Acquisition of ionic copper by a bacterial outer membrane protein

Copper, while toxic in excess, is an essential micronutrient in all kingdoms of life due to its essential role in the structure and function of many proteins. Proteins mediating ionic copper import have been characterised in detail for eukaryotes, but much less so for prokaryotes. In particular, it is still unclear whether and how Gram-negative bacteria acquire ionic copper. Here we show that Pseudomonas aeruginosa OprC is an outer membrane, TonB-dependent transporter that is conserved in many Proteobacteria and which mediates acquisition of both reduced and oxidised ionic copper via an unprecedented CxxxM-HxM metal binding site. Crystal structures of wild type and mutant OprC variants with silver and copper suggest that acquisition of Cu(I) occurs via a surface-exposed “methionine track” leading towards the principal metal binding site. Together with whole-cell copper quantitation and quantitative proteomics in a murine lung infection model, our data identify OprC as an abundant component of bacterial copper biology that may enable copper acquisition under a wide range of conditions. Significance Copper is an essential metal in biology due to its role in the structure and function of many proteins. Despite this, it is not very clear how bacteria acquire copper, especially for Gram-negative organisms. In this study we show that the outer membrane protein OprC has an unusual metal binding site that allows OprC to bind both reduced and oxidised ionic copper near-irreversibly. Given the versatility of OprC, its presence in many Proteobacteria and its abundance during lung infection in mice, our study shows that OprC is an important component of prokaryote copper biology that warrants further study to uncover its regulation and to assess its role in bacterial virulence.


Introduction 59
Metals fulfil cellular functions that cannot be met by organic molecules and are indispensable 60 for the biochemistry of life in all organisms. Copper is the third-most abundant transition metal 61 in biological systems after iron and zinc. It has key roles as structural component of proteins 62 or catalytic cofactor for enzymes(1) , most notably associated with the biology of oxygen and 63 in electron transfer. On the other hand, an excess of copper can be deleterious due to its 64 ability to catalyse production of hydroxyl radicals(2, 3). Excessive copper may also disrupt 4 characterised metal binding via ICP-MS and EPR. In addition, we have confirmed metal 105 uptake by OprC using whole cell metal quantitation. OprC indeed has the typical structure of 106 a TBDT, and differences between the Cu-loaded and Cu-free protein demonstrate changes in 107 tertiary structure that likely lead to TonB interaction and copper import.

157
Copper binding by OprC is highly specific and near-irreversible. 158 Following structure determination of copper-bound OprC, several attempts were made to 159 produce a structure of copper-free OprC. First, the protein was purified and crystallised without 160 added copper; however, this gave a structure that was identical to the one already obtained

206
The fact that it is not possible to obtain copper-free wild type protein, even after taking 207 extensive precautions, suggests that copper binds to OprC with very high affinity. To explore 208 this further, we performed copper extraction assays with a large excess of OprCWT, only 20% copper was removed after 24 hrs at room temperature, and the temperature 211 had to be increased to 60 °C to obtain near-quantitative extraction of copper (~90% after 24 hard to separate from OprC, and BCS-treated OprC did not bind copper anymore, suggesting 214 an irreversible change in the protein due to the harsh incubation conditions. Nevertheless, 215 these results demonstrate that copper is kinetically trapped inside OprC and is, for all intents 216 and purposes, irreversibly bound. This is fully compatible with the consensus transport 217 mechanism of TBDTs, in which the interaction with TonB, occurring after substrate binding, is 218 required to disrupt the binding site, leading to release of the substrate(33).

Conformational changes upon copper binding. 221
The OprCAA structure was solved by molecular replacement using Cu-bound OprC as the  OprCAA lack electron density for a limited number of residues, suggesting increased mobility.

231
Overall, the conformational changes of the loops upon copper binding likely decrease the 232 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 mutant, copper becomes solvent accessible due to the absence of the Met147 side chain (

265
The OprC methionine track and the principal binding site bind Cu(I). 266 We next asked whether OprC also binds Cu (I). Since it is challenging to maintain copper in 267 its +1 state during crystallisation, we used silver (Ag(I)) as a proxy for Cu(I) and determined 268 the co-crystal structure of WT OprC in the presence of 2 mM AgNO3 (Methods). This is To obtain more information on the individual residue contribution to copper binding, we next 12 (C143A, M147A, H323A and M325A), and determined copper binding via analytical SEC and 286 ICP-MS. For all single mutants, the copper content after purification from LB was below 10%, 287 except for M147A (~20 %) (Fig. 3b). Upon incubation with 3 or 10 equivalents Cu (II), various 288 occupancies were obtained. C143A has no bound copper even after incubation with 10 289 equivalents Cu (II), suggesting this residue has a crucial role. H323A (~30%) and in particular 290 M147A (~60%) have relatively high occupancies after copper incubation and SEC, indicating 291 that these residues contribute less towards binding. Of the four ligands, the M147 thioether is 292 the furthest away from the copper in the crystal structure (Fig. 2B), which may explain why it 293 contributes the least to ligand binding. Interestingly, removal of bound copper is much faster 294 in the M147A mutant compared to OprCWT (Fig. 3C), suggesting that solvent exclusion by the 295 M147 side chain (Fig. 4D) is the main reason why copper is kinetically trapped in OprCWT.

297
To shed additional light on the redox state of the bound copper, continuous wave EPR (cw-298 EPR) spectra were recorded on OprCWT. Surprisingly, as-purified OprCWT containing ~0.6 299 equivalents copper was EPR silent (Fig. 3D), demonstrating that the copper species present 300 in the crystal structure is Cu(I). The as-purified M147A protein, with ~0.1 equivalent copper, 301 was EPR silent as well. We next loaded the M147A mutant with CuSO4 to 1 equivalent, and 302 EPR spectra were recorded over time. The observed EPR signal is different from the standard  is not linked to copper uptake.

353
Since copper toxicity assays failed, we decided to determine P. aeruginosa whole cell metal 354 contents using ICP-MS. We observed no differences in copper content between the wild type    to their large size and complex architecture, and no attempts were made to assess the via SEC, suggesting any interaction will be transient.

445
Given that copper, and in particular the more toxic Cu(I), is a known antimicrobial, the 446 presence of bacterial proteins dedicated to copper acquisition such as OprC might be 447 problematic under certain conditions. Indeed, it is thought that, in contrast to iron that is 448 withheld from a pathogen by the host during infection, elevated levels of host-derived copper 449 in e.g. macrophages could be an alternative "nutritional immunity" antimicrobial response(49).

471
OprC is the first example of a TBDT that mediates copper import without a metallophore. The

472
TBDT with the closest substrate specificity to OprC is the ionic zinc transporter ZnuD from 473 Neisseria meningitidis, the structure of which has been solved(52). Large structural 474 differences between OprC and ZnuD exist for the extracellular loops (overall Cα r.m.s.d. ~5.9 475 Å). ZnuD has several discrete low-affinity binding sites that may guide the metal towards the 476 high-affinity binding site (52). In OprC, a distinctive "methionine track" provides low-affinity 477 binding sites to guide copper to the high affinity site. Interestingly, while the extracellular loops between OprC and ZnuD are very different and the overall sequence identity is only 28%, the 479 metal binding sites are located at very similar positions and only 2.8 Å apart (Fig. S8),

480
suggesting that the transport channel formed via TonB interaction may be similar. Inspection 481 of the ZnuD structure shows that the zinc binding site is excluded from solvent, and we 482 propose that the zinc ion in ZnuD is kinetically trapped, analogous to copper in OprC.

484
OprC shares ~60 % identity to NosA from Pseudomonas stutzeri, for which no structure is

488
HxM copper binding motif and some of the methionine track residues are conserved (Fig. S6).

489
NosA is important during denitrification in P. stutzeri JM300 and was proposed to load copper

604
The structure was solved via single anomalous dispersion (SAD) via AUTOSOL in Phenix(62)

605
. Two copper sites were found, one for each OprC molecule in the asymmetric unit (Fig. S1).

606
The phases were of sufficient quality to allow automated model building via Phenix 607 AUTOBUILD, generating ~60% of the structure and using data to 2.0 Å. The remainder of the  Table S1. Subsequently, crystals were also obtained without  Table S1. C143A and H323A proteins (~10-12 mg/ml protein) were 624 incubated with 2 mM CuSO4 at room temperature for 1 hr, followed by co-crystallisation.

625
Diffracting crystals for both C143A and H323A in the presence of copper were obtained in 626 0.34 M Ammonium sulfate/0.1 M Sodium citrate pH 5.5/12 -16 % w/v PEG 4000 and were 627 cryo-protected using mother liquor lacking CuSO4 and with 25% ethylene glycol for ~10 s and

704
The supernatant was combined with the first supernatant and centrifuged at 18'000 x g for 5 705 min. The pellet was washed with 2 ml and again centrifuged at 18'000 x g for 5 min. The

754
For quantitative PRM experiments the resolution of the orbitrap was set to 30,000 FWHM (at 755 200 m/z) and the fill time was set to 50 ms to reach a target value of 1e6 ions. Ion isolation 756 window was set to 0.7 Th (isolation width) and the first mass was fixed to 100 Th. Each 757 condition was analyzed in biological triplicates. All raw-files were imported into Spectrodive 758 (Biognosys AG) for protein and peptide quantification.

Author contributions 761
BvdB designed the study. SPB and BvdB expressed, purified, and crystallized proteins. AB collected the diffraction data. SPB and BvdB analysed the diffraction data and refined the proteomics experiments, supervised by DB. BvdB and SPB wrote the paper.            is Cu (I). Distances between coordinating residues and metal (magenta) are shown. The Table  summarises distances between copper and co-ordinating residues as well as geometry.