Cryo-electron microscopy structure of a zinc uptake ABC transporter

Summary:

Zn²⁺ is an essential micronutrient required for many biological processes. Microorganisms employ high-affinity Zn²⁺ ABC transporters to take up zinc from the environment when it is scarce, yet the mechanism of zinc uptake and its regulation remain poorly understood. Here, we report the cryo-electron microscopy structure of the Zn²⁺ ABC transporter complex ZnuB-ZnuC from Escherichia coli. It contains two ZnuB transport and two ZnuC regulatory subunits. The ZnuB homodimer is in an outward-facing, closed conformation with a large hydrophilic cavity at the dimer interface. Each ZnuC subunit contains an Nterminal nucleotide-binding domain and a C-terminal zinc-sensing domain (ZSD). Zn²⁺ binding to the ZSD locks the transporter in a closed state, whereas under low-Zn²⁺ conditions, the disorder of ZSD permits ATP-driven zinc uptake. Zn²⁺ ABC transporters are ubiquitously utilized by pathogenic bacteria to compete with hosts for Zn²⁺. These findings reveal potential therapeutic targets for disrupting Zn²⁺ homeostasis in antibioticresistant pathogens.

Graphical Abstract

eTOC:

Pang et al. present structural and functional analyses on the bacterial ZnuB-ZnuC heterotetrameric zinc ABC transporter. The ZnuB homodimer contains a large hydrophilic cavity at its interface for Zn²⁺ transport, while each ZnuC subunit features an intracellular Zn²⁺-sensing domain to autoregulate the transport activity.

Introduction

The zinc ion (Zn²⁺) is an essential micronutrient supporting all living organisms. It plays key functions in gene transcription, structure rigidity, development, photosynthesis, immune response, and numerous Zn²⁺-dependent enzymatic activities¹. Zn²⁺ is impermeable to membranes and has to be transported across cell membranes by Zn²⁺ uptake transporters, including Zrt-/Irt-like proteins (ZIPs)^2,3, ATP-binding cassette (ABC) transporters⁴, and natural resistance-associated macrophage proteins (NRAMP)⁵ before being used by cells. While Zn²⁺ is an essential micronutrient, its excessive accumulation is cytotoxic, and its intracellular concentration has to be tightly regulated. As revealed by Zn²⁺ uptake assays, Zn²⁺ ABC transporters are high-affinity transporters whose expression levels are tightly regulated by the Zn²⁺-sensing transcription factor Zur⁶. Many bacterial pathogens use ABC transporters to compete for Zn²⁺ with their hosts - humans, animals, and plants. Conversely, upon sensing a pathogen attack, the hosts produce Zn²⁺-chelating proteins and molecules to limit the access of Zn²⁺ to be used by pathogens, forming an adaptive immunity called nutritional immunity^7–16. Antibiotic resistance is a growing global health crisis. A recent analysis reported that AMR caused 1.27 million deaths directly in 2019, in addition to contributing an additional 5 million indirect deaths¹⁷. These pathogen-specific ABC transporters may be attractive targets for the development of antibiotics and vaccines. However, the structure and mechanism of function of Zn²⁺ ABC transporters, and how their activities are regulated and linked to intracellular Zn²⁺ contents, remain largely unclear.

ABC transporters constitute a large family of transporters that are ubiquitous across all living organisms. They use ATP hydrolysis to drive the transport of ions, metabolites, proteins, nucleotides, and other biological molecules across membranes^18,19. Many human ABC transporters are also important therapeutic targets¹⁸. The ABC transporters contain at least two ATP-binding cassette (ABC) domains (also called nucleotide-binding domains, NBDs) and two transmembrane domains (TMDs). The NBD regulates the transport activity in the TMD through ATP binding and hydrolysis. ZnuABC is an E. coli ABC transporter composed of five subunits: two transmembrane ZnuB subunits, two ATP-binding cassette (ABC) regulatory ZnuC subunits, and one Zn²⁺-binding protein ZnuA²⁰.

In this work, we present the cryo-electron microscopy (cryoEM) structures of the high-affinity Zn²⁺ ABC transporter ZnuB-ZnuC, hereafter referred to as the tetrameric ZnuBC complex from E. coli. The structures are resolved in outward-facing Zn²⁺-bound as well as Zn²⁺-free states. We identify a Zn²⁺-sensing domain (ZSD) at the C terminus of the regulatory subunit ZnuC. The ZSD senses intracellular Zn²⁺ levels and regulates transport activity. Together with functional assays and computational analysis, these structures reveal a mechanism of Zn²⁺ uptake and autoregulation mediated by a built-in intracellular Zn²⁺ sensor.

Results

Structure determination

The expression of ZnuA, ZnuB, and ZnuC in E. coli is regulated by the zur transcription factor, as illustrated in Figure S1A. We cloned the E. coli ZnuA and ZnuBC genes into the pETDuet1 vector with one cassette containing ZnuBC and the other cassette containing ZnuA (Figure S1B). We expressed the complex in bacteria, purified it in detergent, and reconstituted it into amphipol A8–35 using a workflow summarized in Figure S1C. The complex contained ZnuB and ZnuC (ZnuBC), lacking ZnuA (Figure S1D–E). We determined its structure using single-particle cryo-electron microscopy (CryoEM). To obtain the structure, we prepared the ZnuBC particles (Figure S2A) in the presence of adenylyl-imidodiphosphate (AMP-PNP), a non-hydrolyzing ATP analog. Averaged 2D classes showed multiple views of the ZnuBC complex (Figure S2B). The reconstructed 3D map shows a belt of amphipol densities around the transporter (Figure S2C). The cryoEM data analysis workflow is summarized in Figure S3. The reconstructed cryoEM map used 34,187 particles, reaching a resolution of 2.5 Å (Figure S4A). The final reconstruction contains particles with a widespread distribution in Euler space with slightly more views at elevation angles around zero (Figure S5A). The cryoEM density is of high quality, showing all nine transmembrane helices and most side-chain features in a transporter subunit of ZnuB (Figure S6).

Overall structure

The ZnuBC structure is a heterotetramer, consisting of two copies of ZnuB and ZnuC subunits (Figure 1A–B). ZnuB is the zinc transporter subunit. The two ZnuB subunits form a dimer with an interface of 1503 Ų as calculated by PISA²¹. Each ZnuB monomer contains nine TM helices (TM1-TM9) and two short α-helices: α1 in the periplasm and α2 in the cytosol (Figure 1B–C). Most TM helices are bent or tilted inside the membrane. TM5 is long and extends from the membrane into the cytosol, followed by α2, which is known as a coupling helix in ABC transporters¹⁹. TM6 and TM7 do not fully cross the membrane; they form a kink inside the membrane (Figure 1C). The intramembrane loop between TM6 and TM7, L(6,7), has close contact with the periplasmic helix α1, which connects to TM4 and TM5 through two long loops (Figure 1C).

Figure 1. Overall structure of the ZnuB-ZnuC (ZnuBC) transporter.

(A) CryoEM density map of the ZnuBC subunits of the ZnuABC transporter. One subunit is colored rainbow, and the other, gray. (B) Overall structure of the ZnuBC dimer. The coloring is as in panel A. Four Zn²⁺ ions are shown as red spheres. Two AMP-PNP molecules are shown as magenta sticks. (C) Topology diagram of a ZnuB subunit. (D) Topology diagram of a ZnuC subunit. See also Figures S1 and S2.

ZnuC is the regulatory subunit, located in the cytosol. It interacts with ZnuB mainly through the α2 on ZnuB and a Q-loop on ZnuC (Figure 1B). The two ZnuC subunits form a dimer with an interface of 1606 Ų. Each ZnuC monomer contains an AMP-PNP molecule bound to its NBD domain. In addition to the NBD domain, the C-terminus of ZnuC has an extended domain consisting of three β-strands (β11–13), two short α-helices (α6 and α7), and a long loop (Figure 1D). Surprisingly, the extension harbors two Zn²⁺ ions in each of the ZnuC subunits (Figure 1B).

ZnuB zinc transporter

The ZnuB dimer has a two-fold symmetry; TM2, TM4, TM7, and TM9 from each monomer form the dimer interface (Figure 2A–B). A 40 Å-long transport channel from the periplasmic side to the cytosol is at the dimer interface (Figure 2C–D), lined by TM2 and TM4 residues of each subunit. These two helices on each ZnuB subunit bend to expose a vestibule to the periplasm with the α1 helices inserted as a plug to constrict the channel. Therefore, we interpret this structure as an outward-facing, closed state. The channel is wide in the middle and has two restriction bottlenecks. On the periplasmic side, the bottleneck is formed by the hydrophobic residues Met112, Leu115, and Phe116 from α1 and Leu190 from the loop L(6,7). On the cytosol side, the bottleneck is formed mainly by Tyr33 residues from both subunits.

Figure 2. Structure of ZnuB dimer.

(A-B) Two views of the overall structure of the dimer, viewed from the side and the cytosol, respectively. The dimer has a twofold symmetry. (C) The membrane transport channel (magenta surface) is lined by the labeled TM2 and TM4 residues from each monomer. (D) Quantification of the size and hydrophobicity of the transport channel. (E) An elongated cavity at the center of the transmembrane region is formed by histidine and aspartate residues. The surface color shows the conservation level: magenta, more conserved; cyan, relatively less conserved. (F) Functional characterization of zinc uptake and cell viability by removing Zn²⁺ using EDTA. See also Figures S7 and S8.

Between the two bottlenecks, there is a large cavity of about 13 Å long and 6 Å wide as calculated by the program MOLE²². The cavity is hydrophobic on the periplasmic side, formed by residues Leu94 and Leu190 (Figure 2C–D). The hydrophobic residues might help seal the transporter to prevent a leaky uptake of Zn²⁺. The cytoplasmic side is hydrophilic, formed by conserved residues Asp36, His40, and a less conserved residue His91 (Figures 2E and S7). The cavity is elongated and has enough space to accommodate one Zn²⁺ inside. We hypothesized that the side chains of the conserved Asp36 and His40 might contribute to the Zn²⁺ transport activity. We thus produced D36A, H40A, and D36A-H40A mutants, transformed them with bacterial cells, and measured cell growth viability in the presence of Ethylenediaminetetraacetic acid (EDTA) (Figures 2F and S8A). EDTA treatment removed Zn²⁺ from uptake by bacteria. Compared to the plasmid control and wild-type (WT), the D36A-H40A double mutant did not grow well under Zn²⁺-limiting conditions, treated with 2 mM EDTA, suggesting their essential role in Zn²⁺ uptake. To test if His91 is involved in the ZnuB transport function, we produced H91A, D36A-H91A, and H40A-H91A mutants and performed cell viability assays in the presence of various concentrations of EDTA (Figure S8B), we found that cells overexpressing H91A grew similarly to the wild-type; D36A-H91A and H40A-H91A double mutants remain active in Zn²⁺ uptake, suggesting that His91 is not as important as Asp36 and His40 in Zn²⁺ transport. In addition, bacterial cells transformed with WT grew better than the control under the Zn²⁺-limiting condition, implicating that ZnuB was expressed and contributed to the uptake of Zn²⁺ from the Zn²⁺-limiting environment.

ZnuC nucleotide-binding domain

ZnuC is the regulatory subunit of the Zn²⁺ ABC transport system. The solved AMP-PNP-bound ZnuC dimer contains two AMP-PNP molecules and four Zn²⁺ ions (Figure 3A–B). The AMP-PNP molecule binds at the N-terminus of α1, which is sandwiched by two β-sheets. In many ABC transporters, ATP or AMP-PNP promotes NBD dimerization through nucleotide-mediated inter-subunit interactions²³. However, in the ZnuC dimer structure, AMP-PNP has no interactions with the other ZnuC subunit, indicating not involved in dimerization (Figures 3A and S9A).

Figure 3. Structure of the ATP-binding cassette subunit ZnuC.

(A-B) Two views of the AMP-PNP-bound structure of the ZnuC dimer. The coloring is the same as Figure 1B. (C) Interactions between AMP-PNP and the NBD of ZnuC. The cyan sphere is an Mg²⁺ ion, coordinated by S44, Q74, and AMP-PNP. See also Figure S9.

AMP-PNP binds to the NBD primarily via the Walker A motif (Figure 3C). Its phosphate groups form hydrogen bonds with Gly42, Gly43, and Thr45, while Ser44 and two phosphate oxygens coordinate the Mg²⁺ ion that is also liganded by Gln74 in the Q-loop (residues 74–85). The side chain of the Walker B catalytic residue, Glu146, points directly at the γ-phosphate, positioning it for hydrolysis. Meanwhile, the adenine ring of AMP-PNP π-stacks against Phe14 in β2.

We also produced an ATPase-deficient ZnuBC E146Q mutant and solved its cryoEM structure in the presence of ATP at 2.9 Å resolution (Figures S4B and S5B). The solved structure is very similar to the AMP-PNP-bound structure; neither ATP is involved in the dimerization of ZnuC (Figure S9B). All of these contacts shown in the AMP-PNP binding are retained in the ATP-bound E146Q mutant structure (Figure S9B).

To assess nucleotide-binding induced conformational changes, we solved an apo-form, nucleotide-free ZnuBC structure (Figures S4C and S5C) and overlaid it with the AMP-PNP–bound structure (Figure S9C). ZnuB, including the coupling helix α2 (residues 154–162), shows virtually no shift (RMSD 0.337 Å over 260 ZnuB Cα atoms), and ZnuC also remains largely unchanged (RMSD 0.455 Å over 242 ZnuC Cα atoms). The notable conformational changes localize to the nucleotide-binding pocket (Figure S9C). In the ZnuC apo structure, the sidechain of Gln74 in the Q-loop reoriented due to the loss of interaction with the Mg²⁺. However, the Q-loop backbone itself is unaltered. Phe14 in β2 and the associated β-sheet shifted due to the loss of the π stacking with the nucleotide adenine ring. In the ZnuB apo structure, there is a movement of the L(3,4) loop (residues 73–82), likely due to its interactions with the β-sheet in ZnuC. Compared with the apo structure, ATP or AMP-PNP binding at the Walker A site does not trigger significant global rearrangements in either ZnuB or ZnuC.

ZnuC Zn²⁺ binding and autoregulation

In both the apo and nucleotide-bound ZnuBC structures, the C-terminus of each ZnuC monomer harbors two Zn²⁺ sites, M1 and M2 (Figure 4A). M1 is coordinated by Cys193, Cys200, His230 and His232; M2 by His197, Cys199, His234 and His246. These two Zn²⁺ ions lock the ZnuC C-terminus into an ordered conformation, and helices α6–α7 from each monomer pack tightly to form a dimer interface burying 807 Ų (Figure 3A). This Zn²⁺-stabilized dimer interface via helices α6–α7 might prevent the ATP-induced global conformational switch, confining movement to local changes around the nucleotide pocket (Figure S9C). To assess the specificity of ZnuC to Zn²⁺ and other divalent metals, we performed microscale thermophoresis (MST) using Zn²⁺, Co²⁺, Cd²⁺, Cu²⁺, Fe²⁺, and Mn²⁺ (Figure S10). These experiments showed that Zn²⁺ binds with the highest affinity, followed by Co²⁺, while the other metals exhibited no detectable binding under our assay conditions, suggesting that the regulatory mechanism of ZnuBC is highly selective for Zn²⁺.

Figure 4. Intracellular zinc-sensing and transport cycle regulation by ZnuC ZSD.

(A), Zn²⁺ binding sites M1 and M2 at the ZnuC ZSD. The inset shows the cryoEM density for the M1 and M2 sites. (B) Cell growth viability assays of ZnuBC wild-type and mutants. E. coli C43 cells transformed with the indicated plasmids for plating LB agar plates with buffer or 1 mM Zn²⁺ and the presence of 0.2 mM IPTG (final). Compared with the control, which has an empty plasmid, M1 or M1M2 double mutants in ZnuC showed cell toxicity. (C) ATPase activity of ZnuBC WT and the ZSD M1 and M2 mutants in the presence of TPEN or Zn²⁺. WT displays lower activity compared with the mutants, and the addition of Zn²⁺ reduced the activity, compared with Zn²⁺-depletion by TPEN, a Zn²⁺-specific chelator. The error bars indicate the standard deviations of the mean (n=3). (D) CryoEM structure of EDTA-treated Zn²⁺-free state. Compared to the apo structure, the two ZnuC monomers moved away from each other by 3.6 Å. (E) Superimposed apo and EDTA-treated structures to show conformational changes associated with Zn²⁺ sensing and regulation. Arrows show rotations of ZnuC relative to the respective interacting ZnuB. See also Figures S10–S16.

We therefore asked whether M1 and M2 sites regulate Zn²⁺ uptake. We replaced each coordinating Cys/His with alanine, expressed the mutants in E. coli, and compared bacterial growth in media containing Zn²⁺ (Figures 4B and S11). The cells expressing WT were less viable than the control in the presence of 1 mM Zn²⁺, suggesting increased intracellular Zn²⁺ loading and associated cytosolic toxicity. The cells expressing M2 mutant behaved like control, but the cells expressing M1 or M1M2 double mutants showed markedly reduced viability under 1 mM Zn²⁺, consistent with the loss of regulation, leading to intracellular Zn²⁺ overloading and cytosolic toxicity. We therefore hypothesize that the C-terminal domain of ZnuC (residues 193–251) could function as an intracellular Zn²⁺ sensor and designate it as the zinc-sensing domain (ZSD).

To further confirm the role of ZSD as an intracellular Zn²⁺ sensor, we used a Zn²⁺-specific chelator, TPEN, at various concentrations (0, 10, 20, and 30 μM) to deplete Zn²⁺ and performed cell growth assays (Figure S12). Under conditions of no (0 μM) or low (10 μM) TPEN, the growth of the M1 and M1M2 mutants was suppressed, whereas the WT and the M2 mutants were largely unaffected. This observation is consistent with the increased intracellular Zn²⁺ accumulation in the M1 and M1M2 mutants. At higher TPEN concentrations (20 and 30 μM), WT and all mutants exhibited similar growth, likely due to extensive depletion of intracellular Zn²⁺. These results support the conclusion that the M1 site plays a primary role in intracellular Zn²⁺ sensing and negatively regulates Zn²⁺ uptake.

In ABC transporters, the transport activity is coupled with ATP hydrolysis. To directly evaluate the functional role of ZSD, we performed ATPase activity assays using purified ZnuBC WT, M1, M2, and M1M2 mutant proteins in the presence of Zn²⁺ or the Zn²⁺-specific chelator TPEN. These experiments show that Zn²⁺ reduces ATPase activity relative to TPEN-treated samples. In addition, the M1 and M1M2 mutants display higher ATPase activity than the WT or the M2 mutant under both conditions. These results indicate that disruption of Zn²⁺ binding to the ZSD enhances ATPase activity and likely promotes Zn²⁺ uptake. Together, these findings provide direct biochemical evidence that Zn²⁺ binding to the ZSD regulates ZnuBC activity.

Based on the structure of the M1 and M2 sites, we identified a strictly conserved motif, Cys193xxxHisxxCysCys204, in 5,712 ZnuC homologs, including major human and plant pathogens, using PatternSearch²⁴ and MEGA11²⁵ (Figure S13). Alignment of eight representative pathogens confirmed absolute conservation of all M1/M2 residues (Figure S14), suggesting that Zn²⁺-mediated ZnuC dimerization is a common mechanism for controlling zinc uptake and homeostasis across these pathogenic bacteria.

Because Zn²⁺-binding stabilizes the ZnuC zinc-sensing domain (ZSD) structure, we hypothesized that Zn²⁺ removal would destabilize it. To test this, we treated ZnuBC with EDTA to strip Zn²⁺ and solved its cryoEM structure (Figures S4D and S5D). The EDTA-treated ZnuBC dataset is heterogeneous and shows varying degrees of ZSD disorder and loss of the ZnuC dimerization interface, as shown by 3D classification analysis. These analyses reveal increased structural heterogeneity (Figure S15A–D) compared to the untreated dataset (Figure S15E), with varying degrees of ZSD disorder and disruption of the ZnuC–ZnuC interface. In some classes, the ZSD domains remain relatively close (Figure S15B), whereas in others (Figure S15A, C, D) they are clearly separated and partially disordered. Local resolution analysis focused on the ZSD and the ZnuC dimerization interface indicates substantially reduced local resolution in these regions following EDTA treatment, consistent with increased flexibility and partial destabilization upon Zn²⁺ removal (Figure S15F), compared to the untreated reconstruction (Figure S15E). Due to the poor ZnuC density, the modeled EDTA-treated structure doesn’t contain ZSD and Zn²⁺ ions (Figure 4D–E). Alignment with the apo structure revealed broken two-fold symmetry: ZnuC subunits rotated by 6° and 3° relative to their ZnuB partners, reducing the ZnuC–ZnuC interface from 1,606 Ų to 86 Ų, while ZnuB remained largely unchanged (RMSD 0.660 Å) (Figure 4E).

These structural and functional data suggest that Zn²⁺ depletion destabilizes ZSD and increases ZnuC flexibility by disrupting its dimer interface. We hypothesize that ZSD may sense and monitor intracellular Zn²⁺ content and autoregulate the transporter function to prevent over-accumulation of cellular Zn²⁺.

Alternating access transport model

To explore Zn²⁺ transport mechanisms in ZnuBC, we employed anisotropic network model (ANM) analysis^26,27 on cryoEM structures. While molecular dynamics simulations offer atomic detail, they are limited by sampling and timescale, especially for multimeric proteins in membranes. ANM-based normal mode analysis, via the ProDy interface²⁸, efficiently reveals global, cooperative motions. We computed intrinsic dynamics for apo and EDTA-treated ZnuBC, both isolated and in explicit lipid bilayers, using the membrANM extension²⁹. The four lowest-frequency modes (ANM1–4) reflect the transporter’s architecture and are largely unaffected by the membrane (Figure S16A–B). ANM1 involves global twisting of ZnuB and ZnuC dimers; ANM2, bending of ZnuB relative to ZnuC; ANM3, tilting; and ANM4, alternating periplasmic/cytosolic opening (Figure S16C). Mode spectra comparison shows ANM1, ANM2, and ANM4 are conserved (cosine similarity >0.80) despite Zn²⁺-induced changes. ANM4 is particularly insightful, predicting large-scale conformational shifts, enabling Zn²⁺ access from the periplasm and release into the cytosol, consistent with an alternating-access model³⁰. (Figure S16D). Transitions involve exposure and narrowing of the hydrophobic cavity (TM2/TM4) and coordinated rotation of the α2 helix and ZnuC dimer rearrangement, facilitating Zn²⁺ release. However, ANM does not resolve localized motions, which likely occur at higher frequencies.

Discussion

In this work, we determined cryoEM structures of a high-affinity Zn²⁺ uptake ABC transporter ZnuBC in its nucleotide/Zn²⁺, apo/Zn^2+, and apo/Zn²⁺-free states. To capture the ZnuABC complex, our expression construct was designed to contain both ZnuBC and ZnuA. However, the purified complex consisted of only ZnuB and ZnuC, lacking ZnuA (Figure S1). The interaction between ZnuA and ZnuBC is likely transient and may occur only during specific stages of the transport cycle. In particular, ZnuA may interact with ZnuB when the ZnuBC zinc-binding sites are unoccupied or during ATP-driven conformational transitions. Such transient interactions may make the complex difficult to capture in a stable form for purification and structural analysis. The identification of a ZSD at the C-terminus of ZnuC, together with the functional characterization of the M1 and M2 sites, allowed us to propose an autoregulation mechanism for Zn²⁺ uptake and homeostasis. Under Zn²⁺-replete conditions, each ZnuC zinc-sensing domain (ZSD) binds Zn²⁺, dimerizes, and locks ZnuB in an outward-facing, closed conformation that is refractory to Zn²⁺ uptake (Figure 3A). When cellular Zn²⁺ is depleted by enzymes or transcription factors, Zn²⁺ dissociates from the ZSDs, destabilizing their dimer interface (Figure 4C), thereby allowing the transporter to use ATP to drive the uptake of Zn²⁺ from periplasm.

We have identified two well-resolved Zn²⁺-binding sites in ZnuC ZSD (Figure 4A), which link these sites to Zn²⁺-dependent autoregulation. However, our cell-growth assays showed that the M1 and M1M2 mutants disrupted Zn²⁺-dependent regulation, whereas the M2 mutant behaved similarly to the control (Figure 4B, S11). Given the proximity of the two sites in the structure, we propose that Zn²⁺ binding to M1 may influence the affinity of the neighboring M2 site through an allosteric effect. This model could explain why the mutation of M2 alone produces a weaker phenotype despite the structural presence of the site. Additional experiments will be required to fully characterize this potential coupling mechanism.

ZnuABC expression is controlled by the Zur transcription factor, but its post-translational regulation was unknown. We discovered a Zn²⁺ sensor in the C-terminal zinc-sensing domain (ZSD) of ZnuC, defined by the conserved motif CxxxHxCC. This motif creates two high-affinity Zn²⁺ sites (M1 and M2) that prevent Zn²⁺ overaccumulation and support cell growth (Figure 4A–B). Based on the classification of ABC transporters³¹, ZnuABC belongs to type II ABC transporters. Five other type II ABC transporter structures have been reported: ButCD-F³² (PDB code 2QI9), HI1470/1³³ (PDB code 2NQ2), FhuCDB³⁴ (PDB code 7LB8), BhuUV³⁵ (PDB code 5B57), and PsaBC³⁶ (PDB code 7KYP). We performed structural alignments of the ZnuC ZSD with the NBD C-termini of the five ABC importers (PsaC, ButD, HI1470, FhuD, and BhuV), as shown in Figure S17. Unlike the ZnuC ZSD, which comprises two short α-helices (α6–7), a β-strand (β13), and a long loop, the C-termini of the compared NBDs consist of either a single long helix (PsaC, Figure S17B) or a long helix followed by two β-strands (ButD, HI1470, FhuD, and BhuV, Figure S17C–F). These differences highlight the unique structural features of the ZnuC ZSD in Zn²⁺ sensing and uptake regulation.

Our alternating-access transport cycle model is based on ANM normal-mode analysis. Currently, there is no inward-facing structure of ZnuBC, which limits direct validation of our model. The expected inward-facing structure might be dynamic and transient. Future studies are needed to capture an inward-facing structure, including the use of substrate analogs, nanobodies, cysteine disulfide-bond crosslinking, or ZnuA-bound conditions that may stabilize alternative conformational states.

Beyond the Zn²⁺ ABC transporters, we previously proposed that bacterial ZIP-family Zn²⁺ transporter BbZIP has a histidine-rich loop between TM3 and TM4 and functions as an intracellular Zn²⁺ sensor to regulate the transporter activity³⁷. In plants, a histidine-rich loop in Arabidopsis IRT1, also a ZIP member, senses metals and triggers endosomal sorting and vacuolar degradation of IRT1³⁸. A histidine-rich cluster in human ZIP4 mediates its ubiquitination and degradation to prevent Zn²⁺ toxicity³⁹. It appears that all living organisms have evolved versatile ways to regulate the activity of Zn²⁺ transporters post-translationally through inhibition or degradation to prevent excess accumulation of Zn²⁺ and toxicity. Further studies are needed to see if this regulation exists in other essential but toxic metal micronutrients.

Zn²⁺ ABC transporters are bacteria-specific and exist in most high-risk pathogens (Figures S13 and S14). The structures of the Zn²⁺ ABC transporter ZnuBC provide a basis for the development of antibiotics and vaccines targeting the conserved sites, such as the transmembrane cavity in the ZnuB dimer interface. The cavity is elongated and large enough to accommodate more than a single Zn²⁺. A potential inhibitor blocking the cavity may help reduce pathogenicity by preventing the evasion of pathogens from hosts’ nutritional immunity.

Resource Availability

Lead contact

Requests for further information should be directed to the lead contact, Qun Liu (qunliu@bnl.gov).

Materials availability

All materials produced in this study will be made available upon reasonable request.

Data and Code availability

The 3-dimensional cryo-EM density maps have been deposited in the Electron Microscopy Data Bank under the accession numbers EMD-70570 for WT-AMP-PNP/Zn²⁺ state [https://www.emdataresource.org/EMD-70570], EMD-70571 for E146Q-ATP/Zn²⁺ state [https://www.emdataresource.org/EMD-70571], EMD-70572 for apo/Zn²⁺ state [https://www.emdataresource.org/EMD-70572], EMD-70573 for apo/Zn²⁺-free state [https://www.emdataresource.org/EMD-70573]. Atomic coordinates have been deposited in the Protein Data Bank under the accession numbers 9OKO for WT-AMP-PNP/Zn²⁺ state [https://doi.org/10.2210/pdb9oko/pdb], 9OKP for E146Q-ATP/Zn²⁺ state [https://doi.org/10.2210/pdb9okp/pdb], 9OKQ for apo/Zn²⁺ state [https://doi.org/10.2210/pdb9okq/pdb], and 9OKR for EDTA-treated, Zn²⁺-free state [https://doi.org/10.2210/pdb9okr/pdb]. They are publicly available as of the date of publication.

This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request.

STAR★Methods

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Bacterial strain E. coli One Shot MAX Efficiency DH5α-T1R competent cells were obtained from Thermo Fisher Scientific. E. coli OverExpress C43(DE3) chemically competent cells were obtained from Biosearch Technologies. Cells were cultured at 37°C in Terrific Broth or LB medium supplemented with 100 μg/ml carbenicillin.

Method details

ZnuA-ZnuB-ZnuC cloning

The E. coli gene ZnuCB was amplified from genomic DNA (strain DH5α) using the forward primer 5′-GCCAGGATCCGAATTCACAAGTCTGGTTTCCCTGGA-3′ and the reverse primer 5′-GGCGAAGCTTTTAGCTGGCCTGCTTTTTCA-3′, which introduced unique BamHI and HindIII restriction sites flanking the ZnuCB gene. The E. coli ZnuA was amplified from genomic DNA (strain DH5α) using the forward primer 5′-TATACATATGTTACATAAAAAAACGCTTCT −3′ and the reverse primer 5′-CAGACTCGAGATCTCCTTTCAGGCAGCTCG −3′, which introduced unique NdeI and XhoI restriction sites flanking the ZnuA gene. The PCR products were inserted into the pGEM-T-Easy vector according to the manufacturer’s instructions. The sequences were confirmed by DNA sequencing with the sequencing primers M13F 5′GTAAAACGACGGCCAGTG-3’ and M13R 5’-GGAAACAGCTATGACCATG-3’. The ZnuCB was digested with BamHI and HindIII and ligated into the pET-Duet1 vector. Then ZnuA was digested with NdeI and XhoI and ligated into the pET-Duet1-ZnuCB. The sequence was confirmed by DNA sequencing. All primers used are listed in Table S1. The assembled plasmid pET-Duet1_ZnuA-ZnuCB was then transformed into E. coli C43 (DE3) cells.

ZnuB-ZnuC expression and purification

The cells transformed with pET-Duet1_ZnuA-ZnuCB were cultured at 37°C to an OD₆₀₀ of 0.8–1.2 in Terrific Broth (TB) medium supplemented with 100 μg/ml carbenicillin. Protein production was induced with 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG). Cells were harvested 4 hrs after the IPTG induction by centrifugation at 3500xg for 3 min. Cells were resuspended in cold lysate buffer (20 mM HEPES, pH 7.4, 200 mM NaCl, and 20% glycerol) containing 0.1 mM phenylmethylsulfonyl fluoride (PMSF). The cells were lysed by passage through an EmulsiFlex-C3 homogenizer (Avestin, Ottawa, Canada) six times at 15,000 psi. Cell debris was removed by centrifugation at 31,360xg for 1 hr. The supernatant was collected, and the membrane was harvested by centrifugation at 223,608xg for 2 hrs. The membrane pellet was resuspended in resuspend buffer (20 mM HEPES, pH 7.4, 200 mM NaCl, 20% glycerol, and 10 mM imidazole) supplemented with Phenylmethylsulfonyl fluoride (0.5 mM). The membrane was solubilized with 1.5% n-Dodecyl-B-D-Maltoside (DDM) supplemented with 50 μM ZnCl₂ for 2 hrs. The insoluble components were removed by centrifugation at 113,210xg for 1 hr. The supernatant was applied to a pre-equilibrated Ni²⁺-NTA column. The column was washed by 10 column volumes (CV) of wash buffer (20 mM HEPES, pH 7.4, 200 mM NaCl, 20% glycerol, 40 mM imidazole, and 0.05% DDM). The protein was eluted by elute buffer (20 mM HEPES, pH 7.4, 200 mM NaCl, 200 mM imidazole, and 0.05% DDM) and concentrated to 5 mg/ml by using a 100 kDa molecular-weight cutoff concentrator (Millipore Sigma, Burlington, MA, USA). The purified complex contained ZnuBC (Figure S1).

Amphipol reconstitution of the ZnuB-ZnuC complex

The concentrated protein (5 mg/ml) was mixed with amphipol A8–35 (Anatrace, Maumee, OH, Cat# A8–35) in a mass ratio of protein/amphipol of 1:5 (w/w). The mixture was incubated at 4°C for 20 hrs. Then, SM-2 Bio-Beads (Bio-Rad, Cat#1523920) were added at a mass ratio of protein/beads of 1:100 (w/w) to the mixture. The mixture was incubated at 4°C for 3 hrs. Prior to use, Bio-Beads were soaked with buffer containing 20 mM HEPES, pH 7.4, and 200 mM NaCl overnight at 4°C. The Bio-Beads were removed by passing through a micro Bio-Spin chromatography column. The flow-through containing the reconstitution mixture was centrifuged at 20,000xg for 1 hr at 4°C. The supernatant was applied to a Superdex 200 Increase column (GE Healthcare, Inc., Chicago, IL, USA) and eluted in a buffer containing 20 mM HEPES, pH 7.4, and 100 mM NaCl. The fractions containing the ZnuBC complex were collected and concentrated to 2 mg/ml for cryo-electron microscopy analysis. To prepare the EDTA-treated ZnuBC sample, the amphipol-reconstituted ZnuBC sample was incubated with 0.5 mM EDTA for 2 hrs on ice before loading onto the pre-equilibrated Superdex 200 column.

For ATPase and metal-binding assays, we found that the purified WT and mutant proteins were not stable after removing Zn²⁺. So, based on the determined ZnuBC structure, we created a disulfide bond between ZnuB and ZnuC by introducing mutations S204C in ZnuB and T81C in ZnuC. The SEC buffer (HEPES, pH 7.4, 150 mM NaCl) was pretreated with Chelex-100 resin (BioRad, catalog # 1422832) to remove Zn²⁺.

ATPase activity assay

ATPase activity assay was performed using ATPase/GTPase Activity Assay Kit (Sigma-Aldrich, Catalog # MAK113) according to the manufacturer’s protocol. Three replicates were measured for each sample. Buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, 0.1 mM TCEP, pretreated with Chelex-100 resin) with the corresponding concentration of ATP was tested as the control. For wild-type and E146Q mutant, the concentration of each sample was 0.2 mg/ml, and 10 μl was used for each replicate. Reactions were incubated at room temperature for 30 minutes before the reagent was added to terminate the reactions. For wild-type and zinc-binding site mutants, the concentration of each sample was 0.3 mg/ml, and 10 μl was used for each replicate, premixed with ZnCl₂ (20 μM final), TPEN (10 μM final), or no treatment (control). Reactions were incubated at 37°C for 40 minutes before the reagent was added to terminate the reactions. After 30 min incubation for colorimetric product generation at room temperature, absorbance at 620 nm was measured at room temperature using a Tecan Spark 20M multimode reader (Tecan Trading AG, Switzerland). ATPase activity was calculated as phosphate (nmol) produced by 1 mg protein per minute under experimental conditions.

Microscale thermophoresis (MST) assay

Amphipol reconstituted WT and mutant ZnuBC samples were mixed with Alexa Fluor^™ 647 NHS Ester (Succinimidyl Ester) at a 1:3 ratio for protein:dye. The mixture was incubated with gentle rotation at room temperature for 1 hour. Excess dye was removed by size exclusion chromatography with the SEC buffer (HEPES, pH 7.4, 150 mM NaCl, pretreated with Chelex-100 resin). The corresponding peak was harvested and concentrated for the MST assay.

The MST assay was conducted with Monolith NT.115 Microscale Thermophoresis (Nanotemper technologies) with the Monolith NT.115 Series Capillaries Standard Treated (Nanotemper technologies, Catalog # MO-K002). Both proteins and metals were diluted in MST buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 0.005% Tween20, 0.1 mM TCEP, pretreated with Chelex-100). Protein samples were first centrifuged at 20,000xg for 1 hour at 4°C, and the supernatant was diluted to a raw fluorescent count of around 1000 (approximately 115 nM) for each MST assay. Divalent metals (ZnCl₂, CuCl₂, MnCl₂, CdCl₂, CoCl₂, FeSO₄) were prepared at 0.2 mM with the MST buffer. The series dilution was performed with a 1:1 ratio. For each MST sample, 10 μl protein was mixed with an equal volume of metal solution for 15 min at room temperature before MST measurement. The MST setting parameters were LED power 20%, MST power 20%. Data analysis was conducted with MO.Affinity Analysis software.

Bacterial growth viability assay

The mutation expression constructs were created with Q5 Site-Directed Mutagenesis Kit (New England BioLabs, Inc., Ipswich, MA) according to the manufacturer’s instructions. The primers used are listed in Table S1. The empty pETDuet-1 vector was used for vector control.

The expression constructs were transformed into E. coli C43 (DE3) competent cells. Three colonies were inoculated into 5 mL of LB media at 37°C for overnight culture. 100 μl culture was used to inoculate 5 ml of fresh LB media. When the OD₆₀₀ value reached between 0.8 and 1.0, 1 mM IPTG (final) was added to induce the expression at 37 °C for 2 hours. Cells were then diluted to OD₆₀₀ of 0.1, 0.01, 0.001, 0.0001, 0.00001, and 0.000001 with dilution buffer (20 mM HEPES, pH 7.4, 135 mM NaCl, and 5 mM KCl). Two microliters of diluted cells were placed on LB-Agar plates supplemented with 100 μg/ml carbenicillin, 1 mM IPTG, and various concentrations of ethylenediaminetetraacetic acid (EDTA, 0, 1, 2 , and 3 mM), ZnCl2 (0, 1, 2 mM), and N,N,N′,N′-Tetrakis(2-pyridylmethyl)ethylenediamine (TPEN, 0, 10, 20, 30 μM). The plates were placed at 37 °C for bacterial growth and were imaged every 12 or 24 hrs.

CryoEM sample preparation and data collection

Three microliters of the reconstituted nanodiscs containing nucleotide-free ZnuBC were applied to a glow-discharged (15 mA current for 15 sec) 300-mesh R 0.6/1 UltrAuFoil Holey Gold grid (Electron Microscopy Sciences, Hatfield, PA, Cat# Q350AR1A). After waiting for about 60 sec, the grid was plunge-frozen to liquid ethane using a ThermoFisher Mark IV vitrobot (ThermoFisher Scientific, Waltham, MA) with a blotting condition of 3.5 sec blot time, 0 blot force, and 100% humidity at 6°C. For E146Q-ATP and WT-AMP-PNP samples, 5 mM ATP or AMP-PNP was briefly incubated with 10 mM Mg²⁺ and ZnuBC for 60 sec at room temperature before blotting under the same conditions as the nucleotide-free samples.

Cryo-EM data for WT-AMP-PNP/Zn²⁺ state, E146Q-ATP/Zn²⁺ state, apo/Zn²⁺ state, and EDTA-treated/Zn²⁺-free state were collected at the LBMS facility at Brookhaven National Laboratory using a ThermoFisher Titan Krios 300 kV electron microscope equipped with a Gatan K3 camera and a BioQuantum energy filter. The physical pixel size used for all data collection is 0.666 Å. The beam size is 1.45 μm in diameter, covering the entire 0.6-μm hole. For each movie data frame, a total dose of 66 e⁻/Ų was fractioned to 50 frames using the ThermoFisher data acquisition program EPU. All data sets were collected with an energy filter width of 15 eV, which was centered every 1 hr of data acquisition. Data collection statistics are listed in Table 1.

Table 1.

CryoEM data collection, 3D reconstruction, and refinement statistics.

Data Collection	WT-AMP-PNP/Zn2+	E146Q-ATP/Zn2+	WT-Apo/Zn2+	WT-Apo/Zn2+-free
Microscope	Krios G3i	Krios G3i	Krios G3i	Krios G3i
Magnification	130 Kx	130 Kx	130 Kx	130 Kx
Stage type	Autoloader	Autoloader	Autoloader	Autoloader
Voltage (kV)	300	300	300	300
Detector	Gatan K3	Gatan K3	Gatan K3	Gatan K3
Energy filter (eV)	15	15	15	15
Acquisition mode	Super-resolution	Super-resolution	Super-resolution	Super-resolution
Physical pixel size (Å)	0.666	0.666	0.666	0.666
Defocus range (μm)	0.7–2.3	0.7–2.3	0.7–2.3	0.7–2.3
Electron exposure (e−/Å2)	66	66	66	66
Reconstruction
Software	Relion v3.08, cryoSPARC v3.2.0	Relion v3.08, cryoSPARC v3.2.0	Relion v3.08, cryoSPARC v3.2.0	Relion v3.08, cryoSPARC v3.2.0
Particles picked	1,817,498	1,250,170	1,366,732	3,590,067
Particles final	34,187	18,325	32,554	44,314
Extraction box size (pixels)	256	256	256	256
Rescaled box size (pixels)	160	128	128	128
Final pixel size	1.066	1.332	1.332	1.332
Symmetry imposed	C2	C2	C2	C1
Map resolution (Å)	2.5	2.9	2.9	4.1
FSC threshold	0.143	0.143	0.143	0.143
Map resolution range (Å)	341–2.50	341–2.9	341–2.9	341–4.1
Model refinement
Software	PHENIX	PHENIX	PHENIX	PHENIX
Refinement algorithm	Real Space	Real Space	Real Space	Real Space
Clipped box size (pixels)	None	None	None	None
Number of residues	998	994	998	910
Number of Zn2+	4	4	4	0
Number nucleotide	2	2	0	0
R.M.S. deviations
Bond length (Å)	0.005	0.004	0.002	0.006
Bond angle (°)	0.644	0.596	0.470	0.898
Molprobity clashscore	2.71	2.02	2.91	13.16
Rotamer outliers (%)	0.0	0.0	0.0	0.0
Cβ deviations (%)	0.0	0.0	0.0	0.0
Ramachandran plot
Favored (%)	98.20	99.19	97.60	94.51
Allowed (%)	1.80	1.81	2.40	4.84
Outliers (%)	0.0	0.0	0.0	0.66
PDB code	9OKO	9OKP	9OKQ	9OKR

CryoEM data processing

Fractioned movies were motion-corrected and averaged using MotionCor2⁴⁰ with a bin factor of 2. Averaged micrographs were further corrected by CTF estimation using Gctf⁴¹. Micrographs with an estimated resolution better than 4.5 Å were selected for further use. Particles were initially picked in Relion3⁴² and cleaned up in cryoSPARC⁴³. Then the cleaned particles were used for training and particle picking using Kpicker⁴⁴.

For each data set, the picked particles were extracted at 256 pixels and binned to 64 pixels with a pixel size of 2.664 Å. We used cryoSPARC⁴³ for 2D class averaging to very generously select classes and particles for 3D classification and refinement. With three classes each, selected particles were used for two cycles of ab initio reconstruction followed by 3D heterogeneous refinements with a particle size of 64 pixels (2.664 Å) in cryoSPARC. The particles belonging to classes with structural features were centered and re-extracted at 256 pixels and binned to 128 pixels with a pixel size of 1.332 Å. The re-extracted particles were used for 3D heterogeneous refinement using 3 or 4 classes. The process is repeated in cryoSPARC until there is no improvement in the refined map resolution as estimated by the gold-standard Fourier Shell Correlation (GS-FSC). There was no symmetry used during the iterative ab initio model generation and 3D heterogeneous refinements in cryoSPARC.

Particles from the best class or classes in cryoSPARC were auto-refined to convergence in Relion3⁴² followed by non-alignment 3D classification and local refinement without symmetry (Figure S3). Particles from the best 3D class or classes with the best structural feature, as visualized in Chimera⁴⁵, were selected for non-uniform refinement and local refinement in cryoSPARC. With the WT-AMP-PNP/Zn²⁺, E146Q-ATP/Zn²⁺, and apo/Zn²⁺ states, C2 symmetry was justified to be appropriate based on GS-FSC (Figure S4). All final particles are 128 pixels except for the AMP-PNP data, for which we extracted the final particles at 160 pixels. Local resolutions were estimated using BlocRes⁴⁶. Data processing and reconstruction statistics are listed in Table 1.

Model building and refinement

To enhance the features for model building and visualization, the masked and filtered maps were processed by CryoFEM, which is a 3D convolutional neural network (3D-CNN) model for local map sharpening and feature enhancement⁴⁷. The models were built in COOT⁴⁸ and refined in real space using PHENIX⁴⁹. The refined models were validated using Molprobity⁵⁰. The refinement statistics are listed in Table 1.

Evaluation of transporter dynamics

All analyses for examining the dynamics of ZnuBC, including its potential transition to the inward-facing state, were performed using the ProDy API²⁸. The Anisotropic Network Model (ANM) analysis^51,52 on ZnuBC was conducted by constructing a Hessian matrix (3N × 3N) for the tetrameric apo structure (ZnuBC) and the EDTA-treated structure (no ZSD) based on Cα coordinates. Explicit membrane ANM (ex MembrANM)²⁹, a method developed to incorporate lipid membrane in the ANM studies of membrane proteins embedded in an explicit membrane, was also performed on the apo structure to assess the effect of the membrane on protein dynamics. The membrane was constructed using the OPM server⁵³. The Hessian matrices were decomposed into ANM normal modes to extract the lowest frequency (softest) modes, which encapsulate global and cooperative motions. The first four ANM modes from the apo structure were reconstructed into 3D motions to visualize the individual mode motions (Figure S16).

QUANTIFICATION AND STATISTICAL ANALYSIS

For metal specificity analysis, values are reported as mean ± standard deviation from three independent replicates (n = 3). No statistical tests were performed for this analysis. Cryo-EM data collection and refinement statistics are summarized in Table 1.

Supplementary Material

Document S1. Figures S1–S17, Table S1

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
E. coli One Shot MAX Efficiency DH5α-T1R Competent Cells	Thermo Fisher Scientific	Cat#:12297016
E. coli OverExpress C43(DE3) Chemically Competent Cells	Biosearch Technologies	Cat#: 60446-1
Chemicals, peptides, and recombinant proteins
LB broth	Teknova	Cat#: L9145
Terrific broth	Teknova	Cat#: T0320
Carbenicillin (disodium salt)	Goldbio	Cat#: C-103-1
n-Dodecyl-B-D-maltoside (DDM)	Goldbio	Cat#: DDM100
Ampipol A8–35	Anatrace	Cat#: A835 500 MG
SM-2 Bio-Beads	Bio-Rad	Cat#:1523920
Chelex 100 Chelating Resin, biotechnology grade, 100–200 mesh, sodium form	Bio-Rad	Cat#: 1432832
Ethylenediaminetetraacetic Acid (EDTA), Disodium Salt, Dihydrate	Thermo Fisher Scientific	Cat#: 15576028
N,N,N’,N’- tetrakis-(2-Pyridylmethyl)ethylenediamine) (TPEN)	Sigma-Aldrich	Cat#: 616394-100MG
Adenosine 5’-(P,y-imido)triphosphate lithium salt hydrate (AMP-PNP)	Sigma-Aldrich	A2647-25MG
Alexa Fluor 647 NHS Ester (Succinimidyl Ester)	ThermoFisher Scientific	Cat#: A20106
Protein ladder (BLUEstain 2 Protein ladder, 5–245 kDa	Goldbio	Cat#: P008-3000
BamHI	New England Biolabs	Cat#: R3136S
HIndIII	New England Biolabs	Cat#: R31045
NdeI	New England Biolabs	Cat#: R0111S
XhoI	New England Biolabs	Cat#: R0146S
T4 DNA ligase	Promega	M1801
Q5 High-Fidelity 2X Master Mix	New England Biolabs	M04925
Antifoam 204	Sigma-Aldrich	Cat#: A6426-100G
Sodium hydroxide (NaOH)	Sigma-Aldrich	Cat#: 221465-500G
Potassium chloride (KCl)	Sigma-Aldrich	Cat#: P3911-1KG
Zinc chloride (ZnCl2)	Sigma-Aldrich	Cat#: 208086-100G
Copper(II) chloride dihydrate (CuCl2 · 2H2O)	Sigma-Aldrich	Cat#: 307483-100G
Cadmium sulfate (CdSO4)	Sigma-Aldrich	Cat#: 383082-100G
Manganese(II) chloride tetrahydrate (MnCl2 · 4H2O)	Sigma-Aldrich	Cat#: M3634-100G
Cobalt chloride (CoCl2)	Sigma-Aldrich	Cat#: 232696-100G
Iron sulfate heptahydrate (FeSO4 · 7H2O)	Sigma-Aldrich	F7002-250G
Magnesium chloride hexahydrate (MgCl2 · 6H2O)	Sigma-Aldrich	M9272-1KG
Critical commercial assays
ATPase/GTPase Activity Assay Kit	Sigma-Aldrich	MAK113-1KT
HiFi DNA Assembly Master Mix	New England Biolabs	E2621S
Q5 Site-Directed Mutagenesis Kit	New England Biolabs	Cat#: E0554S
ZR Plasmid Miniprep - Classic	ZymoResearch	Cat#: D4054
Zymo Research Gel Recovery Kit	ZymoResearch	Cat#: D4002
Deposited data
WT-AMP-PNP/Zn2+	This paper	PDB 9OKO, EMDB EMD-70570
E146Q-ATP/Zn2+	This paper	PDB 9OKP, EMDB EMD-70571
apo/Zn2+	This paper	PDB 9OKQ, EMDB EMD-70572
EDTA-treated, Zn2+-free	This paper	PDB 9OKR, EMDB EMD-70573
Oligonucleotides
Primer table S1	This paper	N/A
Recombinant DNA
Plasmid: pGEM-T-EASY	Promega	Cat#: A1360
Plasmid: pETDuet1 ZnuABC	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuC E146Q	This paper	N/A
Plasmid: pETDuet1 ZnuABC M1 mutant	This paper	N/A
Plasmid: pETDuet1 ZnuABC M2 mutant	This paper	N/A
Plasmid: pETDuet1 ZnuABC M1M2 mutant	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuCB Crosslinking	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuCB Crosslinking M1 mutant	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuCB Crosslinking M2 mutant	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuCB Crosslinking M1M2 mutant	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuB D36A	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuB H40A	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuB D36AH40A	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuB H91A	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuB D36AH91A	This paper	N/A
Plasmid: pETDuet1 ZnuABC ZnuB H40AH91A	This paper	N/A
Software and algorithms
Relion 3.08	Zivanov et al.42	https://github.com/3dem/relion
Coot 0.9.5	Casanal et al.48	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
cryoSPARC 3.2.0	Punjani et al.43	https://cryosparc.com/
Phenix 1.18.2	Adams et al.49	https://www.phenix-online.org/
Pymol 2.6.0a0	Schrodinger LLC	https://www.pymol.org/
Chimera 1.14	Pettersen et al.45	https://www.cgl.ucsf.edu/chimera/
Molprobity 1.18.2-3874	Williams et al.50	https://molprobity.biochem.duke.edu/index.php
MO.Affinity Analysis	Nanotemper technologies	https://shop.nanotempertech.com/mo-affinity-analysis-software-unlimited-licenses/
Graphpad Prism 10.1.1	Graphpad	https://www.graphpad.com/
MEGA 11	Tamura et al.25	https://megasoftware.net/
iTOL	iTOL	https://itol.embl.de/
Other
TECAN’s Spark 20M multimode reader	TECAN trading AG	https://lifesciences.tecan.com/multimode-plate-reader
Monolith NT.115 MicroScale Thermophoresis	Nanotemper technologies	https://nanotempertech.com/monolith/
Monolith NT.115 Series Capillaries Standard Treated	Nanotemper technologies	Cat#: MO-K002

Highlights:

• High-affinity Zn²⁺ ABC transporter ZnuB-ZnuC structure in an inward-facing closed state

• The ZnuB dimer forms a large hydrophilic cavity that facilitates Zn²⁺ transportation

• ZnuC harbors a conserved Zn²⁺-sensing domain that regulates transporter activity

• Reveal potential therapeutic targets to disrupt Zn²⁺ homeostasis in resistant pathogens