Zinc sequestration by human calprotectin facilitates manganese binding to the bacterial solute-binding proteins PsaA and MntC

Abstract

Zinc is an essential transition metal nutrient for bacterial survival and growth but may become toxic when present at elevated levels. The Gram-positive bacterial pathogen Streptococcus pneumoniae is sensitive to zinc poisoning, which results in growth inhibition and lower resistance to oxidative stress. Streptococcus pneumoniae has a relatively high manganese requirement, and zinc toxicity in this pathogen has been attributed to the coordination of Zn(II) at the Mn(II) site of the solute-binding protein (SBP) PsaA, which prevents Mn(II) uptake by the PsaABC transport system. In this work, we investigate the Zn(II)-binding properties of pneumococcal PsaA and staphylococcal MntC, a related SBP expressed by another Gram-positive bacterial pathogen, Staphylococcus aureus, which contributes to Mn(II) uptake. X-ray absorption spectroscopic studies demonstrate that both SBPs harbor Zn(II) sites best described as five-coordinate, and metal-binding studies in solution show that both SBPs bind Zn(II) reversibly with sub-nanomolar affinities. Moreover, both SBPs exhibit a strong thermodynamic preference for Zn(II) ions, which readily displace bound Mn(II) ions from these proteins. We also evaluate the Zn(II) competition between these SBPs and the human S100 protein calprotectin (CP, S100A8/S100A9 oligomer), an abundant host-defense protein that is involved in the metal-withholding innate immune response. CP can sequester Zn(II) from PsaA and MntC, which facilitates Mn(II) binding to the SBPs. These results demonstrate that CP can inhibit Zn(II) poisoning of the SBPs and provide molecular insight into how S100 proteins may inadvertently benefit bacterial pathogens rather than the host.

Graphical Abstract

This work provides a molecular picture of how Zn(II) sequestration by calprotectin attenuates Zn(II) poisoning of microbial Mn(II) uptake machinery.

Introduction

The interplay between the host innate immune system, microbial pathogens, and zinc highlights how the host can leverage a metal ion as an essential nutrient and, when in excess, as an antimicrobial to prevent microbial infection. During infection, invading pathogens must acquire Zn(II) from the host to colonize and cause disease. To inhibit the growth of microbial pathogens, the mammalian host limits the availability of Zn(II) and other nutrient metal ions (e.g. Mn and Fe) through a metal-withholding innate immune response (Fig. 1A).^1,2 In an opposing strategy termed metal intoxication, the host takes advantage of the inherent toxicity of transition metals and bombards microbial pathogens with high levels of Zn(II) in the phagosome, which can disrupt cellular metal homeostasis and inhibit important metabolic processes in these organisms (Fig. 1B).^3–8 Metal sequestration and metal intoxication are conceptualized as two independent mechanisms of innate immunity that occur in the extracellular and intracellular space, respectively.⁹ Metal intoxication by protein mismetallation may also occur in locales where metal levels create an environment that favors binding of the wrong metal ion. In this work, we consider how the metal-sequestering innate immune protein calprotectin affects mismetallation of bacterial Mn(II)-uptake machinery with Zn(II) as a case study.

Fig. 1

Schematic representations of key metal-dependent mechanisms employed by the innate immune response against microbial pathogens. (A) Model for metal sequestration by CP in the extracellular space. CP is released from neutrophils at sites of infection and undergoes Ca(II)-induced tetramerization. The resulting (S100A8/S100A9)₂ heterotetramer is protease resistant and exhibits enhanced transition metal affinities and enhanced antimicrobial activity. In the extracellular space, CP competes with microbes for bioavailable divalent first-row transition-metal ions. (B) Model for intracellular metal intoxication against S. pneumoniae. The host deploys high levels of Zn(II) against the pathogen to reduce its virulence and inhibit its growth. Zn(II) ions bind to and poison PsaA, the extracellular SBP, and prevent Mn(II) uptake by the PsaABC transporter. Consequently, S. pneumoniae suffers from cytoplasmic Mn(II) deficiency and exhibits higher sensitivity to oxidative stress, and higher susceptibility to leukocyte activity.

The ability of Zn(II) ions to disrupt cellular metal homeostasis by inhibiting microbial metal uptake has been showcased with the Gram-positive bacterial pathogen Streptococcus pneumoniae, the leading cause of community-acquired pneumonia and meningitis.¹⁰Streptococcus pneumoniae is a Mn(II)-centric organism that is unable to grow aerobically in environments deprived of this metal.¹¹ To support its Mn(II) requirements during infection, S. pneumoniae expresses the ATP-binding cassette transporter PsaABC, which is required for the virulence and development of pneumococcal disease.¹² Indeed, deletion of any protein component of the PsaABC transport machinery was found to attenuate the virulence of S. pneumoniae.^13–15 PsaA is a solute-binding protein (SBP) with high Mn(II) affinity (Supplementary Table S1) that scavenges extracellular Mn(II) ions and delivers them to the PsaBC permease for transport into the cytoplasm. High levels of Zn(II) were found to reduce S. pneumoniae virulence and result in growth inhibition, increased sensitivity to oxidative stress, and higher susceptibility to leukocyte activity.^16,17 A proposed mechanism for this toxicity involves Zn(II) poisoning of PsaABC (Fig. 1B). The Mn(II) site of PsaA coordinates Zn(II), which prevents Mn(II) uptake by the PsaABC system and results in cytoplasmic Mn(II) deficiency and the induction of a Mn(II)-starvation response in S. pneumoniae.^17–19

Mn(II) uptake systems are also required for virulence in the Gram-positive opportunistic human pathogen Staphylococcus aureus.^20,21 The ATP-binding cassette transport system MntABC is particularly important to S. aureus and has been implicated as the dominant Mn(II) acquisition machinery during infection.²² MntC is the SBP that binds extracellular Mn(II) ions with high affinity (Supplementary Table S1) and delivers them to the MntAB permease.²¹ Although MntC also binds Zn(II) with high affinity,²³ Zn(II) intoxication has not been reported for S. aureus. Thus, whether elevated levels of Zn(II) also poison the MntABC transporter and prevent Mn(II) uptake in S. aureus warrants evaluation.

Host mechanisms to limit Zn(II) availability at infection sites include the deployment of the Zn(II)-sequestering S100 protein CP (S100A8/S100A9 oligomer, calgranulin A/B oligomer, MRP8/MRP14 oligomer),²⁴ an abundant antimicrobial protein that is released by white blood cells.^25,26 CP sequesters divalent first-row transition metal ions using two binding sites with different coordination motifs (Supplementary Fig. S1C).²⁷ Site 1 is a His₃Asp motif that exhibits high affinity and selectivity for Zn(II); site 2 is a functionally versatile His₆ motif that exhibits high affinity for Mn(II), Fe(II), Ni(II), Cu,²⁸ and Zn(II) (Supplementary Fig. S1H and I and Table S1).^29–35 Each subunit of CP has two EF-hand domains for Ca(II) binding. In the current working model, Ca(II) binding to the CP heterodimer in the extracellular space promotes self-oligomerization into a heterotetrameric complex that exhibits higher protease stability, enhanced transition metal ion affinities, and greater antimicrobial activity compared to the heterodimer.^27,30,^36–38

CP is upregulated in patients with bacterial pneumonia, including pneumococcal pneumonia.³⁹ Antibacterial activity assays revealed that CP inhibits the growth of S. pneumoniae.⁴⁰ In contrast, one study reported that CP-deficient mice were less susceptible to pneumococcal infection and exhibited a reduced bacterial burden and mortality.^39,41 Subsequent in vitro studies revealed that the growth inhibition of S. pneumoniae caused by a high Zn(II) concentration was partially reversed in the presence of CP.³⁹ Based on these observations, a model was proposed in which CP prevents Zn(II) poisoning of S. pneumoniae by sequestering Zn(II) in the extracellular space, decreasing the likelihood of Zn(II) mismetallation of PsaA, and thus providing a more favorable environment for Mn(II) uptake by S. pneumoniae.³⁹

In this work, we aim to provide a molecular picture for the observed reduction of Zn(II)-mediated toxicity in S. pneumoniae by CP. We investigate the Zn(II) sites and Zn(II)-binding properties of streptococcal PsaA and staphylococcal MntC in solution and examine the interplay between these SBPs and CP. Our results support five-coordinate Zn(II) sites within PsaA and MntC, and we show that both SBPs bind Zn(II) reversibly with sub-nanomolar affinities. Moreover, we demonstrate that CP can compete with both SBPs for Zn(II) and facilitate Mn(II) capture by the SBPs. These results provide new perspectives on how the metal-withholding innate immune response may benefit microbial pathogens rather than the host.

Results and discussion

X-ray absorption spectroscopy supports five-coordinate Zn(II) sites in PsaA and MntC

Our interest in understanding the interplay between CP, metal-transporting SBPs, and Zn(II) motivated us to examine the Zn(II) sites of MntC and PsaA in solution using X-ray absorption spectroscopy (XAS). Although structural studies of PsaA and MntC have examined the Mn(II)- and Zn(II)-bound forms of these proteins, the coordination numbers and geometries at the metal centers of both SBPs remain ambiguous.^{16,19,23,42,43} Crystal structures of the apo, Mn(II)-bound, and Zn(II)-bound (Supplementary Fig. S1A) forms of PsaA have been reported.^16,19,42 In apo PsaA, two semirigid domains, referred to as the N- and C-terminal lobes, are connected by a flexible helix linker.¹⁹ Binding of Mn(II) or Zn(II) at the interface of the two lobes induces unfolding of the linking helix and a transition between open and closed conformations. The overall fold and secondary structures of the Mn(II)- and Zn(II)-bound closed states of PsaA are almost identical, and both Mn(II) and Zn(II) coordinate PsaA at the same His₂GluAsp site defined by His67, His139, Glu205, and Asp280 (Supplementary Fig. S1D and E and Table S2).^16,19,42 The crystallographic Mn(II)- and Zn(II)-PsaA sites have been described as four-coordinate with monodentate coordination of Glu205 and Asp280 to each of the metal ions.^16,42 Nevertheless, the M(II)–O bond lengths determined from these structures may suggest a higher coordination number (Supplementary Table S2). In particular, the M(II)–O bond lengths may indicate bidentate ligation of Glu205 and/or Asp280 in Mn(II)- and Zn(II)-PsaA.

Like PsaA, MntC employs a His₂GluAsp motif, defined by His50, His123, Glu189, and Asp264 (Supplementary Fig. S1B, F, and G and Table S2; residue numbering based on the Mn(II)-MntC structure),²³ to bind Mn(II) and Zn(II).The crystallographic Mn(II) and Zn(II) sites of MntC were also described as four-coordinate with monodentate coordination of Glu189 and Asp264.^23,43 Nevertheless, the crystallographic refinement of Mn(II)-MntC indicates bidentate ligation of Mn(II) by Asp264, which would provide a five-coordinate Mn(II) site (Supplementary Table S2). Subsequent high-field electron paramagnetic resonance (EPR) spectroscopic evaluation of Mn(II)-PsaA and Mn(II)-MntC demonstrated that both proteins exhibit similar coordination environments for the bound Mn(II) ions.⁴⁴ This analysis also indicated that the Mn(II) ion is not five-coordinate and provided spectroscopic parameters similar to those previously determined for proteins harboring six-coordinate Mn(II) sites, but the possibility of a four-coordinate Mn(II) center could not be ruled out. Lastly, a recent molecular dynamics simulation supported octahedral Mn(II) and Zn(II) sites in PsaA.⁴⁵ Together, these observations indicate uncertainty about the metal ion coordination spheres in PsaA and MntC; however, different coordination numbers may occur under different sample conditions and during the process of each SBP binding a metal ion and delivering it to the permease. Consequently, additional biophysical studies to address the biological coordination chemistry of both SBPs are warranted.

To interrogate the Zn(II) coordination spheres of PsaA and MntC in solution, we prepared samples of the proteins (Supplementary Table S3) in the presence of 0.9 equiv of Zn(II) and analyzed them by K-edge XAS. The X-ray absorption near-edge structure (XANES) region of the spectra revealed that both Zn(II)-PsaA and Zn(II)-MntC samples share almost identical absorption features (Fig. 2A). The analyses of the XANES spectra suggested a similar coordination environment for the Zn(II) sites within both SBPs, and was most consistent with a four- or five-coordinate ligand environment.

Fig. 2

X-ray absorption spectroscopy of Zn(II)-bound PsaA and MntC. (A) XANES region of the Zn K-edge X-ray absorption spectrum of Zn(II)-PsaA (blank line) and Zn(II)-MntC (red line). (B) The κ³(χ) (left) and the magnitude Fourier transformed (FT) κ³(χ) (right) EXAFS spectra of Zn(II)-PsaA. Solid black spectra, experimental data; dashed blue spectra, simulated data. The EXAFS spectra of Zn(II)-MntC are provided in Supplementary Fig. S2. The fits of the EXAFS spectra are given in Supplementary Tables S4 and S5.

We next analyzed the extended X-ray absorption fine structure (EXAFS) regions of both Zn(II)-SBPs. For Zn(II)-PsaA, the best model for the EXAFS spectrum yielded average Zn–N^His and Zn–O bond lengths of 2.10 and 1.99 Å, respectively (Fig. 2B, Supplementary Table S4). Notably, the Debye–Waller parameters were more consistent with a five-coordinate model, according to the small or negative values obtained when only four donors were considered (five-coordinate: two Zn–N^His σ² = 0.003 Ų, three Zn–O σ² = 0.004 Ų; four-coordinate: two Zn–N^His σ² = –0.0018 Ų, two Zn–O σ² = –0.0017 Ų). The EXAFS regions in Zn(II)-MntC yielded considerable differences in the derived bond lengths of the N- and O-donors (Supplementary Fig. S2 and Table S5). In this case, the best model provided two Zn–N^His scatterers at 2.03 Å, which contributed significant multiple scattering pathways observed between 2.5 and 4.0 Å in the magnitude Fourier transformed κ³ EXAFS spectrum. This model also revealed two Zn–O scatterers at 2.02 Å, which are separable from the Zn–N^His pathways on account of the inclusion of the multiple scattering pathways within the imidazole functions. Nevertheless, the Debye–Waller parameters (five-coordinate: two Zn–N^His σ² = 0.003 Ų, three Zn–O σ² = 0.005 Ų; four-coordinate: two Zn–N^His σ² = 0.0005 Ų, two Zn–O σ² = –0.0017 Ų) were again more consistent with a five-coordinate model (vide supra).

Because both four- and five-coordinate modes provided similar fitting statistics, we performed a bond valence sum (BVS) analysis as previously described.^33,46,47 In an ideal case, the BVS values should match the oxidation state of the bound metal ion [i.e. 2.00 for Zn(II)]. When applied to the EXAFS data for Zn(II)-PsaA and Zn(II)-MntC, the BVS values were more consistent with five-coordinate sites [Zn(II)-PsaA = 2.10; Zn(II)-MntC = 2.14] than with four-coordinate sites [Zn(II)-PsaA = 1.64; Zn(II)-MntC = 1.72]. We also note that a six-coordinate model to the EXAFS data for both Zn(II)-SBPs resulted in physically unrealistic refinement parameters for the inner-sphere scatterers (σ² > 0.01 Ų) and an unacceptably large bond valence sum (BVS ≥ 2.6), suggesting the Zn(II) center is not six-coordinate. Taken together, the XAS data support five-coordinate Zn(II) sites composed of two Zn–N^His and three Zn–O ligands for both Zn(II)-PsaA and Zn(II)-MntC. The Zn–N^His bond lengths determined by XAS for Zn(II)-PsaA are consistent with those reported in the crystal structure,⁴² whereas the Zn–O bond lengths vary between the crystallographic and XAS results (Supplementary Table S2).

Co(II)-binding titrations indicate five-coordinate Co(II) sites in PsaA and MntC

To further characterize the metal-binding sites of PsaA and MntC in solution, we titrated both proteins with Co(II) and monitored complexation by optical absorption spectroscopy (Fig. 3). The electronic configuration of high-spin Co(II) (3d⁷) and the established correlations between Co(II) d–d transitions and coordination geometries allow its use as a spectroscopic probe to interrogate Zn(II) (3d¹⁰) sites.⁴⁸ Addition of 0–5 equiv of Co(II) to PsaA (300 μM) afforded a pink solution, and the optical absorption spectrum of Co(II)-PsaA exhibited ligand–field transitions centered at 515 nm (ε = 177 M^–1 cm^–1) and 562 nm (ε = 181 M^–1 cm^–1) (Fig. 3). These spectroscopic features, particularly the molar extinction coefficients of 250 > ε > 50 M^–1 cm^–1, are reminiscent of five-coordinate Co(II)-bound proteins.^49,50 Both four-coordinate tetrahedral (ε > 300 M^–1 cm^–1) and six-coordinate (ε < 50 M^–1 cm^–1) environments in which Co(II) is coordinated by N- and O-donors are not consistent with the data.⁴⁸ Likewise, addition of 0–5 equiv of Co(II) to MntC resulted in a similar absorption spectra, with d–d transitions centered at 520 nm (ε = 165 M⁻¹ cm^–1) and 552 nm (ε = 179 M^–1 cm^–1) (Fig. 3), supporting a five-coordinate Co(II)-MntC species.

Fig. 3

Co(II) binding to PsaA and MntC monitored by optical absorption spectroscopy. PsaA or MntC (300 μM) were titrated with 0–5 equiv of Co(II) at room temperature (75 mM HEPES, 100 mM NaCl, pH 7.0).

We subsequently evaluated metal substitution in PsaA and MntC by monitoring the dissociation of Co(II) ions from the Co(II)-SBPs upon addition of Zn(II). The addition of 1 equiv of Zn(II) resulted in a gradual disappearance of the absorption features of Co(II)-PsaA and Co(II)-MntC (Supplementary Fig. S3), indicating that both proteins have a higher thermodynamic preference for Zn(II) than Co(II). This behavior is expected based on the Irving–Williams series.⁵¹ The absorption spectra obtained after 1 h of incubation following Zn(II) addition revealed that ≈68% of MntC remained coordinated to Co(II) ions, whereas only ≈23% of PsaA remained Co(II)-bound. This comparison suggests that the metal substitution rate in PsaA is higher than in MntC, at least for Co(II) displacement by Zn(II).

PsaA and MntC bind Zn(II) with sub-nanomolar affinity

To evaluate the Zn(II) affinities of PsaA and MntC, we conducted competition experiments with small-molecule Zn(II) probes with apparent K_{d, Zn(II)} values ranging from the sub-nanomolar to low-micromolar. Colorimetric titrations using the Zn(II) probes Zincon (K_{d, Zn(II)} = 12.6 and 5.8 μM)^52,53 and Mag-Fura-2 (MF2, K_{d, Zn(II)} = 30 nM)⁵⁴ revealed that PsaA outcompetes both probes for 1 equiv of Zn(II) (Fig. 4A and B), indicating the expected binding stoichiometry of 1:1 Zn(II): PsaA and a low- or sub-nanomolar Zn(II) affinity for PsaA. Next, we examined the Zn(II) competition between PsaA and the fluorescent Zn(II) sensor Zinpyr-4 (ZP4, K_{d, Zn(II)} = 650 pM).⁵⁵ The addition of Zn(II) to a mixture of ZP4 and PsaA resulted in a gradual and attenuated increase in ZP4 emission compared to the ZP4-only control until an equimolar ratio of Zn(II) and PsaA was achieved, and maximum ZP4 turn-on occurred following the introduction of additional Zn(II) equivalents, indicating competition between PsaA and ZP4 for Zn(II) (Fig. 4C). The resulting titration curve was fitted to a one-site model, which afforded an apparent K_{d, Zn(II)}= 130 ± 30 pM (Fig. 4C and Supplementary Fig. S4). This value is more than three orders of magnitude lower than an apparent K_{d, Zn(II)}= 231 nM, which was previously reported from a direct Zn(II)-binding titration monitored by isothermal titration calorimetry (ITC) where stoichiometric binding was observed (Supplementary Table S1).^16,56 Competition titrations between MntC and the three Zn(II) probes gave the expected 1:1 Zn(II): MntC binding stoichiometry, and a fit of the ZP4 competition titration curve to a one-site model afforded an apparent K_{d, Zn(II)}= 120 ± 20 pM, similar to that determined for PsaA in the current work (Supplementary Figs. S4 and S5).

Fig. 4

Zn(II) competition titrations between PsaA and Zn(II) probes of varying affinities. (A) Response of Zincon (20 μM) to Zn(II) in the presence of PsaA (10 μM). The Zn(II)–Zincon complex absorbs at 621 nm. (B) Response of MF2 (10 μM) to Zn(II) in the presence of PsaA (10 μM). The Zn(II)-MF2 complex absorbs at 325 nm. (C) Response of ZP4 (1.5 μM) to Zn(II) in the presence of PsaA (5 μM). The increase in ZP4 fluorescence emission with Zn(II) binding (λ_ex = 495 nm) was integrated from 505 to 650 nm. Integrated emission values were normalized to maximum Zn(II)-induced turn-on emission. A one-site binding model was used to fit the normalized integrated emission data with DynaFit (red line).⁵⁷ Simulated fits with higher and lower K_d values are provided in Supplementary Fig. S4. All experiments were performed at room temperature in 75 mM HEPES, 100 mM NaCl, pH 7.0 (mean ± SD; n = 3). Zn(II)-competition titrations for MntC and the three probes are provided in Supplementary Fig. S5.

Zn(II) binding to MntC and PsaA is reversible

To investigate whether Zn(II) ions bound to PsaA are kinetically labile, we performed isotope-exchange experiments using the ⁷⁰Zn(II) isotope (0.62% natural abundance). We prepared Zn(II)-PsaA (20 μM) with the metal ion in its natural abundance (⁶⁶Zn,  27.7%) and incubated the sample with either 1 or 10 equiv of ⁷⁰Zn(II) (≈72% enriched) at room temperature. The Zn(II) isotope % excess was determined from the ⁷⁰Zn/⁶⁶Zn ratio of the protein sample obtained by inductively-coupled plasma mass spectrometry (ICP-MS) (Fig. 5 and Supplementary Table S8). This assay revealed an increase in the ⁷⁰Zn/⁶⁶Zn ratio over time, indicating Zn(II) exchange at the His₂GluAsp site of PsaA. For instance, when Zn(II)-PsaA was incubated with 1 equiv of ⁷⁰Zn(II), an ≈17% and ≈61% excess of ⁷⁰Zn(II) was observed after 1 and 96 h, respectively. In the presence of a 10-fold molar excess of ⁷⁰Zn(II), an ≈32% and ≈87% excess of bound ⁷⁰Zn(II) was observed after 1 and 96 h, respectively. These results indicate that while PsaA binds Zn(II) with high affinity, the bound metal is labile. ⁷⁰Zn(II) exchange experiments with MntC revealed similar behavior, albeit with a lower Zn(II)-exchange rate (Supplementary Fig. S6). For instance, in the presence of 1 equiv of ⁷⁰Zn(II), an ≈11% and ≈47% excess of ⁷⁰Zn(II) bound to MntC was observed after 1 and 96 h, respectively. Thus, Zn(II) ions bound to MntC appear to be slightly less labile compared to those bound to PsaA, in agreement with the differences in Co(II) displacement observed for these two proteins (Supplementary Fig. S3).

Fig. 5

Zn(II) isotope exchange of Zn(II)-PsaA in the presence of 1 (white bars) or 10 (gray bars) equiv of enriched ⁷⁰Zn(II) at room temperature (75 mM HEPES, 100 mM NaCl, pH 7.0). The ⁷⁰Zn(II) % excess was calculated by the ⁷⁰Zn/⁶⁶Zn ratio (Supplementary Table S8, n = 5). Analysis of Zn(II) isotope exchange of Zn(II)-MntC is provided in Supplementary Fig. S6.

Binding of Zn(II) ions to PsaA and MntC was previously described as irreversible based on competition experiments with the the high-affinity Zn(II) chelator ethylenediaminetetraacetic acid (EDTA) (K_{d, Zn(II)} = 26 fM),⁵⁸ where transfer of Zn(II) ions from either Zn(II)-SBP to EDTA was not observed.^19,23 Because EDTA is expected to outcompete both SBPs for Zn(II) on the basis of apparent K_{d, Zn(II)} values, we reasoned that a slow dissociation rate of Zn(II) from Zn(II)-SBPs may impede metal exchange between the proteins and EDTA. To test this notion, we treated Zn(II)-PsaA or Zn(II)-MntC (20 μM) with 100 equiv of EDTA at room temperature and analyzed the amount of Zn(II) retained by both SBPs by ICP-MS (Supplementary Fig. S7). In agreement with prior studies, the total Zn(II) removed from both Zn(II)-bound SBPs by EDTA after 1–4 h of incubation was negligible (<5%).^19,23 However, extending the incubation time to 16 and 96 h resulted in a decrease in the Zn(II) content in both proteins by 16% and 40%, respectively. Thus, both the Zn(II)-isotope exchange and competition experiments with EDTA demonstrate dissociation of Zn(II) from the SBPs. Although a significant exchange of Zn(II) between the SBPs and EDTA was observed only over the course of days, Zn(II) exchange with chelators that exhibit faster binding kinetics may occur on a shorter timescale. Along these lines, a combination of thermodynamic stability and kinetic lability has been demonstrated for zinc-finger proteins and metallothioneins and may also be the case for Zn(II)-bound SBPs.^59–61

Mn(II) is readily displaced from PsaA and MntC by Zn(II) and other first-row transition metals

We employed low-temperature continuous-wave (CW) EPR spectroscopy to investigate whether Zn(II) ions displace Mn(II) ions bound to PsaA and MntC. The low-temperature CW-EPR spectra of Mn(II)-PsaA and Mn(II)-MntC feature a broad signal spanning >600 mT in the g = 4.5 region at X-band frequency,⁶² whereas the signal of Mn(II) ions in buffer is more defined and appears as a six-line pattern centered at g = 2.0. Consequently, the EPR signals of the Mn(II)-SBPs and free Mn(II) in aqueous solution do not overlap and allow both species to be monitored simultaneously. We prepared samples of PsaA and MntC (400 μM) with 0.9 equiv of Mn(II) to yield the Mn(II)-SBPs, added 1 equiv of Zn(II) to each sample, and monitored the Mn(II) EPR signals over 16 h (see the Supplementary Information). We found that Mn(II) ions bound to both PsaA and MntC are readily displaced by Zn(II), evidenced by the disappearance of the Mn(II)-SBP features and the formation of the signal for buffered Mn(II) ions after 5 min of incubation (Fig.6 and Supplementary Figs. S8–S9). Further changes in signal intensities were not observed at later time points. Thus, Zn(II) ions readily replace Mn(II) ions bound to PsaA and MntC, and this Mn(II)/Zn(II) exchange occurs more rapidly than Co(II)/Zn(II) exchange (Supplementary Fig. S3), as expected based on the Irving–Williams series.⁵¹

Fig. 6

Displacement of Mn(II) ions from Mn(II)-PsaA by Zn(II) monitored by low-temperature X-band CW EPR spectroscopy. PsaA (400 μM) was incubated with 0.9 equiv Mn(II) for 30 min at room temperature to form Mn(II)-PsaA (black line) (75 mM HEPES, 100 mM NaCl, pH 7.0). Following 5 min of incubation at room temperature with 1 equiv of Zn(II), the spectra reveal free Mn(II) ions in solution, indicating that Mn(II) ions dissociated from PsaA (blue line). The y-axis scale of panel A is 10-fold enlarged relative to panel B to compare the low-field features of Mn(II)-PsaA to the mid-field features of buffered Mn(II) ions. EPR spectra obtained after 1 and 16 h of incubation is provided in Supplementary Fig. S8. EPR analysis of Mn(II) displacement from MntC is provided in Supplementary Fig. S9. Instrument conditions: ν_mw = 9.38 GHz, 1.5 mT modulation at 100 KHz, power = 0.2 mW, and temperature = 10 K.

To probe the tendency of Mn(II) ions bound to PsaA and MntC to undergo transmetalation in a more complex environment that contains multiple first-row transition metal ions and is representative of conditions employed in microbiology studies, we performed metal-depletion assays based on our prior studies of S100 proteins.^34,63 We employed Tris:TSB growth medium, a 62:38 mixture of Tris buffer and Tryptic soy broth (TSB), which is commonly employed in microbiology studies of CP, for these studies. Tris:TSB contains ≈0.2 μM Mn, ≈5 μM Fe, ≈0.2 μM Ni, ≈6 μM Zn, and <100 nM of Co and Cu. We treated Tris:TSB with 10 μM of (i) apo-PsaA, (ii) Mn(II)-PsaA—PsaA preincubated with 1 equiv of Mn(II), and (iii) Zn(II)-PsaA—PsaA preincubated with 1 equiv of Zn(II). Following 20 h of incubation at 30 °C, PsaA and any tightly-bound metal ions were removed from the medium by spin-filtration, and the metals remaining in the filtrate were quantified by ICP-MS (Fig.7 and Supplementary Fig. S10). This assay revealed that apo-PsaA depleted multiple first-row transition metal ions, including Mn, Fe, and Zn, from Tris:TSB. The ability of Mn(II)-PsaA to deplete Fe and Zn from Tris:TSB was comparable to that of apo-PsaA, which resulted in the release of the Mn(II) ions initially bound to the protein into the growth medium (Fig. 7A). In contrast, Zn(II)-PsaA was ineffective at depleting Mn, and Mn levels of the Zn(II)-PsaA-treated medium were comparable to those of the untreated medium. Similar metal-depletion profiles were obtained with apo-MntC, Mn(II)-MntC, and Zn(II)-MntC (Supplementary Fig. S10). Overall, these data show that Mn(II) ions bound to both SBPs are prone to metal substitution by transition metals later in the first row of the Periodic Table and further highlight the thermodynamic stability of both Zn(II)-SBPs.

Fig. 7

Analysis of Mn (A), Fe (B), and Zn (C) concentrations in Tris:TSB medium treated with 10 μM of PsaA. The metal content of untreated medium (white bars), untreated medium after spin filtration (light gray bars), and medium treated with apo-PsaA (medium gray bars), Mn(II)-PsaA (dark gray bars), and Zn(II)-PsaA (teal bars) was measured by ICP-MS (n = 5; *P < 0.002, **P < 0.05, ***P ≤ 0.0001 for comparison with untreated medium after spin filtration). The Fe and Zn levels in untreated Tris:TSB medium decrease after spin-filtration, presumably because high-molecular Fe- and Zn-containing biomolecules are also retained by the spin-filter. Complete metal analysis is provided in Supplementary Fig. S10 and Tables S6 and S7. Metal analysis of medium treated with apo-MntC, Mn(II)-MntC, and Zn(II)-MntC is provided in Supplementary Fig. S10.

CP sequesters Zn(II) from PsaA and MntC

We reasoned that CP may attenuate Zn(II) toxicity in PsaA and MntC by lowering bioavailable Zn(II) levels in the extracellular space or by directly sequestering Zn(II) from the SBPs. Given the ability of CP to outcompete both PsaA and MntC for Mn(II) under conditions of high Ca(II),⁶² the Zn(II)-sequestering activity of CP,^30,33 and the relative apparent K_{d, Zn(II)} values of CP and these SBPs (Supplementary Table S1), we reasoned that CP may outcompete PsaA and MntC for Zn(II). To evaluate both possibilities, we employed a pull-down assay using biotinylated CP (B-CP).^35,62 We prepared solutions containing equimolar amounts (20 μM) of either SBP and B-CP, added 0.9 equiv (18 μM) of Zn(II), and incubated the resulting samples for 16 h at room temperature. Subsequent pull-down of B-CP with streptavidin resin resulted in <45% of the Zn(II) remaining in solution for all samples (Fig. 8A). Extending the incubation time to 96 h had minimal effect on the Zn(II) speciation between CP and either SBP. These results indicate that the Zn(II) speciated between Zn(II)-CP and Zn(II)-SBPs with slightly more Zn(II) bound to CP. We next investigated the ability of B-CP to sequester Zn(II) from Zn(II)-PsaA and Zn(II)-MntC by preincubating each SBP (20 μM) with 0.9 equiv of Zn(II) before adding B-CP (20 μM). Following pull-down of B-CP, ≈70% of the Zn(II) remained in solution after 4 and 16 h of incubation at room temperature (Fig. 8B). Thus, although CP binds Zn(II) with higher affinity than PsaA and MntC, under these experimental conditions, it only sequestered ≈30% of Zn(II) initially bound to PsaA and MntC. In all cases, addition of 20 equiv Ca(II) had a negligible effect on the Zn(II) speciation between CP and the SBPs. Overall, these results indicate that CP can prevent Zn(II) from saturating the SBPs by chelating available Zn(II) or by sequestering Zn(II) bound to the SBPs.

Fig. 8

Zn(II) competition between B-CP and SBPs. Zn(II) levels in solution were determined by ICP-MS following pull-down of B-CP. (A) Mixtures containing B-CP, PsaA, or MntC (20 μM for all proteins), and Zn(II) (18 μM) with or without 20 equiv Ca(II) (75 mM HEPES, 100 mM NaCl, pH 7.0). The mixtures were incubated for 16 h (light gray bars) or 96 h (dark gray bars) at room temperature before pull-down with streptavidin resin. The white bars are the mixtures at the 0 h and without treatment with streptavidin resin. (B) Mixtures containing B-CP and Zn(II)-SBPs with or without Ca(II) (75 mM HEPES, 100 mM NaCl, pH 7.0). The mixtures were incubated for 4 h (light gray bars) or 16 h (dark gray bars) at room temperature before pull-down with streptavidin resin. (n = 3; *P < 0.03, **P < 0.002 for comparison with the mixtures). Data for the ΔHis₃Asp and ΔHis₄ site variants are provided in Supplementary Fig. S13.

Prior studies demonstrated that both the His₃Asp and His₆ sites of CP bind Zn(II) with high affinity in the absence and presence of Ca(II) (Supplementary Table S1).^30,33 We therefore investigated the ability of the isolated His₆ and His₃Asp sites to sequester Zn(II) from the SBPs to ascertain the contribution of each site. We evaluated the Zn(II) speciation between the biotinylated CP variants B-ΔHis₃Asp³⁵ and B-ΔHis₄ (see the Supplementary Information, Tables S3 and S9–S12, Figs. S11 and S12) and each Zn(II)-SBP. After pull-down of either B-ΔHis₃Asp or B-ΔHis₄, ≈70% of the Zn(II) remained in solution (Supplementary Fig. S13). This result indicates that both the His₃Asp and His₆ sites have a comparable ability to sequester Zn(II) from the SBPs. This result was also surprising because the same concentration of B-CP and the biotinylated site variants was employed in these experiments; thus, the samples treated with B-ΔHis₃Asp or B-ΔHis₄ have 50% of the binding sites as CP but provide a comparable reduction in Zn(II) remaining in solution after pull-down. Lastly, these experiments also revealed that the time scale required for the transfer of Zn(II) ions from each SBP to CP is relatively short (≤4 h), particularly in comparison to transfer from each SBP to EDTA (>16 h).

CP promotes Mn(II) binding to Zn(II)-SBPs

The ability of CP to sequester Zn(II) from PsaA and MntC led us to question whether it can promote Mn(II) binding to the SBPs in the presence of Zn(II). We employed EPR spectroscopy to evaluate Mn(II) speciation in samples that contained mixtures of Zn(II)-SBP, Mn(II), and CP-Ser [S100A8(C42S)/S100A9(C3S), Supplementary Table S3] prepared with or without a Ca(II) supplement.⁶⁴ When CP-Ser was added to samples of Zn(II)-PsaA or Zn(II)-MntC (300 μM for all proteins) that contained 1 equiv of Mn(II), a mixed population of free Mn(II) ions and Mn(II)-CP was observed (Fig.9 and Supplementary Fig. S14). The addition of excess Ca(II) resulted in increased resolution and sharpening of the allowed and semi-forbidden transitions of the Mn(II)-CP signal, in agreement with prior work.^32,65 In all cases, the signals of Mn(II)-PsaA or Mn(II)-MntC were absent from the EPR spectra. Because the His₆ site of CP coordinates Mn(II) ions with high affinity and is proposed to kinetically trap the bound metal ion,⁶⁵ this site may be unable to sequester Zn(II) from the Zn(II)-SBPs under these conditions. Nevertheless, the His₃Asp site has low Mn(II) affinity³² and is expected to remain functional at Zn(II) sequestration when the His₆ site is occupied by a metal ion. Thus, we repeated this assay using 1.5 equiv Mn(II) ions such that this His₆ site is fully bound and excess Mn(II) ions remain in solution. We observed weak low-field signals corresponding to Mn(II)-PsaA and Mn(II)-MntC under these conditions, and the intensity of the Mn(II)-CP signal was unchanged (Fig.9 and Supplementary S14). Thus, the interplay between the His₃Asp site of CP and the Zn(II)-SBPs can result in Mn(II) binding to the SBPs after CP uses the His₃Asp site to sequester Zn(II) from the SBP. Taken together, these data further demonstrate that CP can compete with both PsaA and MntC for Zn(II), and promote binding of thermodynamically less preferred metals ions as Mn(II) to these SBPs.

Fig. 9

Binding of Mn(II) to PsaA that was initially in the Zn(II)-bound form in the presence of CP-Ser monitored by low-temperature X-band CW EPR spectroscopy. Zn(II)-PsaA (300 μM) was incubated with 1 equiv of Mn(II) for 30 min at room temperature (black line). Then, CP-Ser (300 μM) was added in the absence (blue line) and presence (red line) of 20 equiv of Ca(II), and with additional 150 μM of Mn(II) (green line). The resulting mixtures were incubated for 16 h at room temperature before analysis. All samples were prepared in 75 mM HEPES, 100 mM NaCl, pH 7.0. The y-axis scale of the low-field region is 100-fold enlarged to compare the features of Mn(II)-PsaA to the mid-field features of Mn(II)-His₆ or buffered Mn(II) ions. EPR analysis of Mn(II) binding to Zn(II)-MntC in the presence of CP-Ser is provided in Supplementary Fig. S14. Instrument conditions: ν_mw = 9.38 GHz, 1.5 mT modulation at 100 KHz, power = 0.2 mW, and temperature = 10 K.

Conclusion and perspectives

The interplay among microbial pathogens, the host innate immune system, and zinc is multifaceted and involves Zn(II) as a nutrient as well as a toxic element when in excess. During pneumococcal infection, the mammalian host uses Zn(II) to limit S. pneumoniae growth by Zn(II) intoxication.^16,66 The mammalian host also deploys host-defense proteins such as CP against S. pneumoniae, which presumably sequesters Zn(II) and other essential metal nutrients from the pathogen.⁴⁰ Although CP exhibits antimicrobial activity against S. pneumoniae,⁴⁰ one study reported that CP-deficiency in a murine model of pneumococcal infection attenuated pathogenesis.³⁹ That observation raised the possibility that S. pneumoniae benefits from the Zn(II)-sequestering ability of CP and afforded a model in which CP protects S. pneumoniae from Zn(II) intoxication. In this study, we provide a molecular basis for this model and present data that suggest CP can reduce Zn(II) binding to PsaA as well as sequester Zn(II) from PsaA, and thereby facilitate Mn(II) binding to the SBP. Moreover, we extend this study to staphylococcal MntC and obtain similar results. Overall, this work provides new molecular insights into the interplay between bacterial metal-transport systems and the host metal-withholding response.

The molecular model of S. pneumoniae susceptibility to high levels of Zn(II) has been proposed on the basis of irreversible Zn(II) binding to PsaA and the formation of a highly-stable closed conformation that precludes Mn(II) binding to the SBP and its subsequent transport into the cytoplasm.^19,67 Our results provide a more nuanced picture and show that PsaA has a thermodynamic preference for Zn(II) over Mn(II). The apparent K_{d, Zn(II)} value we obtained is approximately three orders of magnitude lower than the K_{d, Zn(II)} value reported previously.¹⁶ Moreover, PsaA was reported in that prior study to bind Mn(II) more tightly than Zn(II) by two orders of magnitude, contrary to the expected binding preferences given by the Irving–Williams series. In contrast to prior work, our experiments show that Zn(II) bound to PsaA is kinetically labile and can undergo a slow exchange with Zn(II) ions present in the buffer or be captured by a competing chelator such as EDTA or CP following dissociation from the SBP. These processes require a reversible transition between closed and open states of PsaA,¹⁹ indicating that metal binding to PsaA is thermodynamically controlled.

Zn(II) intoxication has not been reported for S. aureus to date. Here we show that staphylococcal MntC exhibits very similar Zn(II)-binding properties to PsaA, including a strong thermodynamic preference for coordinating Zn(II) over Mn(II). Consequently, elevated levels of Zn(II) may also poison MntABC and prevent Mn(II) uptake by this transporter in S. aureus. Prior studies of the interplay between S. aureus and CP have focused on extracellular environments, including tissue abscesses in which Zn(II) levels are relatively low.⁶⁸ Nevertheless, S. aureus is a facultative intracellular pathogen and can be present in complex microbial communities, including respiratory co-infections that involve S. pneumoniae.⁶⁹ Thus, it is possible that S. aureus experiences Zn(II) intoxication in certain host contexts. Along these lines, our results also demonstrate that other transition metals, such as Fe, are able to displace Mn(II) ions bound to PsaA and MntC. In particular, both SBPs are unable to retain bound Mn(II) when incubated in microbial growth medium and depleted the medium of Fe and Zn, in agreement with Mn(II) being at the weaker binding end of the Irving–Williams series. Deciphering the physiological consequences of Zn as well as Fe binding to MntC for S. aureus is warranted.

The current work further informs the biological coordination chemistry of PsaA and MntC by evaluating the Zn(II)-His₂GluAsp sites in solution using XAS. Continued investigation of these sites is important because ambiguity regarding the coordination number and geometry of the metal centers exists (vide supra). Our XAS analysis revealed that both proteins have very similar Zn(II) sites, and the spectra are best fitted with a five-coordinate N₂O₃ model, indicating that one of the acidic residues provides bidentate coordination to the Zn(II) center. Several possible explanations for the discrepancies between the various structural and spectroscopic studies of these proteins exist. For instance, the coordination chemistry of these SBPs may be sensitive to the chemical environment and sample conditions, or the coordination number may change depending on whether the SBP is engaged with its permease. Further investigations are needed to substantiate these possibilities, and studies of the fully reconstituted transport systems should be informative.⁷⁰

In closing, CP may prevent Zn(II) poisoning of Mn(II) transporters by lowering the bioavailability of Zn(II) ions or by sequestering Zn(II) from SBPs. Although both the His₆ and His₃Asp sites of CP sequester Zn(II) from PsaA and MntC, only the His₃Asp site promoted Mn(II) binding to the SBPs in the presence of Zn(II) under the conditions employed in this work. In prior studies, the ability of the His₆ site to sequester multiple transition metal ions was proposed to involve metal ion entrapment by the C-terminal tail of S100A9.^27,33,65,71 Indeed, the His₆ site exhibits negligible metal exchange and thus remains populated by the metal ion that binds first. In contrast, the His₃Asp site exhibits fast metal exchange and selects for Zn(II) in the presence of other divalent transition metal ions.⁷² The His₆ site has received more attention than the His₃Asp site because of its unusual amino acid composition and functional versatility. Nevertheless, the His₃Asp site is also important, and the current results highlight that it functions in Zn(II) sequestration. In the biological milieu, we expect that the relative contributions of the His₃Asp and His₆ sites to sequestering Zn(II) and promoting Mn(II) binding to SBPs will be largely determined by metal availability at an infection site, and may also be affected by the local concentration of other metal-withholding proteins such as the Zn(II)-sequestering protein human S100A12.⁷³

Acknowledgements

We thank Ms. Yu Gu for performing initial experiments with CP, PsaA, and Zn(II). The ICP-MS instrument at MIT is maintained by the MIT Center for Environmental Health Sciences (NIH P30-ES002109). CD spectroscopy instrumentation is provided by the MIT Biophysical Instrumentation Facility for the Study of Complex Macromolecular Systems, which is supported by National Science Foundation (NSF) grant 0070319. EPR spectroscopy and Q-TOF-MS instrumentation is provided by the MIT Department of Chemistry Instrumentation Facility. Zn K-edge X-ray absorption spectroscopy was performed at the Canadian Light Source, a national research facility of the University of Saskatchewan, which is supported by the Canada Foundation for Innovation (CFI), the Natural Sciences and Engineering Research Council (NSERC), the National Research Council (NRC), the Canadian Institutes of Health Research (CIHR), the Government of Saskatchewan, and the University of Saskatchewan.

Funding

This work was supported by the National Institutes of Health (NIH) Grants R01 GM118695 (EMN) and R15 GM141650 (JS).

Conflicts of interest

The authors declare no conflicts of interest.

Data availability

The data underlying this article are available in the article and in its online supplementary material.