Metal Ion Chaperone Function of the Soluble Cu(I) Receptor Atx1

See allHide authors and affiliations

Science  31 Oct 1997:
Vol. 278, Issue 5339, pp. 853-856
DOI: 10.1126/science.278.5339.853


Reactive and potentially toxic cofactors such as copper ions are imported into eukaryotic cells and incorporated into target proteins by unknown mechanisms. Atx1, a prototypical copper chaperone protein from yeast, has now been shown to act as a soluble cytoplasmic copper(I) receptor that can adopt either a two- or three-coordinate metal center in the active site. Atx1 also associated directly with the Atx1-like cytosolic domains of Ccc2, a vesicular protein defined in genetic studies as a member of the copper-trafficking pathway. The unusual structure and dynamics of Atx1 suggest a copper exchange function for this protein and related domains in the Menkes and Wilson disease proteins.

Although Cu is an essential cofactor for mitochondrial, cytosolic, and vesicular oxygen-processing enzymes (1), it can be toxic even at low concentrations. Cu(I) and Cu(II) ions can bind with high affinity to adventitious sites in partially folded proteins and catalyze auto-oxidation of lipids, proteins, and nucleic acids. To investigate the mechanisms by which cells overcome the dilemma of maintaining Cu availability while controlling deleterious reactivity of the free ions, we have determined the Cu chemistry and Cu-handling function of Atx1, an intracellular eukaryotic protein implicated in Cu trafficking. Our results indicate that Atx1 functions as a metal ion chaperone, a protein that protects and guides Cu(I) ions to activate target enzymes.

ATX1 is one of several genes in the Cu-dependent, high-affinity iron uptake pathway in yeast. These genes encode Ctr1, a Cu uptake protein in the plasma membrane; Ccc2, an intracellular membrane protein; and the multicopper oxidase Fet3 (2-6). Although the role of Fet3 in iron uptake is unclear, Cu loading into this enzyme occurs in a post-Golgi vesicle and is mediated by Ccc2 (4). The ATX1 gene and its human homolog, HAH1, encode cytosolic proteins implicated in Cu trafficking to these Ccc2-containing vesicles (7-9). The Ccc2 protein in yeast and its human homologs, the Menkes disease protein (10) and Wilson disease protein (11), are members of the P-type adenosine triphosphatase (ATPase) cation transporter family (12) and are present in the membranes of secretory vesicles (13). Each transporter contains two or more Atx1-like cytoplasmic domains (Fig.1), the functions of which are not known. The conserved MTCXXC sequence (X, any residue), a motif observed in several bacterial Hg(II) transport proteins (14), is thought to be a Cu binding site, although it does not correspond to known Cu(I) or Cu(II) sites in structurally characterized proteins. Establishing the Cu oxidation state and coordination environment in Atx1 should provide insight into its function and that of the homologous domains in its partner proteins.

Figure 1

Homology among MXCXXC-containing proteins. Identical regions are shaded dark and regions displaying high homology are light. Areas with internal gaps are indicated by a period. Proteins (with accession number) are as follows: Atx1, Saccharomyces cerevisiae (L35270); Hah1,Homo sapiens (U70660); Ccc2, S. cerevisiae(L36317); Wilson disease protein, H. sapiens(U11700); Menkes disease protein, H. sapiens(L06133); MerP, Pseudomonas aeruginosa (Z00027). Partial amino acid sequences of some proteins are as follows: Ccc2-d1, amino acids 1 to 71, Ccc2-d2, amino acids 76 to 150, Wilson1, amino acids 54 to 125, Menkes1, amino acids 4 to 75, and MerP, 20 to 91. The figure was created with the Pileup program from the Wisconsin Package, Version 9.0, from the Genetics Computer Group (GCG). Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Atx1 was expressed in Escherichia coli and purified as the apoprotein (15). Addition of CuSO4 to apo-Atx1 in the presence of excess thiol reproducibly yielded a complex with a copper/protein ratio of 0.6 to 0.8 (16). Gel filtration experiments indicated that the predominant form of the protein was a monomer under these conditions, regardless of the presence or absence of Cu (15). The Cu oxidation state was investigated by electron paramagnetic resonance (EPR) spectroscopy and x-ray absorption spectroscopy (XAS). No EPR signal was observed at 77 K, suggesting the absence of a mononuclear Cu(II) site. XAS experiments (17) indicated that the bulk sample contained Cu(I). The Cu-Atx1 x-ray absorption near edge structure (XANES) spectrum exhibited a weak shoulder at 8984 eV, which is typical of Cu(I) and inconsistent with comparable edge features for Cu(II) compounds, the energies of which are 3 to 4 eV greater (Fig.2A). The intensity of the 8984-eV transition varies with Cu(I) geometry, ranging from low for tetrahedral sites to high and well resolved for digonal sites (18). The observed transition is typical of those of trigonal Cu(I) model compounds (Fig. 2A). Mononuclear Cu(I) coordination complexes, however, are usually not stable in aqueous solution, and typically undergo auto-oxidation or disproportionation reactions to give Cu(II) (aqueous) and Cu(solid). In contrast, millimolar concentrations of Cu(I)-Atx1 are stable in air at neutral pH for at least 30 min (16), suggesting that the coordination environment in Atx1 stabilizes the Cu(I) state and suppresses disproportionation.

Figure 2

Normalized Cu XANES (A) and EXAFS (B) spectra. (A) XANES spectra. Solid line, Cu-Atx1; dashed line, [(C6H5)4P]Cu(SC6H5)3(23); dotted line, [N(C3H7)4]Cu(SC10H13)2(24); dashed-dotted line, [Cu(II)(imidazole)4](NO3)2. (B) Fourier transform (FT) ofk 3 weighted EXAFS data for Cu-Atx1, where R is the phase-shifted absorber-scatter distance. (Inset) EXAFS data, wherek is photoelectron wave vector. Thin lines represent data; thick lines are the best fit with two shells with 2S at 2.25 Å (σ2 = 3 × 10−3Å2) and 1S at 2.40 Å (σ2 = 5 × 10−3 to 6 × 10−3Å2).

Proteins that stabilize Cu(I) form either polynuclear metal thiolate clusters, as in metallothionein and the transcription factors Ace1, CUP2, and AMT (19), or a constrained His2Cys coordination environment, as in blue copper proteins (20). In contrast, extended x-ray absorption fine structure (EXAFS) measurements indicated that Atx1 stabilization of Cu(I) is achieved in a mononuclear site through an all-sulfur coordination environment (Fig. 2B). The high intensity and relatively high frequency of the EXAFS oscillations were typical of those observed for Cu-S interactions. The data could be modeled with a single three-sulfur (3S) shell with a Cu-S distance of 2.26 Å; however, the Debye-Waller factor, σ2, was somewhat larger than expected (7 × 10−3 Å2). This Cu-S distance is typical of three-coordinate Cu (21-23) and is 0.1 Å longer than that in two-coordinate Cu-thiolate complexes (24). Given that Atx1 contains only two conserved cysteine residues, the third ligand can be either a low- Z (atomic number) ligand (O or N), a methionine sulfur, or an exogenous thiol. A 2S + 1S fit with a shell of two sulfurs at 2.25 Å and a shell of one sulfur at 2.40 Å reproduced the data better than did the single-shell fit, and gave more reasonable σ2 values (3 × 10−3 and 5 × 10−3 to 6 × 10−3 Å2, respectively). Furthermore, attempts to model the data with a shell of 2S and a low- Z shell (1N) gave chemically unrealistic Debye-Waller factors for the N shell (σ2 ≤ 0).

The 2S + 1S result is unexpected for several reasons. The most common coordination environment for Cu(I) complexes is distorted four-coordinate, and there are few precedents among mononuclear Cu(I) complexes for the low–coordination number environment in Atx1 (23, 24). Furthermore, no other native Cu-proteins are known to coordinate a mononuclear Cu center through two cysteines, regardless of the absence or presence of other ligands.

XAS and multinuclear nuclear magnetic resonance (NMR) spectroscopy revealed that the protein conformation does not impose a rigid three-coordinate geometry on the metal ion: Atx1 will also accommodate lower coordination numbers. In these experiments, we used Hg(II), a metal ion with a d 10 electronic configuration that exhibits several similarities to Cu(I). The results were consistent with Hg(II) coordination by two Cys thiolates in Hg-Atx1. A 199Hg single broad resonance was observed at −821 parts per million (ppm) for Hg-Atx1 (Fig.3) (25). This chemical shift is consistent with either a linear dithiolate arrangement, which typically exhibits resonances from −790 to −1000 ppm (26), or a blue copper environment (type I site), consisting of two His, one Met, and one Cys, which have chemical shifts that range from −700 to −890 ppm (27, 28). Given that Atx1 contains only one His, a canonical type I site is not possible. The dithiolate coordination environment for Hg-Atx1 is supported by three additional lines of reasoning: (i) 1H{199Hg} heteronuclear multiple-quantum coherence (HMQC) spectra of Hg-Atx1 revealed coupling only to the Cys β-methylene protons. No coupling was observed for either Met or His protons (29). (ii) The line width in Hg-Atx1 was greater than that observed for Hg-substituted blue copper proteins, which is consistent with a center of lower symmetry. And (iii), EXAFS data for Hg-Atx1 exhibited a single peak that was well modeled by a single shell of two sulfurs at 2.34 Å. Fits with a higher coordination number were poor. Thus, the Hg–S bond lengths, refined coordination number, and chemical shift are typical of two-coordinate mercury thiolate centers in model compounds and Hg-peptides (30). On the basis of the ready accessibility of the two-coordinate state, the similarities between Cu(I) and Hg(II), and coordination chemistry principles delineated below, rapid rates of metal ion transfer from Atx1 to a Cu acceptor protein such as Ccc2 are anticipated.

Figure 3

NMR spectrum of a 2.4-mM solution of199Hg-Atx1. A single resonance at −821 ppm [relative to Hg(CH3)2] and a ν1/2 (line width at half peak height) of 2100 Hz were observed. Experiments were performed on a Bruker 600-MHz spectrometer with a 5-mm tunable broadband probe. The spectrum was collected at 20°C with 110,000 scans, a 90° pulse angle (determined to be 12.5 μs), a sweep width of 83,333 Hz, and a relaxation delay of 150 ms.

Atx1 binds Cu(I) even in the presence of a 20-fold excess of competing thiols (16). This high-affinity Cu binding is consistent with hard-soft acid-base principles and the chelate effect. The strength of the chemical bonds between the soft sulfur atoms and highly polarizable d10 centers such as Cu(I) and Hg(II) is typically so great that it is difficult to measure, which has led to the belief that Cu(I) exchange out of proteins with thiolate-rich coordination sites must be slow (31). In instances in which the metal center can adopt a low coordination number and an incoming ligand forms a bond of similar strength to those of the other metal ligands, however, a facile associative mechanism can operate. A three-coordinate intermediate forms that can subsequently lose one of the original ligands to give a new two-coordinate center. When all three metal–ligand bonds are of similar bond energy, rapid ligand exchange is typically observed. An example of thermodynamically tight binding with a rapid rate of ligand exchange is provided by the chemistry of Hg(II) thiolate complexes; the formation constant for Hg(SR)2 is 1040 and the rate of thiol exchange is almost diffusion controlled (26,32).

To test this direct Cu transfer mechanism, we used two hybrid experiments with Atx1 and Ccc2. The entire 73–amino acid coding region of Atx1 was fused in frame to that of the DNA-binding domain of Gal4, whereas the coding region for the Gal4 activation domain was fused to those for a series of segments of the Ccc2 polypeptide (33). Ccc2 is predicted to possess four cytoplasmic domains interspersed by eight membrane-spanning domains (5). The most NH2-terminal of these cytoplasmic domains (Ccc2-a) contains two MTCXXC Cu-binding motifs (5). Enhanced β-galactosidase (β-Gal) activity was observed only in yeast cells expressing separate fusions of Atx1 and Ccc2-a (Fig.4). These results indicate that Atx1 directly interacts with the NH2-terminal region of Ccc2. The Cu-specific chelator bathocuproine disulfonic acid (BCS) inhibited β-Gal activity only in the strains expressing Ccc2-a and Atx1, indicating that the interaction of Atx1 with Ccc2 is dependent on Cu ions. These data are consistent with a Cu exchange mechanism in which Atx1 directly contacts the homologous domains in Ccc2 and donates Cu(I) through a metal-bridged intermediate.

Figure 4

Two-hybrid analysis of the interaction of Atx1 with Ccc2. Yeast strain SFY526 transformed with vectors based on pGBT9 (Gal4 DNA-binding domain) and pGAD424 (Gal4 activation domain) was subjected to measurements of β-Gal activity (absorbance at 420 nm). Where indicated, cells were grown in the presence of the Cu-chelator BSC. Data are shown for two independent transformants. Plasmids: Atx1, pPS002 (ATX1 in pGBT9); Ccc2-a, pPS001 (CCC2 DNA encoding amino acids 1 to 248 in pGAD424); Ccc2-b, pCF12 (CCC2 DNA encoding amino acids 388 to 527 in pGAD424); Ccc2-c, pCF34 (CCC2 DNA encoding amino acids 594 to 903 in pGAD424); Ccc2-d, pCF56 (CCC2 DNA encoding amino acids 994 to 1004 in pGAD424).

On the basis of the Cu(I)-Atx1 coordination chemistry, the Cu-dependent interaction of Atx1 with Ccc2-a, and first principles, we propose a chemical exchange mechanism and one branch of the Cu-trafficking pathways in yeast (Fig.5). After Cu import into the cell by Ctr1, Cu(I) binds to apo-Atx1 and the complex diffuses throughout the cytoplasm. Cu(I) is coordinated to the two Cys residues of Atx1 and is weakly bound to a third sulfur atom. The latter could be from an adjacent Atx1 molecule, a buffer thiol, or a Met side chain. The Met (M13) can be mutated to Leu without loss of Atx1 activity in vivo, indicating that the latter possibility is unlikely. The distant third sulfur atom is predicted to be in rapid equilibrium between the bound and unbound state. On encountering its target, the Ccc2 molecule, Cu-Atx1 associates with the cytosolic Atx1-like domains. Such an association may be stabilized by a Cu-induced protein-protein interaction or through a Cu(I) bridge between Atx1 and Ccc2-a (or both). The Cu(I) then undergoes a series of rapid associative exchange reactions involving well-precedented two- and three-coordinate intermediates, moving from Atx1 to Ccc2-a [Fig. 5(inset)]. This mechanism provides an avenue for a diffusion-driven movement of the Cu(I) from one site in the cell to another on the basis of small differences in the binding constants of each domain for Cu(I). Such a flow is basically a vectorial process with the Cu(I) ion being directly and rapidly passed down a local thermodynamic gradient, from Cu(I) binding sites of lower affinity to higher affinity sites, without being released to the cytoplasm. If the free energy changes for each of these steps are coupled, hydrolysis of adenosine triphosphate by Ccc2 may provide the overall driving force by moving Cu(I) ions across the vesicular membrane into a separate compartment, where they are ultimately incorporated into Fet3.

Figure 5

Proposed path for the intracellular transfer of Cu(I) by Atx1. Copper destined for incorporation into the vesicular multicopper oxidase Fet3 requires both Ctr1 and Ccc2. Cytoplasmic Cu(I)-Atx1, but not apo-Atx1, associates with the NH2-terminal domain of Ccc2 (purple) and Cu(I) is transferred to the latter. (Inset) A proposed mechanism for the exchange of Cu(I) involving two- and three-coordinate Cu-bridged intermediates. The human homologs of Atx1 (Hah1), Ccc2 (Menkes and Wilson proteins) and Fet3 (ceruloplasmin) are likely to employ similar mechanisms.

By delivering essential cofactors or substrates to apoenzymes, the emerging class of metal ion chaperones facilitates formation of an active state of a protein. These chaperones guide metal ions to their appropriate biological partners and protect them from being trapped at adventitious sites. They may also protect cellular components by sequestering specific ions or inorganic clusters (34) and preventing adverse reactions. Our results underscore the idea that cells make use of elaborate machinery for recruiting, trafficking, compartmentalizing, and, ultimately, inserting into the appropriate enzyme reactive cofactors such as mononuclear Cu ions. The intracellular activation of apo-metalloenzymes by binding of the correct metal ion cofactors is unlikely to proceed by spontaneous self-assembly. Rather, metal insertion is emerging as an orchestrated event controlled by metal ion transport and chaperone proteins whose functions are kinetically and thermodynamically coupled.

Note added in proof: The three-dimensional structures of Hg-MerP (35) and MNK4 (36) have recently been determined by NMR methods. The MNK4 and MerP proteins exhibit the same protein fold, supporting the alignment in Fig. 1. The 199Hg-NMR chemical shift of Hg-MerP (−816 ppm) corresponds to a linear S-Hg-S environment, further corroborating the assignment for Hg-Atx1. In addition, two other Cu chaperone proteins have been found in yeast and humans (37). Lys7 from S. cerevisiae, and Ccs from H. sapiens are required for incorporation of Cu into superoxide dismutase. Both proteins contain a single MXCXXC motif.

  • * To whom correspondence should be addressed. E-mail: t-ohalloran{at}


View Abstract

Navigate This Article