaryotes and eukaryotes for a range of proteins, primarily due to its facile redox chemistry. 1−6 However, this reactivity, and the ability of intracellular Cu(I) to bind tightly to sites for other metals in proteins, results in potential dangers associated with the use of this metal. 3−9 Homeostasis systems involving an array of proteins and some small molecules help cells safely handle Cu, 2−6,10−12 including cytosolic storage by metallothioneins (MTs), mainly in eukaryotes. A family of proteins that can store Cu in bacteria, the Csps, was first identified in the methanotroph Methylosinus trichosporium OB3b (Mt). 13 The predicted twin-arginine translocase (Tat) exported MtCsp1 stores Cu(I) for the membrane-bound (particulate) methane monooxygenase (pMMO). A cytosolic homologue from the same methanotroph (MtCsp3) has been characterized in vitro, 14,15 but its function remains to be determined. Using Escherichia coli as a heterologous in vivo model system has shown that MtCsp3, and the protein from Bacillus subtilis (BsCsp3), can safely store Cu(I) in the cytosol and prevent the toxicity caused by increased intracellular Cu concentrations. 16 Recent work has demonstrated that BsCsp3 stores Cu(I) for the endospore multicopper oxidase CotA, 17 the only known function for a Csp3.
Csps are tetramers of four-helix bundles, [13][14][15]18,19 a common fold for proteins that naturally bind metals (metalloproteins) and one that is also widely used in metal-site design. 19−26 A highly unusual feature of the Csps is the large number of Cys residues they possess, which all point into the center of the bundle and do not form disulfide bonds. This allows the proteins to be filled with Cu(I) ions bound predominantly by Cys thiolates, giving an arrangement of metal sites unlike that seen in any other metalloprotein. This way of storing Cu(I) is therefore very different from how the unstructured Cys-rich apo-MTs sequester cuprous ions by folding around two Cu(I) clusters. 19,27−29 There is limited information about the mechanisms of Cu(I) uptake by, and removal from, Csps. 15,18,19 The kinetics of these processes, along with how tightly Csps bind Cu(I), studied primarily using Cu(I) ligands, are key in vitro characteristics that have been measured to help understand how homeostasis proteins safely bind, store, and chaperone Cu(I) in cells. 13,14,18,29−41 Crystal structures are available for the fully Cu(I)-loaded forms of MtCsp3 and the protein (SlCsp3) from Streptomyces lividans ( Figure 1). 14,42 The structure of apo-BsCsp3 has also been determined. 14 In both Cu(I)-MtCsp3 and Cu(I)-SlCsp3, the main Cu(I) core is made up of three thiolatecoordinated tetranuclear clusters. 15 A number of Cu(I) sites are located closer to the mouth of the bundle and involve coordination from three His residues in the α3-loop-α4 region as well as five Cys residues. 14,42 This His-containing motif is present in MtCsp3, SlCsp3, and BsCsp3 and many other Csp3s ( Figure S1 and refs 14 and 18). As these were the first characterized members of this family of proteins, we refer to them as "classical" Csp3s. Homologues predicted to have the same Cu(I) core structure but with Cu(I)-binding differences elsewhere in the bundle have been identified from sequence alignments ( Figure S1). In particular, the absence of three Cys residues (Cys38, Cys54, and Cys111, MtCsp3 numbering) and the α3-loop-α4 His-containing motif suggest dramatic alterations in the Cu(I) sites bound at the mouth of the bundle. In this work, a nonclassical Csp3 has been analyzed in detail, including determination of its crystal structure fully loaded with Cu(I). Comparisons to classical Csp3s are assisted by new data for MtCsp3, SlCsp3, and BsCsp3.  crystal structure of the protein fully loaded with Cu(I) has been determined. There are Cu(I) ions bound at 17 sites ( Figure 3A, Table 2, and Figure S6), consistent with the above stoichiometry studies. The protein has a helical content of ∼70%, which agrees with far-UV CD data (Table 1 and Figure S7A) and is unaffected by Cu(I) binding, as is the case for other Csps (Table 1, Figure S8A, and refs 13 and 14). Cu(I)-RkCsp3 forms a tetrameric arrangement in crystals ( Figure S6), and both the Cu(I)-and apo-proteins are tetramers in solution (Table 1 and Figure S7B), as found for other Csps (Table 1, Figure S8B, Supporting Results, and refs 13, 14, and 42). The structure of a Cu(I)-RkCsp3 monomer is similar (rmsd for C α atoms of 0.80 Å) to that of the unpublished structure of the apo-Csp3 (NmCsp3) from Nitrosospira multiformis ATCC 25196 (3LMF) ( Figure S9), which belongs to the same subgroup and whose sequence is similar (∼45% identity) to RkCsp3 ( Figure S1).
Three thiolate-coordinated tetranuclear Cu(I) clusters form the main core of Cu(I)-RkCsp3, which is also the case in Cu(I)-MtCsp3 and Cu(I)-SlCsp3 (Figures 1 and 3 and Table  2). 14,15,42 This is due to conservation of the Cys residues that are ligands at these sites ( Figure S1). The only non-Cys ligand at these three clusters varies, with the side-chain thioether sulfur of Met46 coordinating Cu13 (at the Cu11− Cu14 cluster) in Cu(I)-RkCsp3 (Figures 3A and 4A, Table  2). In MtCsp3, the corresponding Asn58 coordinates Cu13 via its side-chain amide oxygen atom (O δ1 ) ( Figure 4B and Table 2). 14 This residue is located at the interface between the main Cu(I) core and sites at the mouth of the bundle, and in SlCsp3 the carboxylate of Asp61 (O δ1 ) ligates not only Cu13 and Cu11 at the same tetranuclear cluster but also Cu15 via its O δ2 atom ( Figure 4B and Table 2). 42 There are differences in how RkCsp3 binds Cu(I) at the two extremes of the bundle compared to classical Csp3s ( Figure 3 and Table 2). The hydrophobic end furthest from the mouth has an additional Cu(I) site (Cu*; Figure 4C and Table 2). This is due to the presence of Cys71 (Leu87/90 in MtCsp3/SlCsp3, see Figure S1), Met60 (Ala72/Leu75 in MtCsp3/SlCsp3), and Met108 (Ala128/Leu131 in MtCsp3/ SlCsp3). The thiolate of Cys11 also coordinates Cu* and bridges to Cu1. The corresponding Cys24/27 in MtCsp3/ SlCsp3 bridges Cu1 and Cu2 ( Figure 4D), while Cys71 provides the second thiolate for Cu2 in Cu(I)-RkCsp3 ( Figure 4C).
Significantly larger differences are seen when comparing Cu(I) sites at the mouth of the Cu(I)-RkCsp3 bundle with those in classical Cu(I)-Csp3s ( Figure 4A,B and Table 2). Cu(I)-MtCsp3 binds a pentanuclear Cu(I) cluster including coordination of Cu18 and Cu19 by His110 and His104, respectively, both from the H 104 AGNH 108 EH 110 α3-loop-α4 motif. In the structure of Cu(I)-SlCsp3, there is an additional Cu(I) site (Cu20) bound by the central His111 residue of its corresponding (H 107 AGMH 111 EH 113 ) motif ( Figure 4B and Table 2). 42 There are only two Cu(I) sites at the mouth of Cu(I)-RkCsp3, and their positions most closely match those of Cu17 and Cu18 in Cu(I)-MtCsp3 ( Figure 4A,B and Table  2). The α3-loop-α4 His-containing motif is missing, but three His residues (highly conserved in the subfamily, see Figure  S1) originating from the α1-loop-α2 region are present at the mouth of Cu(I)-RkCsp3. The N δ1 atom of His37 coordinates Cu18 with the His42 N ε2 atom binding Cu17 ( Figure 4A), while His29 does not act as a ligand. The tautomeric form of the His42 imidazole (N δ1 H) is unusual as in all published Cu(I)-Csp structures His residues exhibit N δ1 atom coordination (see Table 2 for Csp3s). [13][14][15]42 The thiolate of the only Cys residue (Cys30) in this region of RkCsp3 (there are four in MtCsp3 and SlCsp3; see Figure 4 and Table 2) bridges Cu17 and Cu18, with the latter also bound by Met91 (Cu17 is two-coordinate).
Filling the Cores of Csp3s with Cu(I) Ions. The ability of RkCsp3 and the three classical Csp3s; MtCsp3, SlCsp3, and BsCsp3 to bind Cu(I) has been studied. The high-affinity chromophoric chelator bicinchoninic acid (BCA) was used as the surrogate partner (i.e., by monitoring the loss of Cu(I) Cu(I) binding stoichiometry e 18.0 ± 0.6 (n = 5) 17.9 ± 1.0 (n = 8) 14 18.2 ± 0.5 (n = 5) 19.6 ± 0.8 (n = 5) 14 average Cu(I) affinity f SlCsp3 has been studied previously 42 with our analysis carried out only to enable a direct comparison to data for the other Csp3s we have characterized (see the Supporting Results for more information). b Sequence alignments are provided in Figure S1. c Determined using the mean residue ellipticity at 222 nm obtained from far-UV circular dichroism (CD) spectra. 14 d Determined in solution using analytical gel-filtration chromatography. 13 from [Cu(BCA) 2 ] 3− ), and data have been acquired over a range of [Cu(BCA) 2 ] 3− concentrations. This approach is similar to that reported previously for studying the binding of Cu(I) to the initial sites occupied in SlCsp3. 43 The authors of this work state they have discrepancies between the amount of Cu(I) added and the uptake of Cu(I), which appears to become more significant at higher equiv. However, we have not experienced this issue with any of the Csp3s and have focused on reactions adding sufficient Cu(I) to approximately half and fully load the proteins.
The analysis of the Cu(I)-binding data presented is based on experiments with MtCsp3 ( Figure 5A,B), as almost all of the half-and full-loading reactions are monitored for this protein. The kinetics of these reactions are multiphasic (see also ref 43), and data have been fit to two and three exponentials, respectively ( Figure S10). The rates obtained are listed in Table S1. The initial rate of Cu(I) uptake (t 1 ) when half loading MtCsp3 is ∼9-fold faster than the second step (t 2 ). For full loading, very similar t 1 and t 2 values are obtained (Table S1), and a third stage is observed (t 3 ) that is ∼15-fold slower than t 2 . In the crystal structure of MtCsp3 plus ∼2 equiv of Cu(I), the only sites occupied, and presumably the most thermodynamically stable, are those at the tetranuclear cluster closest to the mouth of the bundle (Cu11−Cu14). 15 Adding more Cu(I) results in further occupancy of this and the two other tetranuclear clusters (Cu3−Cu6 and Cu7−Cu10) that form the main Cu(I) core (Figures 1 and 3B and Table 2). 15 The last sites to bind Cu(I) are those at the two extremes of the bundle, whose affinities must be weakest (access to sites at the hydrophobic end of the protein may be sterically hindered). The data therefore appear consistent with t 1 and t 2 being the rates for forming the main Cu(I) core, initially Cu11−Cu14. The slowest step (t 3 ) corresponds to completing the process of filling MtCsp3 with Cu(I), predominantly involving binding by sites at the hydrophobic end and the mouth of the bundle.
The core of RkCsp3 fills much more quickly with Cu(I) ions from [Cu(BCA) 2 ] 3− than does MtCsp3, and the process is complete in <1 h ( Figure 5C). Fitting data in the same way as described for MtCsp3 gives the results shown in Table  S1. Again, the two rates for half loading the protein with Cu(I) agree with the t 1 and t 2 values from fitting the fullloading experiments. We therefore assume these rates are also due to binding mainly at the three conserved tetranuclear clusters of the RkCsp3 Cu(I) core ( Figure 3 and Table  2), 14,42 and both are ∼20-fold faster than for MtCsp3 (Table  S1). The last stage of filling the RkCsp3 bundle with Cu(I) (t 3 ) is ∼30-fold faster than in MtCsp3. This process again  To further investigate the structural features of a Csp3 that are important for the rate of Cu(I) uptake, experiments were performed on two more classical Csp3s. SlCsp3 and BsCsp3 were chosen as reported in vitro data for these proteins (  14 Half and full loading from [Cu(BCA) 2 ] 3− were also fit in the same way as described for MtCsp3. The rates for SlCsp3 and BsCsp3 (Table S1) were all significantly faster (t 1 and t 2 ≈ 20-fold and t 3 ≈ 40fold) than for MtCsp3, and full loading is complete in <1 h for both proteins ( Figure 5D), comparable to the data for RkCsp3. We assume that the rates correspond to the same processes described above, particularly considering the high structural similarity of Cu(I)-MtCsp3 and Cu(I)-SlCsp3 ( Figure 1). It should be noted that the initial Cu(I)-binding cluster in a crystal structure of SlCsp3 is adjacent to that in MtCsp, 15 shifted towards the mouth of the bundle ( Figure  1). 43 However, these crystal structures were obtained at different pH values that could influence Cu(I) binding.
Removing Cu(I) Ions from the Cores of Csp3s. The rates of Cu(I) removal from all four Csp3s have been studied using the higher-affinity chromophoric ligand bathocuproine disulfonic acid (BCS), i.e., by following the formation of the [Cu(BCS) 2 ] 3− complex. This ligand was used rather than BCA as it acquires Cu(I) more rapidly from Csp3s. However, it still takes ∼20 h ( Figure 6A) for all Cu(I) to be acquired from RkCsp3 by BCS (∼2.5 mM). Removal of Cu(I) with BCS is slower for all classical Csp3s that have been tested, and only ∼12% is acquired from MtCsp3 after 20 h (∼20% after 85 h). 14 For SlCsp3 ( Figure 6B) and BsCsp3, 14 ∼70 and ∼80% of Cu(I) are removed after 85 h (∼40 and ∼50% at 20 h), respectively. Most Cu(I) removal data fit better to two exponentials (fits to one and two exponentials are similar for SlCsp3), with an initial faster phase of low amplitude ( Figure S11). The rates for the main (slower) phases are compared in Table S2. Because Cu(I) is removed slowly from MtCsp3, the analysis of data for up to ∼85 h is unreliable, which is confirmed by the much smaller than expected changes in [Cu(BCS) 2 ] 3− concentration obtained from fits. When the reaction was monitored for >600 h (>25 days), the rate was ∼10-fold slower (Table S2). Using this value, Cu(I) removal from RkCsp3 is ∼100-fold faster than that from MtCsp3, while the rates are ∼25-fold quicker for SlCsp3 and BsCsp3 (∼3.5-fold slower than RkCsp3).
Structural Features of Csp3s that Influence the Rates of Cu(I) Binding and Removal. Filling and emptying the core of RkCsp3, the first characterized member of a new subgroup of Csp3s, via exchange with Cu(I) ligands are both relatively fast processes. The rates of Cu(I) uptake are similar to those for SlCsp3 and BsCsp3, but removal is slower in the classical Csp3s. For MtCsp3 (another classical There are two His residues coordinating Cu17 and Cu18, the only Cu(I) sites at the mouth of Cu(I)-RkCsp3, but these are present as different tautomers: His42 exists as the N δ1 -H form (coordination via the N ε2 atom), with His37 in the N ε2 -H form ( Figure 4A). The coordination of a His via its N ε2 atom has not previously been observed in a Csp [the His residues in the α3-loop-α4 motif of classical Csp3s ligate via their N δ1 atoms ( Figure 4B)]. [13][14][15]42 The significance of the tautomeric form of His ligands at biological Cu sites has been discussed for many years without any clear conclusions. His residues always coordinate mononuclear type 1 electrontransferring Cu sites via their N δ1 atoms, whereas N ε2 atom ligation is more common for type 2 and type 3 Cu sites in a range of enzymes. 1,20,21,44−47 This is consistent with enhanced rates of catalysis for an engineered type 2 Cu site within a designed three-helix bundle scaffold when His residues coordinate via their N ε2 atoms. 48 Information about His ligation in Csps is relatively limited, and it is too early to know if the coordination of Cu17 by the N ε2 atom of His42 in RkCsp3 facilitates fast Cu(I) uptake and removal.
Another potential feature of Csp3s that could influence the rate of Cu(I) uptake and removal is the location of the first sites where Cu(I) ions bind within the core. In MtCsp3, the sites forming the Cu11−Cu14 tetranuclear cluster (Figures  3B and 4B and Table 2) at which Asn58 is a ligand (to Cu13, Figure 4B) are the first to be occupied. 15 Classical Csps typically have either an Asn or Asp residue at this position ( Figure S1) located at the interface between the  Table 2, and Figure S1). This coincides with faster rates of Cu(I) binding. Consistent with this observation, the Asn58Asp MtCsp3 variant exhibits faster Cu(I) uptake than the WT protein (full loading occurs in ∼1.5 h), but this is not as fast as for SlCsp3 and BsCsp3.
Whether the location of the initial Cu(I) cluster in MtCsp3 (Cu11-Cu14) 15 has been influenced by the Asn58Asp mutation, matching that in SlCsp3, 43 remains to be clarified. Cu(I) removal is less affected by the Asn58Asp mutation to MtCsp3, with ∼30−40% acquired by BCS after 85 h (∼20% for the WT protein in the same time). A detailed analysis of Asn58Asp MtCsp3 and other variants mentioned herein will be described in a subsequent paper. However, Asn58 is partially responsible for the slow Cu(I) uptake rates of MtCsp3 and presumably other classical homologues with the same residue in this position. If the α3-loop-α4 motif is missing, as in the new subgroup of Csp3s identified in this work, then the residue corresponding to Asn58 of MtCsp3 is usually a Met (Met46 in RkCsp3, see Figure S1). This coincides with faster Cu(I) uptake and removal. A possible reason that Asn58 slows Cu(I) uptake in MtCsp3 is the hydrogen bonds made by its noncoordinating N δ2 H 2 group, including with the thiolate sulfur ( Figure 7A) and backbone carbonyl oxygen of Cys54. 14 These interactions fix the conformation of the Asn58 side chain, which is the same in the structures of Cu(I)-and apo-MtCsp3 ( Figure  7A). 14 In SlCsp3, such interactions are missing and the orientation of Asp61 is different in the apo-and Cu(I)protein structures ( Figure 7B) with this residue coordinating three different Cu(I) sites (Table 2). These observations are consistent with previous findings that a carboxylate side chain can readily exhibit changes in coordination at metal sites in proteins. 49,50 The Met side chain is unbranched, flexible, and cannot be involved in hydrogen bonding. There appears to be space within the RkCsp3 core for the side chain of Met46 to adopt different conformers, which is the case for Met49 in

■ CONCLUSIONS
Sequence alignments indicated the presence of Csp3s with a different arrangement of Cu(I) sites at the mouth of their four-helix bundles compared to previously studied classical homologues ( Figure S1). The crystal structure of the Cu(I)form of such a protein determined herein, confirms this to be the case, and RkCsp3 is the first characterized example of a new subgroup of Csp3s. These proteins possess only two Cu(I) sites at the mouth of their bundles, compared to five or six in classical Csp3s (Figures 3 and 4A,B). This is due to having only a single Cys residue (Cys30) in this region compared to four in classical Csp3s, and the absence of the α3-loop-α4 His-containing motif. RkCsp3 has an additional Cu(I) site at the hydrophobic end of the bundle coordinated by two Met residues as well as two Cys thiolates. RkCsp3 can bind up to 17 Cu(I) ions per monomer ( ■ METHODS Cloning, Overexpression, and Purification of Proteins. The csp3 gene from Methylocystis sp. ATCC 49242 (Rockwell) was amplified from genomic DNA using the following primers; 5′-GGAGGACGCCATATGCACAAAATGTCCAAGG-3′ (forward, NdeI restriction site in bold) and 5′-CGCAAGCGCCCATGG-TCACGCCGCCATCTTTTGTCCC-3′ (reverse, NcoI restriction site in bold, stop codon underlined). The fragment was ligated into pGEM-T (Promega), verified by sequencing, and subcloned into the NdeI and NcoI sites of pET29a (this was carried out in two steps as the gene contains an NdeI site) to give pET29a_RkCsp3. The SlCsp3gene was synthesized with codons optimized for expression in Escherichia coli (GENEWIZ) and subcloned into the NdeI and HindIII sites of pET29a, giving pET29a_SlCsp3.
E. coli BL21 (DE3) transformed with either pET29a_RkCsp3 or pET29a_SlCsp3 was grown in LB media at 37°C (50 μg/mL kanamycin) until an OD 600 of ∼0.5 to 0.8 was achieved. Cells were induced with 0.1 and 1 mM isopropyl β-D-1-thiogalactopyranoside, grown for 24 h, and harvested by centrifugation at 5000g for 10 min, and pellets were stored at −30°C. SlCsp3 was purified using the same procedure as for MtCsp3. 13,14 Purification was more complex for RkCsp3 due to relatively weak binding to ion-exchange columns. The first step typically involved using a HiTrap Q HP column (5 mL, sometimes two columns connected) equilibrated in 2−20 mM tris(hydroxymethyl)aminomethane (Tris) pH 8.5 plus 1 mM dithiotheritol (DTT) and eluted with a linear NaCl gradient of ∼0−500 mM. The RkCsp3-containing fractions, identified by 18% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), were further purified on a HiTrap SP HP column (5 mL) in 1−10 mM 2-(N-morpholino) ethanesulfonic acid pH 6.0 plus 1 mM DTT and eluted with a linear NaCl gradient (∼0 up to ∼500 mM). Regardless of the buffer concentration, some protein did not bind to this column, and further purification of the flow-through and eluted fractions was carried out in 10−20 mM Tris pH 7.0 to 8.5 plus 1 mM DTT on either a HiTrap Q HP column (5 mL) or connected HiTrap Q and HiTrap SP HP columns (5 mL). When anion and cation exchange columns were used in series, the two columns were eluted separately using ∼0−400 mM NaCl gradients, and fractions from the anion-exchange column were further purified on a HiTrap SP HP column in 20 mM HEPES pH 7.0 plus 1 mM DTT. The purest fractions, as determined by SDS-PAGE, were concentrated and exchanged into 20 mM HEPES pH 7.5 plus 200 mM NaCl by ultrafiltration [Amicon stirred cell with a 10 kDa molecular weight cutoff (MWCO) membrane]. The final purification step utilized a Superdex S75 10/300 GL gel-filtration column, equilibrated in 20 mM HEPES pH 7.5 plus 200 mM NaCl.
Purified RkCsp3 and SlCsp3 were analyzed for Cu and Zn(II) by AAS as described previously. 13,14 Both proteins contained a negligible amount of Cu (≤0.1 equiv) and a maximum of ∼0.4 equiv of Zn(II). The molecular masses of proteins were as expected when measured using liquid chromatography−mass spectrometry.  42 and Cu(I)-RkCsp3 (C), colored as in Figures 1 and 3, are compared with those in the apo-proteins (apo-NmCsp3 is used in C) that are all colored light gray. The side chains of Asn58, Asp61, and Met46 ligate Cu13 in MtCsp3, SlCsp3, and RkCsp3, respectively, while Asp61 also coordinates Cu11 and Cu15 (Table 2). Cu(I)-ligand bonds are shown as dashed lines of the appropriate color, and one of the hydrogen bonds between Asn58 and Cys54 in the MtCsp3 structures is shown as a slate dashed line in (A).

Inorganic Chemistry pubs.acs.org/IC Article
Purification of MtCsp3 and BsCsp3. MtCsp3 and BsCsp3 were purified using previously reported protocols. 13,14 Protein Quantification. All procedures, apart from Bradford assays (see below), were carried out under strict anaerobic conditions using an anaerobic chamber (Belle Technology, [O 2 ] ≪2 ppm), gastight syringes (Hamilton), and sealable quartz cuvettes (Hellma). UV/vis spectra were recorded on a λ35 spectrophotometer (PerkinElmer). The concentration of apo-Csp3s was determined using the 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB, Ellman's reagent) assay to quantify thiols, 51 under anaerobic conditions as described previously. 13,14 These measurements assumed 16 and 18 free thiols for RkCsp3 and SlCsp3, respectively. For all of the Cys residues to be accessible in this assay, the proteins are unfolded, which was achieved in 20 mM HEPES pH 7.5 plus 200 mM NaCl containing 6.8 M guanidine hydrochloride (sometimes 8 M urea) and 1 mM ethylenediaminetetraacetic acid. The apo-Csp3s were also quantified using the Bradford assay with the Coomassie Plus protein assay kit (Thermo Scientific) as described previously. 13,14 Bradford concentrations were divided by those obtained from the DTNB assay giving Bradford:DTNB ratios of 1.72 ± 0.22 (n = 28) and 1.18 ± 0.08 (n = 18) for RkCsp3 and SlCsp3, respectively. MtCsp3 and BsCsp3 were quantified in >7.0 M urea as described previously, 13,14 giving the expected Bradford:DTNB ratios.
Investigating Cu(I) Binding. A Cu(I) stock solution (typically 50 mM) prepared using [Cu(CH 3 CN) 4 ]PF 6 (Merck) in 100% anhydrous acetonitrile was diluted into 20 mM HEPES pH 7.5 plus 200 mM NaCl. 13,14,38,52 The Cu(I) concentration of this working stock was determined with BCS that forms the [Cu(BCS) 2 ] 3− complex with an ε value of 12 500 M −1 cm −1 at 483 nm. 38,53 This was routinely compared to the total Cu concentration quantified using AAS. To monitor the effect of Cu(I) binding to proteins, the buffered Cu(I) solution was titrated into apo-Csp3s (typically ∼5 μM) and the formation of S(Cys)→Cu(I) ligand-tometal charge-transfer bands was observed by UV/vis spectroscopy. Fluorescence was also measured during Cu(I) titrations on a Cary Eclipse fluorimeter (Varian), exciting at 280 nm and monitoring the emission in the 400−700 nm range with excitation and emission slits set to 10 and 20 nm, respectively. 13,52 Cu(I) was also added to protein (∼2.5 μM) plus ∼100 μM BCA, which forms the [Cu(BCA) 2 ] 3− complex with an ε value of 7700 M −1 cm −1 at 562 nm. 32,40,53 This experiment was performed as either a titration for RkCsp3 [equilibration for each Cu(I) addition is complete in ∼15 min] or by setting up a series of mixtures that were incubated and measured for up to 48 h for SlCsp3 (slower equilibration).
Investigating to approximately half load the protein to two exponentials providing rates for the first two stages of uptake (t 1 and t 2 ). The full-loading data were fit to three exponentials that also provided the rate for the slowest stage of Cu(I) uptake (t 3 ) as well as t 1 and t 2 .
Far-UV CD Spectroscopy and Analytical Gel-Filtration Chromatography. For far-UV (180−250 nm) CD spectroscopy, Cu(I)-Csp3s binding ∼17 to 18 mol equiv of Cu(I) was prepared by the addition of buffered Cu(I) (∼2−5 mM) to apo-protein (∼55 μM apo-RkCsp3 and 138 μM apo-SlCsp3) in 20 mM HEPES pH 7.5 plus 200 mM NaCl. 13,14 Both the apo-and Cu(I)-proteins were transferred into either 20 or 100 mM phosphate pH 8.0 by desalting on a PD10 column. The apo-Csp3s were quantified using both the Bradford and DTNB assays, while the concentrations of Cu(I)-Csp3 samples were determined from the Bradford assay value corrected using the Bradford:DTNB ratio for the apo-sample. Far-UV CD spectra of the Csp3s (apo: 0.66−0.86 mg/mL, 46−67 μM; Cu(I): 0.68−0.84 mg/mL, 47−65 μM) were recorded aerobically on a Jasco J-810 spectrometer at 20°C using a 0.2-mm-path-length quartz cuvette, as described previously. 13,40,54 Analytical gel-filtration chromatography was performed on a Superdex 75 10/300 GL column (GE Healthcare) equilibrated in degassed 20 mM HEPES pH 7.5 plus 200 mM NaCl purged with nitrogen throughout the experiment. 13,52 Protein concentrations ranged from 30 to 152 μM for apo-Csp3s and from 29 to 101 μM for the Cu(I) forms [binding ∼16−19 mol equiv of Cu(I)], in which the sample volume was 100 μL and the flow rate was 0.8 mL/min with elution monitored at 240 nm. Calibration of the column and the determination of estimated molecular weights were carried out as described previously. 13,52 Average Cu(I) Affinities. The average Cu(I) affinities of RkCsp3 and SlCsp3 were determined using an approach described previously. 13 Cu(I) Removal from Csp3s by BCS. The rate of Cu(I) removal from almost fully Cu(I)-loaded Csp3s was investigated using BCS as described previously. 13,14 Csp3s (1.1−1.3 μM) binding 15.9−17.1 (RkCsp3) and 15.8−18.1 (SlCsp3) mol equiv of Cu(I), prepared as described above, were mixed anaerobically with 2.5 mM BCS in 20 mM HEPES pH 7.5 plus 200 mM NaCl. The absorbance of the sample at 483 nm was monitored over time at 22°C. Plots of [Cu(BCS) 2 ] 3− concentration against time were fit to either one or two exponentials in Origin 7. The removal of Cu(I) was also performed in the presence of 6.8 M guanidine hydrochloride to determine the amount of Cu(I) present in the sample and therefore the expected end point for each reaction.
Crystallization, Data Collection, Structure Solution, and Refinement for Cu(I)-RkCsp3. For crystallization, apo-RkCsp3 (69.5 μM) was incubated with ∼18 equiv of Cu(I) in 20 mM HEPES pH 7.5 plus 200 mM NaCl in the anaerobic chamber. The sample was concentrated using a Vivaspin 500 centrifugal concentrator (10 kDa MWCO). The final sample was quantified with a Bradford assay (17.8 mg/mL, obtained by correction using the Bradford:DTNB ratio), and the Cu(I) concentration was assayed using BCS in the presence of guanidine hydrochloride, giving 22.8 equiv. The Cu(I)-protein was removed from the anaerobic chamber, and screens were set up on a mosquito liquidhandling robot (SPT Labtech). Crystallization plates were immediately returned to the anaerobic chamber, sealed after being flushed with nitrogen for 3 min, and incubated at room temperature. Diffraction-quality crystals were obtained using the sitting drop method of vapor diffusion from 200 mM MgCl 2 , 100 mM HEPES pH 7.5 plus 25% PEG 3350 (100 or 200 nL of protein plus 100 nL of reservoir solution, well volume 80 μL). Crystals were cryoprotected with 25% ethylene glycol, and diffraction data were collected at the Diamond Light Source (Didcot, U.K.). Data were Inorganic Chemistry pubs.acs.org/IC Article integrated with XIA2, 55 using XDS, 56 and scaled with Aimless. 57 The space group was confirmed using Pointless. 58 Phases were solved by molecular replacement using Phaser 59 and PDB file 3LMF as the search model. The model was refined with refmac 60 and manual model building was carried out with COOT. 61 Model validation was performed with COOT 61 and Molprobity. 62 Other software used was from the CCP4 cloud and the CCP4 suite. 63,64 Figures were made with PyMol, 65 and data collection statistics and refinement details are reported in Table S3.

■ ASSOCIATED CONTENT Data Availability Statement
The raw data underpinning conclusions of this article that are not included in the Supporting Information will be made available by the authors.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.2c04490. Supporting results, amino acid sequence alignments of Csp3s, Cu(I) binding data for RkCsp3 and SlCsp3, average Cu(I) affinity for SlCsp3, tetrameric arrangement of crystallized Cu(I)-RkCsp3, influence of Cu(I) on secondary and quaternary structures of RkCsp3 and SlCsp3, an overlay of the structures of Cu(I)-RkCsp3 and apo-NmCsp3, fits of Cu(I) uptake and removal data, and data statistics and refinement for the crystal structure of Cu(I)-RkCsp3 (PDF)

Accession Codes
The crystal structure of Cu(I)-RkCsp3 has been deposited at the Protein Data Bank with the code 6ZIF.