Bacterial cytosolic proteins with a high capacity for Cu(I) that protect against copper toxicity

Bacteria are thought to avoid using the essential metal ion copper in their cytosol due to its toxicity. Herein we characterize Csp3, the cytosolic member of a new family of bacterial copper storage proteins from Methylosinus trichosporium OB3b and Bacillus subtilis. These tetrameric proteins possess a large number of Cys residues that point into the cores of their four-helix bundle monomers. The Csp3 tetramers can bind a maximum of approximately 80 Cu(I) ions, mainly via thiolate groups, with average affinities in the (1–2) × 1017 M−1 range. Cu(I) removal from these Csp3s by higher affinity potential physiological partners and small-molecule ligands is very slow, which is unexpected for a metal-storage protein. In vivo data demonstrate that Csp3s prevent toxicity caused by the presence of excess copper. Furthermore, bacteria expressing Csp3 accumulate copper and are able to safely maintain large quantities of this metal ion in their cytosol. This suggests a requirement for storing copper in this compartment of Csp3-producing bacteria.

Expression and purification of the Csp3 proteins. Escherichia coli BL21 (DE3) transformed with pET29a-MtCsp3 was grown in LB media at 37 ºC (100 μg/ml kanamycin) until an OD 600 of ~0.6 when cells were induced with either 0.1 or 1 mM isopropyl β-D-thiogalactopyranoside (IPTG), harvested after 24 h and stored at -30 ºC. The protein was purified as described previously for MtCsp1 1 , except that MtCsp3 fractions identified by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) from the first anion-exchange column were combined, diluted 10-fold in 10 mM tris(hydroxymethyl)aminomethane (Tris) pH 7.0 plus 1 mM dithiothreitol (DTT), and applied to either a HiTrap Q FF or a HP anion-exchange column (5 ml, GE Healthcare) equilibrated in the same buffer and eluted using a linear NaCl gradient (0-200 mM, total volume ~200 ml).
Essentially pure MtCsp3 could be obtained from this column and was used for crystallisation samples and some in vitro experiments. The protein at this stage can contain a very small amount of a heme-containing protein and further purification was achieved on a Superdex 75 10/300 GL column (GE Healthcare) as described previously 1 . Purified MtCsp3 contained no copper or zinc determined (n = 6) by atomic absorption spectroscopy (AAS, see below), and has a mass of 14524.6 Da determined by Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS), consistent with the recombinant protein having Met1 and His133 at its N-and C-termini respectively (calculated mass 14524.7 Da).
BsCsp3 was over-expressed in E. coli BL21 (DE3) transformed with pET29a-BsCsp3 as described above for MtCsp3, using 0.1 mM IPTG for induction. Pellets were resuspended in 20 mM Tris pH 8.0, sonicated, and centrifuged at 40000 g for 25 min. The supernatant was diluted 5-fold in 20 mM Tris pH 8.0, loaded onto two connected HiTrap Q FF anion-exchange columns (5 ml) equilibrated in the same buffer, and proteins eluted with a linear NaCl gradient (0 to 400 mM, total volume ~240 ml). The BsCsp3-containing fractions were identified by SDS-PAGE, combined, diluted 6 to 8-fold in 20 mM Tris pH 8.0 and loaded onto two connected HiTrap Q FF columns (5 ml). Proteins were eluted using a linear NaCl gradient (0 to 300 mM, ~240 ml) in 20 mM 2-(Nmorpholino)ethanesulfonic acid (Mes) pH 6.5. The purest fractions, identified by SDS-PAGE, were combined, diluted 10-fold in 20 mM 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (Hepes) pH 7.5 and loaded onto a HiTrap Q HP column (1 ml), equilibrated in the same buffer. BsCsp3 was eluted with 20 mM Hepes pH 7.5 plus 500 mM NaCl and further purified by gel-filtration chromatography on a Superdex 75 10/300 GL column equilibrated in the same buffer. Fractions containing BsCsp3, identified by SDS-PAGE, were combined, concentrated as described above, and the NaCl concentration diluted to 200 mM using 20 mM Hepes pH 7.5. Purified BsCsp3 contained very little copper and typically had <1.3 equivalents of zinc when analyzed by AAS (n = 9). The mass of BsCsp3 determined by FT-ICR-MS is 11838.1 Da, in very good agreement with the predicted value of 11838.7 Da for recombinant protein with Met1 and Ala108 respectively at its Nand C-termini.

Cloning the copZ gene
The copZ gene from B. subtilis 168 was amplified from genomic DNA using the primers 5'-GCGCATATGGAACAAAAAACATTGC-3' (forward, NdeI restriction site in bold) and 5'-GCGCCATGGTCACTTGGCTACGTCATAGC-3' (reverse, NcoI restriction site in bold, stop codon underlined) 2 . Fragments were cloned into pGEM-T, verified by sequencing, and sub-cloned into the NdeI and NcoI sites to give pET29a-BsCopZ.

Expression and purification of BsCopZ
BsCopZ was over-expressed and purified as described previously 2 with some modifications. E. coli BL21 (DE3) transformed with pET29a-BsCopZ was grown in LB media at 37 ºC (100 μg/ml kanamycin) until an OD 600 of ~0.6 when cells were induced with 1 mM IPTG, harvested after 5 h and stored at -30 ºC. Pellets were resuspended in 20 mM Tris pH 7.5, sonicated, and centrifuged at 40000 g for 25 min. The supernatant was diluted 5-fold in 20 mM Tris pH 7.5, loaded onto two connected HiTrap Q FF anion-exchange columns (5 ml) equilibrated in the same buffer, and proteins eluted with a linear NaCl gradient (0 to 400 mM, total volume ~240 ml). The BsCopZcontaining fractions identified by SDS-PAGE were combined, diluted ~20-fold in 20 mM Mes pH 6.0 and loaded onto a HiTrap Q HP column (5 ml). Proteins were eluted using a linear NaCl gradient (0 to 1 M, ~180 ml) in 20 mM Mes pH 6.0. The purest fractions, identified by SDS-PAGE, were combined and thoroughly exchanged by ultrafiltration using an Amicon stirred cell with a 5 kDa molecular weight cut-off membrane into 20 mM Hepes pH 7.5 plus 200 mM NaCl.
Concentrated BsCopZ (~500 µM) was incubated with 10 to 30 mM ethylenediaminetetraacetic acid (EDTA), and further purified on a Superdex 75 10/300 GL column, equilibrated in 20 mM Hepes pH 7.5 plus 200 mM NaCl. BsCopZ fractions were combined, concentrated as described above, and stored at -30 ºC. Purified BsCopZ contained no copper or zinc, and has a mass of 7337.6 Da determined by FT-ICR-MS (the sample was treated with tris(2-carboxyethyl)phosphine prior to mass analysis as the two Cys residues were mainly oxidised in the purified protein), consistent with the calculated mass of 7338.1 Da (Met1 and Lys69 at the N-and C-termini respectively).
Atomic absorption spectroscopy. AAS was carried out on an M Series spectrometer (Thermo Electron Corp.) typically with ten standards containing up to 1.8 ppm copper and 1.0 ppm zinc in 2% HNO 3 using the standard calibration method. after approximately 5 to 10 min (typically monitored for 20 min) when it was assumed that all 18 and 19 Cys residues respectively of unfolded apo-MtCsp3 and apo-BsCsp3 had reacted with DTNB (in the absence of denaturant the Cys residues are much less accessible and the reaction of apo-MtCsp3 and apo-BsCsp3 with DTNB after 20 min is ~7% and ~12% respectively of that with urea and guanidine hydrochloride present). An extinction coefficient of 14150 M -1 cm -1 was used for the DTNB reaction in urea 1,4 , whilst a value of 13000 M -1 cm -1 was used in the presence of guanidine hydrochloride 4 . The latter was verified using L-glutathione in 20 mM Hepes pH 7.5 plus 200 mM pH 7.5 with 1 mM EDTA and 6.5 M guanidine hydrochloride (n = 7). Apo-Csp3s (~4 to 20 µM) incubated overnight in an anaerobic chamber (Belle Technology, [O 2 ] <<2 ppm) with DTT (~0.5 to 6.5 mM) in 20 mM Hepes pH 7.5 plus 200 mM NaCl and subsequently desalted on a PD10 column (GE Healthcare), sometimes twice, particularly at high DTT concentrations, were also quantified using DTNB. Some samples that had been treated with DTT were re-quantified at various times after exposure to air at room temperature (MtCsp3 samples left in air for up to 8 days (n = 3), and up to 21 days for BsCsp3 (n = 3)). Apo-Csp3 concentrations were also determined with the Bradford assay (Coomassie Plus protein assay kit, Thermo Scientific) using BSA standards (0 to 1000 µg/ml), allowing a Bradford:DTNB concentration ratio to be determined. Treatment with DTT and exposure to air had no effect on the Bradford:DTNB ratio for both apo-MtCsp3 and apo-BsCsp3. DTT treatment was therefore excluded, particularly as contamination of apo-Csp3 with trace amounts of DTT would influence quantification using DTNB; the preferred approach due to its precision and sensitivity.
BsCopZ has only two Cys residues in its C 13 XXC 16 motif and the purified protein was largely oxidised. To fully reduce apo-BsCopZ, the protein (35 to 350 µM) in 20 mM Hepes pH 7.5 plus 200 mM NaCl was incubated with DTT (4 to 20 mM) overnight in the anaerobic chamber and desalted twice on a PD10 column. The reduced protein was subsequently quantified using the DTNB assay as described above but in the absence of denaturant. Far-UV circular dichroism spectroscopy. Far-UV CD spectra (180 to 250 nm) were recorded using a JASCO J-810 spectrometer as described previously 8,11 . Spectra of apo-MtCsp3 (17.9 to 39.9 µM, 0.26 to 0.58 mg/ml) and protein (22.7  For both proteins stability and unfolding samples analyzed by far-UV CD were in 20 mM Hepes pH 7.5 plus 200 mM NaCl and were prepared and subsequently incubated in the anaerobic chamber. The α-helical content was determined using the mean residue ellipticity at 222 nm 12 .

Investigating Cu(I) binding and the preparation of
Analytical gel-filtration chromatography of the Csp3 proteins. Analytical gel-filtration chromatography of apo-Csp3 (~3 to 140 µM) and protein (~2 to 110 µM) to which ~16 to 21 equivalents of Cu(I) were added was performed on a Superdex 75 10/300 GL column. The elution buffer was 20 mM Hepes pH 7.5 plus either 200 (MtCsp3) or 500 (BsCsp3) mM NaCl (all samples loaded in 20 mM Hepes pH 7.5 plus 200 mM NaCl) and in some cases this was thoroughly degassed and purged with nitrogen throughout the experiment 1,5 . The column was calibrated at the two different NaCl concentrations used allowing apparent molecular weights to be calculated from elution volumes for apo-and Cu(I)-Csp3s. Experiments with apo-Csp3 were also performed on samples (~4 to 100 µM) incubated overnight in the anaerobic chamber with DTT (~40 to 120 µM protein and ~4 to 9 mM DTT), with chromatography performed in the same buffers as above plus 1 mM DTT. Injection volumes were 100 µl, the flow rate was 0.8 ml/min and absorbance was monitored at 240 nm. Diffraction-quality crystals were obtained by mixing 1 µl protein with 1 µl 200 mM MgCl 2 , 100 mM Hepes pH 7.5, and 36% PEG 400 (500 µl well volume) and were directly frozen.
All crystallographic data were collected at Diamond Light Source Ltd, UK, beamline I02. Apo-MtCsp3 data were integrated with MOSFLM 13 . Apo-BsCsp3 and Cu(I)-MtCsp3 data were integrated with XDS 14 . All data were scaled with Aimless 15 and space group determination was confirmed with Pointless 16 . The structure of apo-MtCsp3 was solved by molecular replacement using Molrep implemented via the CCP4 suite 17 with PDB file 3KAW (apo-Csp3 from Pseudomonas aeruginosa, deposited in the PDB but functionally uncharacterized) as the search model. The structures of Cu(I)-MtCsp3 and apo-BsCsp3 were solved by molecular replacement using Molrep with apo-MtCsp3 as the search model. The initial solution for apo-BsCsp3 was improved using the automated model building program ARP/wARP 18 . All models underwent cycles of model building in Coot 19 and refinement in REFMAC5 20 . Five percent of observations were used to monitor refinement. All models were validated using MolProbity 21 and data collection statistics and refinement details are reported in Supplementary Table S1.
Studies using the copA deletion strain of E. coli. E. coli BW25113 and the strain in which the CopA-encoding gene has been inactivated through allelic replacement (herein ∆copA) 22 were obtained from the Coli Genetics Stock Center. The published phenotype for ∆copA relative the wild type (WT) strain 22 was confirmed using overnight cultures grown in LB (plus 50 µg/ml kanamycin for ∆copA) and diluted 100-fold in LB and LB plus copper nitrate (0.5 to 2.0 mM) at 37 ºC (agitation 250 rpm) and measuring the OD at 600 nm at regular intervals up to 6 h (see Fig. 6a). WT   Supplementary Fig. S11 were tested by PCR with the primers used to clone BsCsp3 and also using the following primers; 5'-CATTCATGACAGTGCGACG-3' and 5'-CACAAGAGGACTGGACGC-3', designed to hybridise ~ 300 bp upstream and downstream of the Csp3 gene giving a PCR fragment of 967 bp for WT and longer for the ∆csp3 strain (~1.5 kbp).  Supplementary Figure S1. Multiple sequence alignment of Csp3 from a range of bacteria produced using T-coffee 23 . Asterisks indicate fully conserved sequence positions; the ':' and '.' symbols indicate strongly and weakly similar sequence positions respectively. Cys residues are highlighted in yellow, and the residues corresponding to Asn58 (green), His104, His108 and His110

Supplementary References
(cyan) in MtCsp3 are also highlighted. The organism labels are coloured according to phylum:  Supplementary Figure S4. Structure comparison of the Csp3s. Overlay of the C α atoms (rmsd 1.09 Å for 103 aligned C α atoms) for apo-MtCsp3 (cyan) and apo-BsCsp3 (red). α N (and the loop to α1) is missing in BsCsp3, and the loop between α2 and α3 is significantly shorter than in MtCsp3 (3 rather than 7 residues). There are various degrees of completeness (disorder) in the 6 chains in the asymmetric unit of the apo-BsCsp3 structure, with chain F missing 5 residues whilst 18 amino acids could not be modelled in chain D. The regions affected are the N-terminus (up to 3 residues), the loop between α1 and α2 (absent in chain D), and the loop between α3 and α4 (also absent in chain D). Therefore chain F (missing the first 3 residues plus Ser30 and Val31 from the loop between α1 and α2) has been used for the overlay shown above, and the tetramer in Fig. 1b consists of chains E and F and two symmetry related monomers. The side chains of Cys38 and Cys81 appear in alternate conformations in all chains except E (no evidence of disulfide) and D (complete disulfide formation), with ~30% disulfide estimated in the other four chains (in Fig. 1b the main conformation of Cys38 is shown for chain F). Cys81 is close to the C-terminus of α3 and at the end of the molecule where Cys residues are potentially more solvent exposed. Apo-BsCsp3 was crystallised aerobically over approximately 3 weeks (the Bradford:DTNB ratio did not change over this time for a sample in solution treated with DTT and then incubated in air) in a relatively high concentration of ammonium sulfate at pH 4, which may cause partial unfolding at this end of the molecule. This is consistent with the observation of 100% Cys38-Cys81 disulfide in chain D in which the loop between Lys83 and His88, and Cys89 and Gln90, which are part of α4 in other chains, is disordered.