OprC is a TonB-dependent transporter that binds ionic copper

The structure of OprC, crystallised with an N-terminal His7 purification tag under aerobic conditions in the presence of 2 mM CuCl₂, was solved using single wavelength anomalous dispersion (Cu-SAD), using data to 2.0 Å resolution (Methods; S1 Table, S1 Fig). As indicated by the successful structure solution, OprC contains a single bound copper and shows the typical fold of a TBDT, with a large 22-stranded β-barrel occluded by an N-terminal approximately 15 kDa plug domain that, like in other TBDTs, completely occludes the lumen of the barrel (Fig 1A, 1C, and 1E). The copper binding site comprises residues Cys143 and Met147 in the plug domain and His323 and Met325 in the barrel wall. The CxxxM-HxM configuration, which coordinates the copper in a tetrahedral manner (Fig 1G and 1H), is highly unusual and has, to our knowledge, not been observed before in copper homeostasis proteins. A similar site is present for one of the copper ions of the valence-delocalised Cu_A dimer in cytochrome c oxidase, where the copper ion is coordinated by 2Cys+1Met+1His [30,31]. Other similar sites are class I Type I copper proteins like pseudoazurin and plastocyanin, where copper is coordinated by 2His+1Cys+1Met [32]. Interestingly, and unlike class I Type I copper proteins, concentrated solutions and OprC crystals are colourless in the presence of Cu(II). Another notable feature of the OprC structure becomes apparent when analysing the positions of the methionine residues. As shown in Fig 1J and 1K, out of the 16 visible methionines in OprC, 11 are organised in such a way that they form a distinct “track” leading from the extracellular surface towards the copper binding site. An additional methionine (Met558) is not visible due to loop disorder, but, given their positions, they will be a part of the methionine track. Considering that Cu(II) prefers nitrogen and oxygen as ligands while Cu(I) prefers sulphur, we propose that the methionine track might bind Cu(I) with low affinity and may guide the metal towards the principal binding site, which is at the bottom of the track (Fig 1J). Importantly, the anomalous difference maps of OprC crystallised with Cu(II) do not show any evidence for weaker, secondary copper sites (S1 Fig), demonstrating that the methionine track does not bind Cu(II).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. OprC is a TBDT that binds ionic copper via an unprecedented CxxxM-HxM binding site.

Cartoon representation of (A, C) Cu-loaded OprC and (B, D) Cu-free OprC (OprC_AA). (A, B) are viewed from the OM plane, whereas the views for (C, D) are from the outside of the cell. The N-terminal plug domain is shown separately for both forms (E, F). For Cu-OprC, structures are shown in rainbow from N-terminus (blue) to C-terminus (red); copper is represented as a magenta sphere. Apo-OprC is coloured light cyan, with the plug rendered brown. The arrow in (F) highlights the visibility of the Ton box in apo-OprC. (G) Stick models of copper-coordinating residues Cys143, Met147, Met325, and His323. Electron density in grey mesh (2Fo-Fc map contoured at 2.0σ, carve = 2.0) is shown for the binding site residues C/M-H/M and the copper atom (anomalous difference map shown in magenta, contoured at 3.0σ, carve = 2.25). (H) Distances between coordinating residues and metal show that copper is coordinated via 1 thiolate (from Cys), 2 thioethers (from Met), and 1 imidazole nitrogen from His. (I) Mutation of binding site residues Cys143 and Met147 to alanines in OprC_AA abolishes copper binding (2Fo-Fc map contoured at 2.0σ, carve = 2.0). (J, K) OM plane (J) and extracellular views (K) showing the thioether atoms of all methionine residues in Cu-OprC as yellow spheres. The copper atom, only visible in (J), is shown as a magenta sphere. OM, outer membrane; TBDT, TonB-dependent transporter.

https://doi.org/10.1371/journal.pbio.3001446.g001

Copper binding by OprC is highly specific and near-irreversible

Following structure determination of copper-bound OprC, several attempts were made to produce a structure of copper-free OprC. First, the protein was purified and crystallised without added copper; however, this gave a structure that was identical to the one already obtained and contained bound copper that presumably originated from the LB medium. As expression in rich media always yielded OprC with approximately 45% to 80% copper as judged by ICP-MS, various attempts to lower the copper content were made. Removal of bound copper from purified protein with combinations of denaturants (up to 4.0 M urea) and ethylenediamine tetra-acetic acid (EDTA) were not successful. Expression in minimal medium also yielded copper contents of approximately 45% to 60% (Fig 2A and 2B), but with much lower protein yields compared to rich medium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. OprC binds 1 equivalent copper near-irreversibly.

(A) Metal occupancy of LB-purified WT OprC by ICP-MS shows specific binding only to copper. Each colour indicates individual batches of protein purified from rich media. (B) Copper content of WT OprC and binding site mutant proteins before (blue) and after aerobic incubation with either 3 (pink) or 10 (green) equivalents Cu(II) for 30 min followed by analytical SEC. All proteins contain a N-terminal His7 tag except where stated. (C) Copper is kinetically trapped in OprC. Time course of copper extraction experiments showing % bound copper for OprC_WT (blue), OprC_AA (green), and OprC M147A (red), at RT (open symbols) and 60°C (filled symbols). The inset shows % bound copper in the first few minutes after starting the experiment. OprC_AA served as a control. Dotted lines indicate initial occupancies of OprC_WT and M147A. (D) Comparison of the cw-EPR spectra of OprC_WT and OprC_M147A mutant before (black traces) and after (red traces) addition of 1 equivalent Cu(II) solution. The blue trace shows the EPR spectrum of the Cu(II)SO₄ in Hepes buffer. All EPR spectra have been background subtracted. The double-headed magenta dotted arrows show the difference in the observed g and A tensor of OprC variants. The blue goal posts indicate the ^63,65Cu-hyperfine splitting along the parallel region. Note that the starting copper equivalencies for these proteins were 0.6 (OprC_WT) and 0.1 (OprC_M147A), respectively. (E) EPR time course for OprC_M147A after addition of 1 equivalent Cu (II). (F) Relative intensities of EPR signals at approximately 2,672 G and 3,178 G (black dotted rectangular boxes in the top panel) plotted as a function of time. Values shown are averages from 3 independent time courses. Underlying data for this figure can be found in S1 Data. EPR, electron paramagnetic resonance; ICP-MS, inductively coupled plasma mass spectrometry; RT, room temperature; SEC, size exclusion chromatography; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3001446.g002

Aerobic incubation of purified WT OprC (OprC_WT) with 45% copper occupancy in the presence of either 3 or 10 equivalents Cu(II) followed by size exclusion chromatography (SEC) demonstrate coelution of 1 equivalent copper (Fig 2B). Thus, the His7 tag does not bind Cu(II) with high affinity. Coincubation with 0.5 mM EDTA (approximately 50-fold excess) does not result in copper loading, suggesting that EDTA effectively withholds Cu(II) from OprC (S2 Fig). As-purified OprC does not contain zinc, the most common contaminant in metal-binding proteins, nor does it contain appreciable amounts of any other metals that could have been introduced during purification such as Ni and Fe, indicating that OprC is highly specific for copper (Fig 2A, S2 Fig). Indeed, incubation of purified OprC in the presence of 3 or 10 equivalents Zn does not result in zinc coelution (S2 Fig). To obtain copper-free OprC after purification from rich media, we constructed a variant (OprC_AA) in which the binding site residues Cys143 and Met147 were both mutated to alanines. Even after equilibration of OprC_AA for 30 min with 3 or 10 equivalents Cu(II), no coelution with metal is observed (Fig 2B), indicating that high-affinity copper binding is completely abolished and confirming that the His7 tag does not bind Cu(II) with an affinity high enough to survive SEC, possibly due to the presence of 0.5 mM EDTA in the SEC buffer.

The fact that it is not possible to obtain copper-free WT protein, even after taking extensive precautions, suggests that copper binds to OprC with very high affinity. To explore this further, we performed a copper extraction assay with a large excess of bathocuproinedisulfonic acid (BCS) under reducing conditions (Methods). For copper-loaded OprC_WT, only 20% copper was removed after 24 h at room temperature, and the temperature had to be increased to 60°C to obtain near-quantitative extraction of copper (approximately 90% after 24 h) (Fig 2C). For reasons that are unclear, the orange-coloured [Cu(BCS)₂]⁻³ complex was hard to separate from OprC, and BCS-treated OprC did not bind copper anymore, suggesting an irreversible change in the protein due to the harsh incubation conditions. Nevertheless, these results demonstrate that copper is kinetically trapped inside OprC and is, for all intents and purposes, irreversibly bound. This is fully compatible with the consensus transport mechanism of TBDTs, in which the interaction with TonB, occurring after substrate binding, is required to disrupt the binding site and release the ligand [33].

Conformational changes upon copper binding

The OprC_AA structure was solved by molecular replacement (MR) using Cu-bound OprC_WT as the search model (Fig 1B, 1D, 1F, and 1I). The binding site residues of both structures occupy very similar positions, indicating that the introduced mutations abolish copper binding without generating gross changes in the binding site. Superposition of the structures (S3A Fig and S3B Fig) shows that for the remainder of the protein, structural changes upon copper binding are confined to the vicinity of the copper binding site, with parts far removed virtually unchanged (overall Cα RMSD approximately 1.0 Å). The largest change is observed for loop L11, which undergoes an inward-directed motion of approximately 8.0 Å upon copper binding (S3D, S3E, and S4 Figs). A similar inward-directed but smaller change occurs for loop L8. Some loop tips (e.g., L4, L5, L6) in OprC_AA lack electron density for a limited number of residues, suggesting increased mobility. Overall, the conformational changes of the loops upon copper binding likely decrease the accessibility of the copper binding site. However, the main reason why the bound copper is inaccessible to solvent is that the binding site residues Met147 and Met325, together with Asn145, effectively form a lid on the copper ion in the WT protein. In the double mutant, copper becomes solvent accessible due to the absence of the Met147 side chain (S3D, S3E, and S4 Figs).

The consensus mechanism for TonB-dependent transport postulates that ligand binding on the extracellular side generates conformational changes that are propagated to the periplasmic side of the plug and increase the periplasmic accessibility of the Ton box for subsequent interaction with TonB [34]. In OprC_AA, N-terminal density is visible up to Leu66 (i.e., the first 10 residues of the mature protein are disordered) including the Ton box (⁶⁸PSVVTGV⁷⁵), which is tucked away against the plug domain and the barrel wall. In Cu-OprC, the density between Glu88 and Pro94 is poor and hard to interpret, and, more importantly, no density is observed before Pro79, including the Ton box (Fig 1E and 1F, S3C Fig). Thus, while we cannot say conclusively that the Ton box is accessible to TonB in Cu-OprC, the structures do show that changes occur in the Ton box upon substrate binding. Thus, the structures of OprC in the absence and presence of ligand are consistent with the consensus TBDT mechanism. The observed position of the Ton box in OprC_AA, likely hard to reach from the periplasmic space, would prevent nonproductive interactions of TonB with transporters that do not have substrate bound [34].

The OprC methionine track and the principal binding site bind Cu(I)

We next asked whether OprC also binds Cu(I). Since it is challenging to maintain copper in its +1 state during crystallisation, we used silver (Ag(I)) as a proxy for Cu(I) and determined the co-crystal structure of WT OprC in the presence of 2 mM AgNO₃ (Methods). This is possible because the protein used for crystallisation only had approximately 60% copper occupancy. Data were collected at 8,000 eV, at which energy the anomalous signal of copper is very small (0.6 e⁻, compared to 4.2 e⁻ for Ag). Strikingly, and in sharp contrast to Cu(II) (S1 Fig), the anomalous map of OprC WT crystallised in the presence of silver shows not one but 3 anomalous peaks. The first, very strong peak (Ag1; 23σ) is located at the same site as in OprC crystallised with Cu(II) and is coordinated by the same residues (Cys143, Met147, His323, and Met325; Fig 3A). The other 2 silver sites have lower occupancies (Ag2, approximately 10σ and Ag3, approximately 10σ) and are each coordinated by 3 methionines of the methionine track (Met286, Met339, and Met343 for Ag2; Met341, Met343, and Met348 for Ag3), as seen also, for example, for Cu(I) in the Ctr1 copper transporter [21]. Only 5 Met residues form the 2 sites (Met343 is shared between the two), and this, combined with the fact that the track has 11 or 12 methionines, suggests that there are more than 2 metal binding sites. The distance between Ag2 and Ag3 is approximately 6.5 Å, i.e., large enough for 2 metal ions to be bound simultaneously, consistent with the fact that Ag was present at approximately 10-fold excess during crystallisation (approximately 0.2 mM OprC and 2 mM Ag). Under physiological conditions, however, it seems more likely that only one Ag atom is bound to the track at any one time. We speculate that the sites closer to the irreversible binding site (Ag1) may have a higher affinity, driving metal towards this site. While direct measurement of the affinities of the methionine binding track sites would be challenging, the structural data suggest that the methionine track provides at least 2, and possibly more, binding sites for Ag(I), and, by extension, for Cu(I). Moreover, while the methionine track only binds Cu(I), the high-affinity CxxxM-HxM site likely binds both copper redox states.

Intriguingly, 3 out of the 5 “non-track” methionine residues (Met138 and Met139 in the plug, Met374 in the barrel wall) are located close together and on the same side of the barrel as the methionine track and the principal binding site (Fig 1K). Given that the nature of the TonB-dependent remodelling of the plug is not known, it is unclear whether a potential binding site formed by these residues could be part of the path taken by the metal towards the periplasmic space.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. OprC binds Ag at the high-affinity binding site and the methionine track.

(A-E) Anomalous difference maps of (A) OprC_WT, (B, D) C143A, and (C, E) H323A variants crystallised in the presence of (A, D, E) Ag or (B, C) Cu(II) and collected at different energies. The inset to (A) shows a close-up of the anomalous difference peaks (magenta) near the principal binding site in OprC_WT, with binding residues labelled and represented as stick models. Sulphurs are coloured yellow. For clarity, the metal used in cocrystallisation and the energy used for data collection are shown underneath each panel. The OprC plug domain is coloured blue. WT, wild-type.

https://doi.org/10.1371/journal.pbio.3001446.g003

To obtain more information on the individual residue contribution to copper binding, we next generated the complete set of single alanine mutants of the principal binding site residues (C143A, M147A, H323A, and M325A) and determined copper binding via analytical SEC and ICP-MS. For all single mutants, the copper content after protein purification from LB-grown cells was below 10%, except for M147A (approximately 20%) (Fig 2B). Upon incubation with 3 or 10 equivalents Cu(II), various occupancies were obtained. C143A has no bound copper even after incubation with 10 equivalents Cu(II), suggesting that this residue has a crucial role. H323A (approximately 30%) and, in particular, M147A (approximately 60%), have relatively high occupancies after copper incubation and SEC, indicating that these residues contribute less towards binding. Of the 4 ligands, the M147 thioether is the furthest away from the copper in the crystal structure (Fig 1H), which may explain why it contributes the least to ligand binding. Interestingly, removal of bound copper is much faster in the M147A mutant compared to OprC_WT (Fig 2C), suggesting that solvent exclusion by the M147 side chain (S3D Fig) is the main reason why copper is kinetically trapped in OprC_WT.

To shed additional light on the redox state of the bound copper, continuous wave EPR (cw-EPR) spectra were recorded on OprC_WT. Surprisingly, as-purified OprC_WT containing approximately 0.6 equivalents copper was EPR silent (Fig 2D), demonstrating that the copper species present is Cu(I). The as-purified M147A protein, with approximately 0.1 equivalent copper, was EPR silent as well. We next loaded the M147A mutant with CuSO₄ to 1 equivalent, and EPR spectra were recorded over time. The observed EPR signal is different from the standard CuSO₄ Cu(II) EPR signals, confirming that Cu(II) binds to the protein. The EPR spectra of the OprC-WT and OprC-M147A mutant show nicely resolved ^63,65Cu(II) hyperfine coupling along the parallel region, due to the interaction of an unpaired electron spin (S = ½) of Cu(II) with the nuclear spin of (I = 3/2) of ^63,65Cu nuclei, as indicated by the blue goal post in Fig 2D. Interestingly, the EPR signals decrease slowly upon prolonged incubation, suggesting that added Cu(II) is very slowly reduced to Cu(I) (Fig 2E and 2F). This, together with the possibility that OprC binds Cu(I) directly from the LB media, could be an explanation for the observation that as-purified OprC, expressed under aerobic conditions, contains reduced copper. However, it is clear that the observed reduction of Cu(II) is too slow to be physiologically relevant, obviating the need to find a mechanistic explanation.

Cysteine is essential for high-affinity copper Cu(II) binding

While OprC_WT and most single alanine mutants can be (partly) loaded via Cu(II) incubation, this is not the case for the C143A mutant (Fig 2B). We hypothesised that removal of the cysteine could lead to much lower affinity for Cu(II), so that after SEC, nothing remains bound. To provide support for this, we determined the crystal structures of the OprC C143A mutant cocrystallised with Cu(II) or silver Ag(I). For each crystal, datasets were collected at 8,800 eV and 9,175 eV to distinguish between both metals. Bound copper is expected to give a strong anomalous peak only at 9,175 eV (which is above the copper K edge at 8,979 eV), while bound silver will give comparable peaks at both energies (the silver L-III edge is at 3,351 eV). For C143A cocrystallised with Cu(II), no anomalous peaks are visible at both energies (Fig 3B), showing that Cu(II) binding is indeed abolished. Crucially, in the presence of silver, the same 3 anomalous peaks are visible as for WT OprC (compare Fig 3A and 3D), strongly suggesting that the C143A mutant can still bind Cu(I). Since Cu(II) prefers histidine nitrogen as ligands and Cu(II) binding sites often contain one or more His residues, we also cocrystallised the H323A mutant with Cu(II) and Ag(I). As shown in Fig 3C and 3E, one strong anomalous peak, at the high-affinity binding site, is observed with Cu(II), supporting the SEC data that the histidine is not required for Cu(II) binding. With Ag(I), 2 clear anomalous peaks are observed, suggesting that the H323A mutant can still bind Cu(I) at the principal site and at the methionine track.

OprC mediates copper acquisition in P. aeruginosa

To demonstrate that OprC imports copper, we performed anaerobic growth experiments in P. aeruginosa with added copper. Given that oprC expression is repressed with excess external Cu(II) [26–29], we employed arabinose-inducible overexpression of His-tagged oprC via the broad range pHERD30 plasmid [35]. We complemented the PA14 ΔoprC strain with OprC_WT- and OprC_AA-containing plasmids and performed growth assays in rich media with empty vector as control. S5 Fig shows clear toxicity when OprC_WT is overexpressed, even without Cu(II) addition. Surprisingly, expression of OprC_AA was equally toxic as OprC_WT overexpression, which indicates that the toxicity phenotype is caused by overexpression of OprC per se and is likely not linked to OprC function.

Since copper toxicity assays failed, we decided to determine P. aeruginosa whole-cell metal contents using ICP-MS. We observed no differences in copper content between the WT PA14 and ΔoprC strains in rich media without added copper (S6 Fig), suggesting that OprC is not expressed under these conditions. By contrast, cells expressing OprC_WT from pHERD30 have more associated copper when compared to the empty vector control, under both aerobic and anaerobic conditions (Fig 4A). However, as shown by the toxicity phenotypes presented above, this could also be due to increased leakiness of cells as a result of plasmid-based OMP expression, a possibility that was not taken into account in a recent study [29]. However, cells expressing the OprC_AA inactive mutant have copper levels similar to those of the control. Moreover, OprC_WT and OprC_AA are present at similar levels in the OM (Fig 4B), demonstrating that the different amounts of copper associated with the cells are not due to differences in protein levels. In addition, no substantial differences were detected for other divalent metals, confirming the in vitro experiments that OprC is specific for copper. These data, together with the fact that the OprC structures exhibit all the hallmarks of a bona fide TBDT (Fig 1, S3 Fig), strongly suggest that OprC is a copper importer in P. aeruginosa. Given the relatively modest differences in observed copper contents, cellular fractionation studies to experimentally demonstrate OprC-mediated copper import versus binding will be challenging. In addition, it is not known to which of the 3 P. aeruginosa TonB proteins OprC couples, and how expression of this TonB_OprC is regulated, i.e., it is possible that OprC is fully loaded without copper being imported. It also seems clear that P. aeruginosa can acquire copper in the absence of OprC, either via one of the many OM channels expressed in P. aeruginosa or perhaps via (a) porin-independent pathway(s) as recently shown for antibiotics [36].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Specific acquisition of copper by OprC.

(A) Whole-cell metal content of PA14 ΔoprC cells overexpressing empty vector (white bars) OprC_WT (grey bars) and OprC_AA proteins (black bars) analysed via ICP-MS. Cell associated metal content was determined in cells grown in rich media supplemented with 100 mM sodium nitrate (no added copper) under both aerobic (left panels) and anaerobic conditions (right panels). The 3 biological replicates are plotted separately due to differences in absolute metal levels. Reported values are averages ± SD (n = 3). Significant levels were analysed via unpaired 2-tailed t test. ns, not significant (p ≥ 0.05); *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001. (B) SDS-PAGE gel of pHERD30-overexpressed OprC_WT and OprC_AA proteins in PA14 ΔoprC after IMAC. Masses from the molecular weight marker are shown on the left. Underlying data for this figure can be found in S1 Data. ICP-MS, inductively coupled plasma mass spectrometry; IMAC, immobilised metal affinity chromatography; ns, not significant; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3001446.g004

OprC is abundant in P. aeruginosa during infection

Recent studies have suggested that OprC is important for virulence in P. aeruginosa and Acinetobacter baumannii [37,38]. These papers, however, did not provide a measure of the abundance of OprC during infection in vivo. We therefore decided to determine the abundance of the highly similar OprC proteins (50% sequence identity; S7 Fig) in P. aeruginosa UCBPP-PA14 and A. baumannii ATCC 19606 in infected lung tissues via a sensitive targeted proteomics approach with parallel reaction monitoring (Methods). The results showed that in both mouse and rat pneumonia models, OprC was present at 1,000 to 10,000 molecules per P. aeruginosa cell, making it one of the 5 most abundant TBDTs. As a comparison, the most abundant TBDT, FpvA (involved in iron-siderophore uptake), had 8,000 to 33,000 molecules per cell. In A. baumannii, OprC was less abundant in mouse and rat pneumonia models (40 to 400 molecules per cell), while the most abundant TBDTs BfnH and BauA were present at 500 to 3,000 molecules per cell.

Recombinant production of Pseudomonas aeruginosa OprC

The mature version of the gene coding for oprC of P. aeruginosa PAO1 (UniProt ID; PA3790) [59], starting with His56 as determined by Nakae and colleagues [26], was synthesised to include a 7 × His tag at the N-terminus (Eurofins, United Kingdom), cloned into the pB22 arabinose-inducible expression vector [60] and transformed into chemically competent E. coli DH5α cells. After expression and processing by signal peptidase, the N-terminal sequence of this construct is NVRLQHHHHHHHLEAEEHSQHQ-. A second version of this construct was constructed in a pB22 version containing a tobacco etch virus (TEV) site after the His7-tag. Correct sequences were confirmed by DNA sequencing (Eurofins, UK) using both forward and reverse plasmid-specific primers. The OprC_AA mutant was produced by changing the key amino acids Cys143 and Met147 to alanine residues using the KLD Quickchange site-directed mutagenesis kit (New England Biolabs, UK) and specific primers containing both mutation sites (forward: 5′-tcgcgcggatgcaccaaccagctatattagc-3′; reverse: 5′-ttcggggcggcgccaagcatcatgc-3′). The single mutants C143A, M147A, H323A, M325A, M147H, and M325H were made in similar ways.

OprC recombinant protein production and purification was performed as follows: E. coli C43 Δcyo was electroporated with expression vector, recovered for 60 min in LB (Sigma, UK) at 37°C, and plated on LB agar (Sigma, UK) containing 100 μg mL⁻¹ ampicillin (Melford, UK). Transformants were cultured in LB medium or in LeMasters-Richards (LR) minimal medium with glycerol (2–3 g/l) as carbon source. All media contained 100 μg mL⁻¹ ampicillin. For rich media, cells were grown (37°C, 180 rpm) until OD600 approximately 0.6, when protein expression was induced with 0.1% arabinose for 4 to 5 h at 30°C or overnight at 16°C (150 rpm). For LR media, a small overnight preculture in LB was used at 1/100 v/v to inoculate an LR-medium preculture early in the morning (typically 1 ml preculture for 100 ml of cells was used), which was grown during the day at 37°C. After late afternoon inoculation, large-scale cultures (typically 6 to 8 l) were grown overnight at 30°C until OD 0.4 to 0.7, followed by induction with 0.1% arabinose at 30°C for 6 to 8 h. Cells were harvested by centrifugation (5,000 rpm, 20 min, 4°C), and pellets homogenised in 20 mM Tris (Sigma), 300 mM NaCl (Fisher) (pH 8.00) (TBS buffer), in the presence of 10 mM EDTA (Sigma). Cells were broken by one pass through a cell disruptor (Constant Systems 0.75 kW operated at 23 kpsi), centrifuged at 42,000 rpm for 45 min at 4°C (45Ti rotor; Beckman), and the resulting total membrane fraction was homogenised in TBS buffer containing 1.5% lauryl-dimethylamine oxide (LDAO) (Sigma, UK). Membrane proteins were extracted by stirring (60 min, 4°C), centrifuged (42,000 rpm in 45Ti rotor, 30 min, 4°C), and the membrane extract was loaded on a Chelating Sepharose Fast Flow bed resin (approximately 10 ml; GE Healthcare, UK) previously activated with 200 mM NiCl₂ (Sigma) and equilibrated in TBS containing 0.15% n-dodecyl-beta-D-maltopyranoside (DDM). After washing with 15 column volumes buffer with 30 mM imidazole, protein was eluted with 0.25 M imidazole buffer (Fisher), incubated with 20 mM EDTA (30 min, 4°C), and loaded on a Superdex 200 16/600 size exclusion column equilibrated with 10 mM HEPES, 100 mM NaCl, 0.05% DDM, 10 mM EDTA (pH 7.5). Peak fractions were pooled and concentrated using a 50 MWCO Amicon filter (Millipore, UK), analysed on SDS-PAGE, flash-frozen in liquid nitrogen, and stored at −80°C. Typical yields of purified WT and most mutant OprC proteins ranged between 2 and 5 mg per l media grown at 16°C. All media and buffer components were made in fresh milli-Q water.

Protein preparations intended for crystallisation trials were pooled and buffer-exchanged to 10 mM HEPES, 100 mM NaCl, 0.4% tetraethylene glycol mono-octyl ether (C₈E₄) (Anatrace, United States) (pH 7.5). NaNO₃ was substituted for NaCl for protein preparations intended for crystal trials with silver in order to avoid formation of insoluble AgCl. Protein preparations to be used for metal analysis after removal of the His-tag underwent a slightly different protocol. The elution fraction from immobilised metal affinity chromatography (IMAC) was buffer-exchanged to 50 mM Tris, 0.5 mM EDTA, 0.2 mM TCEP, 100 mM NaCl, 0.05% DDM (Anatrace, US) (pH 7.50) and submitted to TEV protease digestion (ratio 1 mg TEV: 10 mg protein, 4°C, overnight). Samples were submitted to a second IMAC column, where flow-through and wash fractions were combined for the subsequent SEC step in 10 mM HEPES, 100 mM NaCl, 0.05% DDM, 0.5 mM EDTA (pH 7.5). Protein concentration was determined by BCA assay (Thermo Scientific, UK) and by UV/Vis absorbance at 280 nm (considering OprC E_0.1% = 1.6 as determined by ProtParam).

In vitro metal binding assays and inductively coupled plasma mass spectrometry (ICP-MS)

OprC samples intended for metal binding assays were exchanged into respective chelex-treated buffers without EDTA and were equilibrated with different equivalents of Cu(II) or Zn, for 30 min at room temperature (n = 3). Protein concentrations used were in the range of 10 to 20 μM. Samples were loaded on an analytical Superdex 200 Increase 10/300GL (GE Healthcare) column, equilibrated in 10 mM HEPES, 100 mM NaCl, 0.05% DDM, 0.5 mM EDTA (pH 7.5). Size exclusion peaks were pooled, concentrated, and quantified for protein by UV absorbance at 280 nm. Protein samples were diluted 10-fold in 2.5% HNO₃. Analytical metal standards of 0 to 500 ppb were prepared by serial dilution from individual metal stocks (VWR, UK) and were matrix-matched to protein samples. Samples and standard curves were analysed by ICP-MS using a Thermo X-series instrument (Thermo Fisher Scientific). OprC WT samples were screened for the presence of ⁶⁵Cu, ⁵⁵Mn, ⁵⁹Co, ⁶⁰Ni, ⁶⁵Cu, ⁶⁶Zn, and rest were typically screened for the presence of ⁶⁵Cu and ⁶⁶Zn. The increase in copper content in “as-purified” tagless WT protein (Fig 3B) is most likely due to the use of a second IMAC column after tag cleavage. In addition to the additional handling steps that could have increased copper content, the NiCl₂ used for the IMAC column might contain traces of copper.

Copper extraction (demetallation) experiments

OprC_WT, OprC_AA, and M147A samples were incubated with 3 equivalents copper for 30 min and then loaded onto a Superdex S-200 Increase 10/300GL column equilibrated with 10 mM HEPES, 100 mM NaCl, 0.05% DDM, 0.5 mM EDTA (pH 7.5). Peak fractions were pooled, concentrated, and quantified by UV absorbance at 280 nm_. Samples were exchanged into respective chelex-treated buffer without EDTA. For demetallation experiments, 15 to 20 μM of copper-bound proteins were taken in duplicates and incubated with 100-fold excess of the copper chelator BCS (Sigma) and 100-fold excess of the reducing agent hydroxyl amine (NH₂OH) (Sigma) at 60°C and room temperature. BCS is a high-affinity Cu(I) chelator (logβ₂ 20.8) and forms a 2:1 complex with Cu(I), namely [Cu(BCS)₂]⁻³, with a molar extinction coefficient of 13,300 cm⁻¹ M⁻¹ at 483 nm, enabling quantitation of Cu(I) [61].

Protein crystallisation, data collection, and structure determination

Sitting-drop crystallisation trials were set up using a Mosquito crystallisation robot (TTP Labtech) with commercial screens (MemGold1 and MemGold2, Molecular Dimensions) at 20°C. To obtain the initial structure of Cu-bound OprC, the protein (approximately 12 mg/ml) was incubated with 3 mM CuCl₂ for 1 h at room temperature, followed by setting up crystallisation trials. A number of initial hits were obtained and were subsequently optimised by manual hanging drop vapour diffusion using larger drops (typically 1 to 1.5 μl protein + 1 μl reservoir). Well-diffracting crystals (approximately 3 Å resolution at a home source) were obtained in 0.1 M NaCl/0.15 M NH₄SO₄/0.1 M MES (pH 6.5)/18% to 22% PEG1000. Crystals were cryoprotected with mother liquor lacking CuCl₂ containing 10% PEG400 for approximately 5 to 10 s and flash-frozen in liquid nitrogen. Diffraction data were collected at Diamond Light Source (Didcot, UK) at beamline i02. For the best crystal, belonging to space group C222₁, 720 degrees of data were collected at an energy of 8,994 eV, corresponding to the K-edge of copper (S1 Table). Data were autoprocessed by xia2 [62]. The structure was solved via single anomalous dispersion (SAD) via AUTOSOL in Phenix [63]. Two copper sites were found, one for each OprC molecule in the asymmetric unit (S1 Fig). The phases were of sufficient quality to allow automated model building via Phenix AUTOBUILD, generating approximately 60% of the structure and using data to 2.0 Å. The remainder of the structure was built manually via iterative cycles of refinement in Phenix and model building in COOT [64]. Metal coordination was analysed by the Check-my-metal server [65]. The final refinement statistics are listed in S1 Table. Subsequently, crystals were also obtained without any copper supplementation of the protein. These were isomorphous to those described above and obtained under identical crystallisation conditions. MR indicated the presence of copper and an identical structure to that obtained above. OprC_AA crystals (approximately 10 mg/ml protein) were obtained and optimised by hanging drop vapour diffusion as described above, and diffraction-quality crystals were obtained in the same conditions as for Cu-OprC, i.e., 0.1 M sodium chloride/0.15 M ammonium sulphate/0.01 M MES sodium (pH 6.5)/19% (w/v) PEG1000. Interestingly, however, the OprC_AA crystals belong to a different space group (P22₁2₁), most likely as a result of the structural differences between both OprC variants. Diffraction data were collected at Diamond Light Source (Didcot, UK) at beamline I24. Diffraction data were processed in XDS [66]. The structure was solved by MR using Phaser, with WT OprC as the search model. Model building was done in COOT and refinement in Phenix. As for Cu-OprC, the data collection and refinement statistics are shown in S1 Table. C143A and H323A proteins (approximately 10 to 12 mg/ml protein) were incubated with 2 mM CuSO₄ at room temperature for 1 h, followed by cocrystallisation. Diffracting crystals for both C143A and H323A in the presence of copper were obtained in 0.34 M ammonium sulphate/0.1 M sodium citrate (pH 5.5)/12% to 16% w/v PEG 4000 and were cryoprotected using mother liquor lacking CuSO₄ and with 25% ethylene glycol for approximately 5 to 10 s and flash-frozen in liquid nitrogen. M147H and M325H crystals were obtained in the same condition as those for WT OprC. For cocrystallisation with silver, OprC proteins were incubated with 2 mM AgNO3 for 1 h at room temperature, followed by cocrystallisation. Well-diffracting OprC_WT crystals with silver were obtained under the same conditions as in the presence of copper. For the best OprC_WT crystal, belonging to space group C222₁, 999 degrees of data were collected at an energy of 8,000 eV to obtain anomalous signals for Ag. C143A and H323A crystals (approximately 10 to 12 mg/ml protein) with Ag were obtained from 0.2 M choline chloride/0.1 M Tris (pH 7.5)/12% to 16% w/v PEG 2000 MME and 0.5 M potassium chloride/0.05 M HEPES (pH 6.5)/12% to 16% v/v PEG 400, respectively. Crystals were cryoprotected for 5 to 20 s with mother liquor lacking AgNO₃ but containing 25% ethylene glycol for C143A and 20% PEG 400 for H323A. For C143A and H323A crystallised in the presence of copper or silver, datasets of 360 degrees each were collected at energies of 8,800 and 9,175 eV, using different parts of the same crystal (S2 and S3 Tables).

Electron paramagnetic resonance spectroscopy

EPR measurements were carried out using a Bruker ELEXSYS-E500 X-band EPR spectrometer operating in continuous wave mode, equipped with an Oxford variable-temperature unit and ESR900 cryostat with Super High-Q resonator. All EPR samples were prepared in quartz capillary tubes (outer diameter; 4.0 mm, inner diameter 3.0 mm) and frozen immediately in liquid N₂ until further analysis. The experimental setup and conditions were similar to those reported previously [67]. The low temperature EPR spectra were acquired using the following conditions: sweep time of 84 s, microwave power of 0.2 mW, time constant of 81 ms, average microwave frequency of 9.44 GHz, and modulation amplitude of 5 G, T = 20 K. The concentration of OprC_WT and M147A varied from 210 to 260 μM in 10 mM HEPES, 100 mM NaCl, 0.03% DDM (pH 7.5).

Determination of whole-cell metal content

Whole-cell metal content was determined as described previously. Briefly, overnight bacterial cultures of overexpressed OprC_WT and OprC_AA (with empty vector as control) in PA14 ΔoprC background were diluted with 1:100 fresh LB supplemented with 100 mM NaNO₃ and were grown to an OD of around 1.0 at 37°C. Cultures (25 ml) were pelleted and were washed twice in TBS and once in 20 mM Tris, 0.5 M sorbitol, and 200 uM EDTA (pH 7.5). The cell pellets were digested in 1 ml of 68% conc. nitric acid. Digested sample pellets were diluted 10-fold in 2% nitric acid (prepared in chelex-treated milli-Q water) and were analysed by ICP-MS. Results were corrected for ODs and dilution factors. Protein levels in the OM were verified by IMAC. Briefly, 0.5 l of OprC_WT and OprC_AA overexpressing strains (with empty vector as control) in the P. aeruginosa ΔoprC background strain were grown in LB (supplemented with 100 mM NaNO₃ and 0.1% arabinose) for 6 h to OD₆₀₀ approximately 1.0, followed by cell harvesting, cell lysis, and purification as described above for E. coli.

In vivo metal toxicity assays

For metal toxicity assays in P. aeruginosa, overexpressed OprC WT, C143A, and AA strains using broad-range plasmid pHERD30 (with empty vector as control) in PA14 ΔoprC were used and assays were performed in anaerobic conditions. The Cu(II) (CuSO₄) range tested varied from 0 to 7 mM. Cultures in triplicates were inoculated with 1:100 of the precultures grown in anaerobic conditions (LB with 100 mM sodium nitrate). Growth curves of final volume 200 μl were set up in 96-well Costar culture plate (Sigma Aldrich) and sealed inside an anaerobic chamber (Don Whitley Scientific, A35 workstation). Growth was monitored at 600 nm using an Epoch plate reader (Biotek Instruments Ltd) at 37°C. Time points were collected with 30 min intervals and experiments were performed in triplicates.

Animal infection models

Intratracheal instillation model: Specific pathogen-free (SPF) immunocompetent male Sprague–Dawley rats weighing 100 to 120 g or male CD-1 mice weighing 20 to 25 g were infected by depositing an agar bead containing around 10⁷ colony-forming units A. baumannii ATCC 19606 and P. aeruginosa UCBPP-PA14, deep into the lung via nonsurgical intratracheal intubation [68]. In brief, animals were anaesthetised with isoflurane (5%) and oxygen (1.5 l/min) utilising an anaesthesia machine. Depth of anaesthesia was evaluated by absence of gag reflex; if the reflex was present, the animal was placed back under anaesthesia until the reflex disappeared. No animals were utilised until they were fully anaesthetised. Animals were infected via intrabronchial instillation of molten agar suspension (rats- 100 μl) (mice- 20 μl) via intratracheal intubation and then allowed to recover. Animals were returned in their home cages and observed until recovered from anaesthesia. At 24-h postinfection, animals were killed, and lung was homogenised in sterile saline using a lab blender. All procedures are in accordance with protocols approved by the GSK Institutional Animal Care and Use Committee (IACUC) and meet or exceed the standards of the American Association for the Accreditation of Laboratory Animal Care (AAALAC), the US Department of Health and Human Services, and all local and federal animal welfare laws.

Sample workup for proteomics

The sample workup protocol was optimised to deplete host material while maintaining A. baumannii and P. aeruginosa viability until lysis. All buffers and equipment were used at 0 to 4°C to minimise proteome changes during sample workup. The sample volume (maximum of 1 ml) was estimated, and an equal volume of 1% Tergitol in PBS was added followed by vigorous vortexing for 30 s. After centrifugation at 500 × g for 5 min, the supernatant was transferred to a fresh tube, and the pellet was extracted again with 2 ml 0.5% Tergitol in PBS. The supernatant was combined with the first supernatant and centrifuged at 18,000 × g for 5 min. The pellet was washed with 2 ml and again centrifuged at 18,000 × g for 5 min. The supernatant was removed, and the pellet was resuspended in 100 μl 5% sodium deoxycholate, 5 mM Tris (2-carboxyethyl) phosphine hydrochloride, 100 mM NH₄HCO₃. The sample was incubated at 90°C for 1 min and then stored at −80°C. Samples were thawed and sonicated for 2 × 20 s (1-s interval, 100% power). Proteins were alkylated with 10 mM iodoacetamide for 30 min in the dark at room temperature. Samples were diluted with 0.1 M ammonium bicarbonate solution to a final concentration of 1% sodium deoxycholate before digestion with trypsin (Promega) at 37°C overnight (protein to trypsin ratio: 50:1). After digestion, the samples were supplemented with TFA to a final concentration of 0.5% and HCl to a final concentration of 50 mM. Precipitated sodium deoxycholate was removed by centrifugation at 4°C and 14,000 rpm for 15 min. Peptides in the supernatant were desalted on C18 reversed phase spin columns according to the manufacturer’s instructions (Macrospin, Harvard Apparatus), dried under vacuum, and stored at −80°C until further processing.

Parallel reaction monitoring

Heavy proteotypic peptides (JPT Peptide Technologies GmbH) were chemically synthesised for A. baumannii and P. aeruginosa OM proteins. Peptides were chosen dependent on their highest detection probability and their length ranged between 7 and 20 amino acids. Heavy proteotypic peptides were spiked into each sample as reference peptides at a concentration of 20 fmol of heavy reference peptides per 1 μg of total endogenous protein mass. For spectrum library generation, we generated parallel reaction monitoring (PRM) [69] assays from a mixture containing 500 fmol of each reference peptide. The setup of the μRPLC-MS system was as described previously [70]. Chromatographic separation of peptides was carried out using an EASY nano-LC 1000 system (Thermo Fisher Scientific) equipped with a heated RP-HPLC column (75 μm × 37 cm) packed in-house with 1.9 μm C18 resin (Reprosil-AQ Pur, Dr. Maisch). Peptides were separated using a linear gradient ranging from 97% solvent A (0.15% formic acid, 2% acetonitrile) and 3% solvent B (98% acetonitrile, 2% water, 0.15% formic acid) to 30% solvent B over 60 min at a flow rate of 200 nl/min. Mass spectrometry analysis was performed on Q-Exactive HF mass spectrometer equipped with a nanoelectrospray ion source (both Thermo Fisher Scientific). Each MS1 scan was followed by high-collision dissociation (HCD) of the 10 most abundant precursor ions with dynamic exclusion for 20 s. Total cycle time was approximately 1 s. For MS1, 3e6 ions were accumulated in the Orbitrap cell over a maximum time of 100 ms and scanned at a resolution of 120,000 FWHM (at 200 m/z). MS2 scans were acquired at a target setting of 1e5 ions, accumulation time of 50 ms, and a resolution of 30,000 FWHM (at 200 m/z). Singly charged ions and ions with unassigned charge state were excluded from triggering MS2 events. The normalised collision energy was set to 35%, the mass isolation window was set to 1.1 m/z, and one microscan was acquired for each spectrum.

The acquired raw files were converted to the mascot generic file (mgf) format using the msconvert tool (part of ProteoWizard, version 3.0.4624 (2013-6-3)). Converted files (mgf format) were searched by MASCOT (Matrix Sciences) against normal and reverse sequences (target decoy strategy) of the UniProt database of A. baumannii strains ATCC 19606 and ATCC 17978, and P. aeruginosa UCBPP-PA14, as well as commonly observed contaminants. The precursor ion tolerance was set to 20 ppm, and fragment ion tolerance was set to 0.02 Da. Full tryptic specificity was required (cleavage after lysine or arginine residues unless followed by proline), 3 missed cleavages were allowed, carbamidomethylation of cysteins (+57 Da) was set as fixed modification, and arginine (+10 Da), lysine (+8 Da), and oxidation of methionine (+16 Da) were set as variable modifications. For quantitative PRM experiments, the resolution of the orbitrap was set to 30,000 FWHM (at 200 m/z), and the fill time was set to 50 ms to reach a target value of 1e6 ions. Ion isolation window was set to 0.7 Th (isolation width), and the first mass was fixed to 100 Th. Each condition was analysed in biological triplicates. All raw files were imported into Spectrodive (Biognosys AG) for protein and peptide quantification.