A labile sulfur ligand in a three-cysteine-coordinated [2Fe−2S] cluster mediates sulfide sensing in NreB Authors: Chao Tang, Yuemeng Shen, Jingyao Qu, Lu Yu, Honglei Liu, Huaiwei Liu, Luying Xun, Yongzhen Xia Published: 2026 ABSTRACT AbstractIron-sulfur clusters are versatile protein cofactors involved in diverse biological processes, but their role in hydrogen sulfide/hydrosulfide (H2S/HS-) sensing remains largely unexplored. Here, we report that the Bacillus licheniformis sensor kinase NreB contains an unusual [2Fe-2S] cluster within its PAS domain. A 1.52-Å crystal structure reveals a distinct coordination geometry where three conserved cysteine residues and a non-cysteinyl sulfur ligand stabilize the cluster. Biochemical and native mass spectrometry support assignment of the predominant ligand state as -SH and show enrichment of -SSH/SSOH-like state upon sulfide exposure in the presence of O2, correlating with increased NreB kinase activity. Electron paramagnetic resonance spectroscopy shows that the cluster retains its oxidized [2Fe-2S]2+ state during sulfide-sensing. Molecular dynamics simulations further reveal transient solvent and HS- accessibility to the buried cluster, providing a physical basis for ligand entry. Here, we show that bacteria sense sulfide via a three-cysteine-coordinated Fe-S cluster with a labile sulfur ligand.Subject terms: Metalloproteins, X-ray crystallography, Mass spectrometry, Chemical modification, Bacteria Introduction Hydrogen sulfide (H2S), alongside nitric oxide (NO) and carbon monoxide (CO), has emerged as a gaseous signaling molecule over the past two decades1,2. Extensive research has transformed the perception of H2S from a mere toxic gas to a crucial endogenous biomolecule with diverse physiological functions3,4. Recent research indicates that most physiological effects are primarily mediated through protein persulfidation by supersulfide species generated from H2S metabolism, rather than being directly acted on by H2S itself 5–8. Supersulfide encompasses compounds with sulfur in zero or −1 valence states, including persulfides (H2S2 and RSSH), polysulfides (H2Sn, RSnH, and RSSnR, n > 2), and elemental sulfur predominantly present as octasulfur (S8)9,10. Among these, persulfides and polysulfides are widely present in various organisms, with glutathione polysulfide (GSnH, n ≥ 2) and cysteine polysulfide (CysSnH, n ≥ 2) being the most prevalent forms11–14. These compounds exhibit enhanced redox-active properties compared to conventional thiols5,15,16. Supersulfide species typically exert their regulatory functions through their nucleophilic characteristics, either by directly promoting protein S-persulfidation or by facilitating the formation of disulfide bonds and multiple sulfur linkages between cysteine residues17–19. These mechanisms represent the main pathways for supersulfide-mediated regulation. Two heme-containing bacterial regulators were identified as sulfide sensors, involving sulfide oxidation by the heme-bound Fe3+. The Escherichia coli oxygen sensor DosP employs a heme-Fe2+ within its PAS (Per-Arnt-Sim) domain for oxygen detection. However, the purified protein in the heme-Fe3+ form detects sulfide, albeit at relatively high concentrations ranging from 50 to 500 μM20. Similarly, the oxygen and redox sensor DosS from Mycobacterium tuberculosis also detects H2S21. H2S interacts with the DosS heme in its Fe3+ form, leading to enhanced DosS autokinase activity. Besides H2S, these sensors exhibit reactivity with various molecules, including O2, CO, and NO22,23, showing a lack of specificity for H2S detection. Recently, a sulfur oxidation regulatory gene cluster was identified in Bacillus licheniformis24, playing a critical role in the bacterium’s adaptation to feather biodegradation. This cluster comprises 11 genes, beginning with yrkD, which encodes a CstR-like repressor that senses elevated levels of supersulfides and derepresses the entire operon. The cluster includes nreB and nreC, which encode a two-component system (TCS) that specifically senses H2S. Notably, NreBC further activates the transcription of the downstream sqr and pdo genes, encoding sulfide:quinone oxidoreductase (SQR) and persulfide dioxygenase (PDO), respectively. These enzymes convert toxic H2S and GSSH into sulfite. This regulatory cascade enables B. licheniformis to mitigate sulfide/supersulfide stress during feather degradation. Although the involvement of NreBC in H2S sensing has been demonstrated via mutagenesis and whole-cell assays, the molecular mechanism remains to be elucidated. A NreABC system was initially characterized in Staphylococcus carnosus, which functions as a dual sensor of O2 and nitrate and regulates genes involved in nitrate/nitrite respiration25. The active NreBPAS domain contains a [4Fe-4S] cluster coordinated by four Cys residues and undergoes conversion to a [2Fe-2S] form upon oxygen exposure, leading to NreB inactivation. NreA serves as a nitrate sensor that enhances the NreB kinase activity, and NreC is the cognate response regulator. However, the B. licheniformis NreBC system exhibits distinct features: its NreB PAS domain belongs to a separate phylogenetic clade24, and it lacks the nitrate-sensing NreA protein. Whether the B. licheniformis NreBC system uses a Fe-S cluster for H2S sensing was unclear. Iron-sulfur (Fe-S) clusters are ancient and versatile protein cofactors that perform critical tasks, including electron transfer, catalysis, sulfur donation, and sensing26,27. These clusters can exist in different forms, with [2Fe-2S], [3Fe-4S], and [4Fe-4S] being the most common. The interaction between Fe-S clusters and proteins typically occurs through coordination bonds between iron ions and the thiol groups of cysteine residues28–30. Additional coordination can be achieved through histidine, aspartate, and glutamate residues31–33. Fe-S clusters that function in sensing undergo modifications upon exposure to signaling molecules such as O2, NO, iron, or other reactive oxygen species (ROS)25,28,34–36. These modifications may involve alterations in cluster type, iron oxidation state, or Fe-S cluster occupancy, ultimately influencing gene expression. However, whether Fe-S clusters can serve as dedicated sensors for H2S/hydrosulfide (HS−) remains largely unexplored. Here, we show that the sensor kinase NreB from B. licheniformis employs a structurally unusual [2Fe-2S] cluster within its PAS domain for selective H2S detection. Through a combination of crystallography, functional assays, and molecular dynamics simulations, we elucidate a mechanism consistent with our data by which a three-cysteine-coordinated [2Fe-2S] cluster mediates H2S sensing and signal transduction. Our findings expand the repertoire of Fe-S cluster function and provide insight into the molecular basis of bacterial gasotransmitter sensing. Results The [2Fe-2S] cluster is the prosthetic group of the PAS domain in NreB Phylogenetic analysis previously suggested that NreB is a heme-containing histidine kinase24. To obtain soluble, active protein, we optimized fusion tags and expression conditions and found that an N-terminal His tag with a TEV cleavage site yielded purified protein with the expected molecular weight of 45,415 Da (Supplementary Fig. 1). Purified NreB exhibited a yellow color and UV-Vis features with peaks near 330 and 415 nm and a broad band in the 480–550 nm region that are characteristic of [2Fe-2S]2+ proteins rather than heme proteins (Supplementary Fig. 2)28,37,38. Supplementation of the growth medium with L-cysteine and ferric citrate increased the intensity of these features, consistent with enhanced Fe-S cluster incorporation (Fig. 1a, b, Supplementary Table 1). Because some bacterial sensors harbor O2-labile [4Fe-4S] clusters that convert to [2Fe-2S] upon aerobic purification37,39, we performed anaerobic chemical reconstitution to test cluster identity. Reconstitution enhanced the 330/415 nm features without producing the broad 380–420 nm absorption characteristic of [4Fe-4S] clusters (Fig. 1a, b)38, supporting that NreB natively binds a [2Fe-2S] cluster. Spectroscopic characterization of the [2Fe-2S] cluster in NreB. a Visual comparison of 100 μM purified NreB (left) and 100 μM anaerobically reconstituted NreB (right). b UV-visible absorption spectra of 30 μM NreB: aerobically purified (black), DTH-treated under anaerobic conditions (red), and anaerobically reconstituted (blue). c CD spectra of aerobically purified NreB (black) and DTH-treated NreB under anaerobic conditions (red) at 100 μM protein with 5 mM DTH. d Spectra of DTH-reduced NreBPAS were recorded at varying sample temperatures (10-120 K) with the microwave power held constant at 1 mW. e Double-integrated CW-EPR intensity plotted versus 1/T; data at 10 and 20 K (gray) were excluded from the linear fit due to saturation. f X-band CW-EPR spectra of 5 mM DTH (buffer control; black), aerobically purified NreB (blue), and DTH-treated NreB under anaerobic conditions (red) at 400 μM protein with 5 mM DTH. Labeled values indicate g values. g CW-EPR spectrum of DTH-reduced NreBPAS (black) and EasySpin simulations fitted as the sum of two rhombic S = 1/2 components. Component 1: g = [2.0024, 1.9207, 1.8559] (68.6%); Component 2, g = [1.9869, 1.9108, 1.8064] (31.4%). h X-band CW-EPR spectra of 400 μM CuSO4 and anaerobically 5 mM DTH-reduced 400 μM NreB or NreBPAS. EPR conditions: 40 K, 1 mW microwave power, 5 G modulation amplitude, 100 kHz modulation frequency. No buffer subtraction was performed. Source data are provided as a Source Data file. AlphaFold-Multimer modeling suggested that NreB comprises an N-terminal PAS domain (NreBPAS) followed by HisKA_3 kinase domain and HATPase domain, arranged in a dimeric architecture (Supplementary Figs. 3a & 4a, b; Supplementary Data 1). Although the kinase domains share <18% sequence identity with characterized homologs (DesK and KinB) using SWISS-MODEL, Dali structural alignment confirmed significant topological conservation with Z-scores of 4.9 and 11.0, RMSD 2.4 and 1.7 Å, respectively, including conserved catalytic residues (Supplementary Fig. 3b & 5). Given that PAS domains frequently accommodate redox cofactors40, we purified the isolated PAS domain (NreBPAS) (Supplementary Fig. 1), which exhibited spectral properties and color identical to full-length NreB (Supplementary Fig. 3c, d), establishing the PAS domain as the cluster-binding site. We next characterized cluster redox properties. Anaerobic reduction with DTH decreased the 415 nm band and produced a feature near 420 nm (Fig. 1b & Supplementary Fig. 3d). CD spectroscopy of oxidized NreB revealed characteristic features of [2Fe-2S] proteins with three positive bands at 310, 425, and 603 nm and three negative bands at 280, 364, and 500 nm (Fig. 1c)41–43. Anaerobic DTH reduction caused only modest changes in the CD spectrum, with the overall lineshape largely preserved (Fig. 1c), further supporting redox responsiveness of the cluster. A similar visible CD signature was observed for NreBPAS (Supplementary Fig. 3e). To directly probe the oxidation state, we performed CW-EPR measurements. We first recorded temperature- and power-dependent EPR spectra of reduced NreBPAS to establish acquisition conditions and relaxation behavior. The signal remained observable up to 77 K and followed Curie-law behavior (Fig. 1d, e). Power saturation measurements indicated that 1 mW at 40 K is within a non-saturating regime under our settings (Supplementary Fig. 6). Under these conditions, aerobically purified NreB and NreBPAS were EPR-silent. In contrast, anaerobic DTH reduction yielded a rhombic S = 1/2 signal with g values near 1.99, 1.92, and 1.90, characteristic of [2Fe-2S]+ clusters (Fig. 1f & Supplementary Fig. 3f). To evaluate g-tensor anisotropy and potential heterogeneity, we simulated the X-band CW-EPR spectrum of reduced NreBPAS using EasySpin. The spectrum is best reproduced as the sum of two rhombic S = 1/2 components (Fig. 1g), consistent with modest heterogeneity in the local cluster environment. Finally, to quantify the number of paramagnetic centers per protein, we recorded X‑band CW‑EPR spectra of DTH-reduced NreB or NreBPAS together with a CuSO4 spin standard using identical acquisition settings (40 K, 1 mW; Fig. 1h). Double integration of the spectra yielded spin concentrations of 22.8 μM for NreB and 36.3 μM for NreBPAS at 400 μM protein, respectively (Fig. 1h & Supplementary Table 2). Independent ICP‑MS measurements determined [2Fe-2S] cluster incorporation efficiencies of ~22% for NreB and ~28% for NreBPAS (Supplementary Table 2). Taken together, these data correspond to ~0.259 spins per NreB and ~0.324 spins per NreBPAS. Together, these measurements indicated that only a subset of cluster-containing proteins was detected in the EPR-active reduced state under our experimental conditions. Because incomplete chemical reduction and uncertainties inherent to CW-EPR spin quantitation under these conditions can both contribute to the low apparent spin count, these values are treated conservatively and are not used for thermodynamic analysis. The three Cys residues are conserved and critical in the binding of the Fe-S cluster Cysteine residues typically coordinate Fe-S clusters28,32. In NreB, four cysteine residues are present: Cys3, which is not conserved, and Cys63, Cys66, and Cys78, which are highly conserved (Fig. 2a, b, Supplementary Fig. 7). To assess the roles of these specific residues in Fe-S cluster binding, we generated and purified the corresponding NreB mutants: NreBC3S, NreBC63S, NreBC66S, and NreBC78S. Compared to the wild-type NreB, the NreBC3S mutant retained the yellow color, while the NreBC63S, NreBC66S, and NreBC78S mutants were nearly colorless (Fig. 2c) and lacked the characteristic absorbance peak of the Fe-S cluster (Fig. 2d). Identification and functional characterization of conserved cysteine residues in NreB. a Multiple sequence alignment of NreBPAS homologs located near sqr-pdo operons using Clustal Omega. Sequence logos were generated using WebLogo 3, with conserved cysteine residues highlighted in red grids. b Sequence conservation analysis of NreBPAS and its homologs. The 116 homologous sequences (E-value < 1e-5 and identity > 40%) from RefSeq Select proteins were identified by BlastP and aligned using ClustalW Omega. Sequence logos were generated using WebLogo 3, with conserved cysteine residues highlighted in red grids. c Visual comparison of 100 μM wild-type NreB and cysteine mutants (NreBC3S, NreBC63S, NreBC66S, and NreBC78S). d UV-visible absorption spectra of 30 μM wild-type NreB and cysteine mutants. e Transcriptional response of sqr following 200 μM sulfide treatment for 10 min in the wild-type and mutant strains. Data are represented as mean ± SD from three independent experiments. Statistical significance was assessed using a two-sided ordinary one-way ANOVA followed by Dunnett’s multiple-comparisons test. Exact p values are shown in the graph. Detailed statistical results are provided in the Source Data File. f In vitro autokinase activity assay of wild-type NreB and cysteine mutants at 10 μM protein. The upper panel shows the anti-thiophosphate ester blot (P-NreB), and the lower panel shows a Coomassie-stained SDS-PAGE gel run in parallel with the same samples as a sample processing control to indicate comparable sample loading. The faint lower band is a minor co-purifying impurity that is constant across samples. Data are representative of three independent experiments with similar results. Source data are provided as a Source Data file. To investigate the functional consequences in vivo, we expressed NreBC63S, NreBC66S, and NreBC78S in a B. licheniformis yrkD deletion strain and monitored the H2S-dependent transcriptional response. All three mutants failed to show any increase in sqr transcript levels upon induction with 200 μM sulfide (Fig. 2e). The autokinase activity of wild-type NreB and its single-cysteine variants was further evaluated using an optimized kinase assay. To enhance the stability of NreB autophosphorylation, ATP-γ-S was employed as a phosphate donor44. Assay conditions were systematically optimized for ATP-γ-S concentrations and reaction time, with 100 μM ATP-γ-S and a reaction time of 30 s providing optimal linear kinetics (Supplementary Fig. 8a, b). Under these optimized conditions, all three cysteine variants (C63S, C66S, and C78S) exhibited dramatically reduced kinase activity compared to the wild type, whereas the C3S mutation had little effect (Fig. 2f). These results demonstrate that the three conserved cysteine residues are essential for Fe-S cluster binding and for maintaining NreB kinase activity. The structure of NreBPAS reveals distinct [2Fe-2S] cluster coordination To obtain label-free NreBPAS protein, a His-SUMO tag with a higher molecular weight than the His tag was fused to the N-terminus, and subsequently removed by TEV protease cleavage (Supplementary Fig. 9a). The protein was purified to >95% purity using sequential nickel affinity, ion-exchange, and size-exclusion chromatography (Supplementary Fig. 9b). The purified NreBPAS formed red rhombic crystals (Supplementary Fig. 9c), which were subjected to X-ray diffraction. Structure determination revealed NreBPAS harboring a distinct [2Fe-2S] cluster. The 1.52 Å resolution structure was refined with the statistics shown in Supplementary Table 3. NreBPAS comprises two α-helices (α1, Asn16-His27 and α2, Tyr79-Ala82) and seven β-strands (β1, Gly31-Asp36; β2, Arg39-Val43; β3, Phe89-Ser91; β4, Pro94-Thr95; β5, Gly98-Val101; β6, Met103-Asn113; β7, Gln116-Asp125) (Fig. 3a, b). β1, β2, β3, β6, and β7 are arranged antiparallel to form a major β-sheet, while β4 and β5 form a smaller, perpendicular antiparallel sheet. The longest α-helix (α1) and the shorter α-helix (α2) flank the β-strands. High-resolution crystal structure of NreBPAS reveals a [2Fe-2S] cluster-binding domain. a Cartoon representation of NreBPAS structure at 1.52 Å resolution. Secondary structure elements are colored in purple (β-strands) and cyan (α-helices), with their termini indicated. b Alternative view of NreBPAS structure, rotated 90° around the Y-axis relative to (a). c Coordination environment of the [2Fe-2S] cluster. Iron and sulfur atoms of the cluster are shown as orange and yellow sticks, respectively. Coordinating residues are depicted as sticks with atoms colored: carbon (cyan), nitrogen (blue), oxygen (red), iron (orange), and sulfur (yellow). d 2Fo-Fc electron density map (green mesh, contoured at 1.0 σ) surrounding the [2Fe-2S] cluster, showing comparable electron densities for the two metal centers. e Ground state (S = 0) diagrams of the two models, highlighting the substitution site Fe2-S6 and annotating the corresponding Fe-S bond lengths. S6 corresponds to the exchangeable terminal ligand (CH3S- in Model A and HS- in Model B). Notably, the [2Fe-2S] cluster exhibits a distinct asymmetric coordination. One iron is coordinated by Cys63 and Cys66, whereas the other is coordinated by Cys78 and a non-cysteinyl sulfur atom (Fig. 3c, d). The fourth ligand was modeled as sulfur because the Fe-ligand distance (~2.3 Å) and electron density are most consistent with Fe-S coordination. To exclude alternative protein-derived ligands, we measured the shortest distances from Fe1/Fe2 atoms to potential donor atoms (His imidazole N; Asp/Glu carboxyl O). All candidate donor atoms are >8.7 Å from either Fe (Supplementary Fig. 10), far beyond typical Fe-N/Fe-O coordination distances (~2.0–2.3 Å), indicating that His/Asp/Glu residues do not directly ligate the cluster. A Fo-Fc omit map calculated after omitting the [2Fe-2S] cluster from the model further supports assignment of the cofactor at this site (Supplementary Fig. 11). To assess the identity and lability of the fourth ligand, we compared [Fe2S2(CH3S)4]2- (Model A) and [Fe2S2(CH3S)3(HS)]2- (Model B) by DFT-D calculations (computational details in Methods). Spin‑coupling analysis shows both models possess S = 0 broken‑symmetry ground states (Supplementary Table 4 & Fig. 3e). Relative to Model A, Model B exhibits a weakened Fe2-S6 bond: bond length 2.323 Å vs 2.361 Å; Wiberg index 1.333 vs 1.306; Mayer bond order 0.794 vs 0.713 (Supplementary Table 5). Thermodynamics of the substitution Model B + CH3S- → Model A + HS- gives ΔG = – 6.79 kcal/mol (Supplementary Table 6), indicating stronger binding of CH3S- than that of HS-. These geometric, bond order, and energetic results consistently support HS- as a more labile fourth ligand. GSH-conditioned NreB is reactivated by sulfide under aerobic conditions As NreB-regulated SQR-PDO-mediated sulfide oxidation requires oxygen24,45, the sulfide-sensing mechanism of NreB was first examined under aerobic conditions. Untreated NreB was already fully active, and the addition of sulfide alone produced no further increase in autokinase activity (Supplementary Fig. 12a). We therefore screened a set of reducing agents and thiol-containing compounds for their effects on NreB autophosphorylation. Among the tested compounds, only GSH markedly suppressed the autophosphorylation (Fig. 4a). GSH inhibited NreB in a concentration-dependent manner and nearly abolished the activity at 5 mM (Fig. 4b, c). Importantly, removal of GSH by desalting restored autophosphorylation (Fig. 4d), indicating that the inhibition is reversible under our experimental conditions. GSH-conditioned NreB is activated by HS- and shows an HS--dependent optical response. a Autophosphorylation of 10 μM NreB after 10 min pre-incubation with the indicated reductants/antioxidants under aerobic conditions, followed by reaction with 100 μM ATP-γ-S for 30 s. DTT (dithiothreitol), 2 mM; DTH (dithionite), 5 mM; Cys, 5 mM; GSH, 5 mM; NADPH, TCEP, and ascorbic acid, 1 or 5 mM (as indicated). b Kinase activity of 10 μM NreB was tested after 10 min pre-incubation with different concentrations of GSH. c Time course of kinase activity for 10 μM NreB pre-incubated with 5 mM GSH for 10 min, followed by addition of ATP-γ-S at the indicated time points to assess GSH inhibition. d Reversibility of GSH inhibition. NreB was treated with 5 mM GSH for 10 min, and then GSH was removed by desalting under aerobic conditions before the kinase assay. e HS--dependent reactivation of the GSH-inhibited state. NreB was pre-incubated with 5 mM GSH for 10 min and then treated with NaHS at the indicated concentrations for 4 min before initiating the kinase reaction. For (a–e), the upper panels show anti-thiophosphate ester blots (P-NreB), and the lower panels show Coomassie-stained SDS-PAGE gels run in parallel with the same samples as sample processing controls to indicate comparable sample loading. The data in panels a-e are representative of three independent experiments with similar results. f UV-Vis absorbance spectra of 10 μM NreB (black) after addition of 5 mM GSH (red), and after addition of 5 mM GSH plus 100 μM NaHS (blue). The characteristic features at ~415 and ~418 nm are indicated. g UV-Vis absorbance spectra of 10 μM NreB in the presence of 5 mM GSH upon titration with increasing NaHS concentrations. The spectra of NreB with GSH (control) and with GSH plus 100 µM NaHS were identical to those shown in panel f. h Saturation curve derived from panel g by plotting the NaHS-dependent change at 418 nm and fitting with a one-site binding model to obtain Kd, app; errors represent the SE from nonlinear regression. i pH dependence of the NaHS-induced spectral transition. Difference spectra (NaHS-treated minus NaHS-free) were recorded at the indicated pH values, showing increased amplitude of the ~418 nm feature at higher pH. Source data are provided as a Source Data file. Strikingly, GSH-conditioned NreB became HS--responsive: addition of NaHS restored autophosphorylation in a dose-dependent manner, with detectable activation at approximately equimolar NreB at 10 μM each (Fig. 4e). In contrast, when the same workflow was performed under anaerobic conditions, NaHS failed to restore activity (Supplementary Fig. 12b), suggesting that an oxygen-dependent step is required for sulfide-dependent activation in our assay. To further explore the molecular basis of this regulation, we monitored absorbance changes of the Fe-S cluster upon ligand binding. Titrations were started from the oxidized [2Fe-2S]2+ state in sulfide-free buffer without added GSH or NaHS. GSH alone did not measurably alter the characteristic ~415 nm feature of the oxidized [2Fe-2S]2+ state, whereas NaHS addition produced a reproducible shift to ~418 nm (Fig. 4f & Supplementary Table 1). NaHS titration yielded a saturable spectral transition, from which we derived an apparent dissociation constant (Kd, app) of 8.49 ± 2.42 μM (SE) in 20 mM Tris buffer with 5 mM GSH (Fig. 4g, h). The isolated PAS domain exhibited a comparable NaHS-dependent transition with Kd, app = 6.10 ± 1.19 μM (SE) (Supplementary Fig. 13), indicating that the PAS domain is sufficient for this optical response. In an aqueous solution, H2S dissociates into HS- with a pKa of 7.0221. To determine whether NreB recognizes H2S or HS-, UV-Vis spectra were recorded in the presence of NaHS at different pH values. The intensity of the ~418 nm feature increased progressively with rising pH (Fig. 4i), indicating that HS-, rather than neutral H2S, is the primary binding species. NreB responds selectively to sulfide among the tested signals The specificity of NreB for H2S over other Fe-S cluster signals and sulfur compounds was systematically evaluated. The oxygen-sensing capability of NreB was tested as previously described35,39. Transferring aerobically prepared NreB to anaerobic conditions did not affect its autokinase activity (Fig. 5a). Consistently, GSH-mediated inhibition of NreB kinase activity was also independent of O2 levels (Fig. 5a & Supplementary Fig. 12), indicating that NreB does not function as an oxygen sensor. A range of potential signals, including H2O2, NO, CO28,35, and various sulfur compounds, were tested for their effects on GSH-inhibited NreB under aerobic conditions. Of these, only HS- was able to restore autokinase activity (Fig. 5b). Regulation of NreB autokinase activity under different redox conditions. a Effect of 5 mM DTH treatment for 10 min on the autokinase activity of 10 μM NreB. DTH was removed by desalting under anaerobic conditions after treatment. b Effects of the indicated modulators (100 μM, 4 min) on autophosphorylation of GSH-inhibited NreB under aerobic conditions (pre-incubation with 5 mM GSH, 10 min). Upper, ROS/NO and sulfur-containing compounds with NO delivered as GSNO and H2O2 tested as ROS; lower, CO donor CORM-2 with matched DMSO solvent control. c Time-course analysis of ATP-γ-S-dependent phosphorylation of aerobically purified NreB and its DTH-reduced form (5 mM, anaerobic, 10 min). NreB was maintained at 10 μM. For quantitative comparison between samples run on different blots, all samples were derived from the same experiment and blots were processed in parallel. d Quantitative densitometric analysis of phosphorylation levels corresponding to panel c. e Comparative analysis of 5 mM DTH treatment for 10 min on NreB autokinase activity under anaerobic or aerobic conditions. DTH was removed by desalting under anaerobic conditions after treatment. Data in panel d are presented as mean ± SD from three independent experiments. Statistical significance was assessed using a two-sided two-way ANOVA with Sidak’s multiple-comparisons test. Exact p values are shown in the graphs. Detailed statistical results are provided in the Source Data file. For (a–c,e), upper panel shows the phosphoblot; the lower Coomassie-stained SDS-PAGE gel was run in parallel with the same set of samples and is shown as a sample-processing control. The data in panels a-c and e are representative of three independent experiments with similar results. Source data are provided as a Source Data file. Under anaerobic conditions, treatment with DTH altered the oxidation state of the Fe-S cluster (Fig. 1f) and markedly reduced NreB kinase activity (Fig. 5a, c). Densitometric analysis revealed a 10-fold reduction in band density at 30 s (Fig. 5d). Importantly, this inhibition was reversible: DTH removal via desalting fully restored kinase activity (Fig. 5e). Furthermore, GSH retained its inhibitory effect on the NreB after DTH removal (Fig. 5a), confirming that these manipulations did not compromise NreB integrity or function. Notably, transferring the anaerobic, DTH-removed NreB back to aerobic conditions resulted in enhanced kinase activity (Fig. 5e). Mechanism of sulfide sensing by GSH-inhibited NreB To further probe the cluster state during sulfide sensing, we combined EPR, UV-Vis, native MS, and solution-behavior analyses. CW-EPR spectroscopy showed that treatment with either GSH or NaHS did not produce a detectable reduced-cluster signal in either full-length NreB or NreBPAS (Supplementary Fig. 14a, b). Consistent with EPR, only anaerobic DTH treatment produced the spectral changes associated with formation of the reduced state, whereas NaHS caused only a minor 415-to-418 nm shift without an EPR-detectable [2Fe-2S]+ signal (Supplementary Fig. 15 and Supplementary Table 1). These observations support that the cluster remains predominantly oxidized during H2S sensing. Far-UV CD spectra of NreB and NreBPAS showed only minor differences across conditions, suggesting no major changes in global secondary structure accompanying these treatments (Supplementary Fig. 16a, b). We next employed native MS to probe potential cluster-associated adduct states. As NreB aggregated in ammonium acetate buffer, analysis focused on NreBPAS, which remained stable under MS conditions (Supplementary Fig. 17a–f). Native MS of NreBPAS revealed a dominant peak corresponding to the [2Fe-2S]SH form, in which the fourth ligand is HS-, together with minor peaks consistent with [2Fe-2S], [2Fe-2S]OH, [2Fe-2S]SOH, [2Fe-2S]SSH, and [2Fe-2S]SSOH (Fig. 6a & Supplementary Table 7). The raw native mass spectrometry spectra and associated processed output files for all native MS analyses in this study are provided as Supplementary Data 2. Notably, the NreBPAS construct used for native MS (residues 20–135) lacks the non-conserved N-terminal Cys3, retaining only the three conserved ligands (Cys63, Cys66, and Cys78), which are fully engaged in cluster coordination (Fig. 3a–d). This design rules out non-specific persulfidation of free cysteines. Consequently, the observed mass shifts most likely reflect modifications at the labile, non-cysteinyl sulfur ligand identified in our crystal structure (Fig. 3a–d). Mechanism of GSH-mediated regulation of H2S sensing by NreB. a Native mass spectrometry analysis of NreBPAS ( ~ 80 μM) in 200 mM ammonium acetate. Spectra were deconvoluted using Bruker Data Analysis software. b Preservation of NreB activity states following anaerobic desalting. 10 μM NreB was treated with 5 mM GSH alone or with both 5 mM GSH and 200 μM NaHS under aerobic conditions, followed by anaerobic desalting to remove GSH and NaHS prior to kinase activity assays. This demonstrates persistence of activity states despite buffer exchange. c Effect of pH on NreB autokinase activity and sulfide responsiveness. 10 μM NreB was incubated for 4 min with 100 μM NaHS (or vehicle) and assayed in Tris buffer (pH 8.0) or in HEPES buffer (pH 7.0) as indicated. For panels b and c, upper panel shows the phosphoblot; the lower Coomassie-stained SDS-PAGE gel was run in parallel with the same set of samples and is shown as a sample-processing control. The data in b and c are representative of three independent experiments with similar results. d Quantification of native MS peak distributions at pH 7.0 from Supplementary Fig. 22. Peak areas of the six adduct states were normalized to the summed area of all six states and plotted as mean ± SD from three independent experiments. Statistical significance was assessed using a two-sided ordinary one-way ANOVA with Tukey’s multiple comparisons test. Exact p values are shown in the graph. Detailed statistical results are provided in the Source Data file. Source data are provided as a Source Data file. Millimolar free GSH is incompatible with native MS because it strongly suppresses ionization. Moreover, direct addition of GSH to NreBPAS in ammonium acetate reduced sample monodispersity as assessed by DLS (Supplementary Fig. 17g–i). We therefore performed the incubation with ligands in the standard reaction buffer, followed by rapid anaerobic buffer exchange into ammonium acetate immediately prior to MS analysis. Native MS analysis showed that GSH treatment alone did not significantly alter the distribution of Fe-S cluster species compared to the untreated control (Fig. 6a & Supplementary Fig. 18). However, subsequent addition of sulfide markedly shifted the population, increasing the abundance of [2Fe-2S]SOH, [2Fe-2S]SSH, and a [2Fe-2S]SSOH species (Fig. 6a & Supplementary Fig. 18). To ensure that MS-compatible processing preserved relevant functional states, we performed parallel autokinase assays using the identical anaerobic buffer-exchange workflow. Although ligand removal partially attenuated the magnitude of inhibition/activation, the relative differences between conditions were largely retained (Fig. 6b), supporting that the MS spectra report physicochemically meaningful states rather than preparation artifacts. The lack of spectral change by native MS upon GSH treatment stood in apparent contrast to the strong kinase inhibition observed in vitro (Fig. 6a, b). To investigate the basis of this inhibition, we first co-incubated NreBPAS with GSH at molar ratios of 1:30 and 1:300. The resulting crystal structures showed no substantial conformational changes and no unambiguous ligand density (Supplementary Fig. 19). We then examined the physicochemical effects of adding millimolar GSH under the Tris-based assay condition. Addition of 5 mM GSH lowered the bulk pH from 8.0 to ~7.0, and DLS measurements showed that this condition coincided with pronounced self-association/aggregation of full-length NreB (Supplementary Fig. 20a–c), providing a parsimonious explanation for the near-complete loss of autophosphorylation under these conditions. Strikingly, the subsequent addition of NaHS reversed this aggregation, restoring monodispersity (Supplementary Fig. 20d) and recovering kinase activity (Fig. 4e), even while the pH remained near 7.0. To decouple sulfide-dependent effects from aggregation and bulk pH effects, we performed kinase assays in HEPES buffer at pH 7.0, where NreB remained monodisperse (Supplementary Fig. 21a–d). Under these pH-controlled conditions, autophosphorylation was reduced relative to pH 8.0, yet NaHS still partially restored activity (Fig. 6c). Consistently, native MS performed at pH 7.0 showed that lowering pH from 8.0 to 7.0 in the absence of NaHS did not significantly redistribute the detected cluster-adduct states, whereas NaHS treatment at pH 7.0 reproducibly shifted the distribution toward the persulfide/sulfenylated adducts (SSH/SSOH) (Fig. 6d & Supplementary Fig. 22). Collectively, these results indicate that the strong loss of kinase activity in the GSH/Tris assay is dominated by physicochemical effects, especially bulk acidification and self-association, whereas the sulfide-responsive on state correlates with specific cluster-adduct changes consistent with the proposed sulfur-ligand chemistry. MD simulations reveal transient solvent and HS- accessibility to the buried [2Fe-2S] cluster in NreBPAS Structural analysis revealed that the Fe-S cluster is buried within the interior of NreBPAS (Fig. 7a), with no obvious channel connecting the protein surface to the cluster site. To assess whether this Fe-S cluster could become exposed, MD simulations were performed. RMSD (root mean square deviation) analysis indicated that the system reached a stable conformation after 20 ns, suggesting equilibration in the later stages of the simulation (Supplementary Fig. 23a). The radius of gyration (Rg) remained constant throughout, indicating no significant expansion or contraction of the protein (Supplementary Fig. 23b). RMSF (root mean square fluctuation) analysis revealed that most residues displayed low flexibility (RMSF < 0.25 nm), consistent with high structural rigidity, whereas increased fluctuations (RMSF > 0.4–0.6 nm) were limited to the N- and C-termini and random coil regions surrounding the [2Fe-2S] cluster (Supplementary Fig. 23c). Notably, the core atoms of the cluster, including two iron atoms and three cysteine sulfur atoms, exhibited very low RMSF values (<0.02 nm), indicating exceptional stability, while only one peripheral hydrogen atom (atom ID: 1849) showed higher fluctuation (~0.13 nm), likely due to transient solvent interactions (Supplementary Fig. 23d). Molecular dynamics simulations reveal transient solvent and HS accessibility to the buried [2Fe-2S] cluster in NreBPAS. a Surface representation of NreBPAS showing the buried [2Fe-2S] cluster (Fe, orange; S, yellow). Left, mesh surface; right, solid surface in the same orientation. b Number of water molecules within 0.35 nm of the [2Fe-2S] cluster during a 100‑ns simulation. c Number of HS- molecules within 0.35 nm of the [2Fe-2S] cluster during a 100‑ns simulation with 50 HS- molecules. d Representative snapshots from the apo simulation illustrating a transient opening/exposure event near the buried [2Fe-2S] cluster (53.72–55.08 ns). e Representative snapshots from the HS- simulation showing an example trajectory of an HS- ion approaching the cluster through a transient pathway (13.06–13.22 ns). For panels d and e, times are indicated above each snapshot (Fe, orange; S, yellow; H, green). f Electrostatic potential surfaces corresponding to the protein conformations in panel e, highlighting the charged environment along the HS- entry pathway. Red and blue indicate regions of negative and positive potential, respectively. All surface views are shown in the same orientation. Source data are provided as a Source DataSource Data file. To assess the solvent accessibility of the [2Fe-2S] cluster during the simulation, the number of water molecules within 0.35 nm of the cluster was monitored throughout the 100 ns trajectory (Fig. 7b). This cutoff distance reflects direct water-cluster contacts. No water molecules were in contact with the cluster at the start of the simulation, but within the first few nanoseconds, water molecules began to make contact, and their number generally fluctuated between 10 and 40 throughout most of the simulation. Notably, during the 40–60 ns interval, this number increased and remained at higher levels, ranging from 20 to 60, indicating a further opening of the solvent channel. Representative snapshots (Fig. 7d) and trajectory animations (Supplementary Movie 1) further show that local conformational changes progressively increased the exposure of the [2Fe-2S] cluster, facilitating the direct access of HS- ions, in a manner analogous to water molecules, to the [2Fe-2S] cluster. To probe whether bulky thiols can physically access the buried [2Fe-2S] site, we performed a 200 ns all-atom MD simulation of NreBPAS in the presence of 35 GSH molecules. The protein remained structurally stable (Supplementary Fig. 24a, b), and no GSH molecule approached the [2Fe-2S] cluster within 0.35 nm over the entire trajectory (Supplementary Fig. 24c). These data suggest that direct access of GSH to the cluster is sterically disfavored on this timescale. We next assessed whether the small anion HS- can access the cluster. In simulations containing 50 HS- molecules, HS- repeatedly approached the cluster, and the contact frequency increased after ~60 ns (Fig. 7c, Supplementary Fig. 25a, b), consistent with transient opening events that enable ligand entry. Trajectory inspection captured directed migration of HS- toward the cluster (Fig. 7e, Supplementary Movie 2), and electrostatic potential surfaces indicate a predominantly positive environment along the entry pathway (Fig. 7f). Because classical MD does not model the chemical steps of sulfenation/persulfidation, persistent HS- occupancy near the cluster should be interpreted cautiously. In contrast, in simulations with only 10 HS--molecules (matching the 1:10 protein:HS- ratio used in vitro), we did not observe effective cluster contact within the simulated time window (Supplementary Fig. 26a–c), likely reflecting limited sampling at lower ligand abundance. Discussion This study supports a mechanism of H2S sensing through a distinctly coordinated [2Fe-2S] cluster (Fig. 8). The N-terminal PAS domain of NreB harbors the [2Fe-2S] cluster coordinated by three conserved cysteine residues in an atypical CX2CX11C motif. The 1.52 Å structure further reveals a non-cysteinyl sulfur ligand at the fourth coordination position (Fig. 3). Native MS analysis reveals the dominant ligand species is -SH, which is shifted to SSH and SSOH upon HS- exposure in the presence of O2 (Fig. 6a, d), accompanied by characteristic spectral changes and enhanced autokinase activity (Figs. 4f–i, 5c, d, 6b, c). Working model of sulfide sensing and signaling by NreB. The PAS domain of NreB harbors an oxidized [2Fe-2S] cluster coordinated by three cysteine residues and a labile sulfur ligand at the fourth coordination site, assigned predominantly as -SH in the basal state. Upon exposure to sulfide (HS-) in the presence of O2, the ligand population shifts toward -SSH/-SSOH states, correlating with enhanced NreB autokinase activity. The signal is then transmitted by phosphotransfer to NreC, leading to activation of the sqr-pdo regulon. The structural inset shows the experimentally resolved three-cysteine-coordinated [2Fe-2S] cluster in NreBPAS. The physiologically relevant off-state is not depicted because it remains unresolved. The crystal structure reveals that the NreB [2Fe-2S] cluster adopts a coordination geometry involving three conserved cysteine residues (Cys63, Cys66, and Cys78) (Fig. 2). Although NreB comprises an additional N-terminal cysteine (Cys3), structural and biochemical evidence demonstrate this residue is not essential for cluster coordination. This coordination pattern represents a significant departure from canonical Fe-S cluster-binding motifs. Both [2Fe-2S] and [4Fe-4S] clusters require four coordinating residues, primarily involving cysteine residues, with occasional participation of histidine, glutamate, or aspartate27. Recent mutational studies have suggested that RirA from R. leguminosarum might employ three cysteine residues to coordinate a [4Fe-4S] cluster46, but NreB differs in cluster type, motif conservation, and signal specificity. Intriguingly, S. carnosus NreB homologs contain a related CX2CX11CX2C motif to coordinate a [4Fe-4S] cluster25,47. Despite differing by only one cysteine, these proteins share no PAS domain sequence identity and show no direct evolutionary relationship24. It is unclear how similar Fe-S coordination motifs have arisen to support distinct sensing chemistries48. Collectively, these findings expand the known coordination space of [2Fe-2S] cofactors in biological regulation. A distinctive feature of NreB is that H2S detection does not depend on changes in the redox state of the Fe-S core. While many Fe-S-based sensors respond to O2, Fe2+, or NO through redox transitions and/or cluster degradation27,49, some sensors instead rely primarily on ligand binding or metal dissociation50–52. In NreB, only under strictly anaerobic conditions with the strong reductant DTH could the [2Fe-2S] cluster in NreB be reduced from the +2 to the +1 oxidation state (Fig. 1b, f), suggesting that the sulfide-responsive output is not dependent on redox cycling of the cluster. A critical question is how HS- accesses the deeply buried [2Fe-2S] cluster. Our MD simulations indicate that, despite the absence of obvious static channels in the crystal structure (Fig. 7), conformational fluctuations create dynamic pathways permitting water molecules and HS- ions to transiently access this cluster. Recent studies have demonstrated that protein breathing motions can create transient access pathways for gasotransmitters and small ligands to reach buried metallocofactors21,53. The identification of these transient access pathways also contributes to NreB’s selectivity. The size constraints and electrostatic properties of these dynamic channels likely contribute to selective recognition, akin to substrate selectivity mechanisms observed in other metalloproteins46,54. This controlled accessibility represents an additional layer of regulation in the NreB sensing apparatus and exemplifies how protein conformational flexibility can be harnessed for specific signal detection in Fe-S cluster-based sensory systems. Our results highlight that the output of NreB kinase is highly sensitive to assay conditions, particularly pH and solution state. Histidine kinases of the HisKA family exhibit pH-dependent activity, consistent with the protonation requirements of the phosphorylatable histidine and the labile phosphoramidate chemistry of His-P55,56. In this framework, the GSH-associated acidification observed in our Tris-based assay is expected to reduce autophosphorylation efficiency. Bulk pH could theoretically modulate the protonation state and ligation equilibria of sulfur species associated with the PAS-domain cofactor2,15,57, thereby tuning sulfur-adduct chemistry. Persulfides/polysulfides can interconvert between electrophilic (RSSH) and nucleophilic (RSS-) forms with pKa values substantially lower than parent thiols, and protein microenvironments can further tune sulfur reactivity17,19. However, native MS performed at pH 7.0 versus 8.0 in the absence of NaHS showed no major redistribution among detected cluster-adduct states (Fig. 6d & Supplementary Fig. 22), suggesting any pH-dependent effect on these equilibria is subtle under our measurement conditions. Native MS was performed on NreBPAS and required rapid removal of free GSH into a volatile electrolyte before MS analysis, as native ESI-MS is incompatible with high concentrations of nonvolatile additives58,59. Consequently, native MS primarily reports cluster-associated states that persist after processing and cannot exclude labile modification states. Additionally, neutral ammonium acetate solutions can be pH-labile during the electrospray process, potentially affecting short-lived, pH-sensitive states60–62. Consistent with this, kinase assays performed with the same MS-compatible processing preserve relative activity differences but with attenuated amplitudes (Fig. 6b). It should be noted that GSH was used here as a convenient in vitro proxy for a high-thiol background. In Bacillus species, bacillithiol is the principal cytoplasmic low-molecular-weight thiol63. Accordingly, the GSH-mediated inhibition observed in our Tris-based assays should not be interpreted as the physiological off-state of NreB, which remains to be defined. Modification of the labile sulfur coordination site represents a distinct mechanism for coupling reactive sulfur species to bacterial signal transduction. Addition of NaHS to GSH-treated NreBPAS increased the abundance of persulfidated, sulfenylated, and persulfide-sulfenylated Fe-S cluster species (Fig. 6a, d, Supplementary Fig. 18), correlating with UV-Vis transition and changes in kinase activity (Fig. 4). We propose that HS- reacts with sulfenic acid modifications on the Fe-S cluster to generate persulfide species, analogous to H2S-mediated conversion of protein sulfenic acids64. Since sulfenic acids typically form via reaction with two-electron oxidants65, this mechanism reasonably explains the observed O2 requirement for NreB-mediated sensing (Fig. 4e & Supplementary Fig. 12b). The physiological significance of our findings lies in identifying an Fe-S cluster-based sulfide-sensing mechanism that controls the sqr-pdo operon. Although the physiologically relevant off-state of NreB remains unresolved, our data support a model in which sulfide-responsive chemistry at the labile sulfur ligand modulates NreB kinase output under aerobic conditions. This regulation ensures precise control of cellular sulfide oxidation capacity by modulating the sqr-pdo operon24. These results support the presence of a distinct link between sulfur-ligand chemistry at an Fe-S cluster and sulfide homeostasis. This finding may facilitate future studies on understanding bacterial signal transduction and its potential biotechnological or diagnostic applications. The selectivity and sensitivity of NreB’s H2S-sensing mechanism may inform the development of biosensors for environmental H2S detection or provide insights into bacterial adaptation to sulfide-rich environments. Further investigation into the structural dynamics of this sensing mechanism and its conservation across bacterial species may provide broader insights into gasotransmitter sensing and the functional diversity of Fe-S proteins. In conclusion, the PAS domain of NreB is involved in H2S sensing. The structural and MS analyses reveal a three-cysteine, and one-SH coordinated [2Fe-2S] cluster. The -SH ligand is converted to -SSH and -SSOH upon exposure to HS- with enhanced autokinase activity of NreB. This mechanism expands the functional repertoire of [2Fe-2S] cofactors and provides a foundation for dissecting how bacteria integrate sulfide/sulfane-sulfur stress into signal transduction for efficient sulfur metabolism. Methods Bacterial strains, plasmids, culture media, and chemicals The bacterial strains and plasmids used in this study are listed in Supplementary Table 8, and oligonucleotides are listed in Supplementary Table 9. Any materials generated in this study are available from the corresponding author upon reasonable request. E. coli XL1-Blue MRF’ served as the cloning host, while E. coli BL21(DE3) was used for protein expression. E. coli strains were cultured in lysogeny broth (LB). Modified M9 medium for Bacillus licheniformis cultivation was prepared by supplementing the M9 medium with 3.5% (v/v) LB and 1% (m/v) glucose24. Ampicillin (100 µg/mL) and isopropyl β-D-1-thiogalactopyranoside (0.2 mM) were used for selection and induction. Plasmid extraction, DNA purification, and genomic DNA extraction kits were obtained from TianGen (Beijing, China). Restriction enzymes were purchased from New England Biolabs (Ipswich, MA, US). Ampicillin, IPTG, and Tris-(2-carboxyethyl)-phosphine hydrochloride (TCEP) were sourced from Sangon Biotech (Shanghai, China). L-Cysteine, hemin, and sodium dithionite (DTH) were obtained from Sigma-Aldrich (St. Louis, MO, US). Ferric citrate, S-nitrosoglutathione (GSNO), and Tricarbonyldichlororuthenium (II) dimer (CORM-2) were purchased from MedChemExpress (Holmdel, NJ, US). 5-Aminolevulinic acid (ALA) was purchased from Aladdin (Shanghai, China). NaHS was purchased from JiSiEnBei (HongKong, China). Ferrous ammonium sulfate was purchased from Macklin (Shanghai, China). Expression and purification of NreB, NreBPAS, and their variants The nreB gene was amplified and cloned into pMAL-c2X with various N-terminal tags and protease cleavage sites, generating pMAL-MBP-Factor Xa-NreB, pMAL-MBP-TEV-NreB, and pMAL-His-TEV-NreB (Supplementary Table 8). Site-directed mutagenesis was used to generate NreB variants (NreBC3S, NreBC63S, NreBC66S, and NreBC78S) using pMAL-His-TEV-NreB as a template. For protein expression, E. coli BL21(DE3) harboring these plasmids was cultured at 37 °C until OD600 reached 0.6, followed by induction with 0.2 mM IPTG and cultivation at 30 °C for 5 h. Fe-S cluster incorporation was optimized. In vivo, E. coli with the expression plasmids were supplemented with 167 mg/L L-cysteine and 375 μmol/L ferric citrate in the LB medium37. In vitro reconstitution was performed following the established protocol66. 100 μM purified NreB was anaerobically treated with 10 mM dithiothreitol (DTT) on ice for 30 min. Ferrous ammonium sulfate and NaHS were added stepwise to 1.5 mM, and the sample was incubated at 22 °C for 3.5 h. The reaction mixture was centrifuged (5000 × g, 4 °C, 5 min) to remove precipitates, and unbound iron and sulfide were eliminated by buffer exchange into 50 mM Tris-HCl (pH 8.0) by going through a PD-10 column (GE Healthcare, US). NreB and its variants were expressed as MBP or His-tagged fusion proteins. MBP fusion proteins were purified using amylose resin according to the manufacturer’s protocol (NEB, UK), while His-tagged proteins were purified using Ni-NTA resin (Invitrogen, US). Purified proteins underwent buffer exchange into 20 mM Tris-HCl (pH 8.0), 100 mM NaCl, and 6% (v/v) glycerol using PD-10 columns (GE Healthcare, US). Tag removal was performed overnight at 4 °C using either Factor Xa (1 μg per 50 μg protein) or TEV protease (0.5 U per 10 μg protein) according to usage instructions. The MBP tags were removed by using amylose resin. The PAS domain of nreB was cloned into pMAL-c2X with either His-tag-TEV or His-SUMO-TEV as tags for biochemical and structural analyses, respectively. E. coli BL21(DE3) harboring these constructs was grown in an LB medium supplemented with 167 mg/L L-cysteine and 375 μmol/L ferric citrate. Protein expression and purification procedures followed the protocol described above. Anaerobic procedures Unless otherwise stated, procedures were carried out under aerobic conditions. Anaerobic manipulations were performed in an anaerobic glovebox (O2 < 1 ppm). These anaerobic steps include Fe-S cluster reconstitution, dithionite (DTH) reduction, preparation of DTH-treated samples in sealed cuvettes for UV-Vis absorption and circular dichroism (CD) measurements, preparation of electron paramagnetic resonance (EPR) samples after DTH addition, and buffer exchange before native MS analysis. Samples removed from the glovebox were maintained in sealed containers (e.g., septum-sealed cuvettes or sealed EPR tubes) to minimize O2 exposure during measurements. UV/Vis spectroscopy UV/Vis absorption spectra were recorded using a UV-3600 spectrophotometer (Shimadzu, Japan) at 25 ± 0.2 °C in a 1-cm path-length quartz cuvette. For reduction analysis, ~30 μM NreB was treated with 5 mM DTH under anaerobic conditions by following standard procedures35,67. The reaction mixture was then transferred to a septum-sealed quartz cuvette and additionally sealed with parafilm before removing from the anaerobic chamber. Absorption spectra were immediately recorded between 250–600 nm. For GSH and H2S treatments conducted under aerobic conditions, NaHS was introduced to the NreB protein mixture by using a Hamilton syringe, and measurements were performed in a septum-sealed quartz cuvette. EPR measurements Continuous-wave (CW) X-band EPR spectra of NreB and NreBPAS were performed at the Steady High Magnetic Field Facilities, High Magnetic Field Laboratory, Chinese Academy of Sciences, using a Bruker EMX plus 10/12 spectrometer equipped with an Oxford ESR910 liquid helium cryostat. Samples were prepared in 20 mM Tris-HCl buffer (pH 8.0) containing 100 mM NaCl and 10% (v/v) glycerol. DTH was added under anaerobic conditions in a glove box, whereas GSH and NaHS were added under aerobic conditions. For NreBPAS, two series of EPR measurements were collected: (i) temperature-dependent spectra at a fixed microwave power of 1 mW, and (ii) microwave power-dependent spectra at fixed temperatures of 10, 40, and 77 K. Unless otherwise noted, EPR acquisition parameters were: modulation amplitude 5 G, time constant 163.84 ms, and modulation frequency 100 kHz. NreB and NreBPAS samples treated with DTH, GSH, or NaHS were measured at 40 K with 1 mW microwave power using the same parameters. Spin quantitation was performed under non-saturating conditions. The double-integrated EPR signals of the 400 μM samples were compared to those of a 400 μM CuSO4 standard sample acquired under identical parameters. CW-EPR spectra were simulated and fitted using the EasySpin (Version 6.0.6) toolbox operating in the MathWorks MATLAB environment. Buffer-only subtraction was applied to EPR spectra unless otherwise noted; no buffer subtraction was used for the variable-temperature/variable-power series and for spin quantitation. CD spectroscopy CD spectra were recorded using a JASCO J-1500 spectropolarimeter (JASCO Corporation, Japan) with a 0.1-cm path length quartz cuvette in phosphate buffer (PB, pH 8.0). Measurements were performed at a scan speed of 100 nm/min. Addition of DTH to 5 mM was performed in an anaerobic glove box, and the reaction mixture was transferred to the cuvette and sealed with parafilm before spectral scanning. GSH and NaHS treatments were conducted under aerobic conditions. GSH was used at 5 mM, while NaHS was used at a 10-fold molar excess relative to protein concentration. Upon completion of the reaction, 10 kDa and 30 kDa cutoff filters (Amicon Ultra; Millipore) were employed under anaerobic conditions to remove GSH and NaHS, and the filtrate was exchanged with deoxygenated PB buffer (pH 8.0). The data presented are an average of three scans. ICP-MS A 0.10 mL aliquot of each protein sample was added to a polytetrafluoroethylene digestion vessel, followed by microwave digestion in aqua regia (HNO3: HCl = 1:3, v/v). After cooling, the digest was diluted with aqua regia to a final volume of 5.0 mL and further diluted about 5-fold by volume for ICP-MS measurement (NexION 300, PerkinElmer, USA), resulting in a total dilution factor of approximately 250. Protein concentrations were determined using the Bicinchoninic Acid assay with bovine serum albumin as a standard. The Fe concentration measured by ICP-MS, CFe,ICP (ppb; equivalent to μg/L) was converted to the Fe concentration in the original protein solution as CFe,original=250×CFe,ICP. The experimental iron mass fraction in the protein was calculated using the following equation: wFe,original=CFe,originalV106mprotein, where V is the sample volume (L) and mprotein is the protein mass (g). Each protein monomer contains one [2Fe-2S] cluster (2 Fe atoms), and the theoretical iron mass fraction is therefore wFe,th=2MFeMprotein, where MFe= 55.85 g·mol−1 and Mprotein is the molar mass of the protein monomer (g·mol−1). The [2Fe-2S] cluster occupancy is defined as: occupancy=wFe,originalwFe,th. Structure modeling The AlphaFold algorithm68 was used to predict the tertiary structure of proteins. The structural models of proteins were analyzed and visualized with PyMOL, Version 2.5.7 (Schrödinger, LLC)69. A full-length dimer model of NreB from Bacillus licheniformis was generated using AlphaFold (version beta-20231127) with copies set to 2. A full-length monomer model of NreB was generated using open-source AlphaFold (via ColabFold v1.3.0) with copies set to 1. The corresponding UniProt identifier is Q62SG3. PLDDT-colored views from two viewing angles for the monomeric and dimeric AlphaFold models, together with the corresponding PAE plots, are provided in Supplementary Fig. 4. The AlphaFold-predicted monomer and dimer models together with the corresponding PAE files are provided as Supplementary Data 1. Crystallization and structure determination of NreBPAS The His-SUMO tag was removed with TEV protease (Beyotime, China) at 4 °C overnight, and tag-free NreBPAS was purified using a nickel-nitrilotriacetic acid (Ni-NTA) agarose resin. To obtain high-purity protein, the sample was exchanged into Buffer A (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM DTT, 5% (v/v) glycerol) and further purified using a Q-HP column (Cytiva, US). Elution was performed with Buffer B (50 mM Tris-HCl, pH 8.0, 1 M NaCl, 1 mM DTT, 5% (v/v) glycerol) using a stepwise NaCl gradient (100–500 mM). Size-exclusion chromatography was performed on a Superdex 75 (Cytiva, US) column equilibrated in Buffer C (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 5% (v/v) glycerol). The efficiency of enzymatic cleavage and fusion protein purification was assessed via SDS-PAGE analysis. Finally, the high-purity protein was concentrated to ~12 mg/mL. A pre-crystallization test was performed by mixing 1 μL of 12 mg/mL NreBPAS with 1 μL of either Reagent A1 (0.1 M Tris-HCl, pH 8.5, 2.0 M (NH4)2SO4) or Reagent A2 (0.1 M Tris-HCl, pH 8.5, 0.2 M MgCl2, 30% (m/v) Polyethylene glycol 4,000). Crystallization screening of NreBPAS protein was carried out using a Mosquito LCP crystallization robot and the sitting-drop vapor diffusion method with commercial screens (PEG/Ion, Index, and Wizard I/II). Drops contained 100 nL protein + 100 nL reservoir solution in 96-well plates with 40 μL reservoir volume, and plates were incubated at 18 °C. No matrix optimization was performed. Diffraction-quality crystals used for data collection were reproducibly obtained from the initial hit condition Index-G10 (0.2 M MgCl2·6H2O, 0.1 M Tris, pH 5.5, and 25% (m/v) polyethylene glycol 3,350). For scale-up, the same condition was repeated by hanging-drop vapor diffusion (1 μL protein + 1 μL reservoir) equilibrated against 500 μL reservoir solution in 24-well plates at 18 °C. For NreBPAS-GSH co-crystallization, 5 mg/mL NreBPAS was incubated with GSH at molar ratios of 1:30 (final GSH concentration 10 mM) or 1:300 (final GSH concentration 100 mM) for 1 h at 4 °C, followed by centrifugation at 12,000 × g for 2 min at 4 °C. Supernatants were used for crystallization at 18 °C. Crystals were obtained from condition PEGIon-E4-B3 (0.2 M sodium malonate, pH 5.0, 25% (m/v) PEG 3,350) using the 1:30 PAS: GSH mixture and from PEGIon-E4 (0.2 M sodium malonate, pH 5.0, 20% (m/v) PEG 3,350) using the 1:300 PAS: GSH mixture. Crystals were transferred to cryoprotectant (reservoir solution supplemented with 20% (v/v) glycerol) and flash-cooled in liquid nitrogen. X-ray diffraction data were collected at beamline BL10U2 of the Shanghai Synchrotron Radiation Facility (SSRF). Different images were indexed and integrated with XDS70, and scaled with AIMLESS71, as part of the CCP472 suite. The structure was solved by molecular replacement in Phaser73 using a full-length monomeric AlphaFold-predicted NreB structure68 as the search template, which was generated via the open-source ColabFold v1.3.0. To minimize model bias, low-confidence regions (low pLDDT scores) were removed, and only well-predicted core regions (residues 31–142) were retained (Supplementary Fig. 4c, d; Supplementary Data 1). No alternative molecular replacement search models were tested due to the lack of suitable homologous templates. The resulting structure was manually revised in Coot74 and refined in phenix.refine75 using positional refinement with isotropic atomic B-factors and TLS restraints. Model validation was performed with MolProbity76. Data collection and refinement statistics are summarized in Supplementary Table 3. All experiments were conducted with the assistance of ReadCrystal Biotechnology Co., Ltd (Suzhou, China). Quantum-chemical calculations Our computational models were derived from the experimentally determined structure of the NreBPAS protein. In Model A, the [2Fe-2S] cluster is coordinated by four cysteine ligands, each truncated to a methyl thiolate (CH3S-). In Model B, the [2Fe-2S] cluster is coordinated by three cysteine ligands, which were truncated to methyl thiolates (CH3S-), and by one HS- ligand, modeled as an H-S fragment. All density functional dispersion correction (DFT-D) calculations were performed with the Gaussian 16, Rev. A03 programs77. The B3LYP functional78, including Grimme empirical dispersion correction (D3BJ)79,80, with def2TZVP basis set81 was used to optimize the ground state structure. Harmonic vibrational frequency analysis was calculated at the same level to verify that the optimized structures are true minima on the potential energy surface, with no imaginary frequencies. To account for environmental effects, single‑point energies were computed on the gas‑phase geometries using the CPCM model82 with a dielectric constant ε = 4, which approximates a low‑dielectric protein interior. To improve the accuracy of the energy evaluation, the def2-TZVPP basis set was applied to all atoms. The Wiberg bond orders were calculated based on wavefunction data using the Multiwfn program83,84, and the molecular structures were visualized using GaussView 6.0. The optimized atomic coordinates of the computational models used in the DFT-D calculations (Models A and B) are provided as Supplementary Data 3. Construction of gene deletion and site-directed mutants in B. licheniformis MW3 Markerless mutants of B. licheniformis MW3 were constructed using double-crossover homologous recombination with the conjugative shuttle vector pKVM185. To generate the cysteine-to-serine mutations in NreB, DNA fragments containing 1000-bp upstream and downstream homologous arms of the nreB gene were amplified by PCR and assembled into EcoRI-digested pKVM1 plasmid using TEDA assembly. The nreB deletion strain was constructed through conjugation and homologous recombination in B. licheniformis MW3. Subsequently, the nreB gene, along with its 1000-bp flanking regions, was amplified and cloned into EcoRI-digested pKVM1. Site-directed mutagenesis was performed using the QuickChange method to substitute cysteine residues at positions 3, 63, 66, and 78 with serine86. The mutated nreB genes were then reintroduced into their original locus in the nreB deletion strain via conjugation and homologous recombination, generating three site-directed mutant strains. To eliminate transcriptional repression mediated by the sulfane sulfur-sensing repressor YrkD24 and to create a genetic background where the NreBC system is the primary regulator responding to H2S, the yrkD gene was subsequently deleted in each of these strains. This yielded the double mutants C63S-ΔyrkD, C66S-ΔyrkD, and C78S-ΔyrkD. RNA extraction and reverse-transcription quantitative PCR (RT-qPCR) Total RNA was extracted from B. licheniformis strains using TrizolTM reagent (Thermo, US). Bacterial cultures were grown in modified M9 medium to an OD600 of 0.8, followed by treatment with 200 μM NaHS. To enhance cell lysis, samples were pretreated with 0.05% (m/v) SDS (Sangon Biotech, China) and 20 mg/mL lysozyme (Solarbio, China) at 37 °C for 20 min before TrizolTM (Vazyme, China) extraction. Residual genomic DNA was removed with DNase I (Vazyme, China), and complementary DNA was synthesized using All-In-One RT MasterMix (ABM). RT-qPCR was performed on a LightCycler 480II instrument (Roche) using BestarVR SybrGreen qPCR Mastermix (DBI). The thermal cycling conditions were as follows: initial denaturation at 94 °C for 5 min, 40 cycles of 94 °C for 5 s, 62 °C for 15 s, and 72 °C for 10 s; followed by a final extension at 72 °C for 7 min. Melting curve analysis was conducted by decreasing the temperature from 94 °C to 25 °C at a rate of 0.21 °C/s with continuous fluorescence monitoring. Gene expression levels were normalized to gyrB as the endogenous control. Relative quantification was performed using the 2−ΔΔCt method24. Autokinase assays Autophosphorylation of NreB was monitored using ATP-γ-S (Abcam, UK) under the assay conditions described below, with optimization based on a previously described method44. NreB (10 μM) was incubated with ATP-γ-S (100 μM) in kinase reaction buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 5 mM KCl, 5 mM MgCl2, and 5% (v/v) glycerol) at 37 °C for 30 s. The reaction was terminated by addition of 20 mM EDTA, followed by the alkylation of thiophosphorylated NreB using 2.5 mM p-Nitrobenzyl mesylate (PNBM) at room temperature for 1 h. Alkylated proteins were separated by 12% SDS-PAGE and transferred to PVDF membranes (180 mA, 45 min, 4 °C) using transfer buffer (25 mM Tris-HCl, pH 8.0, 192 mM glycine, and 10% (v/v) methanol). Membranes were blocked with 5% (m/v) nonfat milk in TBST (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.05% (v/v) Tween 20) and probed with anti-thiophosphate ester monoclonal antibody [51-8] (Abcam, UK; cat. ab92570, lot 1073552-5; 1:5000). Detection was performed using HRP-conjugated anti-rabbit IgG(H + L) secondary antibody (Beyotime, China; cat. A0208; 1:2500). Protein bands were visualized using an Amersham Imager 680. Dynamic light scattering (DLS) Dynamic light scattering (DLS) measurements were conducted using disposable cuvettes and a DynaPro NanoStar instrument (Wyatt Technology Corporation, USA) with a laser wavelength of 658 nm. Before the experiments, the buffer solution was filtered through a 0.22 µm Millex-GP syringe filter unit to remove particulate contaminants. Native mass spectrometry (MS) Native MS analysis was performed by direct infusion under the conditions described below to preserve labile Fe-S cluster states; the workflow was adapted from previous work32,41,87 with modifications specified here. Protein samples were prepared in 100 mM Tris-HCl buffer (pH 8.0) containing 100 mM NaCl, 5 mM KCl, 5 mM MgCl2, and 5% (v/v) glycerol. Standard reaction samples were prepared using 80 μM protein, which was incubated with 5 mM GSH for 10 min, followed by addition of NaHS at a 10-fold molar excess relative to the protein concentration and further incubation for 4 min. Untreated NreBPAS protein was used as the control. The above steps were performed under aerobic conditions. Upon completion, the samples were transferred to an anaerobic glovebox for buffer exchange into 200 mM ammonium acetate (pH 8.0) using 10 kDa cutoff filters (Amicon Ultra, Millipore) to remove excess GSH and NaHS, and the protein concentration was adjusted to 80 μM. Each condition was analyzed in one biological replicate. Additional untreated samples were prepared by directly buffer-exchanging purified NreBPAS into 200 mM ammonium acetate at either pH 7.0 or pH 8.0. For pH 7.0 treatment experiments, purified NreBPAS samples were prepared in 100 mM Tris-HCl buffer (pH 7.0) containing 100 mM NaCl, 5 mM KCl, 5 mM MgCl2, and 5% (v/v) glycerol, treated with NaHS, and then transferred under anaerobic conditions into 200 mM ammonium acetate (pH 7.0) using the same buffer-exchange procedure described above. Untreated NreBPAS in pH 8.0 served as the control for these experiments. Each condition was analyzed in three independent biological replicates. Statistical significance was assessed using a two-sided one-way ANOVA with Tukey’s multiple comparisons test. In total, 12 native MS samples were analyzed. Mass spectrometric analysis was carried out on a Bruker Impact HD high-resolution Q-TOF mass spectrometer equipped with an Ultimate3000 system. No LC or GC separation was used. Samples were directly injected into the positive-ion-mode ESI source at a flow rate of 0.3 mL/h, and spectra were acquired over an m/z range of 900–4500. Instrument parameters were as follows: dry gas flow rate, 4 L/min; nebulizer pressure, 0.8 bar; dry gas temperature, 180 °C; capillary voltage, 4500 V; and offset voltage, 500 V. The instrument was calibrated using an electrospray calibrant solution (Fluka, Germany). Data acquisition and analysis were performed using OtofControl Version 3.4 and Compass DataAnalysis version 4.2 software (Bruker Daltonik, Bremen, Germany). Spectra were deconvoluted prior to analysis, and peak assignments were made based on the expected mass differences between species, using well-resolved charge states. For quantitative analysis, we normalized the intensity of each peak type to the summed intensity of all six types. Raw native mass spectrometry spectra and associated processed output files are provided as Supplementary Data 2. Molecular dynamics (MD) simulations MD simulations were performed to investigate the conformational dynamics of the [2Fe-2S] cluster in NreBPAS. [2Fe-2S] parameters were generated using Sobtop [http://sobereva.com/soft/Sobtop] (Tian Lu, version 1.0; accessed 10 Mar 2025). HS- was modeled as a two-atom anion (H-S-) using GAFF88, with a net charge of -1 and RESP-derived partial charges89,90. GSH molecules were also parameterized using GAFF with RESP charges, with a net charge of 0. The complex was simulated in GROMACS 2022.5 using Amber14SB91. The system was solvated in a cubic box of TIP3P water molecules, ensuring a minimum distance of 1.0 nm between any protein atom and the box edge. Sodium and chloride ions were added to neutralize the system. System setup and composition are summarized in Supplementary Table 10. Energy minimization was performed using 2500 steps of the steepest descent method followed by 2500 steps of conjugate gradient. Equilibration consisted of 100 ps in the NVT ensemble (constant number of particles (N), system volume (V), and temperature (T)) and 100 ps of NPT ensemble (constant number of particles (N), system pressure (P), and temperature(T)) at 298.15 K. Production MD simulations were run for 100 ns under periodic boundary conditions. Long-range electrostatic interactions were handled using the Particle Mesh Ewald (PME) method, with a cutoff of 1.0 nm applied for non-bonded interactions. To probe accessibility of sulfur-containing species, additional simulations were performed by randomly inserting either 35 GSH molecules (200 ns; ~58 mM in a 10 nm cubic box) or HS- ions (10 or 50 HS- for 100 ns; ~17 or ~60 mM, respectively) into the bulk solvent, followed by addition of Na+ ions to maintain overall charge neutrality. These simulations were used to probe the encounter/accessibility of HS-/GSH to the [2Fe-2S] cluster in the context of non-reactive classical MD. The MD checklist can be found in Supplementary Table 11. Statistical analysis All statistical data were analyzed using GraphPad Prism version 9.0. Experimental data are presented as the mean ± standard deviation (SD) unless otherwise indicated. Comparisons among multiple groups were analyzed by ordinary one-way ANOVA followed by Dunnett’s multiple-comparisons test or Tukey’s multiple comparisons test as specified in the relevant figure legends. Experiments involving two factors were analyzed by two-way ANOVA followed by Sidak’s multiple-comparisons test. All statistical tests were two-sided unless otherwise stated. Exact p values are shown in the figures where applicable. Full statistical results for the relevant null-hypothesis tests, including test statistics, degrees of freedom, adjusted p values, and 95% confidence intervals where applicable, are provided in the Source Data file. Uncropped scans of gels and blots from the main figures are provided in the Source Data file. Uncropped and unedited source images for the Supplementary Information are provided in Supplementary Fig. 27 and are also included in the Source Data file for convenience. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Movie 1 Supplementary Movie 2 Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements We thank the staff members of the Electron Spin Resonance System (https://cstr.cn/31125.02.SHMFF.ESR) at the Steady High Magnetic Field Facility, CAS (https://cstr.cn/31125.02.SHMFF). We thank Prof. Xiang Gao and Pixian Gong from Shandong University for advice on the Western blot experiments. We also thank Guannan Lin and Jing Zhu of the Core Facilities for Life and Environmental Sciences, State Key laboratory of Microbial Technology of Shandong University for LC-MS and DLS analyses. Author contributions Conceptualization, L.Y.X. and Y.Z.X.; Methodology, C.T., Y.M.S., J.Y.Q., and L.Y.; Investigation, C.T., Y.M.S., J.Y.Q., L.Y., H.L.L, and H.W.L; Formal Analysis, C.T., J.Y.Q, and L.Y.; Writing – Original Draft, C.T.; Writing – Review & Editing, C.T., L.Y.X., and Y.Z.X.; Funding Acquisition, L.Y.X. and Y.Z.X.; Supervision, L.Y.X. and Y.Z.X. Peer review Peer review information Nature Communications thanks Nick Le Brun and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding This work was supported by grants from the National Natural Science Foundation of China (Grant No. 32370037 to Y.Z.X and 92351302 to L.Y.X.) and the Shandong Provincial Natural Science Foundation (Grant No. ZR2023MC004 to Y.Z.X.). Additional funding was provided through the Youth Innovation Promotion Association of Chinese Academy of Sciences (Grant No. 2022455 to L.Y.), and the Intramural Joint Program Fund of the State Key Laboratory of Microbial Technology (Project No. SKLMTIJP-2024-10 to Y.Z.X.). All other authors declare no relevant funding. Data availability The atomic coordinates and structure factors have been deposited in the Protein Data Bank database under accession code 9WEX (NreBPAS crystal structure). The molecular dynamics datasets for NreBPAS are available in Zenodo under 10.5281/zenodo.18398974 [https://zenodo.org/records/18398974] (apo and 35GSH conditions), and 10.5281/zenodo.18408668 [https://zenodo.org/records/18408668] (10HS- and 50HS- conditions). The custom force-field parameters and topology files for the Fe-S cluster are available in Zenodo under 10.5281/zenodo.18410016 [https://zenodo.org/records/18410016] (force-field parameters and topology files). The EPR spectroscopy data are available in Zenodo under 10.5281/zenodo.19478184 [https://zenodo.org/records/19478184] (EPR spectroscopy data). The coordinate file and MTZ structure factors for the NreBPAS [2Fe-2S] crystal structure are available in Figshare under 10.6084/m9.figshare.31046923 [10.6084/m9.figshare.31046923] (coordinate file and MTZ structure factors). The mass spectrometry raw and deconvoluted data have been deposited to the ProteomeXchange Consortium [https://www.proteomexchange.org/] via the PRIDE partner repository [https://www.ebi.ac.uk/pride/] under accession code PXD076861 (native and intact protein mass spectrometry raw and deconvoluted data). The previously published PDB structures 3GIG (HisKA_3 domain from B. subtilis DesK) and 3D36 (HATPase domain from G. stearothermophilus KinB) are also available. Molecular dynamics simulations of NreBPAS without and with HS- are provided as Supplementary Movies 1, 2, respectively. The AlphaFold-predicted monomer and dimer models, together with the corresponding PAE files, are provided as Supplementary Data 1. Raw mass spectrometry spectra and associated processed output files are provided as Supplementary Data 2. The optimized atomic coordinates of Models A and B used in the DFT-D calculations are provided as Supplementary Data 3. Unless otherwise stated, all data supporting the results of this study can be found in the article, supplementary, and source data files. Source data are provided with this paper. Code availability No custom code was generated in this study. Molecular dynamics simulations and related analyses were performed using publicly available software as described in the Methods. The custom force-field parameters and topology files used for the Fe-S cluster are available in Zenodo (see Data Availability statement above). No restrictions apply to access. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Luying Xun, Email: luying_xun@vetmed.wsu.edu. Yongzhen Xia, Email: xiayongzhen2002@sdu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73842-2. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Movie 1 Supplementary Movie 2 Reporting Summary Transparent Peer Review file Source Data Data Availability Statement The atomic coordinates and structure factors have been deposited in the Protein Data Bank database under accession code 9WEX (NreBPAS crystal structure). The molecular dynamics datasets for NreBPAS are available in Zenodo under 10.5281/zenodo.18398974 [https://zenodo.org/records/18398974] (apo and 35GSH conditions), and 10.5281/zenodo.18408668 [https://zenodo.org/records/18408668] (10HS- and 50HS- conditions). The custom force-field parameters and topology files for the Fe-S cluster are available in Zenodo under 10.5281/zenodo.18410016 [https://zenodo.org/records/18410016] (force-field parameters and topology files). The EPR spectroscopy data are available in Zenodo under 10.5281/zenodo.19478184 [https://zenodo.org/records/19478184] (EPR spectroscopy data). The coordinate file and MTZ structure factors for the NreBPAS [2Fe-2S] crystal structure are available in Figshare under 10.6084/m9.figshare.31046923 [10.6084/m9.figshare.31046923] (coordinate file and MTZ structure factors). The mass spectrometry raw and deconvoluted data have been deposited to the ProteomeXchange Consortium [https://www.proteomexchange.org/] via the PRIDE partner repository [https://www.ebi.ac.uk/pride/] under accession code PXD076861 (native and intact protein mass spectrometry raw and deconvoluted data). The previously published PDB structures 3GIG (HisKA_3 domain from B. subtilis DesK) and 3D36 (HATPase domain from G. stearothermophilus KinB) are also available. Molecular dynamics simulations of NreBPAS without and with HS- are provided as Supplementary Movies 1, 2, respectively. The AlphaFold-predicted monomer and dimer models, together with the corresponding PAE files, are provided as Supplementary Data 1. Raw mass spectrometry spectra and associated processed output files are provided as Supplementary Data 2. The optimized atomic coordinates of Models A and B used in the DFT-D calculations are provided as Supplementary Data 3. Unless otherwise stated, all data supporting the results of this study can be found in the article, supplementary, and source data files. Source data are provided with this paper. No custom code was generated in this study. Molecular dynamics simulations and related analyses were performed using publicly available software as described in the Methods. The custom force-field parameters and topology files used for the Fe-S cluster are available in Zenodo (see Data Availability statement above). No restrictions apply to access.