Supplying Copper to the Cuproenzyme Peptidylglycine α-Amidating Monooxygenase*

We explored the role of known copper transporters and chaperones in delivering copper to peptidylglycine-α-hydroxylating monooxygenase (PHM), a copper-dependent enzyme that functions in the secretory pathway lumen. We examined the roles of yeast Ccc2, a P-type ATPase related to human ATP7A (Menkes disease protein) and ATP7B (Wilson disease protein), as well as yeast Atx1, a cytosolic copper chaperone. We expressed soluble PHMcc (catalytic core) in yeast using the yeast pre-pro-α-mating factor leader region to target the enzyme to the secretory pathway. Although the yeast genome encodes no PHM-like enzyme, PHMcc expressed in yeast is at least as active as PHMcc produced by mammalian cells. PHMcc partially co-migrated with a Golgi marker during subcellular fractionation and partially co-localized with Ccc2 based on immunofluorescence. To determine whether production of active PHM was dependent on copper trafficking pathways involving the CCC2 or ATX1 genes, we expressed PHMcc in wild-type, ccc2, and atx1mutant yeast. Although ccc2 and atx1 mutant yeast produce normal levels of PHMcc protein, it lacks catalytic activity. Addition of exogenous copper yields fully active PHMcc. Similarly, production of active PHM in mouse fibroblasts is impaired in the presence of a mutant ATP7A gene. Although delivery of copper to lumenal cuproproteins like PAM involves ATP7A, lumenal chaperones may not be required.

Production of amidated peptides by PAM requires copper and molecular oxygen and two reducing equivalents, usually supplied by two molecules of ascorbate. The PHM domain of PAM binds copper at two non-equivalent sites (Cu A and Cu B ), both of which are critical for activity (8). Each copper atom is reduced by transfer of a single electron from ascorbate, generating semidehydroascorbate. Copper is bound by three His residues at the Cu A site (His 107 , His 108 , His 172 ) and by two His residues and a Met residue at the Cu B site (His 242 , His 244 , Met 314 ) (9). The process by which copper is delivered to the PHM domain of PAM in the secretory pathway is unknown. P-type ATPases are likely to play a role in copper transport into the secretory pathway. Two copper transporting P-type ATPase genes, ATP7A and ATP7B (10,11), were identified in studies of inherited disorders of human copper metabolism (12). Mutations in the ATP7A gene lead to Menkes disease, a disorder characterized by copper deficiency. ATP7A is expressed in all tissues except the liver, and mutations in ATP7A prevent normal intestinal absorption of copper and distribution of copper throughout the body. The resulting lack of copper compromises the function of many tissues, resulting in death in early childhood. Mutations in the ATP7B gene cause Wilson disease, which is characterized by failure of the liver to excrete copper into the biliary tract or to deliver copper to ceruloplasmin, a multi-copper oxidase needed for serum iron homeostasis (13). The prevalence of ATP7B in the liver means that protein mutations result in liver toxicity because of the accumulation of copper. Although tissue-specific, both disorders involve a defect in P-type ATPase-mediated export of copper from the cytosol into the secretory/endocytic pathways (14).
Due in large part to the clarity that yeast genetics brings to the analysis of protein function, Saccharomyces cerevisiae is an excellent model organism for studying many fundamental processes of eukaryotic cells (15). The importance of oxidative metabolism to eukaryotic cells has led to conservation of mechanisms of copper metabolism from yeast to human cells. Indeed, it has been shown that the molecular components of copper trafficking pathways are highly conserved between yeast and humans. In S. cerevisiae, the pathway begins with Cu ϩ uptake through the action of copper transporters Ctr1 and Ctr3 (16,17). In the cytoplasm, copper is bound by cytoplasmic copper chaperones that deliver the metal to specific target enzymes (18). The Atx1 copper chaperone delivers copper to Ccc2, the only yeast homologue of the mammalian Wilson and Menkes ATPases (19). Ccc2 is required for transport of copper from the cytosol into the lumen of the trans-Golgi network, where the multi-copper ferroxidase Fet3, the yeast homolog of ceruloplasmin, is loaded with copper (20,21).
The yeast genome does not encode a PHM or a PAL homolog. In the present study, we determined whether active PHM could be produced in the yeast secretory pathway and whether its production requires the assistance of known copper transporters and chaperones. In particular, the roles of the P-type ATPase Ccc2 and the cytosolic copper chaperone Atx1 were investigated. We expressed soluble rat PHMcc (PHM catalytic core) using the yeast ␣-mating factor leader sequence to target the enzyme to the yeast secretory pathway (Fig. 1). We compared the activity of PHMcc produced in yeast to that of PHMcc produced in Chinese hamster ovary (CHO) cells. Using immunofluorescence microscopy and subcellular fractionation, we established that PHMcc was partially co-localized with Golgi markers and Ccc2 in yeast cells. Using gene knockouts, we established an essential role for the secretory pathway copper transporter Ccc2 and the cytosolic copper chaperone Atx1 in producing active PHMcc in yeast. Finally, we demonstrated that active Menkes protein, a mammalian homologue of Ccc2, can support production of active PHM in fibroblasts.
Plasmids-We expressed PHMcc protein using a yeast ␣-mating factor signal sequence. Soluble and integral membrane PAM proteins (PAM-3 and PAM-2) expressed in yeast with the rat PAM signal sequence were not directed to the secretory pathway, remained cytosolic, and exhibited no enzyme activity, even in the presence of added copper (data not shown). pSM703.␣MF-PHMcc was generated by PCR amplification of PHMcc (amino acids 42-356 of rat PAM-1; nucleotides 421-1365) from pBS.PAM-1, which encodes full-length PAM-1 (25). The PCR product was cloned into the pICK9 vector (Invitrogen) using the unique XhoI and NotI sites; pICK9.PHMcc, containing the PHM fragment with the ␣MF pre-pro-sequence (amino acids 1-81) and a LEKR linker, was digested with BamHI and NotI. The digested PHMcc product was ligated into the same sites of the pICK9K vector (Invitrogen). A multicopy ␣MF-PHMcc plasmid, pSM703.␣MF-PHMcc, was constructed by inserting the BamHI-NotI ␣MF-PHMcc fragment of pICK9K.PHMcc into the 2 URA3 plasmid pSM703 (26). As described by Yuan et al. (21), the pDY207.Ccc2-HA vector fused the 3Ј-end of the CCC2 coding sequence with sequences encoding three tandem copies of the influenza hemagglutinin (HA) epitope tag followed by two stop codons and the CYC1 terminator. The PAM-1 recombinant adenovirus (PAM-1 virus) encodes full-length rat PAM-1 (nucleotides 293-3245); its preparation was described previously (27,28).
Cell Extracts and Immunoblotting-For Western blot analysis of PAM-expressing S. cerevisiae cells, the yeast strains transformed with pSM703 vector or vector encoding ␣MF-PHMcc were harvested at midlogarithmic phase (A 600 ϭ 1) in regular SD medium or low copper SD medium (less than 0.25 M copper) minus uracil. Yeast cell extracts were prepared by glass bead homogenization in 20 mM sodium (Ntris(hydroxymethyl)methyl-2-aminoethanesulfonic acid) (TES)/TMT (10 mM mannitol, 1% Triton X-100, pH 7.4) containing protease inhibitors, 1 mM sodium ascorbate, and 1 mM bathocuproinedisulfonate to limit the availability of copper during extraction as described previously (20). To analyze PAM expression in mammalian fibroblasts, the 802-1 and 802-5 cells, grown as described above, were scraped into TMT buffer containing protease inhibitors, frozen and thawed three times, and centrifuged for 5 min to remove cell debris (27). With either yeast or mammalian expression systems, samples were resolved by 8 or 12% SDS-PAGE, transferred to nitrocellulose membranes, and analyzed by Western blot using rabbit polyclonal antiserum to PHM (JH1761) and the ECL kit (Amersham Biosciences).
PHM Assays-PHM activity was measured in yeast and mammalian cell extracts as described previously (8) using ␣-N-acetyl-Tyr-Val-Gly as substrate. Protein (30 to 50 ng) was assayed in the presence or absence of 0.5 M CuSO 4 as indicated, in duplicate, and reactions were carried out for 2 h. PHM specific activity is expressed as picomoles of product formed per h (units) per microgram of protein.

FIG. 1. PAM-catalyzed reactions and PAM proteins.
The enzymatic reactions catalyzed by the PHM and PAL domains of PAM are shown. PAM protein structure is compared with that of the PHM catalytic core (cc) used in this study. The PHMcc protein expressed in mammalian CHO cells and the ␣-mating factor-PHMcc fusion protein expressed in wild-type yeast cells (YPH252) are shown. The yeast ␣-mating factor precursor is diagrammed with ␣-factor peptides (␣F; clear boxes) separated by spacer peptides consisting of KR followed by two or three EA or DA pairs (horizontally cross-hatched boxes) that are used as endopeptidase cleavage sites (60). Subcellular Fractionation-The procedure for subcellular fractionation of yeast cells is based on sedimentation through sucrose density gradients, as described previously (32), but with some modifications. Yeast (YPH252) transformed with pSM703.␣MF-PHMcc were grown to log phase (A 600 ϭ 1) in SD medium minus uracil. After being chilled on ice, cells (500 optical density units) were harvested by centrifugation at 3700 ϫ g for 10 min. The pellet was resuspended in 30 ml of cold 10 mM sodium azide and centrifuged for 10 min at 2000 ϫ g. Cells were resuspended to 50 A 600 units/ml in spheroplast buffer (1.4 M sorbitol, 50 mM potassium phosphate, pH 7.5, 10 mM sodium azide, 40 mM mercaptoethanol), and zymolase 20T (ICN Biomedicals) was added to 1.5 mg/ml; cells were converted to spheroplasts by incubating for 45 min at 30°C with gentle shaking. All further steps were carried out on ice. The spheroplast suspension was adjusted to 1 mM EDTA, layered on top of 8 ml of 1.8 M sorbitol, and spheroplasts were centrifuged through the sorbitol cushion by centrifugation at 500 ϫ g for 10 min. The pellet was resuspended in 5 ml of lysis buffer (0.8 M sorbitol, 10 mM MOPS, pH 7.2, 1 mM EDTA) containing a mixture of the following protease inhibitors: 2 g/ml leupeptin, 16 g/ml benzamidine, and 300 g/ml phenylmethylsulfonyl fluoride. The spheroplast lysate was homogenized with a tissue grinder (Kontes Scientific Glassware) and then centrifuged for 10 min at 500 ϫ g to remove unlysed cells and debris. The resulting supernatant (1 ml) was layered onto a step gradient containing 1 ml each of 18,22,26,30,34,38,42,46,50, and 54% sucrose (w/v) in 10 mM HEPES, pH 7.5, 1 mM MgCl 2 . Gradients were centrifuged for 2.5 h at 174,000 ϫ g in an SW41 rotor (Beckman Instruments) at 4°C. Fractions (660 l) were collected from the top, and an equal aliquot of each fraction was analyzed. For Western blot analysis, 50 l of aliquot of each fraction was used for SDS-PAGE as described above.
Organelle Marker Enzyme Assays-Reactions (25 l of each fraction) were carried out in a total volume of 100 l, except for NADPHcytochrome c reductase assays, which were carried out in a volume of 1 ml. Vacuolar ␣-mannosidase was assayed through the generation of p-nitrophenol from p-nitrophenyl-␣-mannopyranoside, which leads to an increase in A 400 . The reaction mixture consisted of 0.4 mM p-nitrophenyl-␣-mannopyranoside, 40 mM sodium acetate, pH 6.5 (33). GDPase was assayed as described by Abeijon et al. (34). Incubation tubes contained 20 mM imidazole, pH 7.4, 2 mM CaCl 2 , 7 mM GDP, and 0.1% Triton X-100. Liberated phosphate was measured with the Fiske Subbarow kit (Sigma). Nonspecific NDPase plus free P i were determined using CDP as a nonspecific substrate. These values were subtracted from those obtained with the substrate GDP to give GDPase activity. NADPH-cytochrome c reductase was assayed by the absorbance increase at 550 nm according to Feldman et al. (35). Assay mixtures (1 ml) contained 1.0 mg/ml cytochrome c, 100 M NADPH, 50 mM KH 2 PO 4 , pH 7.4, 0.4 mM KCN, 1.0 M FMN, and 0.6 M sorbitol. Reaction rates were determined from the initial linear phase of the assay.

RESULTS
Expression of Active PHM in Yeast-The yeast genome does not encode a protein homologous to PHM or to any of the other members of this family of copper-containing monooxygenases; the amino acid sequence of rat PHMcc was compared with the S. cerevisiae genome using BLASTN 2.2.3 (Apr-24 -2002) (NCBI) (36). To determine whether we could use S. cerevisiae to identify proteins involved in the delivery of copper to PHM, we first expressed soluble PHMcc in S. cerevisiae and analyzed cell extracts and medium for PHM activity. We fused the ␣MF signal/pro-region to the catalytic core of rat PHM (␣MF-PHMcc) to direct the foreign monooxygenase into the yeast secretory pathway. With its pro-region, ␣MF-PHMcc should have a mass of 41 kDa. Transformation of wild-type yeast with vector encoding ␣MF-PHMcc led to expression of a 34-kDa protein recognized by antiserum to rat PHMcc (Fig. 2), indicating that the ␣MF pro-region had been removed. Vector alone yielded only a smaller, nonspecific cross-reactive protein.
PHMcc secreted by stably transfected CHO cells consistently had a slightly higher apparent molecular mass (36 kDa) than the cross-reactive product detected in yeast extracts. Although an LEKR sequence separated the ␣MF pro-region from PHMcc, the exact site of endoproteolytic cleavage was not determined, and this difference in mass may reflect different trimming or nicking of PHMcc in yeast and CHO cells.
Extracts prepared from wild-type yeast contained no detectable PHM activity (Ͻ0.01 units/g protein). Control experiments demonstrated complete recovery of exogenous rat PHMcc activity added to yeast extract (1 g protein) and assayed in the presence of 0.5 M exogenous copper (data not shown). Extracts prepared from yeast expressing ␣MF-PHMcc contained 5.4 Ϯ 0.5 units/g of protein when assayed under the same conditions. To compare the specific activity of PHMcc expressed in yeast with that of PHMcc secreted by stably transfected CHO cells (with PHM specific activity of 9000 Ϯ 500 units/g of protein) (8) (Fig. 2), the amount of PHMcc protein was estimated based on cross-reactivity with an antiserum specific to rat PHMcc (37). PHMcc expressed in yeast is at least as active as PHMcc secreted by mammalian cells; for the same amount of PHM enzyme activity (25 or 125 units), less PHM protein is present in the yeast lysate than in the mammalian cell medium. Despite the fact that the ␣-mating factor-PHMcc construct yields a soluble, active protein, PHMcc is secreted very poorly, accumulating in post-Golgi vesicles. Secreted PHM could be detected in lyophilized yeast growth medium by Western blot analysis; however PHM activity was not detectable (data not shown).
PHMcc Partially Co-localizes with Ccc2-HA in the Yeast Secretory Pathway-Indirect immunofluorescence microscopy was utilized to determine the subcellular localization of PHMcc in yeast. Cells expressing PHMcc were fixed and visualized with a PHM antiserum (Fig. 3A). PHM staining was observed in punctate structures reminiscent of Golgi-like vesicles. The punctate PHM staining pattern was similar to that obtained for proteins thought to be associated with Golgi or post-Golgi vesicles (38 -40), with some puncta significantly larger than others. The staining is specific for PHMcc, as only very faint homogeneous staining was observed in cells transformed with empty vector (Fig. 3A). The 4Ј,6Ј-diamidino-2-phenylindole staining profile (Fig. 3A, middle panel) demonstrated that PHMcc is neither nuclear nor perinuclear and is not associated with mitochondria. Comparison of the distribution of PHMcc to the differential interference contrast optics images (Fig. 3A, right panel) demonstrated that PHMcc is not vacuole-associated.
The yeast copper transporter Ccc2 is known to reside in post-Golgi vesicles (21,41). To evaluate a functional connection between Ccc2 and PHMcc, we compared the subcellular localization of expressed PHMcc with that of an epitope-tagged version of Ccc2 (Fig. 3A). Yeast of the wild-type strain, transformed with vectors encoding both PHMcc and HA-tagged Ccc2, were stained simultaneously with a rabbit polyclonal antiserum to PHM and a mouse monoclonal antibody to the HA tag and examined using confocal microscopy (Fig. 3B). PHM staining (in green) and Ccc2-HA staining (in red) were superimposed. As shown in the merged image (Fig. 3B), PHMcc and Ccc2 were sometimes localized to the same vesicular structures (Fig. 3B, arrows). However, the distributions of Ccc2 and PHMcc were not identical, with some vesicles staining more intensely for PHM (arrowheads) and others staining more intensely for Ccc2 (lines). Based on their subcellular localization, Ccc2 and PHMcc have access to the same subcellular compartments. As might be expected for an integral membrane protein and a soluble, secretory protein, transient co-localization of Ccc2 and PHMcc was observed.
PHMcc Co-migrates with a Golgi Marker during Subcellular Fractionation-To confirm our conclusion that PHMcc is at least transiently localized to the Golgi compartment of the secretory pathway, subcellular fractionation was carried out. Extracts prepared from wild-type yeast expressing PHMcc were subjected to velocity sedimentation through sucrose gradients. The various subcellular compartments were identified in gradient fractions by measuring enzymatic markers charac-teristic of each organelle (Fig. 4). The distribution of PHMcc protein was determined by Western blot analysis. The bulk of the PHM was found in fractions 8 though 11. The higher molecular mass cross-reactive protein localized in fractions 1 through 5 corresponds to a nonspecific band visualized in the same gradient fractions of yeast cells transformed with expression vector alone. The Golgi marker GDPase (34) was also localized to fractions 8 through 11. PHMcc was clearly separated from the endoplasmic reticulum marker (35), NADPH-cytochrome reductase, which was highly localized to fraction 13, and from the vacuolar marker (33), ␣-mannosidase, which was concentrated in fraction 3 (Fig. 4B). Based on both immunofluorescence and subcellular fractionation, the bulk of PHMcc is localized to the Golgi complex in S. cerevisiae at steady state.
Effects of CCC2 and ATX1 Gene Mutations on PHM Enzyme Activity-Because active PHM can be produced in wild-type yeast, mutant yeast were used to assess the role of specific copper transporters and chaperones in providing copper to PHM. In yeast, the cytosolic copper chaperone Atx1 shuttles copper to Ccc2, an intracellular copper transporter that is localized to the membranes of the Golgi compartment (18,19,42,43). This transporter pumps copper into the lumen of the Golgi complex, making it available for insertion into copper-dependent lumenal enzymes. To assess the role of Ccc2 and the cytosolic copper chaperone Atx1 in providing copper to PHM, we expressed PHMcc in yeast bearing mutations in the CCC2 or ATX1 genes (Fig. 5). As shown by Western blot analysis (Fig.  5A), equal expression of PHMcc protein was observed in wildtype yeast and in ccc2 and atx1 mutant cells.
We next compared the catalytic activity of PHMcc produced in wild-type, ccc2, and atx1 mutant cells. Because the addition of exogenous copper or the binding of copper made accessible following homogenization could obscure any deficits present in PHMcc expressed in ccc2 or atx1 mutant cells, we varied the assay conditions. Reduced ascorbate was added to the extrac-FIG. 3. PHM and Ccc2 proteins partially co-localize in yeast cells. YPH252 cells expressing PHMcc, Ccc2-HA, or both proteins were examined by fluorescence microscopy (ϫ1000 magnification). A, PHM was visualized with rabbit antibody to rat PHM and a fluorescein isothiocyanate-tagged antibody to rabbit IgG. Ccc2-HA was visualized with antibody to the HA epitope (HA.11) followed by a Cy3-tagged antibody. Control cells were transformed with an empty vector and subjected to the same staining protocol. In parallel, cells were stained with 4Ј,6Ј-diamidino-2-phenylindole (DAPI) for nucleic acid detection and analyzed using Nomarski optics to visualize the indentations of the yeast vacuole. DIC, differential interference contrast optics. B, simultaneous immunostaining of cells expressing PHMcc and Ccc2-HA. PHMcc (in green) and Ccc2-HA (in red) were visualized using two fluorescently tagged secondary antisera (as described above) and confocal laser scanning fluorescence microscopy. In the superimposed images shown on the right, vesicles co-immunostained with antisera to PHMcc and to Ccc2-HA appear yellow and are indicated by arrows; PHM staining alone is indicated by arrowheads, and vesicles stained only by Ccc2 are indicated by lines.

FIG. 4. Subcellular localization of PHMcc in S. cerevisiae.
A spheroplast lysate prepared from YPH252 cells expressing PHMcc was centrifuged at 500 ϫ g for 10 min. The supernatant was layered onto a 18 -54% sucrose gradient, and subcellular organelles were separated during a 2.5-h spin at 174,000 ϫ g (see "Experimental Procedures"). A, aliquots of the gradient fractions were separated by 8% SDS-PAGE, and the distribution of PHMcc was analyzed by immunoblotting. B, for enzymatic assays, NADPH-cytochrome c reductase was measured as an endoplasmic reticulum marker, GDPase was measured as a Golgi marker, and ␣-mannosidase was measured as a vacuolar marker. Enzymatic activities are expressed on a per fraction basis, for GDPase (1 unit ϭ nmol phosphate produced per 20 min), NADPH-cytochrome c reductase (1 unit ϭ nmol cytochrome C reduced per min), and ␣-mannosidase (1 unit ϭ nmol of p-nitrophenyl-␣-mannopyranoside cleaved per 15 min). Fractions were harvested from the top (fraction 1) to the bottom (fraction 16). tion buffer to promote retention of bound copper (by keeping it reduced), and bathocuproinedisulfonic acid was supplemented to ensure that any free copper was made inaccessible (20). When assayed without added copper ions, no PHM activity was found in yeast cells lacking either Ccc2 or Atx1; in contrast, the specific activity of PHMcc in wild-type cell extracts was 2.04 Ϯ 0.11 units/g of protein. The PHMcc protein expressed in ccc2 and atx1 mutant yeast could be fully reconstituted by in vitro addition of 0.5 M copper (2.09 Ϯ 0.18 and 2.31 Ϯ 0.3 units/g of protein, respectively) (Fig. 5B). Although expression of PHMcc protein does not require Ccc2 or Atx1, expression of active PHM does depend on provision of copper through the trafficking pathway involving Ccc2 and Atx1.
ATP7A Supports the Production of Active PAM Enzyme in Mammalian Cells-Although yeast have a single coppertransporting P-type ATPase, mammalian cells have two, ATP7A and ATP7B. We used a murine cell line defective in the mouse homologue of the Menkes disease gene as a test system for assessing the role of ATP7A in providing copper to PAM (44). The 802-1 mottled cell line was derived from mutant male mice carrying the ATP7a Mo-br allele of ATP7A; the ATP7a Mo-br gene has a 6-bp deletion that eliminates two amino acids in a conserved region of the protein, resulting in a severe phenotype similar to classical Menkes disease (29,45,46). In addition, 802-1 mottled cells are heterozygous for deletion of the metallothionein I and II genes, shown previously to increase the sensitivity of these cells to copper, permitting a rapid functional assay for the Menkes gene (44). A control cell line, 802-5, derived from wild-type male mice, has wild-type ATP7A and is also heterozygous for deletion of the metallothionein I and II genes. Neither the 802-1 nor the 802-5 cells express ATP7B, the other mammalian coppertransporting P-type ATPase (29,45,46).
To determine whether active PAM enzyme can be produced in cells lacking functional ATP7A and ATP7B, we used a recombinant adenovirus encoding full-length PAM-1 to infect both mottled 802-1 cells and wild-type 802-5 cells. The 120-kDa PAM-1 protein was identified in cell extracts using the PHM antibody for Western blot analysis (Fig. 6A). The same level of PAM protein expression was achieved in the mutant and wild-type cell lines. When the expressed enzyme was assayed for activity in the absence of additional copper, the PHM activity in wild-type 802-5 cells was 0.9 Ϯ 0.06 units/g of protein, whereas no PHM activity was detected in extracts of the mottled 802-1 cells (Ͻ0.01 units/g protein) (Fig. 6B). Because expression of PAM-1 protein did not require the presence of active ATP7A, we added exogenous copper to try to activate the PAM-1 protein pro- tional ATP7A protein is required for production of active PHM in these mouse fibroblasts, which lack ATP7B.

DISCUSSION
Secretion of Cuproproteins-Highly conserved copper trafficking pathways have evolved to control levels of copper and provide copper to the limited number of enzymes that need it while protecting the rest of the cell (47). Several cytosolic and mitochondrial cuproproteins acquire copper only when it is provided by a specific chaperone protein. Most of these copper chaperones share sequence homology with their target protein, facilitating a specific metallochaperone-cuproprotein interaction (18). Delivery of copper to secreted proteins like Fet3 and ceruloplasmin, multicopper oxidases needed for high affinity iron uptake in yeast and mammals, respectively (41), requires cytosolic chaperonemediated delivery of copper to a P-type ATPase. Hah1 (or Atox1), like its yeast homologue, Atx1 (47)(48)(49), shares an MTCXXC metal binding site with its target P-type ATPases. Mice genetically engineered to lack Atox1 Ϫ/Ϫ failed to thrive immediately after birth, with surviving animals exhibiting growth failure, skin laxity, hypopigmentation, and seizures (50). The activity of cuproenzymes such as cytochrome oxidase and tyrosinase was decreased significantly in Atox1 Ϫ/Ϫ mice. Although PAM function was not analyzed, decreased levels of amidated neuropeptides in the Atox1 Ϫ/Ϫ mice may contribute to the neurodegeneration and growth retardation observed.
The number of copper-dependent secretory pathway proteins is small, but growing. Fet3/ceruloplasmin, laccases (multi-copper enzymes that catalyze the oxidation of phenolic and nonphenolic compounds), and secretory pathway enzymes like tyrosinase, PAM, dopamine ␤-monooxygenase, and lysyl oxidase are catalytically inactive in the absence of copper (8,20,(51)(52)(53). Other secretory pathway proteins like Alzheimer precursor protein (APP) and prion protein (PrP) bind copper, but the function of the bound copper is unknown (54 -56). A role for secretory pathway chaperones in delivering lumenal copper provided by Ccc2, ATPA, or ATP7B to secretory pathway cuproproteins has not been explored. We reasoned that expression in yeast of a mammalian secretory pathway enzyme with no yeast homolog would allow us to address this issue. Yeast lack genes encoding any of the members of the PHM/dopamine ␤monooxygenase family of monooxygenases. To take advantage of the genetic tools available, we expressed PHMcc in S. cerevisiae.
Expression of Active PHM in Yeast-When expressed in yeast using the rat PAM signal sequence, both membrane and soluble forms of PAM remained cytosolic and were not directed to the secretory pathway (data not shown). In contrast, the secretion signal from yeast ␣MF targeted PHMcc to the secretory pathway. Secretion of biologically active epidermal growth factor (57), ␤-endorphin, calcitonin (58), and somatostatin (59) by yeast also required use of the ␣MF signal. In S. cerevisiae, cleavage of pre-pro-␣MF occurs in two steps. The N-terminal 20-amino acid signal sequence is removed in the endoplasmic reticulum by the signal peptidase complex (60,61). After transit to the Golgi, a membrane-bound endoprotease, Kex2, removes the 60-amino acid pro-region and separates the four copies of ␣MF peptide (62,63). Although expression of PHMcc was readily detected in yeast cell extracts, secreted PHMcc could only be detected after lyophilization of a large volume of medium and Western blot analysis (data not shown). A soluble insulin-containing fusion protein encoding the ␣MF leader fused to a single chain insulin variant was also secreted inefficiently (64). Intracellular retention of insulin required endoproteolytic cleavage of the fusion protein by Kex2 in the Golgi. Expression of an ␣MF leader/glycosylated erythropoietin fusion protein in yeast identified several rate-limiting steps, including cleavage of the ␣MF-erythropoietin precursor protein by Kex2 and transport of the protein through the secretory pathway (65).
Based on both immunofluorescence and subcellular fractionation, PHMcc is partially localized at steady state to the Golgi, where Kex2 and Ccc2 are located. Intact ␣MF-PHMcc is not detected by Western blot analysis, suggesting that cleavage of the precursor is not a rate-limiting step. The transient colocalization of Ccc2 and PHMcc in the same subcellular compartment should facilitate delivery of copper to PHM. The PHMcc localized to non-Ccc2-containing vesicles (Fig. 3B) may represent enzyme in the process of being secreted or degraded. The multi-copper oxidase, Fet3, and Ftr1, a permease involved in high affinity iron uptake in yeast, assemble into a complex in a cellular compartment early in the secretory pathway. The complex progresses to a post-Golgi compartment (21), where Ccc2 mediates copper delivery to Fet3. Finally, the copper-loaded Fet3 protein-permease complex is delivered to the plasma membrane and becomes competent for iron transport (41).
Transient co-localization of PHMcc and Ccc2 in a post-Golgi compartment is consistent with an important function for Ccc2 in providing copper to PHM (see below). Cytosolic copper export is restricted to a late Golgi compartment in the S. cerevisiae secretory pathway (21). Yeast sec mutants blocked at pre-Golgi and Golgi sites in the secretory pathway fail to deliver copper to Fet3. In addition, yeast mutants like vps33, with defects in a post-Golgi compartment, are also deficient in Fet3 activity. Thus, Ccc2 distribution is compromised when either early secretory pathway integrity and/or post-Golgi sorting are disturbed, leading to copper transport deficiency and Fet3 inactivity.
Directed to the secretory pathway, PHMcc expressed in wildtype yeast is at least as active as PHMcc secreted by mammalian CHO cells. Despite the fact that yeast S. cerevisiae does not express a protein homologous to PAM or to any of the members of the family of copper-containing monooxygenases, copper is made available to the newly synthesized PHMcc protein traveling through the yeast secretory pathway.
Ccc2 and Atx1 Are Required for Production of Active PHMcc in S. cerevisiae-We expressed PHMcc in wild-type, ccc2, and atx1 mutant yeast to determine whether production of active PHM were dependent on the copper trafficking pathways involving these genes. Some newly synthesized metalloenzymes require the presence of metal for proper folding and subsequent exit from the endoplasmic reticulum (66). However, expression of PHMcc protein was unaltered in yeast lacking secretory pathway copper. These data agree with the fact that an inactive PHMcc mutant unable to bind copper at Cu A or Cu B was still efficiently synthesized and secreted by CHO cells (8). When assayed in the absence of exogenous copper, no PHM enzymatic activity is detected in yeast lacking Ccc2 or Atx1. Nevertheless, following addition of exogenous copper, PHMcc produced in ccc2 or atx1 mutant yeast was as active as PHMcc expressed in wild-type yeast. Thus production of active PHMcc in yeast requires functional Atx1 and Ccc2.
ATP7A and Production of PAM in Mammalian Cells-Expression of membrane PAM-1 protein was similar in fibroblast lines lacking functional ATP7A and ATP7B or having functional ATP7A. Although PAM enzyme produced in cells expressing ATP7A was partially metallated, PAM enzyme produced in cells lacking both copper-transporting P-type ATPases was inactive until addition of exogenous copper. ATP7A was shown previously (49) to complement the ccc2 yeast mutant and provide copper to Fet3, the yeast ceruloplasmin homologue. ATP7A is also required for the production of active tyrosinase, a copper-dependent enzyme involved in melanogenesis within the secretory pathway (52). Thus copper delivered to the secretory pathway by ATP7A is available to PAM. Consistent with these data, our research group demonstrated recently that the ability of PAM to produce bioactive amidated peptides was compromised in the mottled brindled male mouse (Atp7a mouse). Although normal levels of PAM protein were found in Atp7a mice, levels of amidated Joining Peptide and ␣-melanocyte stimulating hormone were diminished in the pituitary, and levels of amidated cholecystokinin were diminished in the cerebral cortex (67). Peptide amidation was reduced, but not eliminated, in the Atp7a mouse. Taken together with our data on fibroblasts lacking both ATP7A and ATP7B, this suggests that ATP7B may also support the delivery of copper to PAM.
Models-Although our findings demonstrate that ATP7A delivers copper into the secretory pathway, allowing PAM to acquire copper, the precise mechanism by which this process occurs is unknown. Two models for the incorporation of copper into copper-dependent enzymes of the secretory pathway need to be considered. In the first model, ATP7A transfers copper into the lumen of the secretory pathway, where the metal is directly incorporated into apo-PHM and other copper-dependent lumenal enzymes. Alternatively, copper pumped into the secretory pathway by ATP7A equilibrates with low and high affinity binding sites in the lumenal compartment, allowing secretory pathway cuproenzymes like PHM to bind copper. Based on the fact that yeast, which do not produce a protein homologous to PHM, can synthesize fully active enzyme, we argue against the existence of a cuproenzyme-specific lumenal chaperone system similar to that described in the cytosol (26,68). However, a copper-binding protein capable of retaining copper in the Golgi and making it available to lumenal cuproenzymes could play a role. The lower pH and oxidizing environment of the lumenal compartment may dictate different schemes for handling cytosolic and lumenal copper.
In summary, our data suggest that loading of copper onto secretory pathway enzymes like PAM is fundamentally different from loading of copper onto cytosolic and mitochondrial enzymes. The sequestered environment of the secretory pathway may eliminate the need for specific copper chaperones. Measurement of total and free copper levels within the secretory pathway will be informative.