Cysteine Mutants of the Major Facilitator Superfamily-Type Transporter CcoA Provide Insight into Copper Import

ABSTRACT CcoA belongs to the widely distributed bacterial copper (Cu) importer subfamily CalT (CcoA-like Transporters) of the Major Facilitator Superfamily (MFS) and provides cytoplasmic Cu needed for cbb3-type cytochrome c oxidase (cbb3-Cox) biogenesis. Earlier studies have supported a 12-transmembrane helix (TMH) topology of CcoA with the well-conserved Met233xxxMet237 and His261xxxMet265 motifs in its TMH7 and TMH8, respectively. Of these residues, Met233 and His261 are essential for Cu uptake and cbb3-Cox production, whereas Met237 and Met265 contribute partly to these processes. CcoA also contains five Cys residues of unknown role and, remarkably, its structural models predict that three of these are exposed to the highly oxidizing periplasm. Here, we first demonstrate that elimination of both Met237 and Met265 completely abolishes Cu uptake and cbb3-Cox production, indicating that CcoA requires at least one of these two Met residues for activity. Second, using scanning mutagenesis to probe plausible metal-interacting Met, His, and Cys residues of CcoA, we found that the periplasm-exposed Cys49 located at the end of TMH2, the Cys247 on a surface loop between TMH7 and THM8, and the C367 located at the end of TMH11 are important for CcoA function. Analyses of the single and double Cys mutants revealed the occurrence of a disulfide bond in CcoA in vivo, possibly related to conformational changes it undergoes during Cu import as MFS-type transporter. Our overall findings suggest a model linking Cu import for cbb3-Cox biogenesis with a thiol:disulfide oxidoreduction step, advancing our understanding of the mechanisms of CcoA function.


RESULTS
Either M 237 or M 265 residues of CcoA are required for Cu import. Earlier studies showed that mutants lacking CcoA were unable to accumulate 64 Cu in a CcoA-specific (i.e., temperature-dependent) manner (12). Indeed, mutagenesis of the M 233 and H 261 residues of CcoA conserved motifs (M 233 XXXM 237 and H 261 XXXM 265 ) ( Fig. 1, left panel) completely abolished cbb 3 -Cox activity (Table 1), and cellular 64 Cu accumulation, while mutating M 237 or M 265 only partially decreased these activities (17). The results indicated that the conserved M 233 and H 261 residues of CcoA are essential for its function, likely forming its intramembrane Cu binding site. However, this study was less informative about the role(s) of the remaining M 237 and M 265 residues of the CcoA conserved motifs (17). To further pursue this issue, a double mutant (M 237 A1M 265 A) lacking both of these Met residues was obtained, and its properties were compared to the corresponding single mutants. Both Escherichia coli ( Fig. 2A) and R. capsulatus (Fig. 2B) cells harboring the double mutant M 237 A1M 265 A produced CcoA variant proteins at wildtype levels. The direct effects of these mutations on CcoA-dependent Cu uptake were determined by monitoring radioactive 64 Cu accumulation in whole cells (see Materials and Methods). Both E. coli and R. capsulatus cells producing this CcoA variant were deficient for 64 Cu uptake ( Fig. 3A and B), similar to those mutants lacking CcoA. The cbb 3 -Cox activity of the double mutant was also very low (;2% of wild type), in contrast to ;73% and ;35% of the single M 237 A and M 265 A mutants, respectively ( Table 1). The R. capsulatus strain lacking a chromosomal copy of ccoA and complemented with a plasmid-borne wild type allele (DccoA/plasmid-born ccoA) (see Table S1 in the supplemental material) was used as a control and exhibited a cbb 3 -Cox activity of 846 6 32 mmol of tetramethyl-p-phenylenediamine (TMPD) oxidized/min/mg of total membrane proteins (referred to as 100% in Table 1). Considering that the CcoA variant lacking both M 237 and M 265 residues was unable to import Cu and produce active cbb 3 -Cox, we concluded that at least one additional Met residue (preferentially M 265 , suggested by its more severe phenotype) located three residues apart from the M 233 or H 261 on TMH7 or TMH8, respectively, is required for Cu import, probably as a Cu binding ligand.
Additional possible metal-liganding CcoA residues of functional importance. CcoA is rich in Met, His, and Cys residues that can act as potential metal ligands (10), and those that are not parts of the conserved motifs were examined for their possible role(s). Two different amino acid sequence alignments were used with the proteobacterial homologs of CcoA that contributed to building the protein similarity network of the CalT subfamily (13). The R. capsulatus CcoA sequence was first aligned with its closely related CcoA homologs among the Rhodobacterales (mostly from node 1 [see reference 13]) (see Fig. S1). This comparison included R. sphaeroides CcoA known to provide Cu to cbb 3 -Cox (16). The R. capsulatus CcoA sequence was also aligned with its more distant homologs among the different proteobacterial orders, including Rhizobiales, Burkhorderiales, Pseudomonales, Rhodospirales, Vibrionales, Oceanospiralles, , and the Cys (C 49 , C 109 , C 225 , C 247 , and C 367 ) (right) residues of CcoA examined in this study. The Met, His, and Cys residues are colored in red, green, and yellow, respectively.
CcoA Thiol:Disulfide and Cu Import ® Synecchocales, Alteromonadales, and Chromatiales (see Fig. S2). This group included Ochrobactrum anthropi (Rhizobiales) CalT (CcoA ortholog) shown to transport Cu (13) and possibly required for the maturation of cuproproteins distinct from cbb 3 -Cox. Based on sequence alignments and topological locations (i.e., TMH or loop; Table 1 (H 249 and H 274 ), and five Cys (C 49 , C 109 , C 225 , C 249 and C 367 ) residues were retained for this study (Fig. 1, middle and right panels). They were substituted with Ala using L-ara-inducible ccoA, and the mutants obtained were introduced into appropriate E. coli and R. capsulatus strains (see Materials and Methods; Table S1). Their cbb 3 -Cox and 64 Cu uptake activities were determined analogously to the conserved motifs mutants (17).
Properties of CcoA Met and His mutants.  (17). The His 249 and H 274 are near the periplasmic and cytoplasmic (p and n) sides of the membrane, respectively, and only the latter residue is conserved in Rhodobacteriales (see Fig. S1). The corresponding mutants had cbb 3 -Cox activities slightly lower than the wild type (ca. 87 and 71%, respectively) ( Table 1). Thus, unlike M 233 and H 261 , none of these Met and His residues were required for CcoA function, except M 73 located close to the cytoplasm, as its substitution significantly reduces cbb 3 -Cox activity.
Properties of CcoA Cys mutants. Of the five Cys residues of CcoA, four (except C 109 ) are well-conserved especially among the Rhodobacteriales (see Fig. S1). Based on the homology model of CcoA (CcoA YajR ) obtained using its highest homolog E. coli YajR (PDB 3WDO) as a template, the C 49 and C 367 residues are at or near the periplasmic ends of the TMH2 and TMH11, respectively, while C 247 is located on a periplasmic loop between TMH7 and TMH8 ( Fig. 1, right panel) (10,17). The nonconserved C 109 on TMH3, and the partly conserved C 225 on TMH7 are more deeply embedded into the membrane, closer to the p and n sides, respectively. Substitution of each of these Cys residues with Ala did not affect the production of mutant proteins either in E. coli or R. capsulatus ( Fig. 2; only C 49 , C 109 , and C 247 are shown). However, it impaired the cbb 3 -Cox activity of the mutant strains to different extents (Table 1). While the effects of C 109 A, C 225 A, and C 367 A mutations were milder (ca. 75, 70, and ;69% of wild-type activity, respectively), those of the C 49 A and C 247 A were more severe (ca. 37 and ;6%, respectively). In respect to Cu import, upon L-arabinose (L-ara) addition radioactive 64 Cu accumulation in whole cells of a control strain lacking CcoA but harboring a plasmid-borne inducible CcoA (DccoA1plasmid-borne ccoA) steadily increased, unlike the DccoA strain lacking CcoA. With the C 49 A or C 247 A mutants, 64 Cu accumulation was very low, similar to the strain lacking CcoA, indicating that these residues are critical for CcoA function (Fig. 3C). On the other hand, the C 109 A and C 367 A mutants accumulated 64 Cu markedly more slowly to a slightly lower level than the control cells, indicating that C 109 and C 367 also contribute to Cu uptake but are not essential. Overall, the data indicated that these mutants fall into two groups (C 49 A and C 247 A versus C 109 A and C 367 A) with distinct kinetics behaviors, suggesting likely different functions.
Topological locations of periplasm-facing Cys residues of CcoA. Currently, no 3D structure of CcoA is available beyond the CcoA YajR homology model (Global Model Quality Estimate [GMQE] 0.51) based on its most pronounced homolog, which is the E. coli YajR (PDB 3WDO) (17). Fortunately, additional homology models of CcoA of similar GMQE yielding similar outcomes can be generated using available X-ray structures, including the iron exporter BpFPN (21). Here, we opted for two of its importer homologs, LacY (PDB 6C9W, CcoA LacY , GMQE 0.46) and GlpT (PDB 1PW4, CcoA GlpT , GMQE 0.44), captured in different conformations than CcoA YajR . While an outward-open conformation (i.e., ready to receive the substrate from the p side of the membrane) is seen with CcoA YajR , the CcoA LacY and CcoA GlpT models provide the occluded and the inward-open (i.e., ready to  Table S1) expressing wild-type or indicated CcoA mutant variants were probed with anti-Myc monoclonal antibodies. pBAD and CcoA (wt) correspond to membranes prepared from E. coli (A) or R. capsulatus (B) strains harboring empty pBAD (E. coli) or pRK-pBAD (R. capsulatus) expression plasmids, and their derivatives containing Myc-tagged versions of wild-type and mutant ccoA alleles, as appropriate. All native and mutant proteins were produced adequately in both backgrounds.
CcoA Thiol:Disulfide and Cu Import  64 Cu uptake kinetics observed with the CcoA M 237 A1M 265 A double mutant and C 49 A, C 109 A, C 247 A, and C 367 A single mutants using appropriate L-ara induced E. coli (LMG194) (A) and R. capsulatus DccoA* (DccoA DcopA, SE24) (B and C) cells, expressing L-ara-inducible native and mutant ccoA alleles. pBAD and DccoA refer to control strains, as appropriate. Uptake assays were performed as described in Materials and Methods. Activities measured in cells kept on ice were subtracted from those measured in cells incubated at 37°C, and assays were repeated at least three times using multiple independent cultures. CcoA (wt) refers to (DccoA1plasmid-borne ccoA) strain carrying L-ara-inducible native CcoA, where DccoA* is DccoA DcopA (SE24), used to avoid frequent CopA revertants seen with a ccoA deletion (see Table S1) (12). Error bars correspond to the standard deviations around the mean values. In each case, at least three biological and three technical repeats were performed. release the substrate to the n side of the membrane) conformations, respectively (Fig. 4A). Top views of these models clearly show that the distances separating the periplasm facing Cys residues change drastically depending on the conformations (Fig. 4B; see also  Table S3, which lists all appropriate a-C-to-a-C distances). When CcoA YajR is in the outward-open conformation (Fig. 4B top), C 49 and C 109 located on the N-ter domain are very close to each other (C 49 -C 109 , 12 Å) and distant from C 247 (C 49 -C 247 , 32 Å; C 109 -C 247 , 39 Å) and C 367 (C 49 -C 367 , 22 Å; C 109 -C 367 , 32 Å) located on the C-ter domain. In the occluded conformation of CcoA LacY (Fig. 4B, middle), C 49 moves closer to C 247 (C 49 -C 247 , 23 Å) and C 367 (C 49 -C 367 , 16 Å), while C 109 shifts closer to both C 247 and C 367 (C 109 -C 247 , 36 Å; C 109 -C 367 , 28 Å). In the inward-open conformation of CcoA GlpT (Fig. 4B, bottom), C 49 and C 109 approach even closer to C 247 (C 49 -C 247 , 16 Å; C 109 -C 247 , 27 Å) and C 367 (C 49 -C 367 , 10 Å; C 109 -C 367 , 22 Å). In all conformations, the N-ter-located C 49 -C 109 pair stays within 12 to 14 Å, and the C-ter-located C 247 -C 367 pair remains within 19 to 22 Å of each other. Indeed, these distance estimations are approximations in the absence of 3D structures. Nonetheless, they depict the progressive movement of the N-ter domain C 49 toward the C-ter domain C 247 -C 367 pair during the transition from the outward-open to the inward-open conformations. This observation enticed us to inquire whether the predicted distance changes are related to the Cys residues that are exposed to the oxidizing periplasm undergoing thiol:disulfide oxidoreduction during CcoA function.
Disulfide bonds formed between the Cys residues of CcoA. The occurrence in vivo of disulfide bond(s) in CcoA was probed using E. coli cells expressing native CcoA or its Cys mutant variants and the thiol-reactive alkylating agent monomethyl-(PEG)24-maleimide (mPEG) (22). Alkylation of free Cys thiols of CcoA by mPEG is expected to increase its molecular weight (MW) by ;1.2 kDa per free thiol. In the case of disulfide bonds, alkylation occurs only after reduction by dithiothreitol (DTT), and then mPEG further increases the MW by ;2.4 kDa per reduced disulfide bond. The relative MW changes (M r ) in native and Cys mutant variants of CcoA were followed by SDS-PAGE/immunodetection ( Fig. 5 and 6).
Under our conditions, native CcoA (predicted MW of 37.4 kDa) runs as a band of ;35-kDa M r in its oxidized or reduced forms (Fig. 5, left panels), which is not uncommon for membrane proteins. In the absence of DTT, alkylating by mPEG increased native CcoA M r by ca. 3 to 4 kDa to ca. 38 to 39 kDa, suggesting that it contained three free Cys thiols (predicted M r of 38.6 kDa after three mPEG addition). Moreover, alkylating native CcoA after DTT reduction further increased its M r by another ;2-3 kDa to ca. 40 to 41 kDa, indicating that the native protein contained one disulfide bound in vivo (Fig. 5, left panels). Although detecting the ca. 1-to 2-kDa M r differences with hydrophobic membrane proteins was challenging, following TCA precipitation all CcoA Cys residues appeared accessible to alkylation, including the Cys 109 and Cys 225 , which are more buried into the lipid bilayer according to the CcoA structural models ( Fig. 1 and 4).
Similar mPEG alkylation experiments were repeated using single Cys mutant variants of CcoA (Fig. 5, right panels). Without DTT reduction, all single Cys mutant variants exhibited mPEG-induced M r shifts similar to native CcoA, and the largest shift was seen with C 367 A mutant. In all cases but C 367 A, the shifts were consistent with the likely presence of at least two free thiols, but not four as would have been expected upon elimination of any Cys residues already engaged in a disulfide bond in native CcoA. After DTT reduction, all single Cys mutants, except C 367 A, showed the additional mPEG-induced M r shifts, indicating that they still contained a disulfide bond formed among the remaining Cys residues. This observation suggested that native CcoA has more than two Cys residues that could form a disulfide bond(s). The C 367 A mutant was intensely alkylated but did not exhibit any readily detected mPEG-mediated M r shift after DTT reduction, indicating that it contained no more disulfide bonds, and suggested that this residue provides one of the thiol groups forming a stable disulfide bond in native CcoA (Fig. 5, right panel, last row). Although occasionally additional minor bands were also seen in some cases (e.g., native CcoA, Fig. 5 left panel, second row, or C 225 A, Fig. 5 right panel, third row), the data showed that the C 49 A, C 109 A, C 225 A, or C 247 A single Cys mutants behaved similarly to each other and to native CcoA, which precluded identification of partner cysteines for forming a disulfide bond.
To identify the disulfide bond forming partner(s) in native CcoA in vivo, a set of double Cys mutants were examined (Fig. 6). All CcoA double Cys variants were produced adequately in E. coli and in R. capsulatus and exhibited low cbb 3 -Cox activities like their cognate single Cys mutants. In the absence of DTT, mPEG alkylation data showed that the double mutants C 49 A-C 247 A (with C 109 , C 225 , and C 367 intact) and C 247 A-C 367 A (with C 49 , C 109 , and C 225 intact) had M r shifts similar to each other, and to native CcoA, containing free thiols (Fig. 6A and B). However, like the C 367 single mutant (Fig. 5, bottom row), these two double Cys mutants did not exhibit any additional mPEG-induced M r increase upon reduction by DTT, indicating that the remaining Cys residues did not CcoA Thiol:Disulfide and Cu Import ® form disulfide bonds. Conversely, the double mutants C 49 -C 109 (with C 225 , C 247 , and C 367 intact) and C 109 -C 367 (with C 49 , C 225 , and C 247 intact) showed no or slight mPEG-induced M r shifts in the absence of DTT (although the absence of this shift was less clear in the latter double mutant) but exhibited clearer M r shifts upon mPEG alkylation after DTT treatment ( Fig. 6C and D). Since the C 225 residue is near the n side and remote from the other periplasm-exposed Cys residues on the p side of the membrane, it is likely that in the C 49 A-C 109 A and C 109 A-C 367 A double mutants, the C 247 and C 367 and the C 49 and C 247 residues, respectively, formed disulfide bonds (although the latter pair might form a less stable disulfide bond) ( Fig. 6C and D, far right). The slight M r shifts seen with these double mutants in the absence of DTT reduction were consistent with the poor alkylation of C 225 , still intact in these mutants. The remaining C 109 A-C 247 A double mutant (with C 225 , C 49 , and C 367 intact) behaved essentially like the latter mutants, except that the CcoA population appeared heterogenous in the absence of DTT (Fig. 6E). A small fraction contained free thiols that was alkylated by mPEG without DTT treatment, whereas a large fraction contained a disulfide bond that could only be alkylated after DTT reduction. Again, assuming that C 225 is too far from the other Cys residues to participate in disulfide bond formation, a large fraction of the C 109 A-C 247 A double-mutant  population comprises a less stable disulfide bond between the C 49 and C 367 residues (Fig. 6E). In summary, the overall data indicated the formation of disulfide bonds between C 49 ;C 247 , C 49 ;C 367 , and C 247 ;C 367 (with the last one forming the most stable bond) and the clear absence of disulfide bonds between C 109 ;C 367 and C 49 ;C 109 . This suggests that Cys 109 is redox inactive, unlike the other periplasm-exposed residues, possibly due to its membrane-buried location in all conformations of CcoA (Fig. 4). Consequently, in cells producing native CcoA, any pair among the C 49 , C 247 , and C 367 residues could form a single disulfide bond in vivo, leaving behind three free thiol groups, including C 109 and C 225 . This finding raised the possibility that native CcoA in vivo might exist as a heterogenous population with different conformations, presumably due to the import of spurious Cu presumably present in the growth medium. How the initial binding of Cu changes the conformation of CcoA and shuffles the free thiols and disulfide bonds between its three active Cys residues remains to be determined in future studies.

DISCUSSION
This study focused on the role of plausible metal-liganding residues Met, His and Cys of CcoA, a member of the CalT (CcoA-like Transporters) subfamily of MFS-type transporters (13) and the prototype of proteobacterial Cu importers (10,15). The CalT subfamily is characterized by two well-conserved motifs (M 233 xxxM 237 and H 261 xxxM 265 in R. capsulatus CcoA) of which the first Met and His residues are absolutely required for Cu import (17). Here, we show that mutating concomitantly the M 237 and M 265 residues also abolishes CcoA activity, unlike the corresponding single mutants. Thus, the presence of at least one additional Met residue together with Met 233 and His 261 is required for Cu import. This finding further supports the Cu binding role of the conserved motifs that are the hallmark of the CalT subfamily of MFS-type transporters (13,17).
We examined the distribution and topological location of additional possible metalliganding residues of CcoA that are often conserved among its homologs, in particular those from the Rhodobacterales within the Proteobacteria. Of these residues, mutating C 225 , M 227 , or H 274 located at the TM7 and TM8 on the C-ter domain of CcoA near the membrane Cu-binding site, had little effect on CcoA activity. This finding was similar to that seen with the M 237 A or M 265 A single mutants, suggesting that they were either not critical for function, or partly substituted by surrogate residues. Intriguingly, mutating M 73 , but not M 69 , of the putative "Met" motif (M 69 xxxM 73 in R. capsulatus) (20) had a stronger effect on cbb 3 -Cox activity. Homology models of the different conformations of CcoA do not seem to suggest that these N-ter residues come very close to the C-ter Cu binding residues. However, how Cu is released from CcoA is not yet known, leaving the possibility open that the C 225 , M 227 , and H 274 residues or the putative Met (M 69 xxxM 73 ) motif, or both, all positioned closer to the n side of the membrane, might play a role in this process.
Remarkably, mutating the periplasm-exposed C 49 , C 247 , and C 367 residues affected CcoA activity to different degrees. These residues are well-conserved among the Rhodobacterales, but either less (;50% for C 247 ) or not (0% for C 49 and C 367 ) conserved in other proteobacterial orders (see Fig. S1 and S2). The basis of this conservation is not obvious, but it might relate to the ultimate destination of Cu (e.g., cbb 3 -Cox in R. capsulatus and other cuproenzymes in O. anthropi) and the different Cu donors and acceptors of CcoA and its homologs. Of the periplasm-facing Cys (C 49 , C 109 , C 247 , and C 367 ) residues of CcoA, mutating C 109 slowed Cu uptake (Fig. 3C), slightly affected cbb 3 -Cox activity (Table 1), and mPEG alkylation indicated that C 109 does not form a disulfide bond with either C 367 or C 49 . Intriguingly, C 109 residue is not conserved among the Rhodobacterales (0%) but is better conserved (;70%) among the other orders of proteobacteria where CalT is thought to provide Cu to other cuproproteins distinct from the cbb 3 -Cox (13).
Alkylation data of the single and double Cys mutants revealed that in native CcoA, two of the three periplasm-facing C 49 , C 247 , and C 367 residues form a disulfide bond, while the remaining two remain as free thiol in vivo. Moreover, all possible disulfide and free thiol combinations among these residues (i.e., C 247 ;C 367 leaving C 49 free, C 49 ;C 247 leaving C 367 free, and C 49 ;C 367 leaving C 247 free) were observed in appropriate Cys double mutants. However, the levels of stability of these bonds seem to be different, with the C 247 ;C 367 bond being most stable. Although the data in Fig. 6 tend to suggest that the C 49 ;C 247 bond might be formed, yet the data with the single C 367 A mutant (Fig. 5) suggest that it certainly must not be stable to be readily detected in this mutant. Assuming that CcoA undergoes conformational changes like any MFS-type transporter, the disulfide bond formation patterns suggest a hypothetical model linking Cu binding and conformational changes (Fig. 7). Accordingly, in the outward-open conformation of CcoA (state 1), C 247 and C 367 would contain a disulfide bond, far away from C 49 . Binding of Cu would convert CcoA into its occluded conformation (state 2), bringing C 49 near the C 247 ;C 367 disulfide bond, and a nucleophilic attack would yield either C 49 ;C 367 ( Fig. 7, left) or C 49 ;C 247 ( Fig. 7, right) disulfide bond while freeing the remaining thiol of C 247 or C 367 . We note that if no such disulfide bond is formed or is extremely unstable, then the occluded conformation (state 2) may not have a disulfide bond (not shown in Fig. 7). In the exponentially growing cells used in this study, Cu import is not synchronized; thus, different conformations of CcoA must coexist, rendering impossible to discriminate between these possibilities at this stage. The more defective phenotype and the periplasmic location (i.e., increased solvent exposure) of C 247 as well as the weaker nature of C 49 ;C 247 (as suggested by its absence in C 367 A single mutant) and the detection of C 49 ;C 367 (as seen with C 109 A-C 247 A double mutant) disulfide bonds might argue that the C 49 -C 367 disulfide bond may be more favorable at the inward open conformation (state 3) (Fig. 7, right). In any case, further progression of Cu within CcoA from the periplasm toward the cytoplasm would trigger the remaining free thiol (C 247 or C 367 ) to attack the disulfide bond involving C 49 (C 49 ;C 247 or C 49 ;C 367 ) at the inward open conformation (state 3). The subsequent resolution of this bond would then reestablish the initial C 247 ;C 367 disulfide bond and free C 49 thiol, returning CcoA to its outward-open conformation (state 1). This model attributing more critical roles to C 49 and C 247 is also consistent with the highly defective 64 Cu uptake seen with the C 49 A and C 247 A single mutants (Fig. 3C). Conceivably, the three periplasm-facing C 49 , C 247 , and C 367 residues that are highly conserved in Rhodobacterales may also play additional and perhaps different roles (e.g., liganding Cu) instead of those ascribed here. However, this hypothetical model suggests a link between the binding of Cu, ensuing conformation changes, and medium supplemented with appropriate antibiotics (Amp, 100 mg/ml; Tet, 10 mg/ml) with shaking at 180 rpm. The next day,100 to 200 ml of these cultures were subcultured into 10 ml of fresh LB medium containing 1% L-ara and appropriate antibiotics at 37°C with shaking (180 rpm) until they reached an optical density at 600 nm (OD 600 ) of 0.5. At this stage, two aliquots of 0.9 ml each were taken out and kept on ice, while the remaining culture (8.2 ml) was reduced by addition of 82 ml of 1.0 M DTT (10 mM final concentration) and further incubated for 10 min at 37°C with shaking. Two additional aliquots of 0.9 ml each were taken and placed on ice. All four samples were precipitated by addition of 100 ml of 100% ice-cold TCA (final concentration, 10% [vol/vol]) and incubated on ice for 30 min. Precipitated materials were centrifuged at 13,000 rpm at 4°C for 12 min, and supernatants were removed without disturbing the pellets, which were washed with 300 ml of ice-cold acetone to eliminate TCA, and recentrifuged under the same conditions. The pellets were partially dried at 30°C for ;10 min to evaporate acetone, one untreated pellet and one DTT-treated pellet were resuspended in 30 ml of PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2.0 mM KH 2 PO 4 [pH 7.0]) supplemented with 0.1% SDS. Similarly, the remaining one untreated and one DTT-treated pellets were resuspended in 30 ml of mPEG-MAL solution (20 mM mPEG-MAL dissolved in PBS buffer) supplemented with 0.1% SDS. The pellets were vortexed vigorously for 1 min for complete dissolution, followed by incubation in the dark at room temperature for 2 h under constant shaking (1,000 rpm) to label the accessible thiol groups of CcoA. At the end of the incubation, 10 ml of 5Â Laemmli buffer (10% SDS [vol/vol], 0.05% bromophenol blue [wt/vol], 60% glycerol [vol/vol], 300 mM Tris-HCl [pH 6.8]) was added to each sample, followed by further incubation at room temperature for 15 min. Then, 20 ml of each sample was loaded on a 12% nonreducing SDS-PAGE gel, run at 200 V, and subjected to immunoblot analyses using a-myc monoclonal antibodies (1:5,000 dilution) as primary antibodies and horseradish peroxidase conjugated anti-mouse IgGs as secondary antibodies (1:3,000 dilution). The addition of mPEG-MAL, specific to free thiol groups, increases the M r of alkylated mPEG-MAL derivatives of CcoA, with the increases being proportional to the number of free thiol groups. Comparison of untreated and DTT-treated samples prior to mPEG-MAL alkylation visualized the DTT-reduced disulfide bonds of CcoA in vivo under the growth conditions used. cbb 3 -Cox activity. The in situ cbb 3 -Cox activity of R. capsulatus colonies was assessed qualitatively using the "NADI" staining solution, which is made by mixing in a 1:1 (vol/vol) ratio 35 mM a-naphthol and 30 mM N,N,N9,N9-dimethyl-p-phenylene diamine (DMPD) dissolved in ethanol and water, respectively (33). Colonies producing cbb 3 -Cox stain blue, while those lacking it remain unstained. The in vitro cbb 3 -Cox activity was measured quantitatively using R. capsulatus chromatophore membranes and TMPD by monitoring spectrophotometrically in a stirred cuvette its oxidized form at 562 nm (l 562 = 11.7) at room temperature. Briefly, 10mg of R. capsulatus chromatophore membranes was added to 1 ml of assay buffer (25 mM Tris-HCl [pH 7.0], 150 mM NaCl), and the enzymatic reaction was initiated by addition of TMPD at a final concentration of 1 mM. The TMPD oxidation specifically due to cbb 3 -Cox activity was controlled by incubating the chromatophore membranes with 200 mM KCN for 2 min prior to TMPD addition. The cbb 3 -Cox activity was calculated by subtracting from the TMPD oxidase activity the fraction that was KCN insensitive (15).
Radioactive 64 Cu uptake assays. Cellular Cu uptake assays were performed as previously described (12), using whole cells and radioactive 64 Cu (1.84 Â 10 4 mCi/mmol specific activity) obtained from Mallinckrodt Institute of Radiology, Washington University Medical School. The. E. coli strains harboring appropriate pBAD/Myc-His derivatives with L-ara-inducible ccoA wild-type and mutant variants (see Table S1) were grown overnight in 10 ml of LB medium supplemented with 0.5% L-ara and appropriate antibiotics. Cells were pelleted, washed with the assay buffer (50 mM sodium citrate [pH 6.5], 5% glucose), and resuspended in 1 ml of the same buffer. All cultures were normalized to the same number of cells (7.5 Â 10 8 cells/500 ml of assay volume) based on their absorbance (1 OD 600 = 5 Â 10 8 cells/ml for E. coli and 1 OD 630 = 7.5 Â 10 8 cells/ml for R. capsulatus strains). Cells to be assayed for 64 Cu uptake were preincubated at 35 or 0°C for 10 min before the assay. The uptake activity was initiated by addition of 10 7 cpm of 64 Cu, determined immediately before use (half-life of 64 Cu isotope ;12 h), and 50-ml aliquots were taken at 0, 1, 2, 5, and 10 min of incubation and immediately mixed with ice-cold 50 ml of 1 mM CuCl 2 and 50 ml of 50 mM EDTA (pH 6.5) to stop the uptake reaction. All aliquots were kept on ice until the end of the assay; the cells were then pelleted, and the pellets were washed twice with 100 ml of ice-cold 50 mM EDTA solution, resuspended in 1 ml of scintillation liquid, and counted using a scintillation counter (Coulter-Beckman, Inc.) with a wide-open window. For each time point, the background 64 Cu uptake activity seen at 0°C was subtracted from that at 35°C and plotted as a function of time to compare CcoA-specific Cu uptake of wild-type control (DccoA/plasmid-born ccoA) and mutant derivatives of CcoA.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.