Visualizing Biological Copper Storage: The Importance of Thiolate‐Coordinated Tetranuclear Clusters

Abstract Bacteria possess cytosolic proteins (Csp3s) capable of binding large quantities of copper and preventing toxicity. Crystal structures of a Csp3 plus increasing amounts of CuI provide atomic‐level information about how a storage protein loads with metal ions. Many more sites are occupied than CuI equiv added, with binding by twelve central sites dominating. These can form [Cu4(S‐Cys)4] intermediates leading to [Cu4(S‐Cys)5]−, [Cu4(S‐Cys)6]2−, and [Cu4(S‐Cys)5(O‐Asn)]− clusters. Construction of the five CuI sites at the opening of the bundle lags behind the main core, and the two least accessible sites at the opposite end of the bundle are occupied last. Facile CuI cluster formation, reminiscent of that for inorganic complexes with organothiolate ligands, is largely avoided in biology but is used by proteins that store copper in the cytosol of prokaryotes and eukaryotes, where this reactivity is also key to toxicity.

Cu(I)-MtCsp3 (~9.8, 12 and 13.5 mg/mL by Bradford assay, to which ~2, 9 and 17 eq. of Cu(I) respectively had been added) were crystallized using the sitting drop method of vapor diffusion.
Cu(I)-protein samples were removed from the anaerobic chamber to use a crystallization robot, and trays were transferred back to the chamber as quickly as possible and sealed (Cu(I)-MtCsp3 exposed to oxygen for approximately 20 mins). Diffraction-quality crystals of protein to which 2 and 9 eq. of Cu(I) were added formed from protein (600 nl) mixed with 300 nl of 200 mM MgCl2, 100 mM Hepes pH 7.5, 30% PEG 400 (80 L well volume) and were frozen directly. Diffractionquality crystals of MtCsp3 to which 17 eq. of Cu(I) had been added were obtained using the same condition (100 nl) mixed with 200 nl of protein (80 l well volume) and were frozen directly.

Data Collection, Structure Solution and Refinement
All crystallographic data were collected at Diamond Light Source Ltd, UK, beamline I02. Data were integrated either with iMosflm or XDS, [6,7] scaled with Aimless [8] and space group determination was confirmed with Pointless. [9] Structures were solved by molecular replacement using Molrep implemented via the CCP4 suite [10] with apo-MtCsp3 (5ARM) [2] as the search model. [11] and refinement in REFMAC5 [12] were performed. Occupancies of copper sites were adjusted manually in 5% increments based on observed peaks in difference maps. The agreement between total copper occupancies in structures and the number of Cu(I) eq. added is within the errors of determinations using these two approaches. Five percent of observations were used to monitor refinement. All models were validated using MolProbity [13] and data collection statistics and refinement details are reported in Table S5.

Importance of the Structures for the in vitro Cu(I) Binding Properties of Csps
Upon the addition of Cu(I), MtCsp3 gives rise to fluorescence at around 600 nm, [2] which reaches a maximum at ~9-11 eq.. This emission must be due to the three buried [14][15][16][17] thiolate-coordinated tetranuclear clusters ( Figure 2). Further Cu(I) binding causes emission to decrease, probably due to the population of the solvent accessible sites [14][15][16][17] near the mouth of the bundle (Figure 3).
Tetranuclear Cu(I) clusters like those seen in MtCsp3 are not present in MtCsp1, [1] due to the absence ( Figure S3) of at least one of the Cys ligands required (MtCsp1 has 13 Cys residues compared to 18 in MtCsp3 and only binds 13 Cu(I) ions). Thus MtCsp1 exhibits little emission at 600 nm upon Cu(I) binding. [1] The relatively independent formation in MtCsp3 of the three tetranuclear Cu(I) clusters, the Cu15-Cu19 arrangement at the mouth of the bundle and the final two Cu(I) sites (Cu1 and Cu2), results in overall non-cooperative Cu(I) binding (cooperativity within the Cu3-Cu14 core is possible). [2] A more disordered core-formation mechanism is likely to be operative in MtCsp1, with Cu(I) ions accessing a greater proportion of sites within the core (initial NMR studies indicate this to be the case), giving rise to overall positive cooperativity. [1] These different core-formation, and presumably also release, mechanisms must contribute to the very different Cu(I) removal kinetics for MtCsp3 and MtCsp1. For example, a large excess of the high affinity ligand BCS can completely strip MtCsp1 of Cu(I) in ~30 min, [1] whereas this process is much slower for MtCsp3 under identical conditions, occurring over days. [2]   Detailed structural information about the sites is provided in Tables S1-S3.         [13]