Crystal Structures of Multicopper Oxidase CueO Bound to Copper(I) and Silver(I)

Background: The multicopper oxidase CueO allows Escherichia coli to survive in aqueous solutions with a high copper concentration. Results: Cu(I) and Ag(I) ions coordinate to a flexible, methionine-rich sequence. Conclusion: The methionine-rich insert and a substrate-binding site ensure that only Cu(I) is oxidized. Significance: Understanding how bacteria survive in the presence of normally toxic concentrations of metal ions may lead to better antibacterial agents. The multicopper oxidase CueO oxidizes toxic Cu(I) and is required for copper homeostasis in Escherichia coli. Like many proteins involved in copper homeostasis, CueO has a methionine-rich segment that is thought to be critical for copper handling. How such segments function is poorly understood. Here, we report the crystal structure of CueO at 1.1 Å with the 45-residue methionine-rich segment fully resolved, revealing an N-terminal helical segment with methionine residues juxtaposed for Cu(I) ligation and a C-terminal highly mobile segment rich in methionine and histidine residues. We also report structures of CueO with a C500S mutation, which leads to loss of the T1 copper, and CueO with six methionines changed to serine. Soaking C500S CueO crystals with Cu(I), or wild-type CueO crystals with Ag(I), leads to occupancy of three sites, the previously identified substrate-binding site and two new sites along the methionine-rich helix, involving methionines 358, 362, 368, and 376. Mutation of these residues leads to a ∼4-fold reduction in kcat for Cu(I) oxidation. Ag(I), which often appears with copper in nature, strongly inhibits CueO oxidase activities in vitro and compromises copper tolerance in vivo, particularly in the absence of the complementary copper efflux cus system. Together, these studies demonstrate a role for the methionine-rich insert of CueO in the binding and oxidation of Cu(I) and highlight the interplay among cue and cus systems in copper and silver homeostasis.

In Escherichia coli, two chromosomally encoded systems, cue and cus, respond to copper stress by expressing proteins that either oxidize Cu(I) to less toxic Cu(II) or expel excess copper from the cell (1,2). The cue regulon codes for the Cu(I) oxidizing multicopper oxidase CueO, the subject of this report (3,4), and the Cu(I) translocating P-type ATPase, CopA (5). The expression of both CopA and CueO are up-regulated by CueR, a transcription factor activated by Cu(I) (6,7). Under anaerobic conditions where CueO is rendered inactive, or when the cue system is overwhelmed by high copper concentrations, the cus system is activated (1). The cus system functions to remove periplasmic Cu(I) (3,8) and codes for four proteins, a threecomponent CusCBA copper efflux pump, and the periplasmic copper chaperone CusF (9,10). Under aerobic conditions, the cus system appears redundant for copper homeostasis (1,11).
Ag(I), a Cu(I) mimic, also induces the cus and cue systems and the gene products can bind and transport Ag(I) efficiently (5,(11)(12)(13). The cus system, in fact, was originally identified as a Ag(I) inducible, Ag(I) detoxifying system, rather than a copper detoxifying system (11). Cu(I)-binding sites in proteins can often bind Ag(I), raising the possibility that either ion may compromise the detoxification of the other ion (see for example, Ref. 14). The possibility for combined Cu(I)/Ag(I) stress is common, occurring, for example, in hospitals (15), public health water systems (16), and copper mines, which are often also a source of silver (AgCl or Ag 2 S). How bacteria cope with combined Cu(I) and Ag(I) stresses, and how the handling of either ion is influenced by the presence of the other, is not known. The apparent redundancy of the cus system in aerobic copper detoxification and its ability to provide moderate Ag(I) detoxification raises the possibility that the cus system may function to overcome Ag(I) poisoning of the cue system.
CueO is a member of the multicopper oxidase (MCO) 2 family, which includes ascorbate oxidase, laccase, ceruloplasmin, Fet3p, and PcoA (17,18). All MCOs couple four one-electron substrate oxidation steps to the four-electron reduction of dioxygen to water (19). Cu(I) is the best substrate so far identified for CueO and the likely major substrate for CueO in vivo (20,21). MCOs contain four copper atoms, designated type 1 (T1), type 2 (T2), and two type 3 (T3). The T1 copper gives rise to an absorption peak at 610 nm in UV-visible spectra and the intense blue color typical of MCO proteins. The T2 and two T3 copper atoms form a trinuclear center (TNC) and give rise to a peak in the region of ϳ330 nm (19). Substrates are oxidized near the T1 copper site, releasing an electron that is shuttled through the T1 site to the TNC, where dioxygen binds and is reduced to water.
The T1 copper in CueO, unlike with laccases (22), is buried in the protein interior (23,24) (Fig. 1). A 45-residue insert (residues 355-399) containing 14 methionines and five histidines, blocks solvent access to the T1 site and contributes ligands to an additional copper-binding site that must be occupied for full CueO activity (24). Organic compounds and Fe(II) are therefore only substrates for CueO in the presence of excess copper. We originally termed this additional site the regulatory copper site (rCu) for its role in stimulating CueO activity (24), but have renamed it the substrate copper site (sCu) in view of its Cu(I) substrate binding role. Much of the methionine-rich insert has been disordered and unseen in previous CueO structures (23)(24)(25). Similar methionine-rich regions are found in other proteins involved in copper homeostasis and are thought to have roles involving copper binding and transport. Prominent examples include the copper importer Ctr1 in Saccharomyces cerevisiae (26), PcoA and PcoC, both from the plasmid-based copper resistance system in E. coli (17). In Ctr1, a methionine-rich motif occurs at the N terminus and is required for Cu(I) transport across the cytoplasmic membrane, whereas in PcoC, a methionine-rich motif lies at the dimerization interface, suggesting that methionine-rich regions are involved in proteinprotein interactions (27). It has also been suggested that PcoC transfers Cu(I) ions bound at its solvent-exposed methionine motifs by docking with the methionine-rich region of the MCO PcoA, which then oxidizes the transferred Cu(I) (21). The MCO CopA from the plasmid encoded cop system in Pseudomonas syringae binds up to seven additional copper atoms (28).
Although these examples provide strong evidence supporting copper binding and transport by methionine-rich motifs, structural characterization of Cu(I) binding at these sites has been lacking.
Here, we have determined the complete structure of the CueO methionine-rich insert and its binding to Cu(I) and Ag(I).
We also make use of mutagenic, kinetic, and cell growth experiments to uncover the role for these residues in the handling of both copper and silver.

EXPERIMENTAL PROCEDURES
Strains, Growth Media, and Copper Tolerance Assay-Wildtype E. coli strain W3110 and cus-deleted strain W3110 ⌬cusCFBA::cat (GR6) (4) were used for copper tolerance assays using Luria Bertani (LB) media lacking NaCl as the growth media. Strain BL21-DE3 was used for protein expression and grown in regular LB media. Ampicillin (100 g/ml) was added to growth media as appropriate. For copper tolerance assays, overnight cultures of W3110 or GR6 were diluted 1:500 with fresh LB media (without NaCl) and grown for 2 h at 37°C, with shaking at 200 rpm. This culture was diluted 1:500 using fresh LB media (without NaCl) in the presence or absence of 5 M AgNO 3 and different concentrations of CuSO 4 , and grown at 37°C, with shaking at 200 rpm. After 15 h, the optical density at 600 nm wavelength was measured to monitor growth.
Mutagenesis and Protein Isolation-Three forms of wildtype CueO and two site-specifically mutated CueO proteins were used in this study. Of these, four proteins have been produced recombinantly in fusion with a C-terminal Strep-Tag II affinity epitope and isolated from E. coli cells, as previously described (3,20). These are wild-type CueO (tagged) (3), apo-CueO (same as wild-type CueO but without copper cofactor), C500S CueO (containing a Cys 500 to Ser mutation, leading to loss of the type 1 copper) (24), and ⌬Met CueO (containing Met to Ser mutations for methionines 358, 361, 362, 364, 366, and 368). We also produced CueO without any purification tag or extraneous linker sequences, referred to herein as untagged CueO.
C500S CueO was constructed using a PCR-based approach as previously described (20,24). Untagged CueO and ⌬Met CueO were generated from the wild-type CueO insert (EcoRI and PstI) in plasmid pASK-IBA3 (IBA, Göttingen, Germany) as the template in a PCR-based approach using the following primers. The forward and reverse primers for untagged CueO were 5Ј-AAA TCT AGA TAA CGA GGG CAA AAA ATG CAA CGT GAT TTC TTA AAA TAT TC-3Ј, and 5Ј-AAA CTG CAG TTA TAC CGT AAA CCC TAA CAT CAT CCC-3Ј. The forward and reverse primers for ⌬Met CueO were 5Ј-TAT GGA TCC GAG TCT CGA TTC GTC CGG GAG CCA GTC GCT AAG TGA GAA ATA TGG CGA TCA GGC GAT G-3Ј and 5Ј-ATC GGA TCC ATA GAG AGT TGC AGC TTG-3Ј. Restriction sites are underlined, the stop site is in italic, and base changes are in bold. The PCR-amplified untagged CueO fragment was inserted into expression plasmid pASK-IBA3 between the XbaI and PstI restriction sites.
All proteins were expressed in the absence of chloride ion, which was previously shown to bind CueO at the TNC (23). Accordingly, Tris-SO 4 buffer was used at all steps of purification in place of Tris-Cl. Proteins were loaded with the Cu(II) cofactor by addition of 5 mM CuSO 4 to the cell lysate, except for apo-CueO. Untagged CueO was purified using a new procedure. The protein was precipitated from the cell lysate by addition of (NH 4 ) 2 SO 4 to 60% saturation followed by centrifugation (32,000 ϫ g, 4°C, 15 min) and redissolving the pellet in a minimal amount of 10 mM Tris-SO 4 buffer, pH 9.0. Excess salt was removed by extensive dialysis against 10 mM Tris-SO 4 buffer, pH 9.0, and the dialyzed protein was loaded onto a HighPrep 16/10 Sepharose Fast Flow (Amersham Biosciences) ion exchange column equilibrated with 10 mM Tris-SO 4 buffer at pH 9.0. Unbound protein was removed by washing with buffer (10 mM Tris-SO 4 , pH 9.0) and bound protein was eluted with a linear gradient of 0 -500 mM Na 2 SO 4 in 10 mM Tris-SO 4 buffer, pH 9.0. Eluted fractions with A 610 Ͼ 0.05 (Varian Cary 300) were pooled, concentrated, and loaded on to a Sephacryl size exclusion column to remove remaining contaminants. Eluted fractions displaying a 610-nm absorption band were pooled. The final material displayed a single band on an SDS-polyacrylamide gel upon staining with Coomassie dye and was estimated to be greater than 95% pure. UV-visible spectra of purified protein showed a ratio of A 280 /A 610 ϭ 11.0, identical to that of CueO with intact Strep-tag. Apo-CueO was isolated and purified in the absence of added copper, resulting in a colorless, copper-free protein. Protein concentrations were determined spectroscopically (Varian Cary 300) using an ⑀ 280 of 63,036 M Ϫ1 cm Ϫ1 , which was estimated based on the protein sequence.
Steady-state Oxygen Consumption Kinetics-Steady-state Cu(I) and Fe(II) oxidase activities were measured as rates of oxygen consumption using an oxygraph (Hansatech, Cambridge, UK) as previously described (20). Stock solutions of the substrates were prepared in an anaerobic chamber. The substrate, Cu(I), was added as a complex, [Cu(I)(MeCN) 4 ]PF 6 (Sigma), and Fe(II) was added as the complex Fe(NH 4 ) 2 (SO 4 ) 2 ⅐6H 2 O. Fe(II) oxidase activity was seen only in the presence of 1 mM Cu(II) added to the reaction mixture as CuSO 4 . Steady-state Ag(I) inhibition kinetic measurements were made similarly to the Cu(I) oxidase activity measurements, except that the Cu(I) substrate concentration was held constant at 500 M and the concentration of AgNO 3 was varied. Small (25 l) aliquots of AgNO 3 (10 nM to 1 mM) were added into the reaction mixture (950 l) containing 10 g/ml of protein in 100 mM Tris acetate buffer, pH 5.0. A 25-l aliquot of Cu(I) substrate, stored as a 20 mM stock solution of [Cu(I)(MeCN) 4 ]PF 6 , was added to initiate the reaction. All kinetic measurements were made at 23°C and pH 5.0. Activity measurements were plotted as a function of substrate or inhibitor concentration using SigmaPlot 7.0 (SSI, Richmond, CA) and kinetic constants, K m and k cat , were evaluated using a least-squares fit of a Michaelis-Menten curve. Enzyme velocities were plotted after subtracting the background contribution of enzyme-independent Cu(I) or Fe(II) oxidation.
X-ray Data Collection and Structure Determination-Crystals were picked up in a small loop (Hampton), cryo-cooled in liquid nitrogen, and diffraction data were measured at 100 K. Data for ⌬Met CueO ϩ Cu(I) and untagged CueO were collected at SSRL beamline 9-2 on a MAR325 detector. C500S CueO ϩ Cu(II) and C500S CueO ϩ Cu(I) data were collected at APS, beamline 14BM-C (BioCARS) on a Q315 detector. Diffraction data for apo-CueO and untagged CueO ϩ Ag(I) were measured at 100 K on a Rigaku RUH3-R/R-Axis IV ϩϩ imaging plate system. All crystals were monoclinic, space group P2 1 , with one molecule per asymmetric unit and cell parameters a ϭ 50 -51 Å, b ϭ 91-92 Å, c ϭ 54 Å, ␤ ϭ 103°. All data were processed with CrystalClear (29); data reduction statistics are given in Table 1. The structure of a tagged wild-type CueO (PDB codes 1KV7 (23) or 1N68 (24)) was used as a starting model for each of the new structures. Models were rebuilt with COOT (30), refined with REFMAC5 (31), and figures were prepared using PyMOL. The 1.1-Å structure was refined using anisotropic temperature factors; all others were refined isotropically. Other calculations were performed using the CCP4 package (32).
where ͗I͘ is the mean intensity of all symmetry-related reflections I h . c R cryst ϭ (⌺͉F obs Ϫ F calc ͉)/⌺F obs . R free as for R cryst , using a random subset of the data (5%) not included in the refinement. Reflections with F Ͼ 0 F were used in the refinement. OCTOBER 28, 2011 • VOLUME 286 • NUMBER 43

RESULTS
Complete Structure of CueO at 1.1-Å Resolution-Previous structures of CueO have only a portion of the methionine-rich insert (residues 355-400) sufficiently well ordered to be seen in the crystal (23)(24)(25). We produced a recombinant CueO without the C-terminal Strep-tag that was used in our previous studies in the hopes of finding a crystal form with the missing residues resolved and to ensure that the purification tag was not affecting activity. Spectra and catalytic activity for the highly pure untagged protein were indistinguishable from that of the tagged version and readily crystallized in the same crystal form as for the tagged protein. One such crystal yielded diffraction to 1.1-Å resolution, allowing for the interpretation of the missing residues.
The N-terminal portion of the methionine-rich insert (residues 355-379) contains Met 355 and Asp 360 , which are part of the sCu-binding site, a helical region that blocks access to the T1 and sCu sites (residues 356 -371), and eight additional well ordered residues (372-379), as previously described (23,24). The remainder of the insert (residues 380 -399) and residues 400 -402 were previously unseen but are now shown to form a loosely constrained loop lying across the methionine-rich helix and having few contacts with the rest of the protein (Fig. 1). Electron density, including side chains, was clear for residues 379 -384 and 391-402, and somewhat discontinuous, but present for residues 385-390 (supplemental Fig. S1). The loop winds through a solvent-occupied region between CueO molecules in the crystal. In solution, this region is likely to be intrinsically disordered and only crystal packing constraints allow for the present conformation to be seen, even at 1.1-Å resolution. This 22-residue region contains 5 methionine and 5 histidine residues, and may provide additional Cu(I)-or Ag(I)-binding sites in solution when metal concentrations are high.
The remainder of the structure was very well defined. The C terminus was ordered and formed specific hydrogen bonds with the nitrogen atom of Ala 490 and several water molecules; this region was essentially unchanged from that seen with the Strep-tag present. The T2 copper atom was slightly depleted in the structure and built as 75% occupied. An unknown ligand, possibly ethanediol, a PEG byproduct, was found bound to T2 and modeled in two orientations. A single oxygen atom was attached to one of the T3 copper atoms, consistent with the 1-electron reduced resting oxidized state and possibly resulting from hydrated electrons generated in the intense synchrotron x-ray beam. This same mixed oxidation state species has been seen in several laccase structures (PDB entries 1V10 (33), 1HFU (34), and 1A65 (35)). The trinuclear center geometry will be discussed in detail elsewhere.
Crystal Structures of C500S CueO ϩ Cu(I) or Cu(II)-Cysteine 500 is one of the ligands of the T1 (blue) copper in CueO (Fig. 1). The C500S mutation results in a colorless protein that presumably lacks the T1 copper and has no catalytic activity (20). The UV-visible solution spectrum of C500S showed a complete loss of absorbance at 610 nm, consistent with the loss of blue color and catalytic activity (supplemental Fig. S2). MCO proteins with mutations that lead to loss of the T1 copper are often referred to as "T1D" for T1 copper depletion. To examine the role of the T1 copper in the loading of substrate copper at the sCu site and along the rest of the methionine-rich insert, we determined the crystal structures of the C500S mutant protein in the presence of either Cu(II) or Cu(I), the likely true CueO substrate.

C500S
CueO crystallized under the same conditions as the wild-type protein. The Cu(II) complex was obtained by soaking C500S crystals in 25 mM CuSO 4 for 50 min followed by cryocooling. The Cu(I) complex was obtained by reducing the Cu(II) complex for 5 min with sodium dithionite, followed by cryo-cooling. Structures were determined at 1.5-and 1.8-Å resolution for the Cu(II) and Cu(I) complexes, respectively, and both are very similar to the wild-type protein, except at the copper-binding sites.
The C500S Cu(I) complex contained a copper atom at the sCu site much as previously described for the wild-type protein (24), consistent with Cu(I) binding as substrate (Fig. 2). However, unexpectedly, the sCu site was unoccupied on addition of Cu(II). We previously identified the sCu site through the soaking of Cu(II) into crystals of the wild-type protein (24). That it no longer binds to C500S suggests that binding of Cu(II) at the sCu site requires a functional T1 site. The reason for this is not yet clear.
Ser 500 was well resolved in the Cu(II)-soaked crystal and the T1 site completely empty of copper, as expected (Fig. 2, e and f).
The T1 site was largely unchanged except for a slight expansion and the appearance of a well ordered water molecule that was ϳ0.9 Å from the position for the T1 copper in the wild-type protein and hydrogen bonded to Ser 500 , His 505 , and His 443 . These three residues along with Met 510 coordinate the T1 copper in the wild-type protein. The trinuclear copper centers in the Cu(I) and Cu(II) structures were much like that described above for the untagged protein, except that they both appeared to be incompletely photoreduced.
Binding of Cu(I) and Ag(I) to the Methionine-rich Insert-We and others have speculated that the methionine-rich insert might be used for copper binding as part of the response to copper toxicity (Fig. 1). The C500S Cu(I) structure, but not the Cu(II) structure, displayed two additional copper atoms bound to the methionine-rich helix (Fig. 3), supporting this hypothesis. Importantly, binding was limited to Cu(I), the more toxic of the copper ions and the probable CueO substrate.
Two Cu(I) ions, Cu6 and Cu7, were bound along the methionine-rich helix that blocks access to the sCu and T1 sites. A third, very low-occupancy copper site is near Met 361 (Cu8, omitted in Fig. 3). Cu6 was coordinated through Met 358 and Met 362 and another ligand was presently modeled as a water molecule. The methionine ligands lay on the same face of the helix and were perfectly arranged to bind copper. Although we have placed a water molecule in the third ligand site for Cu6, the electron density for it is distinctly too large (Fig. 3). This, combined with the fact that water is an unlikely Cu(I) ligand, suggests that the third ligand was a methionine or a histidine side b, geometry of Cu(I) ligation. The water molecule shown was previously found to be coordinated with Cu(II), but is too far from Cu(I) in the present structure for coordination (3.6 Å). c, the sCu site is fully occupied with Ag(I) in CueO ϩ Ag(I). d, geometry of Ag(I) ligation. e, the T1 site is occupied by a water molecule in C500S CueO. f, water molecule hydrogen bonding geometry. Electron density maps have been contoured at 1.5 . Distances are in Å. chain from the disordered region (residues 380 -403). Cu7 was coordinated through Met 368 and Met 376 , and a third, partially disordered residue, Met 364 (Fig. 3).
There was no evidence of copper coordination at copper sites Cu6 -8 in the Cu(II) structure. However, two additional copper ions, one bound to His 488 on the protein surface and another bound to Met 417 and His 145 , were seen in both the Cu(I) and Cu(II) structures. These sites lie far from both the T1 and sCu sites and seem unlikely to be of functional importance.
To further explore the role of Cu(I) binding, and to ensure that the oxidation states for the copper atoms were correctly assigned, we turned to Ag(I), which has similar binding properties as Cu(I) but is relatively redox inert. The structure of untagged CueO co-crystallized with 3 mM Ag(I) revealed Ag(I) binding to the sCu site (Ag in Fig. 2) and to both major Cu(I)binding sites along the methionine-rich helix (Ag6 and Ag7, Fig.  3). All three sites expanded to accommodate the larger Ag(I) ion (ϳ0.5 Å for the sCu site) and the coordination numbers decreased. The sCu site was fully occupied by silver, whereas Ag6 and Ag7 were only ϳ50% occupied, unlike Cu6 and Cu7, which were both fully occupied in the Cu(I) structure. Interestingly, Ag(I) did not displace copper at either the T1 or TNC sites. The Ag(I) crystals were blue, indicating that Cu(II) and not Ag(I) occupied the T1 site, and the ligation distances and electron densities at both the T1 and TNC sites were consistent with copper, not silver ligation. Taken together, the Cu(I), Cu(II), and Ag(I) results confirm that the methionine-rich region is able to bind additional Cu(I) ions.
Ag(I) Inhibition of CueO-To investigate whether Ag(I) could inhibit Cu(I) oxidase or Cu(II) stimulated Fe(II) oxidase activity in CueO, steady-state oxygen consumption rates were measured for untagged CueO at a saturating substrate concentration (500 M Cu(I) or 500 M Fe(II) ϩ 1 mM Cu(II)) in the presence of increasing amounts of AgNO 3 . Ag(I) sharply inhibited CueO: activity fell to 50% at ϳ1 M Ag(I) and was essentially eliminated at 10 M Ag(I) (Fig. 4a). No change in absorbance at 330 or 610 nm was seen until the Ag(I) concentration reached 1 mM (Fig. 4b), indicating inhibition was not due to displacement of the T1 or TNC copper atoms, consistent with the Ag(I) co-crystallization results described above, which also indicate that Ag(I) does not displace the T1 or TNC copper atoms. Taken together, these data indicate strong binding of Ag(I) at the sCu site abolishes CueO activity.
Copper Tolerance Assay: Inhibitory Effect of Ag(I) on Bacterial Copper Tolerance-Because Ag(I) binding at the sCu site makes CueO catalytically inactive, Ag(I) should also make bacteria dependent on the cus detoxification system for survival in the presence of environmental copper. To examine this, we measured cell growth for wild-type and cus-deleted E. coli strain W3110 in the presence and absence of Ag(I) (Fig. 5). When Ag(I) was absent, there was no difference in the growth of wildtype and cus-deleted strains, consistent with previous findings (1). However, addition of just 5 M Ag(I) to the growth media led to marked copper sensitivity in the cus-deleted strain, whereas the wild-type strain survived copper stress nearly as well as in the absence of Ag(I). These results suggest that the cus system plays a specific role in aerobic copper tolerance by overcoming Ag(I) poisoning of the cue system.   mutated to serine to disrupt Cu(I) binding at these locations. The crystal structure of this mutant protein in the presence of Cu(I), and containing a reduced T1 copper site, was determined at 2.1-Å resolution, revealing a protein with an overall structure essentially identical to that of the wild-type protein (C␣ root mean square deviation ϭ 0.3 Å). The helical fold at the site of the mutations remained intact; however, Cu(I) binding along the helix was completely abolished (supplemental Fig. S3), whereas Cu(I) occupancy at the sCu site was reduced to ϳ50%. No notable differences in copper binding were found at the TNC, except for a greater depletion of the T2 copper, the most labile of the three, which was only ϳ50% occupied.
⌬Met CueO Kinetics and Spectra-We investigated Cu(I) oxidase activity in ⌬Met CueO to evaluate the role of the methionine-rich helix in catalysis. The steady-state Cu(I) and Fe(II) oxidase activities of the wild-type and ⌬Met proteins were measured in terms of rates of oxygen consumption (Fig. 6), and the resulting K m and k cat values were tabulated as shown in Table 2. Although K m was largely unchanged between the wildtype and ⌬Met proteins, k cat for Cu(I) oxidase activity decreased 4-fold, and Fe(II) oxidase activity 2-fold, in the ⌬Met mutant. It is important to note that the measured K m in these experiments should be viewed as an upper limit, because it does not directly measure the free concentration of Cu(I), which is highly labile, but rather the concentration of Cu(I) released from the acetonitrile delivery complex (20). Kinetic parameters for untagged and Strep-tagged CueO proteins were essentially identical (data not shown).
The spectrum of ⌬Met CueO as isolated showed a roughly 2-fold higher ratio of A 280 /A 333 and A 280 /A 613 as compared with wild-type CueO, suggesting depletion or reduction of copper centers in this mutant (supplemental Fig. S4). However, the ⌬Met CueO ϩ Cu(I) structure showed that although the T2 and sCu coppers were partially depleted, the T1 copper, which is responsible for the A 613 band, was fully occupied. Because the protein was isolated in the presence of excess copper, and all kinetic measurements were performed with excess copper, the weaker A 613 and A 333 bands in the mutant protein spectra may be due to partial T1 reduction caused by partial depletion of T2.
Crystal Structure of Apo-CueO-Many MCO crystal structures display partial depletion of the four catalytic copper ions. In particular, the T2 copper is easily lost (25,(35)(36)(37) and a crystal structure of CueO with T2 removed, and T1 and T3 partially depleted, has been described (erroneously referred to as apo-CueO (25)). Apo-CueO prepared in the absence of added copper to cells can be reconstituted by titration with Cu(II) in the form of copper sulfate (38); reconstitution of other MCOs can also be achieved, but apparently only with Cu(I) (39). Thus, despite the structurally critical locations of the catalytic copper atoms, it appears that most MCOs either fold properly in the absence of copper or remain properly folded when copper is chemically extracted. The exception is ceruloplasmin, for which the loss of the T3 copper atoms, located at the interface between domains 1 and 3, leads to partial unfolding (40).
We determined the structure of apo-CueO to further explore this issue (2.0-Å resolution, Table 1). The protein was prepared without addition of copper to the cell lysate during purification, leading to a colorless purified protein. The resulting structure is essentially identical to that for the holoprotein (PDB entry 1KV7 (23), C␣ root mean square deviation ϭ 0.16 Å), except for the complete absence of copper at any of the copper sites. A single water molecule occupies the TNC and is located at a position halfway between the two T3 copper positions in the native protein at distances ranging from 2.7 to 3.8 Å from the eight histidine residues normally coordinating copper. The T1 and sCu copper sites are empty, with no evidence of bound water. Cys 500 , Met 510 , and His 443 and His 505 , which normally coordinate the T1 copper, are unchanged in position in comparison with the wild-type protein (PDB entries 1KV7 (23) and 1N68 (24)). The 24-residue region generally disordered in CueO structures (residues 379 -403) was somewhat better ordered in the apocrystal, allowing for residues 395-402 to be added to the model.
CueO is transported to the periplasm using the twin-arginine protein translocase, which transports fully folded proteins (41).  Table 2).

TABLE 2 Kinetic parameters for CueO proteins
Because CueO readily folds into its native state in the absence of copper and will apparently take up either Cu(I) or Cu(II), both of which are kept in low concentrations in the cytosol, it may be that CueO is secreted into the periplasm in its folded but copper-free apo form and becomes catalytically competent there only as copper ions appear. However, because CueO is only expressed when copper ions are present, loading may take place before transport. That each copper-binding site is preformed in the apoprotein, and without well ordered water molecules that might compete with copper loading, suggests activation in either compartment is likely to be a rapid process.

DISCUSSION
Although methionine-rich sequences are common in bacterial copper detoxification systems, molecular details of both structure and function are lacking. Here, we present the 1.1-Å crystal structure of CueO with its entire methionine-rich region revealed (Fig. 1). We also present crystal structures of the Cu(I) CueO complex, which confirms Cu(I) binding to the sCu site and reveals Cu(I) binding to two new sites, Cu6 and Cu7. In each of these Cu(I)-binding sites, the metal ion is ligated by methionine residues from the methionine-rich region (Figs. 2 and 3). Removal of the Cu6 and Cu7 sites through mutation leads to a protein with 4-fold reduced catalytic rates for oxidation of Cu(I) ( Table 2), while retaining its native fold (supplemental Fig. S3). Ag(I) binds to all three Cu(I)-binding sites, inhibiting CueO at low micromolar concentrations in vitro and disabling the cue system in vivo, necessitating the use of the cus system for copper tolerance in E. coli exposed to silver.
The MCOs found in many bacterial copper detoxification systems (28,42,43) have both methionine-rich sequences and cuprous oxidase activity. We have shown that Cu(I) binds to at least two sites along the CueO methionine-rich sequence near the sCu site, and these binding sites are specific for the cuprous ion; no metal binding was found at the sCu, Cu6, and Cu7 sites in crystals of C500S CueO soaked with Cu(II). In the wild-type protein, loading of the sCu site with Cu(II) does occur (24), but this is apparently a redox-dependent event and requires a functional T1 site. We showed previously that removal of the sCu site leads to a dramatically impaired enzyme (24). This, taken together with reduced CueO activity when the methionine residues comprising the Cu6and Cu7-binding sites are mutated to serine, strengthens the argument that the methionine-rich insert facilitates Cu(I) oxidation. The mechanism for this is not yet clear, but is likely to involve either direct transfer of Cu(I) from the Cu6 and Cu7 sites into the sCu site, or provision of a path for electron transfer from Cu6 and Cu7 to sCu, and from there to the catalytic copper atoms (Fig. 1). The methionine-rich insert provides for substrate discrimination because it buries the T1 site; its deletion results in a protein with decreased Cu(I) oxidase activity and an increased ability to oxidize organic substrates in the absence of added copper, presumably through increased access to the T1 site (44).
A recent detailed kinetic analysis of Cu(I) oxidation by CueO, where Cu(I) was provided as a [Cu(Bca) 2 ] 3Ϫ complex, concluded that Cu(I) must be bound to the sCu site to be oxidized and the authors argued for a direct transfer of Cu(I) from the complex to the sCu site (21). They also report a subpicomolar dissociation constant for Cu(I) to the sCu site, a value similar to that for the (Bca) 2 complex itself. Because the sCu site is not solvent exposed, it is more likely that direct transfer occurs to substrate-binding sites in the methionine-rich region and from there to the sCu site (Fig.  1). Although we have identified two Cu(I)-binding sites in this region, the 24 residues unseen in the Cu(I)-soaked C500S CueO structure may provide additional Cu(I)-binding sites.
Cu(I) is bound along the methionine-rich region using thioether ligation. Cu7 displays a clear Met 3 -Cu(I) configuration, whereas Cu6 displays a Met 2 X-Cu(I) configuration with the third ligand (X) uninterpretable in the electron density map and modeled as water (Fig. 3). This third ligand is most likely Met 393 , His 395 , Met 396 , or His 398 from the disordered portion of the methionine-rich region, based on the location of these residues in the unliganded high resolution structure. Although uncommon, structures of proteins with Met 2 -Cu(I) or Met 3 -Cu(I) are known, including Met 2 His-Cu(I) binding in PcoC (45) and Met 2 His-Cu(I) binding in CusF (9). The latter is further stabilized by a unique cation-interaction between the metal ion and a tryptophan ring (14,46). The low coordination number, solvent exposure, and disorder of the methionine-rich binding sites in CueO make them suitable for transient copper binding and substrate procurement.
Ag(I) is a potent inhibitor of CueO oxidase activity: as little as 5 M Ag(I) in the growth media suppresses copper resistance in E. coli (Fig. 5) and nearly eliminates Cu(I) oxidase activity for CueO in vitro (Fig. 4). Ag(I) binds to CueO at all three identified Cu(I) sites (Fig. 3), blocking Cu(I) substrate binding and preventing electron transfer from Cu(I) to the T1 copper atom. This provides a simple, clear explanation for silver inhibition of CueO activity.
Copper detoxification systems must function in the presence of silver because these metals often occur together in nature and their similar ligand-binding chemistries make them competitors (11)(12)(13). The cue and cus systems of E. coli, and copper detoxification systems in general, bind and transport both Ag(I) and Cu(I). CopA, the second component of the E. coli cue system, is induced by Ag(I) and transports Ag(I) across the cytoplasmic membrane almost as efficiently as Cu(I). However, for E. coli to survive mixed copper/silver stress, Ag(I) must be removed not only from the cytoplasm but also from the periplasm to prevent CueO inhibition. Inclusion of the cus system, which serves to export Ag(I) from the periplasm (11,47), allows for full copper tolerance, whereas its removal greatly sensitizes E. coli to copper stress in the presence of even small Ag(I) quantities (Fig. 5). Thus, it appears that even in aerobic conditions the cus system is important for copper tolerance, becoming necessary when silver is present.