Structure and Metal Loading of a Soluble Periplasm Cuproprotein*

A copper-trafficking pathway was found to enable Cu2+ occupancy of a soluble periplasm protein, CucA, even when competing Zn2+ is abundant in the periplasm. Here, we solved the structure of CucA (a new cupin) and found that binding of Cu2+, but not Zn2+, quenches the fluorescence of Trp165, which is adjacent to the metal site. Using this fluorescence probe, we established that CucA becomes partly occupied by Zn2+ following exposure to equimolar Zn2+ and Cu2+. Cu2+-CucA is more thermodynamically stable than Zn2+-CucA but k(Zn→Cu)exchange is slow, raising questions about how the periplasm contains solely the Cu2+ form. We discovered that a copper-trafficking pathway involving two copper transporters (CtaA and PacS) and a metallochaperone (Atx1) is obligatory for Cu2+-CucA to accumulate in the periplasm. There was negligible CucA protein in the periplasm of ΔctaA cells, but the abundance of cucA transcripts was unaltered. Crucially, ΔctaA cells overaccumulate low Mr copper complexes in the periplasm, and purified apoCucA can readily acquire Cu2+ from ΔctaA periplasm extracts, but in vivo apoCucA fails to come into contact with these periplasmic copper pools. Instead, copper traffics via a cytoplasmic pathway that is coupled to CucA translocation to the periplasm.

Metals diffuse through porins in the outer membrane of Gram-negative bacteria such that the periplasm is exposed to metals from the external medium (1,2). This raises questions about how the correct metals are acquired by proteins such as CucA, which are exported in an unfolded state by the general secretory pathway and obtain metal after membrane translocation. We previously identified the most abundant Cu 2ϩ and Mn 2ϩ proteins, CucA and MncA, respectively, in the periplasm of the cyanobacterium Synechocystis PCC 6803 (3). MncA has Mn 2ϩ -dependent oxalate decarboxylase activity, and CucA shows some Cu 2ϩ -dependent quercetin dioxygenase activity. Folding of MncA in an extract of the soluble compo-nents of the periplasm aberrantly generates Zn 2ϩ -MncA, but MncA is a Tat substrate, which folds prior to export (4 -7), and so MncA can acquire Mn 2ϩ in the cytoplasm but not the periplasm. Electron paramagnetic resonance spectra of Cu 2ϩ -MncA and Cu 2ϩ -CucA are closely similar, consistent with similarly determined and deduced metal-binding sites, respectively (3). Because CucA is exported by the general secretory pathway, if its metal-binding site is identical to MncA, then CucA might be predicted to erroneously acquire Zn 2ϩ upon folding in the periplasm. Here, we confirm the prediction that CucA will readily bind Zn 2ϩ and would be liable to aberrantly acquire Zn 2ϩ rather than Cu 2ϩ from the exchangeable metal pools in the periplasm.
It has been proposed that copper was recruited during evolution after the appearance of atmospheric dioxygen and then exploited in the development of multicellular eukaryotes (8). It is unclear whether or not copper is needed in the cytosol of most bacteria. Moreover, few if any copper importers have been identified in bacteria. In vivo evidence that some P-type ATPases import copper (9 -14) has not yet been substantiated with biochemical data in vitro. In eukaryotes, copper proteins are extracellular or located within cytoplasmic compartments such as the trans-Golgi, vesicles derived from this network, mitochondria, or chloroplasts (15)(16)(17)(18)(19)(20). Copperzinc superoxide dismutase is a rare example of a copper protein found within the eukaryotic cytosol, and it has a copper-delivery protein CCS (21,22). It is hypothesized that trafficking copper to its destinations by ligand exchange via copper metallochaperones (23,24), coupled with compartmentalization, minimizes deleterious side reactions in the cytosol (3). A zeptomolar copper affinity of the copper sensor CueR implies that copper is either excluded or tightly bound and buffered within the Escherichia coli cytoplasm (25). In bacteria, copper-zinc superoxide dismutase is always extracellular or located within the periplasm, whereas other bacterial copper proteins are membranous, periplasmic, or extracellular. One known cytoplasmic bacterial copper enzyme is particulate methane monooxygenase (26); however, particulate methane monooxygenase is housed in unusual internal membranes (27). Bacteria appear to have avoided the challenges associated with handling copper in the cytoplasm by excluding copper proteins. However, here we discover that the periplasmic copper protein CucA does not obtain its metal from periplasmic pools but requires that copper be first trafficked through the cytoplasm.
The first crystal structures of CucA now confirm a monocupin fold and the coordination of Cu 2ϩ via a His 3 Glu 1 ligand set, analogous to the Mn 2ϩ -binding residues of MncA. Trp 165 is proximal to the metal-binding site, and its fluorescence is quenched by Cu 2ϩ but not by Zn 2ϩ , allowing metal binding to be followed spectrally. In vitro metallation studies reveal ϳ50% initial occupancy with Zn 2ϩ after exposure of apoCucA to an equimolar mixture of Cu 2ϩ and Zn 2ϩ , followed by a slow (Ͼ24 h to equilibrium) exchange to 100% Cu 2ϩ -CucA. CucA was previously identified via the fractionation of periplasm extracts using two-dimensional native liquid chromatography followed by metal analysis using inductively coupled plasma mass spectrometry (ICP-MS) 3 (3). By this approach, it is possible to quantify the number of atoms of copper bound to CucA per cell. Here, the same method has been used to quantify the number of atoms of copper associated with CucA in mutants deficient in genes known to encode proteins of copper homeostasis (9,12,28,29). Accumulation of Cu 2ϩ -CucA in vivo requires the same three proteins PacS, CtaA, and Atx1, which supply copper to plastocyanin within internal thylakoid compartments. Low M r Cu 2ϩ pools increase in abundance in periplasm extracts of ⌬pacS, ⌬ctaA, and ⌬atx1 mutants, in which Cu 2ϩ -CucA fails to accumulate, showing that CucA does not directly acquire Cu 2ϩ from these periplasmic pools. The abundance of CucA protein, but not transcripts, declines in a copper-trafficking mutant. Thus, the copper pathway does not act on expression of the cucA gene but assists the metal supply to CucA itself, and moreover, the copper supply is linked to CucA accumulation in the periplasm.

EXPERIMENTAL PROCEDURES
General Reagents and Bacterial Cultures-Synechocystis PCC 6803 was cultured at 28°C under constant light in standard BG-11 medium. Mutant strains ⌬ctaA (9), ⌬pacS (9), and ⌬atx1 (28) were described previously and were maintained in the presence of kanamycin (50 g ml Ϫ1 ) or chloramphenicol (7.5 g ml Ϫ1 ), but antibiotics were omitted from the final experimental cultures to prevent metal chelation effects.
CucA Production, Purification, and in Vitro Metal Binding Studies-Cloning of the cucA gene for heterologous expression was described previously (3). Site-directed mutagenesis of Trp 165 to Phe was achieved using the QuikChange method (Stratagene) with the primers 5Ј-CGGATAAAACATTACC-CATTGTTTTTGTTTTCATGCGTAATGAAGTTGCAC-CGG-3Ј and 5Ј-CCGGTGCAACTTCATTACGCATGAAA-ACAAAAACAATGGGTAATGTTTTATCCG-3Ј. Resulting colonies were screened for the loss of a BtsCI restriction site, and a successful clone was verified by sequencing. The expression, refolding, and purification of recombinant CucA protein was performed essentially as described previously (3), except that the final size exclusion chromatography step utilized a HiLoad 26/60 Superdex 200 column (GE Healthcare). The anion exchange-concentrated refolded protein, in 50 mM Tris, pH 7.5, 500 mM NaCl, 10 mM EDTA, was resolved on the size exclusion column in 10 mM HEPES, pH 7.0, 150 mM NaCl at 0.5 ml min Ϫ1 at 4°C, collecting 5-ml fractions. CucA-containing fractions were assayed for protein by Coomassie assay (Pierce), and CucA was quantified by UV-visible spectroscopy (Cary 4E) using the calculated ⑀ 280 nm ϭ 32,890 M Ϫ1 cm Ϫ1 for wild type and 27,930 M Ϫ1 cm Ϫ1 for the W195F mutant. Protein was verified to be metal-free (designated as containing Ͻ0.05 mol eq total) by ICP-MS.
Fluorescence spectra were collected using a Cary Eclipse spectrofluorimeter at 20°C. The excitation wavelength was 280 nm throughout with a 5-nm slit width, and emission was monitored between 300 and 450 nm with a 5-nm slit width, a scan rate of 120 nm min Ϫ1 , 1-nm data interval, and 0.5-s averaging time. All samples contained ϳ5 M CucA in 10 mM HEPES, pH 7.0, 150 mM NaCl in quartz cuvettes (Hellma). All samples were thermally equilibrated for 10 min prior to the experiment. For titrations, metals were added by pipette from concentrated aqueous stock solutions of CuSO 4 , ZnSO 4 , or MnCl 2 , mixed vigorously, and then incubated at 20°C for 60 s before the spectra were collected. All metal stocks were calibrated by ICP-MS before use. For kinetics experiments, metal was added directly to the cuvette within the spectrometer and rapidly mixed by pipetting. Emission at 320 nm was measured at 1 Hz for the duration of the experiment.
Determination of CucA Crystal Structures-Purified recombinant CucA (apo or metallated) was concentrated to ϳ10 mg ml Ϫ1 and buffer-exchanged into 10 mM HEPES, pH 7.0, 50 mM NaCl using a YM-10 centrifugal concentrator (Centricon, Amicon). Initial crystallization trials were set up as sitting drops using a Mosquito robot (TTP Labtech) and commercially available precipitant screens. Conditions yielding crystals were scaled up and optimized in standard 24-well plates using the hanging drop method of vapor diffusion. All crystallization experiments were maintained at 20°C. Diffraction quality crystals were produced from multiple conditions containing 0.1 M either HEPES or Tris, pH 7.5-8.8, 18 -24% (w/v) PEG 8000, at a protein to precipitant solution ratio of 2 l of protein solution to 1 l of precipitant solution. For metallated protein, 0.5 mM CuSO 4 or 0.5 mM ZnSO 4 was included in the precipitant solution or 1 mM EDTA for apoprotein. Prior to data collection, crystals were flash-frozen in liquid nitrogen using the precipitant solution supplemented with 25% glycerol as a cryoprotectant. Data collection statistics and beamlines are listed in supplemental Table 1.
Single wavelength anomalous dispersion was used to solve the crystal structure of the zinc-containing form of SeMetsubstituted recombinant CucA to 2.2 Å; crystals of Cu 2ϩ -SeMet-CucA were obtained but were much smaller and did not diffract as far. Although sufficient anomalous data were collected from the copper form to permit structure solution (data not presented), the zinc structure was used as a molecular replacement model for the other CucA structures, including Cu 2ϩ -SeMet-CucA. The structure of Zn 2ϩ -CucA-SeMet was solved using the SHELX C/D/E programs as part of CCP4i. The resulting phases were used as starting phases in RESOLVE, which built the majority of the protein. The model building was completed manually in COOT with refinement carried out using Refmac5. Molecular replacement was subsequently carried out using MOLREP. Models were validated using the MOLPROBITY webserver. Structures were aligned for comparisons using LSQMAN or THESEUS. The majority of structures were obtained in the closed form; however, particularly the apoprotein or Cu 2ϩ -CucA crystallized in the absence of excess metal crystallized readily in the open form. No correlation was observed between crystal forms and the pH of the precipitant solution. The presence of copper or zinc was confirmed by calculation of anomalous difference maps from data collected at 1.378 or 1.282 Å, respectively, and by the collection of energy scans across the copper/zinc region at ESRF beamline ID29-1.
Analysis of Copper Complexes in Periplasmic Extracts-Isolation of periplasmic extracts from Synechocystis PCC 6803, two-dimensional liquid chromatography, ICP-MS metal analysis, SDS-PAGE, and MALDI-TOF-MS peptide mass fingerprinting used methods described previously (3), except that HEPES buffer was used throughout in place of Tris buffer used previously.
Isolation of RNA and Reverse Transcriptase-PCR-Total RNA was isolated from logarithmically growing wild-type, ⌬ctaA, ⌬pacS, and ⌬atx1 cells using an established method (30). RNA was treated with DNase I (Sigma), and then 1 g of DNase-treated RNA was used for the production of cDNA using an Im-PromII reverse transcription kit (Promega). Reverse transcriptase was omitted from reaction mixtures that were used as negative controls in the RT-PCR analysis. For the PCR, specific primers for petE (forward 5Ј-TGCCG-CTGCCAATGCAACAG-3Ј and reverse 5Ј-GTTCACAGTA-GTAGGTGTAG-3Ј), petJ (forward 5Ј-CTGTGCCGCTTGT-CATAATG-3Ј and reverse 5Ј-AGCACGTAGGCCGCCAC-ATC-3Ј), cucA (forward 5Ј-TATTTTCAAGCTGTGGGGC-3Ј and reverse 5Ј-CGGTTAAAAGCTTCAATTAC-3Ј), and rps1 (forward 5Ј-CTCTGATTGACATTGGGGCG-3Ј and reverse 5Ј-GAGCGCTGATGTGGGAGCCG-3Ј) were used, which were designed to amplify an ϳ300-bp region internal to each gene. Cycling conditions, using Taq polymerase (New England Biolabs), included an initial denaturation step at 95°C for 2 min and then 25-30 cycles of denaturation at 95°C for 1 min, annealing at 60°C for 1 min and extension at 72°C for 20 s, followed by a final extension step at 72°C for 5 min. Products were analyzed on 1% (w/v) agarose gels stained with ethidium bromide.
Metal Association with CucA Using Periplasm Extracts-Periplasm extracts were prepared from 1 to 2 liters of wild-type or ⌬ctaA Synechocystis PCC 6803 culture, into a final volume of 100 ml of water, as described previously (3). This extract was divided equally into two portions, to which was added 1 ml of 10 mM HEPES, pH 7.0, 150 mM NaCl buffer (denoted "control") or 1 ml of 1.5 M CucA in the same buffer (denoted "CucA"). The samples were mixed and incubated for 1 h at 4°C, and then 1 M HEPES, pH 8.8, was added to each sample to a final concentration of 20 mM. Each sample was independently loaded onto a 1-ml HiTrap Q HP anion exchange column (GE Healthcare), washed in 20 ml of 20 mM HEPES, pH 8.8, and then eluted in 1 ml of 20 mM HEPES, pH 8.8, 400 mM NaCl. Aliquots (200 l) of each eluant fraction were resolved by HPLC size exclusion chromatography (TSK SW3000, Tosoh Biosciences) in 5 mM HEPES, pH 7.0, 50 mM NaCl at 0.5 ml min Ϫ1 collecting 0.5-ml fractions. Each fraction was analyzed for metal by ICP-MS.

Visualization of the CucA Metal Site Adjacent to Trp 165 -
The structure of CucA was solved, initially for the purpose of visualizing its metal site to identify the metal ligands and to establish whether or not it is a cupin, as predicted through bioinformatics. Diffraction quality crystals of Cu 2ϩ -SeMet-CucA (four molecules per asymmetric unit (ASU)), Cu 2ϩ -CucA (four and one molecule per ASU), Zn 2ϩ -SeMetCucA (two molecules per ASU), and apoSeMetCucA (four molecules per ASU) were analyzed (supplemental Table 1). The Sec-dependent signal sequence of CucA is missing from the recombinant protein, and in addition, four amino-terminal residues were not detected in the electron density maps so that the structures commence at residue 35. CucA is confirmed to exhibit a monocupin fold with the metal-binding site positioned within a relatively hydrophobic cavity (Fig. 1a). In metalbound forms, Cu 2ϩ or Zn 2ϩ , the metal ion is coordinated to the same four residues, three histidines, His 88,90,149 , plus Glu 95 (Fig.  1b), identical to the protein metal-binding ligands of Mn 2ϩ -MncA (3). Bonds to Cu 2ϩ and Zn 2ϩ are ϳ2.1 Å (respective metal identities confirmed at edge wavelengths of 1.378 and 1.282 Å) (Fig. 1, c and d). Trp 165 is proximal to the metal at 4.4 Å (Fig. 1b), and therefore, its fluorescence is liable to be quenched through Förster dipole coupling of the Trp excited state with metal-derived charge transfer transitions of suitable energies (31).
The metal binding geometry of both Cu 2ϩ -CucA and Zn 2ϩ -CucA to protein ligands approximates to distorted tetrahedral arrangement, but one or two water molecules provide additional fifth and possibly sixth ligands at variable distances. One water molecule is at a comparable distance to the other ligands (1.9 -2.4 Å), and the second tends to be longer, perhaps not coordinating at all (2.4 -3.2 Å), generating a distorted square pyramidal arrangement. Cu(II) has a preference for square planar or square pyramidal Zn(II) for tetrahedral geometry, and hence CucA imposes a geometry that may influence its metal-binding preferences. The surrounding protein structure is comparable between the two metal forms (Cu 2ϩ versus Zn 2ϩ ). More than one crystal conformer of CucA was observed (supplemental Fig. 1), with some analogy to the open and closed forms of oxalate decarboxylase (32,33). Both open and closed conformations of Cu 2ϩ -CucA have been detected. The open forms contain a channel to the metal-binding site that is expected to allow substrate entry and product release. Although the endogenous substrates of Cu 2ϩ -CucA are unknown, the enzyme exhibits some quercetin dioxygenase activity that is absent in Zn 2ϩ -CucA (3). However, the channel to the CucA metal site and the presumed substrate cavity is small compared with known quercetin dioxygenases (34), suggesting that the in vivo substrate is likely to be smaller than quercetin. The entry pocket is lined by Glu 58 , Phe 60 , Pro 85 , and Phe 209 with a single string of water molecules connecting the metal site to the surface (supplemental Fig. 2). In the alternative form of CucA, the channel is occluded by the movement of residues 52-60, resulting in Glu 58 and Ala 56 and the loop 1 region projecting into the channel (supplemental Fig. 3). The B-factors are higher in open CucA, notably in the regions that close the channel in the alternative form (supplemental Fig. 4), and several residues (203-207 inclusive) from helix ␣2 and loop 11 could not be modeled in this open form.
Binding of Cu 2ϩ and Zn 2ϩ to CucA in Vitro-In vitro metalbinding experiments previously suggested that MncA might prefer Zn 2ϩ relative to Cu 2ϩ , with approximately 1 order of magnitude difference in stability constants (3). Because the CucA ligands are identical to those of MncA (Fig. 1b) its preference for Cu 2ϩ relative to Zn 2ϩ was therefore tested (Fig. 2). Intrinsic fluorescence of CucA is quenched by 1 eq of Cu 2ϩ (Fig.  2a) but not Zn 2ϩ (Fig. 2b), although CucA does bind 1 eq of Zn 2ϩ (Fig. 2e). Cu 2ϩ -dependent quenching was lost when Trp 165 was substituted with Phe, and this CucA variant had a low level of fluorescence similar to that of wild-type CucA saturated with Cu 2ϩ (Fig. 2c). These data support Trp 165 as the source of fluorescence that is quenched by Cu 2ϩ . Unlike Zn 2ϩ , which is spectrally silent due to its d 10 electron configuration, charge transfer transitions are common with Cu 2ϩ . Quenching of fluorescence by Cu 2ϩ , but not Zn 2ϩ , is thus consistent with a Förster dipole coupling mechanism.
Metallation of apoCucA with equimolar Zn 2ϩ and Cu 2ϩ initially causes only partial quenching of fluorescence (Fig. 2c). The initial generation of a mixture of Zn 2ϩ -CucA and Cu 2ϩ -CucA was confirmed by analyzing bound metal by ICP-MS after fractionation by size exclusion chromatography (Fig. 2f). FIGURE 1. Crystal structure of Cu 2؉ -CucA. a, structural elements (␣, ␣-helices; ␤, ␤-strands; l, loops), colored in order from the amino (blue) to carboxyl (red) terminus, are numerically labeled. b, residues (green) surrounding the copper atom (blue sphere) of Cu 2ϩ -CucA are shown in the closed conformation. c, residues surrounding the metal atoms within CucA with corresponding composite omit electron density maps each contoured at 1.5 for Cu 2ϩ -CucA, and d, Zn 2ϩ -CucA. FIGURE 2. Copper and zinc both bind to CucA. a, quenching of the fluorescence emission spectra of CucA (5 M in 10 mM HEPES, pH 7.0, 150 mM NaCl) exposed to increasing concentrations of Cu 2ϩ , 280 nm excitation. b, lack of quenching of the fluorescence emission spectra of CucA, as in a, exposed to increasing concentrations of Zn 2ϩ . c, fluorescence emission of CucA at 320 nm as a function of added Zn 2ϩ (diamonds), Cu 2ϩ (triangles), and simultaneous addition of equimolar Zn 2ϩ plus Cu 2ϩ to 5 M apoCucA (circles). Fluorescence was measured 60 s after metal addition and mixing. Fluorescence emission of a CucA W165F variant was also determined (closed squares). Prolonged incubation for 24 h leads to further quenching approaching that observed following the addition of Cu 2ϩ alone (Fig. 3a). This implies slow Zn 2ϩ to Cu 2ϩ exchange, with Cu 2ϩ ultimately forming the more thermodynamically stable complex at equilibrium. The rate of replacement of Zn 2ϩ by Cu 2ϩ is concentration-dependent (Fig. 3b). EDTA (1 mM) out-competes apoCucA (5 M) for Cu 2ϩ (Fig. 4a), and a dissociation rate for Cu 2ϩ -CucA of k Cu,OFF ϭ 7.5 ϫ 10 Ϫ7 s Ϫ1 was thus estimated from the rate of increase in Cu 2ϩ -CucA fluorescence in the presence of a similar excess of EDTA (Fig. 4b). The initial formation of an approximately equal mixture of Zn 2ϩ -CucA and Cu 2ϩ -CucA (Fig. 2f) implies that the two metals have similar association rates, and therefore a dissociation rate for Zn 2ϩ -CucA of k Zn,OFF ϭ 4.6 ϫ 10 Ϫ5 s Ϫ1 was estimated from the rate of metal exchange in the presence of surplus Cu 2ϩ (Fig. 3b). A similar value of k Zn,OFF ϭ 5.0 ϫ 10 Ϫ5 s Ϫ1 was also estimated by detecting loss of Zn 2ϩ from CucA by ICP-MS after 48 h (data not shown).
Thus, the metal site of CucA prefers Cu 2ϩ to Zn 2ϩ , but if Zn 2ϩ binds first, the subsequent replacement with Cu 2ϩ is slow. Under these biochemical/biophysical conditions, partial occupancy with Zn 2ϩ might be predicted to occur in vivo at times when the periplasm does not contain a sufficient excess of Cu 2ϩ relative to Zn 2ϩ . In natural environments periplasm Cu 2ϩ levels are likely to be much more limiting than in nutrient-rich BG-11 culture medium that contains 0.3 M copper. Conversely, BG-11 medium contains a surplus of zinc relative to copper, yet CucA is occupied by copper.
Copper Transporters CtaA and PacS, Plus the Atx1 Metallochaperone, Are Required for Formation of Cu 2ϩ -CucA-Copper is required for plastocyanin and cytochrome oxidase in this bacterium, both of which are located within internal membrane-bound compartments called thylakoids (Fig. 5a). Mutants deficient in either of two P 1 -type ATPases, CtaA and PacS, or a soluble metallochaperone Atx1 (Fig. 5a), show phenotypes consistent with impaired copper supply to these cytoplasmic enzymes (9,28). Here, we have used native two-dimensional liquid chromatography followed by inductively coupled plasma-mass spectrometry of periplasm extracts to compare the number of atoms of Cu 2ϩ bound to CucA per cell, in the ⌬atx1, ⌬ctaA, and ⌬pacS mutants (Fig. 5, b and c). The Cu 2ϩ pool in periplasm extracts corresponding to CucA was identified previously by mass fingerprinting after using principal component analysis to correlate Cu 2ϩ with proteins that had been further resolved using denaturing PAGE (3). Here, we find  that the loss of any of the three proteins that supply copper to plastocyanin also leads to a decline in Cu 2ϩ -CucA (Fig. 5, b and c). The phenotype is weakest for mutants lacking the copper metallochaperone Atx1, implying that its role in the pathway is partly redundant, although in the absence of either of the copper ATPases, CtaA, or PacS, negligible Cu 2ϩ reaches CucA. This has some analogy to the loss of cbb 3 oxidase in Rhodobacter capsulatus on deletion of the hypothetical copper-ATPase CcoI (13). Notably, the cbb 3 oxidase is a membrane protein with an intramembrane Cu B site and hence dissimilar to the periplasm localization of CucA.
Expression of the cucA Gene Does Not Respond to Copper-Loss of Cu 2ϩ -CucA in ⌬ctaA, ⌬pacS, or ⌬atx1 mutant cells could either result from altered expression of the cucA gene due to loss of copper supply to a metal-sensing transcriptional regulator or impaired Cu 2ϩ delivery to apoCucA. The plastocyanin gene petE is switched off in low copper and the cytochrome c 6 gene petJ is switched on, to allow the substitution of copper in plastocyanin with heme iron in cytochrome c 6 for photosynthetic electron transport (35,36). These two genes were therefore used as controls in an analysis of copper-responsive gene expression. No change in the abundance of cucA transcripts was detected in any of the mutants, although a decrease in petE and an increase in petJ transcripts was most evident in cells missing either copper transporter (Fig. 6a). This implies that the two transporters and the metallochaperone are involved in Cu 2ϩ supply to CucA itself rather than to a transcriptional regulator of the cucA gene.
The CucA protein is undetected in a crude periplasm extract from the ⌬ctaA mutant (Fig. 6b). There are at least two similarly migrating bands in the region of CucA on Fig. 6b, and mass fingerprinting confirms that these are not CucA in ⌬ctaA mutants (supplemental Table 2). Loss of CucA protein implies that some facet of the production of CucA, its localization or stability, depends upon acquisition of the correct metal that in turn depends upon a copper-trafficking pathway.
Purified CucA Can Acquire Cu 2ϩ from ⌬ctaA Periplasm Extracts in Vitro-Additional low M r Cu 2ϩ -complexes accumulate in the periplasm of each of the transport and trafficking mutants (Fig. 5b). Presumably these represent a source of copper that would normally be imported across the cytoplasmic membrane and trafficked to CucA via the actions of CtaA, Atx1, and PacS. The chemical nature of these low M r periplasm copper complexes is unknown, but they might include copper-glutathione in the light of evidence of periplasmlocalized glutathione (37). We have examined whether or not these complexes could provide a source of copper for CucA by adding purified apoCucA to periplasm extracts and determining the metal status of the recombinant protein. By using extracts from wild-type cells, CucA does acquire some zinc (0.22 mol eq assuming full recovery of CucA) (Fig. 7), supporting the concept that some mechanism will indeed be required to achieve full copper occupancy in vivo, especially after growth in natural environments with less copper availability. Most importantly, using similar extracts from ⌬ctaA mutants, recombinant CucA became fully populated with copper (0.95 eq assuming full recovery of CucA) (Fig. 7), suggesting that the low M r copper complexes that accumulate in these cells are suitable copper donors yet fail to form Cu 2ϩ -CucA in ⌬ctaA mutants (Fig. 5). Thus, in living cells apoCucA does not directly acquire Cu 2ϩ from the low M r pools in the periplasm, rather these ions must be trafficked through the cytoplasm for Cu 2ϩ - FIGURE 5. CucA obtains less Cu 2؉ in trafficking mutants. a, cuproproteins CucA, plastocyanin and cytochrome oxidase (CcO), are located in the periplasm and thylakoid compartments, respectively. Copper metallochaperone Atx1 is cytosolic and interacts with amino-terminal regions of copper ATPases CtaA and PacS. PacS is predominantly located in thylakoid membranes. b, profile of periplasmic copper complexes from wild-type and mutant cells, separated using two-dimensional liquid chromatography based on relative molecular mass (M r ) and charge (pI). c, number of atoms of Cu 2ϩ cell Ϫ1 bound to CucA in wild-type and mutant cells calculated from the volumes under CucA peaks (as in b). Data are the means of three independent replicates with S.D. of triplicates. FIGURE 6. Loss of CucA protein but not transcripts in Cu 2؉ -trafficking mutants. a, RT-PCRs using RNA isolated from the four genotypes and primers corresponding to the four transcripts as shown. All products are reverse transcriptase-dependent (supplemental Fig. 5). 25 cycles, plus 30 cycles for cucA, were used. Similar trends were observed in two further experiments using independent RNA isolates. b, SDS-PAGE of crude periplasm extracts of wild-type and ⌬ctaA mutants after concentration using ion exchange chromatography. OCTOBER 15, 2010 • VOLUME 285 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 32509
CucA to accumulate in the periplasm implying that transmembrane copper transport is coupled to protein production and/or delivery to the periplasm.

DISCUSSION
Ab initio structures of CucA reveal a monocupin that binds divalent metals via three histidine imidazoles and a glutamate carboxylate within a hydrophobic pocket (Fig. 1). Dissociation rates for Zn 2ϩ and Cu 2ϩ are slow, in the region of k Zn,OFF ϭ 4.6 ϫ 10 Ϫ5 s Ϫ1 and k Cu,OFF ϭ 7.5 ϫ 10 Ϫ7 s Ϫ1 , respectively, allowing only slow replacement of Zn 2ϩ with Cu 2ϩ (Figs. 2-4). Association rates for Zn 2ϩ and Cu 2ϩ are both rapid (k Zn,ON Ϸ k Cu,ON ) such that incubation of apoCucA with equimolar Zn 2ϩ and Cu 2ϩ generates a mixed population of Zn 2ϩ -CucA and Cu 2ϩ -CucA that only gradually (t ϭ days) resolves to the more thermodynamically stable Cu 2ϩ form (Figs. 2 and 3). Cu 2ϩ acquisition in vivo is facilitated by a copper-trafficking pathway involving two P 1 -type ATPases and copper metallochaperone (Fig. 5), the same proteins also supply copper to plastocyanin within thylakoids (9,28).
Why is CucA Cu 2ϩ acquisition assisted by a trafficking pathway in vivo? Maintaining a pool of exchangeable Cu 2ϩ in the periplasm, sufficient to correctly populate CucA, would be challenging in natural environments with low concentrations of Cu 2ϩ . Presumably, the trafficking pathway encourages occupancy of CucA with Cu 2ϩ and not some other metal such as Zn 2ϩ . Notably, partial occupancy of apoCucA with Zn 2ϩ occurs when apoCucA acquires metal from periplasm extracts of wild-type cells cultured in standard medium that is copper-rich relative to natural habitats (Fig. 7).
CucA is missing from the periplasm of copper-trafficking mutants revealing that cofactor insertion into CucA is necessary for the protein to accumulate (Fig. 6b and supplemental Table 2). This is not a function of Cu 2ϩ -dependent expression of the cucA gene because cucA transcripts show normal levels of abundance in both ⌬ctaA and ⌬pacS cells (Fig. 6a). It is well established that many apoproteins are more susceptible to proteolytic degradation in vivo than their metal-bound holo-forms. A reasonable hypothesis might have been that in ⌬ctaA, ⌬pacS, and ⌬atx1 mutants, CucA fails to acquire sufficient Cu 2ϩ and hence is rapidly degraded in the periplasm. However, mutants in either of the copper-transporting ATPases accumulate additional low M r copper complexes in the periplasm (Fig. 5b). Exchangeable copper pools in extracts of the periplasm can readily donate Cu 2ϩ to apoCucA in vitro, and this is especially true of the pools in ⌬ctaA cells (Fig. 7). Periplasm extracts of ⌬ctaA cells grown in standard (copper-replete) culture medium provide sufficient Cu 2ϩ to saturate an aliquot of apoCucA, unlike the partial Zn 2ϩ occupancy generated using analogous extracts from wild-type cells (Fig. 7). This strongly argues that apoCucA fails to reach the periplasm in copper-trafficking mutants because any apoCucA that had been translated and translocated in ⌬ctaA cells should have readily acquired copper and become refractory to proteolysis, yet the protein was not detected.
Cyanobacteria are unlike most bacteria in possessing the internal membrane compartments, thylakoids, that house the copper proteins plastocyanin and cytochrome oxidase (Fig. 5a). There is some evidence that vesicle trafficking occurs between the plasma and thylakoid membranes (38,39), which might suggest that CucA could acquire copper in thylakoids before being trafficked to the periplasm. However, multiple proteomics studies have established that the product of open reading frame sll1785 (namely CucA) is uniquely localized to the periplasm, never the thylakoids; and furthermore, the Sec-dependent signal sequence of CucA has been exploited in multivariate analyses to distinguish periplasm targeting signals from those of proteins targeted to thylakoids (40). Thus, the implication is that translation and translocation of CucA across the plasma membrane is somehow linked to outward copper transport. Future studies should address the nature of this linkage and why it requires both CtaA and PacS. In the related cyanobacterium Synechococcus PCC 7942, PacS was mainly localized to thylakoid membranes by Western blotting using an anti-PacS antiserum, with a smaller subpopulation potentially located at the plasma membranes (29).
Accumulation of low M r Cu 2ϩ pools in the periplasm of mutants missing components of the trafficking pathway (Fig. 5b) implies that these pools are the source of copper for the pathway, but crucially copper from these pools is routed in and out of the cytoplasm to form Cu 2ϩ -CucA in vivo. Prokaryotes, including E. coli, that are devoid of known cytosolic cuproproteins and devoid of homologues of known cytosolic copper chaperones (such as CopZ or Atx1) must nonetheless have alternative mechanisms to inhibit deleterious side reactions of copper in the cytosol if these ions must pass through this compartment to supply cuproproteins in the periplasm. . In vitro apoCucA acquires copper from ⌬ctaA but zinc and copper from wild-type periplasm extracts. Purified CucA protein (1 ml of 1.5 M in 10 mM HEPES, pH 7.0, 150 mM NaCl) (open symbols) or buffer control (closed symbols) was added to periplasm extracts, which were then concentrated by anion exchange and eluted metal complexes resolved by gel filtration HPLC (TSK SW-3000) and eluant analyzed for copper (triangles) and zinc (diamonds) by ICP-MS.