|Home | About | Journals | Submit | Contact Us | Français|
Tetrathiomolybdate (TM) is an orally active agent for treatment of disorders of copper metabolism. Here we describe how TM inhibits proteins that regulate copper physiology. Crystallographic results reveal that the surprising stability of the drug complex with the metallochaperone Atx1 arises from formation of a sulfur-bridged copper-molybdenum cluster reminiscent of those found in molybdenum and iron sulfur proteins. Spectroscopic studies indicate that this cluster is stable in solution and corresponds to physiological clusters isolated from TM-treated Wilson’s disease animal models. Finally, mechanistic studies show that the drug-metallochaperone inhibits metal transfer functions between copper-trafficking proteins. The results are consistent with a model wherein TM can directly and reversibly down-regulate copper delivery to secreted metalloenzymes and suggest that proteins involved in metal regulation might be fruitful drug targets.
Excess dietary molybdate (MoO42−) uptake was first linked to a fatal disorder in cattle known as “teart” pastures syndrome (1) and later to a neurological disorder in sheep known as “swayback” (2). Both disorders arise from Mo-induced copper deficiency, and the symptoms are readily reversed with copper supplementation. Although molybdate itself has little or no affinity for copper ions, the active copper-depleting agent, TM (MoS42−), is formed in the ruminants’ digestive track and readily reacts with CuI or CuII to form insoluble compounds. These zoogenic studies inspired the development of molybdenum compounds to treat copper-dependent diseases in humans (3). The potent chelating and antiangiogenic activities of orally active formulations of TM, such as the ammonium salt [(NH4)2(MoS4)] (4–6) and the choline salt (ATN-224) (7, 8), have been used in treatment of Wilson’s disease, where copper accumulation leads to hepatic and neurological disorders, as well as in the inhibition of metastatic cancer progression in a number of clinical trials (9–11). TM inhibits several copper enzymes, including ceruloplasmin (Cp), ascorbate oxidase, cytochrome oxidase, superoxide dismutase (SOD1), tyrosinase, and the Enterococcus hirae adenosine triphosphatase (ATPase) (CopB) (12, 13), and also down-regulates the expression of cytokines, such as the vascular endothelial growth factor, as well as transcription factors, such as nuclear factor κB, involved in angiogenesis signaling pathways (14, 15). Although TM can bind to Cu-Cp (12), copper–bovine serum albumin (Cu-BSA) (16), and Cu-containing metallothioneins (Cu-MT) (17) and has been proposed to inhibit SOD1 by partially removing copper from the enzyme (8, 18), the reaction chemistry and structures of these complexes have not been resolved.
Metallochaperones constitute a particular kind of protein that delivers metal ions to specific cytoplasmatic targets in the cell (19). The prototypical metallochaperone, yeast Atx1, transfers CuI along a trafficking pathway via electrostatic interactions with structurally homologous N-terminal domains of the ATPase, Ccc2 (20, 21). Likewise, the closely related human copper metallochaperone, antioxidant 1 (Atox1), can transfer copper to N-terminal domains of the copper-transporting ATPases 7a and 7b, also known as the Menkes and Wilson disease proteins. All three of these proteins are important in mammalian copper homeostasis and provide copper to secreted enzymes that are important in vascular integrity such as Cp and extracellular SOD (ecSOD). We anticipated that TM would readily remove CuI from its binding site in Atx1 with subsequent formation of a typical polymeric CuMo sulfide precipitate. We found instead a robust TM-metallochaperone complex with metal sulfur ratios reminiscent of the FeMo cofactor complex in nitrogenase (22) and elucidated how this anti-angiogenic drug affects the structure and function of a canonical metal-trafficking domain.
Direct reaction of TM with Cu-Atx1 leads to rapid formation of an air-stable purple complex that can be readily isolated by size-exclusion chromatography (23). Crystals of this complex diffract to 2.3 Å (fig. S1), and the x-ray structure reveals the presence of 12 Cu-Atx1 molecules in the asymmetric unit arranged as four TM-Cu-Atx1 noncrystallographic trimers (fig. S2). The overall structure of each Atx1 monomer is similar to previously determined structures, retaining the “ferredoxin-like” βαββαβ fold (24), with two cysteines involved in copper binding (Cys15 and Cys18) located at the protein surface. Superposition of the coordinates of Hg-Atx1 (PDB code 1CC8) (24) and Cu-Atox1 (human analog of Atx1, PDB code 1FEE) (25) on the monomers in the complex (Fig. 1, C and D) reveals that the peptide fold around the metal-binding loop is unperturbed by TM binding, with an average root mean square deviation for the Cα atoms of ~0.67 Å (Hg-Atx1) and ~1.3 Å (Cu-Atox1). In the structure, each Atx1 trimer coordinates four copper atoms and one TM molecule, with the stoichiometry [TM][(Cu)(Cu-Atx1)3], which is corroborated by independent elemental analysis of the complex (23). The Cu x-ray absorption near-edge structure of the complex indicates that the copper remains in the CuI oxidation state, whereas the Mo K near-edge spectrum strongly resembles that of tetrathiomolybdate (MoVI) (fig. S3). Aside from a few H-bonding interactions between monomers, the dominant forces stabilizing the trimer are the coordinate covalent bonds between the protein CysS atoms and the metal cluster.
A “nest-shaped” copper-molybdenum cluster, unprecedented in metalloproteins, is located at the center of the Atx1 trimer (Fig. 1, A and B) on the threefold axis. The cluster consists of four CuI ions, [MoS4]2−, and three pairs of Atx1 CysS atoms to give a [S6Cu4MoS4] cluster (Fig. 2). The Mo atom remains tetrahedrally coordinated by four sulfide ions with Mo–S distances in the range 2.18 to 2.26 Å (mean: 2.22 Å), as expected for Cu–S–Mo cluster interactions and commensurate with the ones observed in the parent drug (2.17 to 2.20 Å, mean: 2.19 Å) (7). Three of the copper atoms bind to the sulfur atoms of cysteines 15 and 18, and each of these atoms also binds two sulfides from [MoS4]2−, which results in a distorted tetrahedral coordination environment for the coppers with similar distances for the Cu-S bonds to protein side chains (2.21 to 2.44 Å, mean: 2.30 Å) or the sulfides of TM (2.24 to 2.40 Å, mean: 2.29 Å). The Mo-Cu distances are in the range of 2.74 to 2.82 Å (mean: 2.77 Å). The fourth sulfide of TM does not coordinate copper or interact with protein. On the other side of the complex, the fourth copper atom is bound by three (Cys15)Sγ atoms (2.22 to 2.30 Å, mean: 2.26 Å) and exhibits a trigonal planar coordination. Thus, three of the four sulfide ions in TM form a μ3-S bridge between the Mo atom and two tetrahedral Cu atoms, whereas each of the (Cys15)Sγ atoms of three Atx1 behave as a bridging ligand between one tetrahedral and one trigonal planar copper center. In the tetrahedrally coordinated coppers, the (Cys15)Sγ–Cu–Sγ(Cys18) bond angles are larger (118° to 125°, mean: 122°) than the (TM)S–Cu–S(TM) bond angles (99° to 103°, mean: 101°), consistent with a distorted tetrahedral site. The geometry at the Mo atom is only slightly distorted from tetrahedral, with (TM)(μ3-S)–Mo–(μ3-S)(TM) and (TM)(μ3-S)–Mo–S(TM) bond angles of 103° to 110° (mean: 106°) and 109° to 116° (mean: 112°), respectively. Protein-TM interactions partially neutralize the negative charge delocalized over the [Cu4MoS4]4− cluster (fig. S4 and Fig. 2A). Three positively charged lysines (Lys65), one from each Atx1 monomer, form hydrogen bonds with the sulfides from TM and the thiolates of Cys18. Strong interactions are observed for the only terminal S thiolate ligand in the cluster (Cys18–S–Lys65–Nζ = 3.3 Å) relative to μ3-S bridging sulfide ligands (TM–μ3-S–Lys65–Nζ = 3.8 Å). In addition, H bonds from backbone amides (Thr14 and Gly17) at the amino terminus of α helix 2 to metal-bound thiolates further neutralize the negative charge of the buried cluster.
Although this type of cluster has not been previously reported in metalloproteins, analogous nest-shaped [Cu3MoS3O] inorganic units (with P- and N-donor ligands) are components of larger clusters (26). The closest fragment analog of the protein-drug adduct is a component of the [Bun4N]4[Cu12Mo8S32] complex. Here, a [S6Cu3MoS4] unit exhibits similar cluster framework with Mo–Cu distances from 2.69 to 2.75 Å, Mo–S distances from 2.06 to 2.25 Å and Cu–S distances from 2.29 to 2.36 Å (27) (fig. S5). Another structurally distinct CuSMo center is observed in the Cu-Mo-pterin enzyme carbon monoxide dehydrogenase from Oligotropha carboxidovorans, where a single diagonally coordinated Cu atom is bound via a bridging sulfide to a Mo active site forming a [CuSMo(=O)OH] cluster (28).
To determine whether TM interaction with Atx1 inhibits its copper chaperone activity, we developed a native gel–based copper transfer assay that monitors metal occupancy in a mixture of TM-Cu-Atx1 trimer and Ccc2a, the physiological partner of Atx1 (fig. S6). The assay takes advantage of the fact that apo- and Cu-Atx1 are clearly distinguishable from Ccc2a and TM-Cu-Atx1 in a native agarose gel system (23), where the protein and metal content of the bands are characterized by a variety of analytical techniques to establish the metallation state of each protein (fig. S7 to S12 and table S1). The assay was validated by a combination of electrospray ionization protein mass spectrometry (ESI-MS) and quantitative elemental analysis via inductively coupled plasma MS (ICP-MS) of samples extracted from gel slices, as well as by qualitative laser scanning elemental analysis, that is, laser ablation with ICP-MS (LA-ICP-MS) of the electrophoresis gel itself. Three key lanes are shown in Fig. 3A. The TM-Cu-Atx1(SeMet) migrates as a positive species containing copper and molybdenum (lane I). Mixing of apo-Ccc2a and Cu-Atx1 (SeMet) results in the transfer of copper from Cu-Atx1(SeMet) to Ccc2a (lane II). The transfer of copper from Atx1(SeMet) to Ccc2a is almost completely abolished by the presence of TM (lane III). Both native Atx1 and the SeMet analog give similar results. It is intriguing that protein analysis indicates formation of a new Cu-TM protein complex that contains the Ccc2 domain, as well as TM and Cu-Atx1. The formation of this heteromeric protein complex suggests that other proteins with a surface-exposed MxCxxC copper-binding motif will be able to form similar complexes with TM.
These results suggest a new model for how a drug can disrupt a key protein-protein interaction for metal-trafficking pathways. Support for the physiological occurrence of this type of metal-protein cluster is shown in Fig. 3B by the highly similar Cu and Mo K-edge extended x-ray absorption fine structure analysis of the [TM][(Cu)(Cu-Atx1)3] complex, and a kidney sample extracted from TM-treated LPP rats (animal model of Wilson’s disease), where a similar [(CuSR)3S4Mo]2−-type interaction is proposed (29). The stoichiometry of three chaperone molecules and four copper atoms per drug molecule has several physiological implications. By sequestering multiple copper chaperones and the metal cargo destined for trafficking to the trans-Golgi, TM may suppress Cu incorporation into secreted copper enzymes, including those involved in modification of the vasculature such as ecSOD, copper amine oxidases, lysyl oxidase, and Cp. The TM-mediated sequestration of copper-loaded metallochaperones may perturb other proposed roles of Atox1 in regulation of copper-related tumor angiogenic factors (30).
The structure and biochemistry of the TM-Cu-Atx1 complex also provides chemical insights into the puzzling stoichiometry of the dietary Cu-Mo antagonism (31) and suggests why ternary complex formation between TM and specific Cu proteins can have pronounced physiological consequences (32). A relatively small amount of dietary molybdenum clearly perturbs the timely dissemination of a larger pool of copper in deficiency disorders such as swayback and teart pasture syndrome. Our results raise the possibility that the active agent, TM, functionally suppresses copper trafficking domains that control the secretion of the active forms of copper-dependent enzymes. Finally, our results suggest that proteins involved in such metallation pathways may be targets for the development of new classes of pharmaceutical agents.
This manuscript is dedicated to the memory of E. Stiefel and his contributions to the field of molybdenum sulfide chemistry. This work was supported by grant GM54222 and GM38784 (T.V.O.) and GM38047 (J.E.P.-H.) from the NIH. The Robert H. Lurie Comprehensive Cancer Center provided a Malkin Fellowship (H.M.A.) and support for Structural Biology Facility. Use of the Advanced Photon Source [Structural Biology Center-Collaborative Access Team (CAT) and Industrial Macromolecular Crystallography Association CAT] and the Stanford Synchrotron Radiation Laboratory (SSRL) was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, with additional support (at SSRL) from the National Center for Research Resources, NIH. Use of the Chicago Biomedical Consortium (CBC)–University of Illinois at Chicago Proteomics Facility was supported by The Searle Funds at the CBC, and use of LA-ICP-MS was supported by a National Aeronautics and Space Administration grant to the Quantitative Bioelement Imaging Center in the Chemistry of Life Processes Institute at Northwestern University. We thank P. Focia for assistance with x-ray diffraction collection, M. Clausén for assistance in the early stages of XAS measurements, Y. Wang for assistance with the protein MS, A. Davis for providing apo-Ccc2a, and A. Mazar for helpful discussions. The atomic coordinates have been deposited at the Protein Data Bank with code 3K7R.
Materials and Methods