The bacterial strains used in this study are E. coli DH5α and E. coli BL21 codon plus (DE3). These strains were grown in LB broth at 37°C, 150 rpm. Antibiotics were added as required for the culture at the indicated concentration—chloramphenicol (20 μg ml⁻¹), tetracycline (20 μg ml⁻¹), and kanamycin (50 μg ml⁻¹). Both the larger and shorter isoforms of the SeCobB gene were cloned using genomic DNA of Salmonella enterica subspecies I serovar Enteritidis str. P125109. The pET 28a (+) plasmid was used for both cloning and recombinant protein expression. The detailed protocol for cloning is given in Supplementary Material.

Both shorter and full-length isoforms of the SeCobB protein (SeCobB_S and SeCobB_L, respectively) were purified using Ni²⁺–NTA affinity chromatography (Supplementary Figure S2). The oligomeric status of SeCobB isoforms was determined by size exclusion chromatography (SEC) using a Sephacryl S-200 16/60 GPC column attached to AKTA Pure (GE Healthcare). The column was pre-equilibrated with a protein elution buffer (25 mM Tris pH 8, 200 mM NaCl, 2 mM β-mercaptoethanol, and 5% glycerol) at a flow rate of 0.5 ml/min. It was calibrated using SEC standard protein markers (Gel Filtration Markers Kit, Sigma-Aldrich: MWGF200), and the void volume (Vo) was determined by passing blue dextran under the same conditions. Each of the eluted peaks were analyzed by SDS-PAGE to identify the presence of SeCobB_S and SeCobB_L. The apparent molecular weight of SeCobB_S and SeCobB_L was determined by interpolating the peak elution volume (Gupta et al., 2000) on the SEC standard plot. The SEC standard plot is a linear calibration curve produced by plotting the logarithms of the known molecular masses (log MW) of protein standards versus their respective Ve/Vo values, where Ve is the elution volume and Vo is the void volume.

The secondary structure of recombinant SeCobB_S and SeCobB_L was determined using circular dichroism (McDevitt et al., 2011) spectra using a Chirascan CD spectrometer (Applied Photophysics) at the Central Research Facility of Institute of Life Sciences, Bhubaneswar, India. A graph was plotted between wavelength (λ) and molar ellipticity (ϴ) after baseline correction. All the data are represented as an average of three scans, with the protein purified in three independent batches.

To gain insights into how the solution-state structure of the protein changes in the presence of a divalent metal, DLS experiments were conducted using a multi-angle particle size analyzer from Photocore Ltd. (Russia). DLS was utilized to determine the size distribution of the protein in both the absence and presence of ZnCl₂.

Ni–NTA-purified recombinant SeCobB in the protein elution buffer (25 mM Tris, pH 8, 200 mM NaCl, 2 mM β-mercaptoethanol, and 5% glycerol) was subjected to high-speed centrifugation at 12,000 rpm, 10 min at 4°C to avoid any possible air bubbles. The supernatant was used to prepare suitable dilutions using the protein elution buffer as the solvent. They were transferred to a clean and dry cylindrical glass vial of 10 mm diameter for carrying out DLS. The outer surface of the cuvette was gently wiped with lint-free tissue before placing into the instrument to remove dust or dirt to avoid unnecessary scattering from the glass wall.

In the DLS study, the protein suspension at a very dilute concentration is illuminated with a laser light of wavelength (λ) 654 nm, and the scattering intensity is collected at a scattering angle θ (=90^o) using a photon detector. The intensity auto-correlation is calculated as (Lima et al., 2011) g 1 τ = g 2 − 1 β . (1)

In Eq. 2, I(t) is the correlation function at time t and I t + τ is the correlation of the same signal with a delay time. The angular brackets denote the average over time t. The intensity auto-correlation is related to the field correlation as (Lima et al., 2011) g 2 t = < I t I t + τ > < I t > 2 . (2)

The correlation functions are analyzed using a CONTIN-based method, and the corresponding size distributions are obtained by fitting the correlation function with Eq. 2 (see Figure 1). g 1 τ = ∫ 0 ∞ G Γ e − Γ τ d Γ , (3) where Γ is the characteristic decay rate, which relates the translational free diffusion coefficient D _o as Γ = D o Q 2 . (4) Here, Q is the scattering vector and is related to the scattering angle θ by Q = 4 π μ   sin ( θ 2 ) λ , and D _o is given by the Stokes–Einstein equation: D o = k B T 6 π η R h . (5) Here, k_B is the Boltzmann constant, T is the absolute temperature, η is the viscosity of the solvent, and R _h is the hydrodynamic radius.

The primary sequence of the NAD⁺-dependent deacylase protein of Salmonella Enteritidis PT4 (strain P125109) was retrieved from the UniProtKB database (accession ID: A0A6C7HR52); the protein had a length of 273 aa (SeCobB_L) (Consortium, 2015). However, we deleted the first 37-amino acid (aa) sequence to check their binding affinities against Zn²⁺ in a comparative way, where the 236-aa sequence was referred to as the shorter isoform (SeCobB_S) and the 273-aa sequence was termed the longer isoform (SeCobB_L). The three-dimensional (3D) structure of both SeCobB_L and SeCobB_S was determined using online servers like I-TASSER (Roy et al., 2010). The best I-TASSER-generated structure was selected for refinement based on its C-score value. Qualitative analysis of the best modeled structure of both protein isoforms was done through different online servers.

Servers like PROCHECK (Roman et al., 1993), Verify3D (Lüthy et al., 1992), and ERRAT (Colovos and Yeates, 1993) were used for the quality factor analysis of the model protein from the SAVES meta-server. The ProSA webserver (Wiederstein and Sippl, 2007) was used for the analysis of the Z-score of the target protein. However, the best modeled structure was obtained from the I-TASSER server, with 2.1% residues for SeCobB_L and 1.5% residues for SeCobB_S in the outliers and 72% in the case of SeCobB_L and 74% in the case of SeCobB_S in the favorable region. These structures were then selected for refinement based on their respective C-score values in the GalaxyRefine webserver (Ko et al., 2012). This course of action was repeated until the quality of the structural conformation failed to increase any further.

The in vitro histone deacetylase assay was standardized in the laboratory (Mishra et al., 2022). In brief, HCT-116 cells were treated with sodium butyrate (NaBU) for 24 h to hyperacetylate all the proteins including histones in the cells. Mammalian hyperacetylated histones were enriched by TCA precipitation and used as a substrate for HDAC assays. Then, 50–200 ng of recombinant SeCobB was incubated with 10 μg hyperacetylated, acid-extracted core histones in the presence of NAD⁺ (5 mM) as a cofactor in the HDAC assay buffer (50 mM Tris-Cl, 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂, 1 mM DTT, 5% glycerol, and 0.2 mM PMSF) for 60 min at 37°C.

We also performed an in vitro histone deacetylase assay in the presence of zinc (ZnCl₂). A measure of 0.5 M ZnCl₂ stock was prepared by dissolving 681.45 mg ZnCl₂ (SRL-87288) in a minimum of 2 N HCl and then increasing the volume up to 10 ml with distilled water. A measure of 100 ng recombinant SeCobB was incubated with 10 μg hyperacetylated, acid-extracted core histones in the presence of NAD⁺ (5 mM), along with varying concentrations of ZnCl₂ (0.25–2.5 mM) under the same conditions as mentioned above. The immunoblots were probed with anti-acetyl lysine and anti-H3 antibodies. The in vitro deacetylase assay for both SeCobB_S and SeCobB_L was performed separately.

To check the binding mode interaction between Zn²⁺ and SeCobB_S/SeCobB_L, docking studies were performed using AutoDock v4.2.6 software (Olson and Olson, 2010). It was initialized with protein and ion preparations. The grid scale was positioned at 40 × 40 × 40 xyz points with a grid spacing of 0.375 Å, and the grid core was chosen at dimensions (x, y, and z) 54.535, 58.898, and 50.521, respectively. The docking analysis was carried out using a rigid protein and genetic algorithm with the following default parameters: the maximum number of generations = 2700, maximum number of seeds = 2,500,000 runs, population size of 150, and 100 GA runs. Following that, docking was done by setting the parameters to default values, followed by using the command-line interface for autogrid and autodock applications.

We performed a specific docking approach on both sites and a blind docking approach to check the binding affinity of Zn²⁺ toward SeCobB_S using AutoDock Vina software and the online server Metal Ion-Binding site prediction and modeling server (MIB2), respectively. A webserver that was used to create the expected metal ion-bound 3D structure and prediction of metal ion-binding residues was defined. To create a binding template, areas that bind 12 different types of metal ion-binding residues were taken into consideration. The query protein and the template were compared structurally using the fragment transformation approach without any data training. The template contained residues that are within 3.5 Å of the metal ions by modifying the scoring algorithms based on structural and binding residue similarity. The prediction of binding residues for 18 different types of metal ions, namely, Ca²⁺, Cu²⁺, Fe³⁺, Mg²⁺, Mn²⁺, Zn²⁺, Cd²⁺, Fe²⁺, Ni²⁺, Hg²⁺, Co²⁺, Cu⁺, Au⁺, Ba²⁺, Pb²⁺, Pt²⁺, Sm³⁺, and Sr²⁺, is supported. The metal ion docking after prediction is also provided using MIB (Lu et al., 2012; Lin et al., 2016).

SeCobB_S and SeCobB_L (5 µM in the protein elution buffer, pH 8.0) were titrated with ZnCl₂ (0–100 μM, at a rate of 5 µM per addition). Tryptophan fluorescence intensity of this bivalent metal ion-bound SeCobB_S and SeCobB_L was recorded with an excitation wavelength of 295 nm using a spectrofluorometer (FLS1000, Edinburgh Instruments, United Kingdom). The temperature of the samples was maintained at 25°C by using Peltier attached to the spectrofluorometer. Stern–Volmer plots and Scatchard analysis were done using corrected fluorescence data considering the effect of dilution. The linear fit of the data was obtained using the Stern–Volmer equation, F 0 F = 1 + K S V Q , (6) and the Scatchard equation, log F 0 − F F = l o g K + n l o g Q , (7) where F ₀ and F are the emission intensities of SeCobB_S and SeCobB_L in the absence and presence of the zinc ions, respectively, provided the Stern–Volmer quenching constant (K_SV), the binding constant (K_b), and the number of binding sites (n). Here, [Q] stands for [zinc ions].

The thermal stability of SeCobB_S and SeCobB_L was determined using a thermal-induced denaturation experiment. In brief, both protein isoforms (5 µM in the protein elution buffer, pH 8.0) were incubated in the absence and presence of ZnCl₂ (0–100 µM). Intrinsic tryptophan fluorescence spectra of all the samples were recorded in the 310–400-nm region using an excitation wavelength of 295 nm. The change in tryptophan fluorescence at 334 nm was recorded stepwise between 25°C and 90°C. As mentioned previously (Nandi et al., 2013), Vant Hoff enthalpy (ΔHVH) and entropy (ΔS) were calculated from the thermal melting data.

The schematic domain organization of SeCobB, illustrating both shorter and larger isoforms, the predicted NAD⁺-binding domain, Zn²⁺-binding site, substrate-binding site, and active site, is shown in Figure 1A. To further characterize Salmonella CobB (SeCobB), we cloned it in E. coli using a pET-28a (+) vector, and the N-terminal 6×His-tagged protein was extracted and purified by Ni–NTA agarose metal affinity chromatography (Figure 1B, lanes 2–3 and 4–5). Size exclusion chromatography showed that SeCobB_S exists as a monomer, whereas SeCobB_L exists as an oligomer in solution (Figure 1C). Circular dichroism spectroscopy revealed that both isoforms have a similar secondary structure that predominantly harbors alpha helices, illustrated by peaks at 208 and 222 nm, respectively (Figure 1D).

We determined the oligomeric status of SeCobB isoforms by multi-angle dynamic light scattering. The concentration-dependent correlation function for SeCobB_S (denoted as CobS_C1-CobS_C4) and SeCobB_L (denoted as CobL_C1-CobL_C4) is shown in Supplementary Figure S3. The data are fitted with the distribution of decay rates G (Γ) using Eq. 2. The experimental function (Eq. 1) agreed well with the theoritical fit function, and the obtained distribution of the hydrodynamic radius (Eq. 5) was plotted as shown in Figures 1E, F for SeCobB_S and SeCobB_L, respectively. It is worth noting that for all concentrations ranging from 93 to 750 μg/ml of SeCobB_S, the size distribution curve has multiple peaks that vary from 5 nm to several microns, which clearly demonstrates the heterogeneous nature of protein particles with a large variation in their sizes coexisting in the solution state (Figure 1E). SeCobB_L gives a single distribution peak that overlaps with each other for the entire range of concentrations (Figure 1F), demonstrating a homogeneous particle size distribution within the protein solution.

The bar diagram representation corresponding to the hydrodynamic radius plot is demonstrated as a function of the percentage (%) of size distribution for SeCobB_S and SeCobB_L in the inset of Figures 1E, F, respectively. The major contribution (∼75%) in CobS_C1 to CobS_C4 comes from the main peak that corresponds to a mean radius of 147 nm ± 60 nm, whereas the radius of the small peak varies from 2 to 11 nm with a population of ˂ 5% of total proteins in the solution coexisting with a small % of micrometer-sized aggregates. These proteins, which are monomers with a radius range of 2–11 nm, can be aligned to their molecular weight estimated in GPC. In addition, we also observe large aggregates of the protein at a high concentration (CobS_C3 and CoBS_C4) amounting to a small population (<5%). SeCobB_L, which is an oligomer in solution, displays a single peak corresponding to a mean radius of 49.99 nm ± 15 nm that coincides with the molecular weight estimated in GPC.

A deacetylase protein of Archaeoglobus fulgidus, Sir2-Af2 deacetylating C-terminal of p53 peptide, and yeast Hst2 deacetylating acetylated histone H4 peptide have been reported (Tanny et al., 2004). Thus, we investigated whether SeCobB can deacetylate core histones. To establish the same, we performed in vitro histone deacetylase assay on acid-extracted, hyperacetylated core histones with SeCobBs and SeCobB_L separately in the presence and absence of NAD⁺, followed by Western blotting with anti-acetyl lysine and anti-H3 antibodies. The acetyl lysine antibody detected the concentration-dependent deacetylation of core histones by both isoforms in the presence of NAD⁺ (Figures 2A, B, lanes 2–4), whereas no deacetylation occurred in the absence of NAD⁺ (lane 5, top panel, Figures 2A, B). A gradual decrease in the band intensity was representative of concentration-dependent deacetylation. An equal amount of substrate was used in each reaction, as indicated by the uniform band intensity of H3 using the anti-H3 antibody (Figures 2A, B, bottom panel).

No structural information is available for NAD⁺-dependent protein deacylase (CobB) of Salmonella Enteritidis PT4 (strain P125109). Due to a lack of validated structural information about the target protein, Zn²⁺-binding site and NAD⁺-binding site residues were predicted using UniProt (Figure 1A) and I-TASSER servers. The I-TASSER-modeled structure of SeCobB_L was compared with the AlphaFold database-predicted structure to check their structural similarity. The structural alignment showed that the RMSD between these two structures was 1.549 Å (Supplementary Figure S5), which is less than the default cut-off of RMSD of 2 Å, and that the extent of dissimilarity between two structures were not greater, as per their structural comparison. Likewise, the predicted NAD⁺-binding sites and zinc ion-binding sites are shown in Figure 1. The protein 3D structure generated through the I-TASSER server represents the good quality after validation through Ramachandran plot analysis, Verify3D, ERRAT, and the ProSA server (Tables 3, 4). The Ramachandran plot analysis of the SeCobB_S model structure before refinement revealed that 74.0% of residues were in the allowed region and 1.5% in the disallowed region. However, after protein structure refinement using the GalaxyRefine webserver, the number of amino acid residues in the allowed region increased to 92%, and only 1% of residues were in the disallowed area. Similarly, the number of residues in the SeCobB_L modeled structure was shifted from 72.0% to 89.8% in the allowed region, whereas it was shifted from 2.1% to 1.7% in the disallowed region. The validation of our selected model through the Verify3D program (checks for the compatibility of a three-dimensional atomic model with its own one-dimensional amino acid sequence) revealed that there were 74.15 residues for SeCobB_S and 72.53 residues for SeCobB_L with an average 3D–1D score >0.2 (Supplementary Figures S4E, F). The server “ERRAT” provided the overall quality factor (expressed as the percentage of the protein for which the calculated error values fall within the 95% rejection limit) of the model as 93.1818 for SeCobB_S and 94.024 for SeCobB_L (Supplementary Figures S4C, D). We studied the quality of the target protein using the ProSA webserver, and it reported a Z-score value of −7.05 (SeCobB_S) and −7.92 (SeCobB_L), which is well within the range of the native conformation of the crystal structure. The ProSA analysis of the model structure revealed an improvement in the Z-score after structural refinement.

The predicted 3D model structure of Salmonella CobB (Figure 3C) shows that it contains a zinc-binding motif. Most of the Sirt2 family proteins contain a Cys-X-X-Cys-(X)15–20-Cys-X-X-Cys sequence as a characteristic feature of the Zn²⁺-binding motif within their conserved domain. The predicted zinc-binding sites in SeCobB according to “UniProt” software are cysteine at positions 155 and 174 (Figure 1A). To explore the effect of the Zn²⁺–cysteine interaction on the activity of Salmonella CobB, we performed an in vitro deacetylase assay at different concentrations of ZnCl₂ with SeCobB_S and SeCobB_L separately (Figures 3A, B). The anti-acetyl lysine antibody detected a gradual decrease in the band intensity with decreasing ZnCl₂ concentration, representative of the increase in the deacetylase activity (Figures 3A, B, lanes 2–6). The band intensity was the lowest in the absence of ZnCl₂, indicative of maximum deacetylation by restoring the deacetylase activity (Figures 3A, B, lane 7). An equal loading of the substrate was confirmed by re-probing the blots with the anti-H3 antibody (Figures 3A, B, bottom panel).

Since a predicted zinc-binding motif exists within SeCobB, whose function is unknown, and an in vitro HDAC assay illustrated that zinc mediated inhibition in the deacetylase activity, we performed a docking analysis of Zn²⁺ ions with the predicted 3D model structure of SeCobB_S (Figures 3C, D). The predicted Zn²⁺-binding sites in the case of SeCobB_L are Cys-155 and Cys-174, while it is Cys-118 and Cys-137 in the case of SeCobB_S. The site-specific docking study of SeCobB_L–Zn²⁺ revealed that Zn²⁺ showed three H-bond interactions (Leu-150, Lys-152, and Pro-185) (Figure 3E), none of which is present within the predicted Zn²⁺-binding site. In the case of SeCobB_S, the H-bond interactions are Lys-136, Cys-137, and Leu-146 (Figure 3C), among which only CYS-137 is present within the predicted Zn²⁺-binding sites. These H-bond interactions showed a binding affinity of −1.2 kcal/mol toward Zn²⁺ for both isoforms. The blind docking study revealed that Zn²⁺ showed two H-bond interactions (Cys-174 and Cys-177) for SeCobB_L and four H-bond interactions (Cys-118, Ser-121, Cys-139, and Cys-140) for SeCobB_S (Figures 3D, F), of which Cys-174 for SeCobB_L and Cys-118 in the case of SeCobB_S are the only residues present within the predicted Zn²⁺-binding sites. The hydrogen bond interactions showed a binding score of 3.222 for the longer isoform and 3.272 for the shorter isoform with zinc-binding sites.

Divalent cations are known to induce protein aggregation in solution, leading to a change in particle size, which can be studied by DLS and other techniques. This, coupled with fluorescence spectroscopy and other techniques, helps further elucidate the complex interplay of factors that influence a protein’s size and aggregation behavior in the presence of metal ions and salt. In the DLS experiment, the protein solution was treated with 250 μM ZnCl₂, and we monitored changes in the size and evolution of size aggregates over 600 s (Figure 4). Upon the introduction of the salt, both SeCobB_S and SeCobB_L exhibited two types of size distributions in the solution state (Eq. 5), designated as A and B in Figure 4. For both SeCobB_S and SeCobB_L, within distribution 1, the hydrodynamic radius remained relatively stable at approximately 30 nm throughout the experiment. In contrast, within distribution 2, the hydrodynamic radius showed a remarkable increase from 90 to 300 nm over the observed time frame for SeCobB_L. Conversely, for SeCobB_S, within distribution 2, the variation in the hydrodynamic radius was much narrower, spanning from 35 to 50 nm. This substantial change in the hydrodynamic radius for SeCobB_L within distribution 2 suggests the progressive formation of larger protein complexes. This phenomenon is likely attributed to the strong interaction between the protein molecules and ZnCl₂. Importantly, this observation strongly supports the findings of the experimental studies in fluorescence spectroscopy, which indicated the presence of more available zinc-binding sites within SeCobB_L than within SeCobB_S. The average scattering intensity is known to be proportional to the size distribution of proteins in the solution. Therefore, we investigated its time-dependent variation immediately after the induction with ZnCl₂. Figure 4C shows the average intensity as a function of time, denoted as I_t (normalized with the intensity at time zero), for SeCobB_S and SeCobB_L following ZnCl₂ induction. It is evident from the plot that the average value of I_t/I_o for SeCobB_S is lower than that for SeCobB_L and remains relatively constant throughout the observed time period. In contrast, for SeCobB_L, the average intensity experiences a rapid increase immediately after induction, peaking at 50 s, after which the rate of increase in intensity slows down with time. The higher average scattering intensity and the rapid initial increase observed in SeCobB_L, compared to SeCobB_S, are consistent with the presence of a larger size variation within SeCobB_L.

The proteins containing an intrinsic fluorophore tryptophan can be used to monitor the binding of metal ions (Mattocks et al., 2021). The binding of Zn²⁺ ion with SeCobB_S and SeCobB_L was monitored by the quenching of tryptophan fluorescence upon the addition of zinc ions. A measure of 5 µM of SeCobB_S and SeCobB_L was titrated with zinc ions from 0 to 100 µM. Quenching of fluorescence intensity was observed upon the addition of Zn²⁺, which is suggestive of the interaction between Zn²⁺ and both protein isoforms (Figures 5A, 6A). It is observed from Figures 5A, 6A that the quenching in SeCobB_S is ∼6%, while that in SeCobB_L is ∼14%. This shows that zinc interacts strongly with SeCobB_L compared to SeCobB_S.

The intrinsic fluorescence intensity of both protein isoforms decreased gradually upon increasing the concentration of zinc ions, but the saturation due to zinc quenching was observed earlier in SeCobB_S than that in SeCobB_L (Figures 5A, 6A). Stern–Volmer quenching (Eq. 6) and Scatchard analysis (Eq. 7) revealed that the binding affinity of Zn²⁺ with SeCobB_L is more than that with SeCobB_S (Table 1). The Stern–Volmer quenching constant for Zn²⁺ with SeCobB_S and SeCobB_L is 5.20 × 10² M⁻¹ and 2.84 × 10³ M⁻¹, respectively (Figure 5B, 6B and Table 1). The extent of fluorescence quenching signifies the association of zinc ions with both protein isoforms. The state of equilibrium between free and bound proteins upon binding with small molecules is defined by the Scatchard equation.

The association binding constant (K_b) for SeCobB_S and SeCobB_L with zinc is 2.29 × 10 M⁻¹ and 1.8 × 10¹⁰ M⁻¹, with the binding stoichiometry/site (n) 0.60 and 2.75, respectively (Figures 5C, 6C and Table 1). This shows that the tendency of Zn²⁺ to bind with SeCobB_L is much higher than that with SeCobB_S, which is also reflected in the number of binding sites. The in vitro deacetylase assay demonstrates that there is maximum inhibition in the deacetylase activity of SeCobB_S and SeCobB_L at 0.75 and 1 mM ZnCl_2, respectively. Beyond this concentration, saturation is attained in the band intensity. According to the relative intensity graph, 1.98% and 2.79% residual deacetylase activity of SeCobB_S and SeCobB_L was observed in the presence of 0.75 mM ZnCl₂ compared to the activity in the absence of ZnCl₂ (Figures 3A, B and Supplementary Figure S4). This result also corroborates with the fact that the number of zinc-binding sites in SeCobB_S is less than that in SeCobB_L.

The structural stability and integrity of a protein are essential to exhibit its function. Therefore, to assess the stability of SeCobB_S and SeCobB_L, thermal denaturation experiments were conducted. The thermal denaturation of SeCobB_S and SeCobB_L carried out in the absence of zinc ions showed a bi-dose response model fitting, which revealed two thermal melting (T_m) values for both protein isoforms (Figure 7). The T_m values for SeCobB_L are 31.20°C and 71.32°C, and those for SeCobB_S are 28.67°C and 67.94°C (Table 2) in the absence of zinc ions (Figure 7) (Table 2 and Figures 7A, B).

A similar bi-dose response fitting was also reported by Anand et al. (2019) for the Eis protein from M. tuberculosis and M. smegmatis. This shows that both protein isoforms have a sequential thermal denaturation mechanism. The thermal denaturation data clearly show that SeCobB_L is thermally more stable than SeCobB_S at ∼ 3°C–5°C. In the presence of Zn²⁺, the thermal stability of both SeCobB_S and SeCobB_L decreases but more significantly for SeCobB_L. The addition of 100 µM Zn²⁺ reduces the T_m1 value of SeCobB_S from 28.67°C to 27.75°C (ΔT_m1 = 0.92°C) and the T_m2 value of SeCobB_S from 67.94°C to 64.85°C (ΔT_m2 = 3.09°C). In SeCobB_L, an increase in the zinc ion concentration decreases the T_m1 value from 31.20°C to 24.72°C (ΔT_m1 = 6.48°C) and the T_m2 value from 71.32°C to 58.91°C (ΔT_m2 = 12.41°C), respectively. The ΔT_m1 (0.92°C and 6.48°C for SeCobB_S and SeCobB_L, respectively) and ΔT_m2 values (3.09°C and 12.41°C for SeCobB_S and SeCobB_L, respectively) in the presence of 100 µM ZnCl₂ revealed that the interaction or binding of zinc ions decreased the thermal stability of SeCobB_L to a greater extent than that of SeCobB_S.

The higher decrease in the thermal stability of the former in the presence of zinc ions may be due to the greater binding of the zinc ions with SeCobB_L than with SeCobB_S. The calculation of the van’t Hoff enthalpy of the thermal denaturation curves revealed that the van’t Hoff enthalpy for SeCobB_L for the first thermal melting transition decreased significantly from 89.4 to 41.1 kJ mol⁻¹ (Table 2), while for the second thermal melting transition, it decreased from 110.8 to 100.6 kJ mol⁻¹ (Table 2 and Figure 7) upon the addition of zinc ions. The greater decrease in the van’t Hoff enthalpy for the first denaturation transition may be due to the perturbation in the oligomeric assembly of SeCobB_L upon the addition of zinc ions. Since SeCobB_S is monomeric in nature, the change in van’t Hoff enthalpy upon the addition of zinc ions is not very significant.

Prokaryotic NAD⁺-dependent deacetylases, predominantly known as CobB, the mammalian SIRT5 isoforms, are fairly conserved in Gram-negative bacteria and have been extensively investigated in S. enterica and E. coli (Mishra et al., 2022). Gram-negative species like Vibrio, Mycobacterium, Shigella, and Klebsiella are of clinical importance due to their ability to cause infection in the host. The sirtuin core domain in SeCobB is evolutionarily more conserved in archaea than in Eukarya (Buck et al., 2004). Numerous biological processes including cellular metabolism, transcriptional repression, and epigenetic alterations are influenced by the sirtuin-mediated regulation of acetylation. The deacetylase protein, frequently mentioned as CobB in the majority of the bacterial operations, has profound relevance in Salmonella physiology, like survival under stress conditions and regulation of virulence (Li et al., 2010; Liu et al., 2018). Escalante-Semerena (2010) reported predominant deacetylase activity of SeCobB_S over SeCobB_L in removing the lysine residue (K609) from acetylated acetyl CoA synthase (Acs^Ac). Our data are the first report to biochemically prove SeCobB to be a NAD⁺-dependent histone lysine deacetylase (Figure 2), justifying them to be a class III deacetylase. We found the complete absence of acetylated H3 and H4 bands at 200 ng of SeCobB_S (Figure 2A), which are still visible in the case of SeCobB_L (Figure 2B). The higher deacetylase activity of SeCobB_S than that of SeCobB_L is consistent with an earlier report. Only two such previous reports demonstrated eukaryotic histones as a substrate for a bacterial deacetylase (Zhao et al., 2004; Mishra et al., 2022). In vitro deacetylation of core histones by SeCobB clearly indicates the ability of a bacterial deacetylase to alter host proteins during an infection.

Zinc metalloenzymes are therapeutic targets in cancer, cardiac disease, bacterial infection, and Alzheimer’s disease. The majority of these enzymes are targeted by a potential drug candidate by investigating the interaction with the Zn²⁺ bound to them. As a result, the precise prediction of the protein–Zn²⁺ interaction is a key part of computational docking and virtual screening against Zn²⁺-binding proteins.

The Ramachandran plot was used to define the best modeled structure of SeCobB_S based on the C-score, and software applications like PROCHECK, ERRAT, and Verify3D were used for further refinement (Supplementary Figure S4). The polished ERRAT server output of the SeCobB structure is shown as a function of error values versus amino acids in Supplementary Figure S4. With a resolution of 2–3 Å, the protein structure received a score >90%. The red and yellow sections of the ERRAT graph reflect uncertain parts of the structure, while the white portions show the definite parts. This plot analysis identifies residues with error values >95% and 99% in very less time (Tables 3, 4).

Predictability of an anticipated model is determined using the ProSA webserver. It reveals the accuracy of the modeled protein with respect to the experimentally crystallized structure. It also generates a Z-score that represents the overall model quality. Its value is reflected in a plot of all experimentally determined protein chains in the current PDB. Different colors differentiate groupings of structures from different sources (X-ray and NMR) in this figure.

The software application used here to determine the amino acid residues involved in the SeCobB–Zn²⁺ interaction follows different algorithms and force fields issued for docking purposes. The residues that most often bind to metal ions are CYS, HIS, GLU, and ASP (Auld, 2001; Golovin et al., 2005) because the atoms of their charged side chains can coordinate with metal ions. We performed site-specific docking using AutoDock, in which Cys 118 and Cys 137 were defined in a grid box to facilitate the specific interaction of Zn²⁺ toward these two predicted sites for SeCobB_S. The same exercise was also performed for SeCobB_L. Cys 118 and Cys 137 are annotated as Cys 155 and Cys 174 in the larger isoform. However, no such specific Zn²⁺-binding sites were defined in the case of blind docking; instead, the designed server had specified the grid box based on their template structure. Since the parameters used for both the techniques are different, Zn²⁺ might have more binding affinity toward one cysteine than the other. Further experiments are required to ascertain the binding affinity of Zn²⁺ toward either of the cysteine residues.

NAD⁺-dependent protein deacetylase activity is critical to the physiological function of proteins belonging to the SIR2 family (Imai et al., 2000; Landry et al., 2000; Smith et al., 2000). Zinc has two major functions; either it can serve as a cofactor of enzyme catalysis, or it can act as an inhibitor of enzyme function. Zinc-mediated inhibition of enzyme activity can occur through three different mechanisms, namely, direct binding to the enzyme active site, allosteric mode of inhibition, and inhibition induced by binding to Zn²⁺ consecutive to catalytic zinc. The binding affinity of Zn²⁺ toward a protein, which is necessary to establish the physiological significance of zinc-mediated inhibition, varies from micromolar to picomolar concentration. Zinc-mediated inhibition of the enzyme phosphoglucomutase has been reported, which uses magnesium as its cofactor. The concentration of Zn²⁺ required for its inhibition is > 32 pmol/L (Magneson et al., 1987). The Zn²⁺ concentration beyond 1 mM is cytotoxic to the cells (McDevitt et al., 2011), hence determining the pathophysiological concentration. We performed in vitro histone deacetylase assay using 100 ng of purified SeCobB_S and SeCobB_L separately with various concentrations of ZnCl₂ (0.25–1.25 mM). With the increasing concentration of ZnCl₂, there was a sequential inhibition in the catalytic activity of both isoforms, which was restored in the absence of ZnCl₂ (Figures 3A, B). The inhibition and restoration of deacetylase activity were ascertained by the enhanced and diminished band intensity captured from chemiluminescence data. This manifested the role of Zn²⁺ ions in inhibiting the catalytic activity of both isoforms. However, the extent of inhibition in the biochemical activity was different for both. The maximum inhibition of SeCobB_S activity was achieved with 1.25 mM ZnCl₂, whereas for SeCobB_L, it was at 1 mM (Supplementary Figure S6).

Interesting insights into the importance of Zn²⁺ in the field of nutrition, enzyme catalysis, protein biochemistry, and cellular biology have been obtained with the development of analytical techniques, like fluorescence spectroscopy, with a high sensitive level of detection (Maret et al., 2001; Maret, 2013b). This has shed light on the zinc-mediated alteration in the structure–function regulation of enzymes involved in various cellular processes. Since SeCob_B contains three tryptophan residues, we used fluorescence spectroscopy to record any changes in the thermal and structural stability of the protein due to the Zn²⁺–SeCobB interaction. A quenching study was done to define the binding affinity of Zn²⁺ with both SeCobB_S and SeCobB_L. Data obtained from fluorescence quenching studies displayed a reasonable interaction between Zn²⁺ and SeCobB. The K_sv value for SeCobB_S is 10 times lower than that of SeCobB_L (Table 1), suggesting the requirement of a comparatively higher ZnCl₂ concentration for interaction equivalent to SeCobB_L. The binding constant (K_b) for SeCobB_S is 10⁹ times less than that for SeCobB_L, also supported by the number of Zn²⁺-binding sites (Table 1). The binding affinity of Zn²⁺ toward SeCobB_L is higher than that toward SeCobB_S. SeCobB_L is an oligomer in solution, whereas SeCobB_S is a monomer (Figure 1C). Thus, being an oligomer, SeCobB_L confers a greater number of available zinc-binding sites, resulting in adequate binding and relevant interactions.

The thermal denaturation graph of both isoforms illustrates the biphasic melting curve both in the absence and presence of Zn²⁺ (Figure 7). This may be because of the independent melting of the NAD⁺-binding Rossmann fold domain and Zn²⁺-binding domain, which are the characteristic domains of the SeCobB predicted structure (Gao et al., 2020).

The initial 37-amino acid stretch in the N-terminus of SeCobB_L is 37% hydrophobic, which is a potential reason for its oligomerization. Moreover, DLS results demonstrate the narrow size distribution in SeCobB_L (Figure 1F) that facilitates more Zn²⁺-binding sites than in SeCobB_S (Figure 1E). However, we conducted more detailed DLS studies to understand the effect of Zn²⁺ on the structural properties of SeCobB and their dependence on temperature (Figure 4) in order to correlate with fluorescence spectroscopy studies. Induction with 250 µM ZnCl₂ at 20°C led to an increase in the scattering intensity of both SeCobB_S and SeCobB_L within 50 s of salt addition. However, the time-dependent leap in the scattering intensity (I_t) upon ZnCl₂ addition was ∼9 times more in SeCobB_L than that in SeCobB_S, implying that zinc mediated a profound impact on the larger isoform than on the shorter isoform.