Calculating metalation in cells reveals CobW acquires CoII for vitamin B12 biosynthesis while related proteins prefer ZnII

Abstract

Protein metal-occupancy (metalation) in vivo has been elusive. To address this challenge, the available free energies of metals have recently been determined from the responses of metal sensors. Here, we use these free energy values to develop a metalation-calculator which accounts for inter-metal competition and changing metal-availabilities inside cells. We use the calculator to understand the function and mechanism of GTPase CobW, a predicted Co^II-chaperone for vitamin B₁₂. Upon binding nucleotide (GTP) and Mg^II, CobW assembles a high-affinity site that can obtain Co^II or Zn^II from the intracellular milieu. In idealised cells with sensors at the mid-points of their responses, competition within the cytosol enables Co^II to outcompete Zn^II for binding CobW. Thus, Co^II is the cognate metal. However, after growth in different [Co^II], Co^II-occupancy ranges from 10 to 97% which matches CobW-dependent B₁₂ synthesis. The calculator also reveals that related GTPases with comparable Zn^II affinities to CobW, preferentially acquire Zn^II due to their relatively weaker Co^II affinities. The calculator is made available here for use with other proteins.

Fig. 1. Co^II binding to CobW is enhanced by guanine nucleotides.

a Homology model of CobW (generated with SWISS-MODEL using E. coli YjiA PDB entry 1NIJ as template; image generated using CCP4 Molecular Graphics software) showing sulfur atoms from conserved CxCC motif (in yellow) and nucleotide-binding (GxxGxGKT, hhhExxG, SKxD*) motifs^, (in red). *Ordinarily NKxD but [ST]KxD observed in some COG0523 proteins. b Purified CobW analysed by SDS-PAGE (full image in Supplementary Fig. 1; n = 1 under these conditions). c ESI-MS analysis (de-convoluted spectra) of purified CobW. d Representative (n = 2) apo-subtracted spectra of Co^II-titrated CobW (26.1 µM); feature at 315 nm (inset) shows a non-linear increase. e Representative (n = 2) elution profile following gel-filtration of a mixture of CobW (10 µM) and Co^II (30 µM) showing no co-migration of metal (red) with protein (black). Fractions were analysed for protein by Bradford assay and for metal by ICP-MS. f Structures of nucleotides used in this work (generated using ChemDraw software). g As in d for a mixture of CobW (24 µM) and GMPPNP (60 µM); feature at 339 nm (inset) showing a linear increase saturating at 2:1 ratio Co^II:CobW (n = 2). h As in e for a mixture of CobW (10 µM), Co^II (30 µM) and GMPPNP (30 µM) showing co-migration of 1.8 equivalents Co^II per CobW (mean value from peak integration, n = 2 independent experiments). Data replicates are shown in Supplementary Fig. 1.

Fig. 2. Mg^II and the γ-phosphate group of GTP are necessary for high affinity Co^II binding.

a Absorbance (relative to Co^II-free solution) of Co^II-titrated CobW (20 µM) with GMPPNP (60 µM) in competition with EGTA (40 µM); titrations in the absence (black) or presence (red) of Mg^II (2.7 mM, i.e. concentration in a bacterium^,). Data shown are representative of n = 3 independent experiments (with varying [competitor] and/or identity; see Supplementary Figs. 3c, d and 4a, b). b Absorbance (relative to Co^II-free solution) of Co^II-titrated CobW (20 µM) with GMPPNP (60 µM) and Mg^II (2.7 mM) in the absence of competing ligand; feature at 339 nm (inset) showing linear increase saturating at 1:1 ratio Co^II:CobW (n = 2; see Supplementary Fig. 3e). c–e Representative K_Co(II) quantification for CobW in the absence or presence of nucleotides (n = 3 independent experiments, details in Supplementary Fig. 4 and Supplementary Table 1). c Fluorescence quenching of Co^II-titrated fura-2 (10 µM) in competition with CobW alone (37 µM). d Fluorescence quenching of Co^II-titrated fura-2 (8.1 µM) in competition with CobW (20 µM) with Mg^II (2.7 mM) and GDP (200 µM). e Absorbance (relative to Co^II-free solution) of Co^II-titrated CobW (18 µM) in competition with EGTA (2.0 mM) with Mg^II (2.7 mM) and GTP (200 µM). Solid traces in a, c, d, e show curve fits of experimental data to a model where CobW binds one molar equivalent Co^II per protein monomer. Dashed lines show simulated responses for K_Co(II) tenfold tighter or weaker than the fitted value. f Analysis of GTP hydrolysis by anion-exchange chromatography. Control samples of GTP and GDP elute as distinct peaks (red traces) measured by absorbance at 254 nm. Black traces show the extent of hydrolysis of GTP (200 µM) incubated with CobW (20 µM), Mg^II (2.7 mM) and Co^II (18 µM) over time. g Analysis of data from f showing % GTP remaining over time. After 390 min incubation, nucleotides remain primarily (>75 %) unhydrolysed. Equivalent data using 4:1 ratio GTP:CobW is shown in Supplementary Fig. 6.

Fig. 3. Binding of Mg^IIGTP-CobW to Fe^II, Ni^II, Cu^I and Zn^II.

a Absorbance upon Fe^II titration into a mixture of Tar (16 µM), Mg^II (2.7 mM) and GTP (500 µM) in the absence (red crosses) or presence (black circles) of CobW (50 µM). Dashed lines show simulated responses for specified K_Fe(II) of Mg^IIGTP-CobW, providing limiting K_Fe(II) ≥ 10⁻⁶ M. Control Fe^II titration into a solution of Tar (16 µM) in buffer only (Supplementary Fig. 8a) confirmed that Mg^II and GTP did not inhibit stoichiometric Fe^IITar₂ formation. b Absorbance change (relative to Ni^II-free solution) of Ni^II-titrated Tar (20 µM) in competition with CobW (30 µM) in the presence of Mg^II (2.7 mM) and GTP (300 µM). c Absorbance of Cu^I-titrated Bca (1.0 mM) in competition with CobW (20 µM) in the presence of Mg^II (2.7 mM) and GTP (200 µM). In a–c, solid traces show curve fits of experimental data to models where CobW binds one molar equivalent of metal per protein monomer. Supplementary Table 2 contains mean ± standard deviation (SD) K_metal values from n = 3 independent experiments (full details in Supplementary Figs. 8–12 and Supplementary Table 1). In b, c, dashed lines show simulated responses for K_metal tenfold tighter or weaker than the fitted value. d Absorbance (relative to probe-free solution) upon titration of Zn^II into a mixture of quin-2 (10 µM), Mg^II (2.7 mM), GTP (100 µM) and CobW (10 µM).

Fig. 4. Mg^IIGTP-CobW binds Zn^II with sub-picomolar affinity.

a Representation of the equilibrium for exchange of Co^II and Zn^II between ligand (L = NTA) and protein (P = Mg^IIGTP-CobW). b–e Absorbance (relative to metal-free solution) of solutions of CobW (17.9–20.4 µM), Mg^II (2.7 mM), GTP (200 µM) and NTA (0.4–4.0 mM) upon (b–d) first addition of Co^II (black trace) then Zn^II (blue trace) or (e) the reverse, at equilibrium (n = 1 for each panel). The absorbance peak at 339 nm corresponds to Co^II-bound protein. An excess of ligand NTA was used to buffer both metals in each experiment: varying the ratios of ligand-bound metal ions ([Co^IINTA]/[Zn^IINTA] = 28–167) shifted the ratios of Co^II- and Zn^II-bound protein as predicted by the equilibrium exchange constant in a. Consistent K_Zn(II) values for Mg^IIGTP-CobW were generated at all tested conditions (Supplementary Table 4). Dashed red lines show expected A_{339 nm} peak intensities for K_Zn(II) of Mg^IIGTP-CobW tenfold tighter or weaker than calculated values.

Fig. 5. Mg^IIGTP-CobW is predicted to acquire Co^II or Zn^II in a bacterial cell.

Free-energy change (ΔG) for metal binding to Mg^IIGTP-CobW (red circles) plotted against the intracellular available free energies for metal binding in a reference bacterial cytosol (values correspond to Salmonella) under idealised conditions (i.e. where each metal sensor undergoes half of its transcriptional response; black squares). Intracellular available ΔG_Zn(II) is the mean of the values determined from the two Zn^II sensors ZntR (a) and Zur (b). Bars show the change in intracellular available ΔG as cognate sensors shifts from 10–90% of their responses. Free energy differences (ΔΔG) which favour acquisition of metals by Mg^IIGTP-CobW in vivo are indicated in blue. ΔG values for Co^II complexes of CobW alone (open red triangle) and Mg^IIGDP-CobW (closed red triangle) are also shown. For Fe^II binding to Mg^IIGTP-CobW, arrow indicates limiting ΔG > −34.2 kJ mol⁻¹.

Fig. 6. Mg^IIGTP-YeiR and Mg^IIGTP-YjiA preferentially acquire Zn^II.

a Free-energy change (ΔG) for metal binding to Mg^IIGTP-YeiR (red circles) plotted against the intracellular available free energies for metal binding (as described in Fig. 5; black squares and bars). b As a for Mg^IIGTPγS-YjiA (red circles). Arrows indicate where only a limiting ΔG was determined (thus ΔG > plotted value).

Fig. 7. Mg^IIGTP-CobW outcompetes Mg^IIGTP-YeiR for Co^II.

Elution profile of YeiR (10 μM), CobW (10 μM) or both proteins following incubation with GTP (100 μM), Mg^II (2.7 mM) and Co^II (8 μM) resolved by differential elution from an anion exchange column. Fractions were analysed for Co^II by ICP-MS and protein by SDS-PAGE (YeiR alone black, CobW alone blue, both proteins red; n = 1). Arrow denotes flow through fractions. Full gel images and SDS-PAGE analysis of flow through fractions shown in Supplementary Fig. 26.

Fig. 8. Calculations of conditional Co^II availabilities in B₁₂-producing E. coli*.

a Calculated relationship between intracellular Co^II availability and normalised DNA occupancy (θ_D) by RcnR. θ_D of 0 and 1 are the maximum and minimum calculated DNA occupancies. The dynamic range (within which RcnR responds to changing intracellular Co^II availability) has been defined as θ_D of 0.01–0.99 (i.e. 1–99% of RcnR response). The calibrated maximum and minimum fold changes in rcnA transcript abundance (i.e. boundary conditions, see Supplementary Fig. 27) therefore correspond to θ_D of 0.01 and 0.99 in these calculations (red circles). θ_D for each growth condition (black circles) was calculated from the qPCR response in b, assuming a linear relationship between change in θ_D and change in transcript abundance (Eq. (10)). Corresponding Co^II availabilities are listed in Supplementary Table 8. b Transcript abundance (relative to untreated control) of the RcnR-regulated gene rcnA following 1 h exposure of E. coli* to increasing [Co^II], measured by qPCR. Data are the mean ± SD of n = 3 biologically independent replicates. Triangle shapes represent individual experiments (some data points overlap, experimental values are available in Source Data files).

Fig. 9. B₁₂ production follows predicted metalation of Mg^IIGTP-CobW.

a Predicted metalation of Mg^IIGTP-CobW with Co^II and Zn^II (open and grey bars, respectively) in samples treated with defined media [Co^II]. Intracellular ΔG_Co(II) for each condition was calculated from rcnA expression (Fig. 8 and Supplementary Table 8). b B₁₂ produced by E. coli* strains with and without cobW (open and grey bars, respectively) following 4 h exposure to defined [Co^II]. B₁₂ was detected using a Salmonella AR2680 bioassay (detects corrins, expected to be predominantly B₁₂; see “Methods”) and quantified by automated analysis of growth areas (Supplementary Fig. 29 and Supplementary Software 1). Inset shows original image and detected areas (each false coloured) for representative (n = 3) bioassay plate of B₁₂ calibration standards. All data are the mean ± SD of n = 3 biologically independent replicates (with errors in a propagated from qPCR data in Fig. 8b). Triangles represent individual experiments (some data points overlap, experimental values are available in Source Data files).