Crystal structure of toxin HP0892 from Helicobacter pylori with two Zn(II) at 1.8 Å resolution

Abstract

Antibiotic resistance and microorganism virulence have been consistently exhibited by bacteria and archaea, which survive in conditions of environmental stress through toxin–antitoxin (TA) systems. The HP0892–HP0893 TA system is one of the two known TA systems belonging to Helicobacter pylori. The antitoxin, HP0893, binds and inhibits the HP0892 toxin and regulates the transcription of the TA operon. Here, we present the crystal structure of the zinc-bound HP0892 toxin at 1.8 Å resolution. Reorientation of residues at the mRNase active site was shown. The involved residues, namely E58A, H86A, and H58A/ H60A, were mutated and the binding affinity was monitored by ITC studies. Through the structural difference between the apo and the metal-bound state, and using a homology modeling tool, the involvement of the metal ion in mRNase active site could be identified. The most catalytically important residue, His86, reorients itself to exhibit RNase activity. His47, Glu58, and His60 are involved in metal binding where Glu58 acts as a general base and His47 and His60 may also act as a general acid in enzymatic activity. Glu58 and Asp64 are involved in substrate binding and specific sequence recognition. Arg83 is involved in phosphate binding and stabilization of the transition state, and Phe90 is involved in base packing and substrate orientation.

Introduction

Toxin–antitoxin (TA) systems have been shown to help bacteria and archaea survive in conditions of environmental stress¹ by modulating the global level of biological processes such as translation and DNA synthesis.^2–8 The mechanism of TA systems depends on a stable toxin and a proteolytically unstable antitoxin, encoded by a single TA operon, in which antitoxin is generally being located upstream of toxin.^2,6,9,10 In healthy survival conditions, the antitoxin forms a tight complex with the toxin, neutralizing the cytotoxic effect of a toxin, and the complex remains in a stable dormant stage.^11,12 However, a change in temperature, oxidative stress conditions, or nutritional deprivation give rise to stress conditions, which trigger the host proteases, enabling them to degrade the structurally unstable antitoxin more rapidly than the toxin. The unbound or free toxin in the cell causes cytotoxic activity resulting in the inhibition of cellular processes, leading to cell death.^1,2,5,13,14 This process is commonly known as post-segregational killing.¹⁵ Although the first proposed role of these TA systems was to arrest cell growth, enabling bacteria to survive in unfavorable environmental conditions, many other functions have also been reported, such as gene regulation, persistence, and programmed cell death.^2,6,16

TA pairs are very common in bacteria and archaea, most of which have been identified on the Escherichia coli K-12 chromosome,¹⁷ such as the relBE, yefM-yoeB, mazEF, dinJ-yafQ, yafNO, hipBA, chpBK, hicA, mqsAR, prlF-yhaV, ygjNM, and ygiUT.^{4,6,11,18–23} These TA systems have been classified into three classes. Class I includes ribosome-independent sequence-specific endoribonucleases such as MazEF.¹⁷ Codon-dependent ribonucleases such as RelBE belongs to Class II.^1,6 Class III includes the RNA antitoxins which neutralizes the toxin activity as in YafN-YafO.¹¹ Of RelBE superfamily toxins, RelE, YafQ, YoeB, HigB, and YhaV are most widely studied.^8,24–28 These proteins inhibit translation through related but distinct mechanisms. The RelE toxin is a ribosome-associated endoribonuclease and has a preference for pyrimidines in the first position of the codon and purines in the second and third position of the codon.²⁹ The RNase activity of RelE generally occurs through a three step process, as described by Neubauer et al. First, the abstraction of a proton by a general base activates the 2′-OH of RNA, allowing it to act as a nucleophile, which enables the RNA to reorient itself for an inline attack at the phosphate group. Stabilization of the trigonal bipyramidal transition state must then take place and, lastly, a general acid donates a proton to stabilize the leaving 5′-OH group.³⁰ YafQ is also a ribosome-associated endoribonuclease and has a preference for an AAA RNA codon followed by an A or G in the subsequent codon²⁷ and its RNase activity is neutralized by the DinJ antitoxin through the formation of a strong and stable protein-protein complex.¹¹

Although both the toxin and antitoxin of class II TA systems are proteins, the involvement of metal ions in TA systems has not been discussed much. It has been estimated that, one-fourth to one-third of all proteins requires metal, although in variable quantities.^31,32 The general tendency of some common metal ions in biological system is as follows. Iron is involved in electron transfer and oxygen metabolism, in the form of heme or iron–sulfur clusters.^32,33 Zinc is associated with ribosomes, representing a huge proportion of the total cellular zinc quantity.³² Polymerases such as RNA and DNA polymerases often require zinc for proper functionality, and zinc is also associated with the organization of protein structures, as in zinc fingers or driving catalysis by acting as a Lewis acid. Copper is known to be involved in cytochrome oxidases and periplasmic enzymes. DNA polymerases, such as ATPases and kinases, also require magnesium.³² However, magnesium is generally found to be associated with protein complexes rather than directly interacting with proteins.³² Cobalt is bound to vitamin B12, while nickel is bound to enzymes such as ureases and hydrogenases.³²

Many databases and bioinformatics servers provide analysis of protein sequences and prediction of their structural fold. However, very few are available for the analysis of the interaction between proteins and metal ions.^34,35 When considering the involvement of metal ions in TA systems, even fewer studies have been concluded. One example is VapBC TA system, in which the VapC toxin is proposed to have metal-ion-dependent nuclease activity, and when TA system dissociates efficient binding between metal ion and the VapC toxin occurs.^36–38

Because of the direct involvement of TA systems in cell death, these systems have been targeted in the search for alternative antibiotics against multidrug-resistant bacteria.^39–41 However, in Helicobacter pylori, only two TA systems, HP0892–HP0893 and HP0894-HP0895, have been identified to date using bioinformatics⁴² or determined experimentally.^43,44 HP0892 was reported as a toxin that is homologous to the HP0894 toxin. Given the vast activity of these TA systems, very little is known about the TA systems of H. pylori. For instance, the involvement of general acid and/or general base residues in a trigonal bipyramidal transition state, which is required for efficient RNase activity, as seen in E. coli RelE toxins,³⁰ has not been discussed in detail for HP0892. This study was undertaken to investigate the effect of metal ions on toxins, specifically the HP0892 toxin from HP0892–HP0893 TA system, and its interaction with the substrate-binding residues. In addition, we investigated the binding of zinc metal ions with the HP0892 toxin through NMR perturbation, circular dichroism, and isothermal titration calorimetry (ITC). Taken together, the data from these experiments reveal the location of binding sites of zinc ions and the nature of their interaction with the HP0892 toxin.

Results and Discussion

Protein preparation

The HP0892 toxin was overexpressed with the pET-21a(+) vector containing an intentionally inserted non-native thrombin cleavage site just before the C-terminal hexahistidine tag, and was successfully crystallized. After thrombin cleavage, four non-native amino-acids (LVPR) remained attached to the C-terminus of the HP0892 protein.

Analysis of secondary structure and thermal unfolding by circular dichroism

CD spectra were measured in the far-UV region (200–250 nm) to visualize the effect of different metal ions (Ca²⁺, Cd²⁺, Co²⁺, Cu²⁺, Mg²⁺, Mn²⁺, Ni²⁺, and Zn²⁺) on the HP0892 toxin [Fig. 1(A)] at 20°C. The CD spectra measured for 20 μM of untreated HP0892 displayed a minimum at 209 nm and at 222 nm, which is characteristic of α-helices, and the change in the secondary structure as a result of the addition of 60 μM of metal ions was measured. Analysis of the spectra upon addition of the different metal ions revealed the finer details of the secondary structure that occurred. The spectra showed the maximum change in intensity upon addition of zinc ions and displayed the maxima at 210 nm and at 218 nm. To determine the effect of these metals ions on the stability of the HP0892 toxin, a thermal unfolding experiment was performed. The thermal unfolding with 1°C/min increments from 20 to 80°C was determined for apo HP0892 as a control, and the melting temperature (T_m) was determined to be 63°C [Fig. 1(B)]. The T_m upon the addition of different metal ions was also determined. Interestingly, only copper and zinc ions showed a substantial effect on the stability of protein. Copper ions destabilized the protein, and the T_m appears to be 58.5°C, but the copper ions showed no effect on the secondary structure of HP0892 [Fig. 1(A)]. Figure 1(B) and Supporting Information Table 2 show that the addition of zinc ions resulted a higher T_m (72°C) compared to the apo HP0892, which means that HP0892 binds zinc ions and a slightly higher temperature is required to completely unfold the protein. The stabilization of the HP0892 protein fold by zinc ions [Fig. 1(B)] was also observed when analyzing the effect of metal ions on HP0892 protein [Fig. 1(A)]. Because these results were consistent, only zinc ions were used for the crystallization of metal-bound HP0892 (see below).

Figure 1.

Secondary structure analysis and thermal refolding by circular dichroism. (A) Effect of metal ions on HP0892. HP0892 protein (20 μM) was titrated with 60 μM metal ions. The change in CD (mdeg) was determined by comparing the CD spectrum of apo-HP0892 (black) and upon addition of metal ions. The maximum change was observed upon addition of zinc ions (blue). (B) Unfolded fractions were plotted against increase in temperature to see the effect of metal ions on the stability of HP0892. Thermal unfolding (T_m) of apo-HP0892 occurs at 63°C (black) and at 72°C upon addition of zinc (blue) and is shown in an inset for clarity.

Structure determination of HP0892

The dataset for zinc-bound HP0892 was collected to a resolution of 1.8 Å. To determine the structure of the metal-bound form, we initially used the NMR structure of HP0892 (PDB code 2OTR)⁴⁵ as a search model, and molecular replacement was performed. Unfortunately, molecular replacement did not yield satisfactory results. Therefore, the structure of HP0894 L66S (sequence identity 53.33%, PDB code 4LS4)⁴⁶ was used as a search model, and molecular replacement was performed successfully (we used this mutant protein structure because at this time native HP0894 [PDB code 4LTT] was not available yet).

Metal-bound structure of HP0892

The NMR structure of the apo form of HP0892 previously reported by our group⁴⁵ (PDB code 2OTR) was used for the comparison with the metal-bound form of HP0892. The apo form of HP0892 is a 10.4 kDa toxin characterized by the presence of five antiparallel β-strands (β1, β2, β3, β4, and β5) and three α-helices (α1, α2, and α3) [Fig. 2(A)]. In the metal-bound form, the first β-strand (β1) is located more closely to the last β-strand (β5), and the resulting core β-sheet is flanked by four α-helices [Fig. 2(B)]. The antiparallel β-sheet primarily contains hydrophobic residues, especially β4 and β5; notable exceptions are Tyr69 and Gln70 in the β4 strand and Arg82 in the β5 strand [Fig. 2(C,D)]. However, β2 and β3 strands, which are closest to the zinc ions, are made up of only hydrophilic residues (except Leu60 in the β3 strand). The electrostatic (Coulombic) surface charge resulting from this arrangement, which were calculated using Chimera,⁴⁷ is shown in Figure 2(E,F). The presence 3₁₀ alpha helix (η1) was observed in zinc-bound HP0892 (Fig. 2). In addition to the presence and absence of an α-helix in the apo and zinc-bound forms, respectively, minor differences in the orientation of the C-terminal helix region and loops, with a root-mean-square deviation (RMSD) of 1.97 Å, according to the program COOT,⁴⁸ were also observed. Overall, the apo structure of HP0892 determined by NMR studies and the metal-bound form of HP0892 determined by crystallography are very similar to each other. A topological representation calculated by Pro-origami interface⁴⁹ is shown in Figure 2(G,H).

Figure 2.

Comparison of the NMR structure of apo-HP0892 and crystal structure of metal-bound HP0892. (A) The NMR structure of apo-HP0892 exhibited β1-α1-α2-β2-β3-β4-β5-α3 topology. (B) The crystal structure of the zinc-bound form of HP0892 at 1.8 Å resolution (PDB code 4NRN) exhibited similar β1-α1-α2-β2-β3-β4-η1-β5-α3 topology. Two zinc ions are shown as orange spheres. Minor changes were observed in the terminal α-region and in loop regions. Hydrophobic residues in the metal-bound form are shown in (C) and (D). Electrostatic (Coulombic) surface charge shown in (E) and (F) is colored at a level of +10 (blue) and −10 (red) kcal/mol·e as shown by Chimera.⁴⁷ The orientation of the molecule in (C) and (E) is the same as that in (A), and the molecule is rotated 180° about the y-axis in (D) and (F). (G and H) Topological models of NMR and crystal structures, respectively, generated by Pro-origami interface.⁴⁹ The figures were made using PyMOL.⁵⁰

To identify any metal-coordination sites present in the HP0892 toxin, we incubated the HP0892 toxin with zinc ions (1 mM HP0892:3 mM ZnCl₂), and the mixture was then screened for crystallization. HP0892 has two monomers in the asymmetric unit and the two monomers are related by twofold symmetry. Both monomers of HP0892 contained two zinc atoms positioned in a similar pattern and close to the β2 and β3 strands (Supporting Information Fig. S1). When investigating the interaction of zinc ions with HP0892 and comparing the structure with the apo state of the HP0892 protein [Fig. 3(A)], it was found that both of these zinc ions are important to the interaction. One zinc ion interacts with His47, His60, and Glu58 [Fig. 3(B)] while other zinc ion interacts with His86 and Glu58 [Fig. 3(C,D)]. The two zinc ions were 3.4 Å apart.

Figure 3.

Zinc-bound state of HP0892. (A) The apo- and zinc-bound HP0892 were superimposed for comparison. Both structures show a similar fold. (B and C) His47, Glu58, His60, and His86 in the zinc-bound (blue, zinc in orange) and in apo-states (green). The interaction of one zinc ion is shown in (B), and the interaction of second zinc ion is shown in (C). The distances of the interacting atoms are shown in corresponding colors. (D) Simplified and magnified image of the side chains of all interacting groups HP0892 with the zinc ions. The distances are divided into two color groups for ease of understanding.

The nitrogen [NE2] of the imidazole ring of His60 in chain A of HP0892 forms a partial bond with the first zinc ion at a distance of 1.9 Å, while the nitrogen atom [NE2] of His47 interacts within 2.1 Å distance. A reorientation of His60 and His47 is also observed when the apo and zinc-bound structures were compared. Glu58 is most likely the only residue that interacts with both zinc ions, most likely due to the strong negative charge of the glutamate residue. A partial bond is formed between the terminal carboxylic group [OE2] and the first zinc atom [Fig. 3(B)]. Along with [OE2], the other oxygen atom present in Glu58 [OE1] interacts with the second zinc atom. All these electrostatic interactions occur between Glu58 and HP0892 at a distance of 2.4–3.2 Å. His86 [ND1] exhibits a strong interaction with zinc and shifted considerably by the difference of 7.8 Å. Although the C-terminal end of a protein in NMR structure may be flexible, the 7.8 Å shift between the apo and zinc-bound forms of HP0892 indicate the importance of His86.

Of all the zinc-interacting residues of HP0892, histidine residues play a major role in binding zinc ions. The large change in the orientation of His86 may minimize the distance between itself and zinc ions, thereby facilitating the catalysis. The reorientation of other histidine residues, namely His47 and His60, may also facilitate the reaction and acts as general acid in the acid–base catalysis reaction. Glu58 also showed some reorientation upon binding with metal ions but to a much lesser extent. Given the strong negatively charged carboxylic groups of Glu58, the role of Glu58 might be to act as a general base in acid–base catalysis reaction.

Thermal analysis and binding stoichiometry using ITC data

ITC was used to determine the exact binding stoichiometry between HP0892 and zinc ions as shown in Figure 4. The mutational protein, namely E58A HP0892, H86A HP0892, and double mutational construct E58A/H60A HP0892 were also subjected to ITC analysis to look for the effect of interacting residues as suggested by zinc-bound crystal structure of HP0892 (Supporting Information Fig. S2). The HP0892 protein contains a single site for two zinc ions to bind and the binding affinity (K) along with the heat change (ΔH) during this reaction was recorded. The data exhibited a single-site binding isotherm, which indicates that both zinc ions have an equal opportunity to interact and bind with the HP0892 pocket. The details of the calculations are provided in Supporting Information Table SI. H86A HP0892 exhibited threefold decrease in binding stoichiometry as compared to native HP0892 titration with zinc ions (Supporting Information Fig. S2 and Supporting Information Table SI). The single mutational construct E58A HP0892 and double mutational construct E58A/H60A HP0892 did not exhibit any binding with zinc ions, hence the ITC data could not be fitted and binding affinity could not be derived. Taken together, the data suggest that the E58A and H60A are responsible for the binding with metal ions while Glu58 acts as a general base and His60 acts as a general acid.

Figure 4.

Thermal analysis and binding stoichiometry from ITC data. Top panel: heat change occurred when zinc chloride was added to the chamber containing HP0892. Bottom panel: the area underneath each deflection is integrated and represents the total heat exchanged (black squares). A single-site binding isotherm model was used to fit the data.

Resonance assignment and chemical shift perturbation experiments

A series of two-dimensional [¹H–¹⁵N] HSQC spectra of 400 μM U-¹⁵N HP0892 alone and upon addition of 200 μM, 400 μM, 800 μM, and 1200 μM of ZnCl2 were recorded. Clear chemical shift changes were observed in slow exchange mode on the NMR time scale for many residues of HP0892, while chemical shifts of some of the residues were observed in fast exchange mode (Fig. 5). Chemical shift changes in the residues upon addition of Zn ions in the slow exchange mode were saturated at a 1:2 molar ratio (HP0892:Zn²⁺), while the chemical shift changes in fast exchange mode continued above a 1:2 molar ratio in a zinc concentration-dependent manner. As observed in the zinc-bound HP0892 crystal structure (Fig. 3), two zinc ions are required to fill the HP0892 binding pocket. A plot of the chemical shift changes (Δδ_{avg (ppm)}) upon addition of 800 μM zinc against HP0892 residues is shown in Figure 6(A). As observed in the plot, the signals of Leu17, Lys45, Asp46, His47, Leu61, Leu65, His86, and Ser87 disappeared, indicating their direct or indirect involvement in zinc ion binding. The plot also shows that the His60 residue was shifted to a large extent in the [¹H–¹⁵N] HSQC spectra.

Figure 5.

NMR titration of HP0892 with zinc. A series of two-dimensional [¹H–¹⁵N] HSQC spectra of 400 μM U-¹⁵N HP0892 alone (black) and upon addition of 200 μM (red), 400 μM (blue), 800 μM (green), and 1200 μM (magenta) ZnCl₂ are shown. The overlapping regions in the HSQC spectra are shown in the inset for clarity. Clear chemical shift changes were observed in the slow exchange mode on NMR time scale for many residues, while chemical shifts of some of the residues were observed in fast exchange mode.

Figure 6.

Mapping of NMR chemical shift perturbations of HP0892 upon addition of zinc ions. (A) A plot of the chemical shift changes upon addition of 800 μM zinc. The peaks that exhibited the most movement are shown in dark gray, and those that disappeared are shown in light gray. The three catalytically important histidines are shown. (B) The perturbed residues mapped onto the structure. The perturbed residues are mapped onto the structure to visualize their position in the three-dimensional structure of metal-bound HP0892. The image on the right is a 180° rotation about the y-axis of the image on the left. The figure shows the major involvement of the β-sheet and the position of zinc ions.

His47, Glu58, His60, and His86 were expected to show some chemical shift perturbation in the NMR experiment due to their involvement in zinc binding as shown by the crystal structure. However, the unexpected involvement of Leu17, Lys45, Asp46, Leu61, Leu65, and Ser87 was also observed. The position of all these residues in the metal-bound state is represented in Figure 6(B). These unexpectedly affected residues are mostly hydrophobic leucine and lysine, which have long, flexible hydrocarbon side chains. Upon interaction of HP0892 with zinc ions, the long side-chain of Leu17 might be affected by the large conformational change in the side chain of His86 [Fig. 6(B)], resulting in its own perturbation in chemical shift. Similarly, the combined effect of His86 and His60, might affect the orientation of Leu65. The addition of zinc ions and the movement of His86, His47, Glu58, and His60 to effectively interact with the zinc ions might affect the rigidity of the binding pocket, resulting in the indirect movement of the above unexpectedly perturbed residues. The movement of Lys45 and Asp46 could also be the result of an inductive effect exerted by His47. Similar behavior is also observed in Leu61, caused by His60, and in Ser87, caused by His86. However, it appears that the structural change induced by the addition of zinc ions is limited to the binding pocket itself and does not affect the global fold of the protein [compare Figs. 1(B), 3(A), 3(B), and 6(B)].

Sequence similarity and structural homology of the HP0892 toxin and its homologs

The multiple sequence alignment of HP0892 and its homologs, HP0894 from H. pylori (53.41% identity), YafQ from E. coli (37.50% identity), YoeB from E. coli K12 (19.28% identity), RelE from E. coli K12 (18.29% identity), and RNase T1 from Aspergillus oryzae (12.79% identity), was generated by the Clustal Omega⁵¹ webserver tool (http://www.ebi.ac.uk/Tools/msa/clustalo/) and visualized with Jalview2.8⁵² [Fig. 7(A)]. The sequence alignment showed that all these RelE superfamily members share high similarity in the C-terminal region. These proteins possess a highly conserved catalytic histidine residue, His86 in HP0892,⁴³ His84 in HP0894,^44,46 His87 in YafQ,¹¹ His83 in YoeB,¹ His92 in RNase T1,⁵³ and Arg81 in RelE,⁵⁴ which has been shown to be essential to the RNase activity. As discussed earlier, His86 shows the maximum reorientation and H86A HP0892 exhibited threefold less binding affinity with zinc ions (Supporting Information Fig. S2 and Supporting Information Table SI). Mutation of His87 in YafQ resulted in complete loss of activity,²⁷ and the movement of His83 of YoeB is necessary for catalytic action while interacting with YefM.¹ Both these histidines are in a location structurally similar to the His86 of HP0892, suggesting that His86 of HP0892 is a catalytic residue. His47 and His60 in HP0892 resemble His47 and His60 in HP0894, and His50 and His63 in YafQ. Mutation of His60 in HP0892 resulted in complete loss of metal binding activity (Supporting Information Fig. S2). Similar results were observed where mutation of His50 and His63 in YafQ also resulted in the complete abolishment of mRNase activity in vivo.¹¹ The Tyr38 and His40 in RNase T1 serve as a general acid in RNase activity,⁵³ and they are in proximity to His47 and His60 in HP0892. However, His47 and His60 in HP0892 do not colocalize structurally with any of the catalytic residues of YoeB. Highly positive charged residues (His47 and His60 in case of HP0892) might be required for efficient protonation in an acid–base catalysis reaction and thus serve as a general acid in the reaction.^11,43,44,46

Figure 7.

Sequence alignment of HP0892 and structural comparison. (A) Sequence alignment of HP0892 and its homologs. The highly conserved residues among RelE superfamily members are colored. Catalytically important histidine residues that act as general acids are colored red. Glutamates and aspartates that function as general bases are in green. Arginine residues involved in phosphate binding and stabilization of the transition state are in blue, and residues involved in substrate orientation and catalysis (phenylalanine) are in black. (B) Left side: Structural homology between zinc-bound HP0892 (PDB code: 4NRN), copper-bound HP0894 (PDB code: 4LSY), and E.coli YafQ model. The E. coli YafQ model was obtained by computer modeling using the Swiss-Model protein homology server.^55–57 E. coli YafQ is in cyan, zinc-bound HP0892 in blue, and copper-bound HP0894 in green. The zinc ions are in orange, while copper ion is in magenta. The citrate located in copper-bound HP0894 is in black. Right side: The overlay and zoom view of interacting residues of HP0892 and HP0894. The figure explains the similarity in the location of metal ions in the binding pockets of H. pylori toxins.

Mutation of Glu58 in HP0892 (E58A HP0892 and E58A/H60A HP0892) resulted in complete loss of metal affinity towards HP0892 protein (Supporting Information Fig. S2 and Supporting Information Table SI). This Glu58 along with Asp64 in HP0892 coincides with Asp61 and Asp67 in YafQ, and Glu58 and Asp64 in HP0894, suggesting that they may have similar roles. Asp61 and Asp67 of YafQ and Glu58 of HP0894 were reported to be general bases in the acid–base catalysis reaction,^11,46 while Lys52 and Asp64 of HP0894 were proposed to be involved in substrate binding and specific sequence recognition.⁴⁴ Taken together, these data indicate that Glu58 and Asp64 of HP0892 are important for substrate binding and specific sequence recognition. The C-terminal arginines of RelE superfamily members are very well conserved. Arg83 of HP0892 aligns with Arg80 of HP0894 and Arg83 of YafQ in sequence, structurally aligns with Arg65 of YoeB and Arg77 of RNase T1. Mutation of Arg83 in YafQ resulted in a decrease in RNase activity,¹¹ and Arg80 of HP0894 and Arg65 of YoeB are required for efficient phosphate binding and stabilization of the transition state.⁴⁶ Considering the sequential and structural homology, it appears that Arg83 in HP0892 also serves a similar role to that of Arg80 in HP0894 and Arg83 in YafQ. Glu58 also acts as a general base in the acid–base catalysis reaction, and Arg83 of HP0892 is responsible for phosphate binding and the stabilization of transition state.

The interaction of nucleotides with the C-terminal end of a toxin is required in ribosome-dependent RelE superfamily members.³⁰ This critical reorientation of nucleotides is caused by the conserved basic side chain of aromatic residues. The aromatic side-chain of Tyr87 in RelE,³⁰ Tyr84 in YoeB,¹ and Phe88 in HP0894^44,46 seem to be involved in this type of interaction. Thus, highly conserved aromatic residues at the extreme C-terminus of toxins, Phe90 in case of HP0892, are expected to be involved in base packing and substrate orientation and support the acid–base catalysis reaction.

As mentioned above and as shown in Figure 7(A), HP0892 shares high sequence similarity with HP0894 and YafQ, and their structural homology was used to determine similarities and differences among the RelE superfamily members. However, the E. coli YafQ structure is currently unknown, so structural modeling, as described earlier, was performed to compare the structures.^11,22,46 The E. coli K12 mRNA interferase YafQ sequence was used for modeling with the Swiss-Model protein homology server.^55–57 The YafQ model based on the HP0892 NMR structure (PDB code 2OTR)⁴⁵ showed a modeling score comparable to that of experimentally determined protein structures (−1.796). Structural quality assessment was performed using the ProSA-web protein structure analysis server,⁵⁸ and the overall quality of the model was analyzed (score −3.74). The sequence similarity between YafQ and HP0892 could be attributed to their structural similarity, because the HP0892 model was used for E. coli YafQ modeling. The generated model was used for structural homology comparison between YafQ, zinc-bound HP0892 and copper-bound HP0894 (PDB code 4LSY)⁴⁶ [Fig. 7(B)]. All these RelE superfamily members show high similarity in their structural assemblies. The antiparallel β-sheet is particularly similar, while the arrangement of α-helices differs in each case. The C-terminal α-helix of all these proteins appears to be the most flexible, and there are some minor changes in the loop regions. The interacting groups, namely Glu58, His47, His60, and His86 in HP0892, are located in similar position as those of Glu58, His47, His60, and His84 in HP0894, indicating that all these proteins share a considerable similarity in their binding pockets.

Conclusion

The apo- and zinc-bound HP0892 essentially show a similar structural fold [Fig. 3(A,B)], indicating that the local residues involved in RNase activity rather than the global structural fold are important. Similar structural aspects have been observed in the apo- and copper-bound forms of HP0894.⁴⁶ Sequence similarity [Fig. 7(A)] and structural homology [Fig. 7(B)] of the HP0892 toxin and its homologs revealed that the distribution of the active residues responsible for RNase activity is similar to that of HP0894 and E. coli YafQ, but differs from that of E. coli YoeB, E. coli RelE, and A. oryzae RNase T1.⁴⁶ In the RNA cleavage site, HP0892, HP0894, and YafQ have a preference for purine over pyrimidine bases, located immediately downstream from the cleavage site in vitro.^11,27,44,46 In contrast, E. coli RelE shows a preference for pyrimidines over purines, especially pyrimidines in the first position and purines in second and third position of the codon.^29,30 This uneven selection of RNA cleavage codon sites among RelE superfamily members indicates that all RelE superfamily members do not possess similar catalytic centers but rather share a general way of organizing active residues in the structure. Considering the sequence homology, structural assembly, and organization of the RNA cleavage site, HP0892, HP0894, and YafQ appear to share similar mechanism of RNase enzymatic action.

The circular dichroism experiments suggest that the protein is more stable upon binding with zinc compared to any other metal ion (Fig. 1 and Supporting Information Table SII). The ITC experiments demonstrated that two zinc ions were required in the binding pocket of HP0892 (Fig. 4). Fast exchange rates are observed in the NMR perturbation experiment [Figs. 5 and 6(A)] and the saturation limit of 1:2 molar ratio (HP0892:Zn²⁺) was observed. All these results are consistent with the zinc-bound HP0892 crystal structure. The reorientation of His86 (Fig. 4) upon zinc binding and threefold decrease in binding affinity upon mutation (Supporting Information Fig. S2 and Supporting Information Table SI) indicate the importance of this residue in HP0892 and suggest its catalytic role. Similarly the reorientation of Glu58 and His60 and the loss of binding affinity upon mutation suggest that Glu58 acts as a general base, whereas His60 acts as a general acid. Considering the similarity in position and structural organization of His47 and His60, His47 also looks like to behave as a general acid. It is possible that all this reorientation might take place to minimize the distance between the interacting residues and the introduced metal ion to exhibit the catalytic activity. As discussed by Han et.al.,⁴³ a perturbation experiment in titration of the synthetic 30-residue C-terminus of the HP0893 antitoxin with the HP0892 toxin suggested that Glu58, Arg82, and His86 were putative catalytic active residues. Interestingly, zinc-bound form of HP0892 showed direct interaction of zinc ions with His47, Glu58, His60, and His86 (Fig. 3). By comparing these two studies, it can be suggested that zinc metal ions compete with the antitoxin for binding to the toxin. However, the involvement of metal ion in toxin–antitoxin interaction is unclear and further study needed to be done.

We have presented a zinc-bound structure of HP0892 and showed changes in the orientation of active site residues that interact with zinc. These results, with our previous metal-binding studies with another H. pylori toxin, HP0894, create a platform that provides information on the involvement of metal ions in toxins and suggest possible binding sites for toxin–antitoxin interactions. The insight into the metal dependency for efficient toxin functionality and the response by antitoxin await further investigation.

Materials and Methods

Cloning and expression of HP0892

The ORF of HP0892 was amplified using genomic DNA from the H. pylori 26695 strain (ATCC, Manassas) as a template. The following primer pair was used: HP0892Fwd (5′-GAATTCCATATGCTGACGATTGAAACCAG) and HP0892Rev (5′-CCGCTCGAGCTAATGATGATGATGATGATGGCTGCCGCGCGGCACCAGAAACAGCTCGCTATGACTGCC); the NdeI and XhoI enzyme restriction sites used for cloning are underlined, and the intentionally inserted non-native thrombin cleavage site is shown in underlined italic letters in the reverse primer. After amplification, the DNA fragments were digested with NdeI and XhoI (NEB, UK) and ligated with predigested pET-21a(+) vector (Novagen, Germany). The resulting construct contained a C-terminal thrombin cleavage site just before the hexahistidine tag (LVPRGSHHHHHH). The recombinant plasmid was then overexpressed in E. coli BL21(DE3)pLysS competent cells. The EZchange Site-directed Mutagenesis kit (Enzynomics) was used to generate point mutations in HP0892 recombinant plasmid. The mutations resulted in separate recombinant plasmids, namely E58A HP0892, H86A HP0892, and E58A/H60A HP0892. The mutational recombinant plasmids, E58A HP0892 and H86A HP0892, were then overexpressed in E. coli Rosetta(DE3) competent cells, whereas E58A/H60A HP0892 was overexpressed in E. coli C41(DE3) competent cells.

Purification of HP0892

E. coli BL21(DE3)pLysS cells containing the recombinant plasmid were allowed to grow at 37°C until the OD₆₀₀ reached 0.7. Protein overexpression was induced by the addition of isopropyl-β-d-1-thiogalactopyranoside (IPTG) to 0.5 mM and the cells were grown for an additional 4 h at 37°C after induction. The cells were harvested by centrifugation at 4293g for 15 min. The harvested cell pellet was then resuspended in lysis buffer (20 mM Tris-HCl, 500 mM NaCl, pH 7.9, and 10% [v/v] glycerol) and lysed at 4°C using an ultrasonic processor (Cole Parmer) for 10 min, maintaining a pulse cycle rate of 3 s on and 6 s off, after which the lysate was centrifuged at 6708g for 1 h at 4°C. The supernatant was filtered through a 0.45 μm membrane (Millex-HV filter unit, Millipore) to remove any insoluble particles. The flow-through was applied to an open Ni²⁺-NTA column (His-bind resin, Novagen; 3 mL of resin per liter of cell culture) pre-equilibrated with buffer containing 20 mM Tris-HCl, 500 mM NaCl at pH 7.9. The sample was applied to a Ni²⁺-NTA column with a constant flow rate of no more than 0.5 mL/min under gravitational force. The column was washed with a 5× excess column volume of loading buffer containing 60 mM imidazole and the protein of interest, HP0892, was then eluted with a stepwise imidazole gradient (10 mL each of 80, 100, 150, 200, 250, 300, 350, 400, and 500 mM imidazole in loading buffer). The fractions containing the HP0892 protein as determined by SDS-PAGE analysis were dialyzed using a cellulose membrane with a 3000 molecular weight cut off (MWCO) against 20 mM Tris-HCl, 500 mM NaCl at pH 7.9, after which thrombin digestion was performed at 20°C for 24 h to remove the C-terminal hexahistidine tag. The protein was then applied to an open chelating sepharose column (Qiagen) to further remove protein that has uncleaved hexahistidine-tag. Buffer exchange was performed as described above against 20 mM MES at pH 6.0, and the protein was run on a gel filtration column (Superdex 75 [10/300GL], GE Healthcare sciences, Germany) to achieve good purity. The purity of the protein was judged to be more than 95% by SDS-PAGE analysis. The thrombin cleavage resulted in four additional non-native amino acids (LVPR) were attached to the C-terminal of HP0892. A protein concentration of 10 mg/mL was used to screen for crystallization conditions. The process for the purification of E58A HP0892, H86A HP0892, and E58A/H60A HP0892 mutational constructs remains same as that of native HP0892 purification.

Expression and purification of uniformly labeled ¹⁵N-HP0892

The E. coli BL21(DE3)pLysS cells containing the recombinant plasmid were grown overnight at 37°C in 10 mL of M9 media that contained uniformly labeled ¹⁵N ammonium chloride [U-¹⁵N] (Cambridge Isotope Laboratories) as a nitrogen source. This culture was then inoculated into fresh 1 L M9 media supplemented with [U-¹⁵N] and the cells were allowed to grow at 37°C until the OD₆₀₀ reached 0.7. Protein overexpression was induced by adding IPTG to 0.5 mM, and the cells were grown for an additional 24 h at 20°C after induction. The rest of the process for protein isolation and purification remained the same as described above. Approximately 400 μM of U-¹⁵N labeled HP0892 was prepared in 20 mM MES buffer at pH 6.0, and 10% (v/v) D₂O was added to provide the NMR internal lock signal.

NMR spectroscopy, resonance assignment, and chemical shift perturbation experiments

The phase sensitive 2D-[¹H–¹⁵N] HSQC spectrum with sensitivity improvement for U-¹⁵N labeled apo HP0892 at final concentration of 400 μM was recorded on Bruker Avance 500 MHz spectrophotometer at 30°C. Chemical shifts were externally referenced to DSS. The δ ppm values of the backbone N and HN resonances of HP0892 were assigned from the data obtained in an earlier study (PDB code 2OTR),⁴⁵ and all peaks in the [¹H–¹⁵N] HSQC spectrum of HP0892 were successfully identified. Chemical shift perturbation experiments were performed by titrating HP0892 with Zn²⁺ and the disappearance and/or shifting of peaks were monitored. An increasing concentration of Zn²⁺ ions, specifically 200, 300, 400, 800, and 1200 μM, was used for perturbation experiments. All NMR spectra were processed using the program NMRPipe and nmrDraw⁵⁹ and were analyzed with the program NMRView.⁶⁰ The averaged chemical shift changes were calculated using the following equation^43,61:

Circular dichroism

The buffer condition for all circular dichroism (CD) experiments was 20 mM Tris–HCl and 150 mM NaCl buffer at pH 7.4. Metal ions in threefold higher concentration than the protein concentration were used for titration as well as melting temperature studies. CD spectra were recorded using a 1 mm path length quartz cuvette (Hellma, Germany) in a J-715 spectropolarimeter (JASCO Corporation, Japan) equipped with a Peltier temperature control system (Model PTC-348WI). All samples were scanned three times with a bandwidth of 1 nm and a response time of 0.5 s in the far-UV region (190–250 nm) at a scanning speed of 50 nm/min. All spectra were recorded as the mean of three scans. Protein (20 μM) was titrated against 60 μM metal ions, and the data were processed by blank subtraction and smoothed before analysis. To examine the thermodynamic properties of HP0892 upon the addition of metal ions, temperature scanning was conducted with a temperature increase of 1°C/min in the range of 20–80°C at 220 nm. A band width of 2 nm and a response time of 4 s were used. The data were normalized and smoothed before analysis. The normalization of the melting temperature data was performed using GraphPad Software (GraphPad Prism version 3.03 for Windows, GraphPad Software, La Jolla, California, http://www.graphpad.com).

Isothermal titration calorimetry

The protein was dialyzed extensively in 20 mM MES buffer at pH 6.0, and zinc chloride was dissolved in same buffer, homogenized with repeated stirring, and stored at 4°C until use. Protein and metal samples were degassed and filtered before use. Isothermal titration calorimetry measurements were performed with a MicroCal iTC₂₀₀ instrument (MicroCal, GE Healthcare Sciences, Germany). The protein was concentrated to 40 μM and added to the experimental chamber. The injection syringe was filled with 1 mM zinc chloride in appropriate buffer. The experiment consisted of 30 injections into the experimental chamber, and titrations used an initial delay of 60 s, a 0.4 μL initial injection volume followed by a 5 s delay, then an injection volume of 1.5 μL followed by 150 s delay for the remainder of injections. The experimental chamber was kept under constant stirring at 1000 rpm at 25°C (±0.1°C). The obtained heat signals from the raw ITC data were integrated with the Origin software supplied by MicroCal, and the background heat from the reference buffer was subtracted to give corrected heats. The data were fit into a single-site binding isotherm and the binding affinity (K), change in enthalpy (ΔH), change in entropy (ΔS), and binding stoichiometry (N) were calculated. The values are shown in Supporting Information Table SI.

The ITC measurement for all the mutational proteins remains same as described above, except that 90 s delay was used instead of 150 s delay. E58A HP0892 and E58A/H60A HP0892 did not show any binding with zinc ions, so the ITC data could not be fitted and binding affinity could not be derived.

Crystallization

Threefold higher amount of Zn²⁺ divalent metal ions was titrated with HP0892 to obtain the metal-bound structure of HP0892. ZnCl₂ (3 mM) was added into 1 mM of HP0892 protein, and the mixture was briefly vortexed then left to sit for 30 min. The mixture was centrifuged at 13,000 rpm for 1 min, and only the supernatant was used for crystal screening. The crystallization conditions for metal-bound HP0892 were initially obtained using commercially available crystallization kits from Hampton Research and Emerald BioSystems by the hanging-drop vapor-diffusion method. One microliter of protein solution was mixed with an equal volume of crystallization buffer on a siliconized cover slip, and the mixture was equilibrated over 500 μL of reservoir solution at 4°C. Zinc-bound HP0892 was crystallized in 20% (w/v) polyethylene glycol (PEG) 3350 and 0.2M lithium sulfate monohydrate in 30 days. The cubic-shaped crystals grew to the largest dimension of 0.1 × 0.1 × 0.1 mm³. A suitable cryoprotectant was determined to be the mother liquor supplemented with 20% (v/v) glycerol, and the crystals were flash-frozen in liquid nitrogen before data collection.

X-ray data collection, model building, and refinement

X-ray diffraction data from a single crystal of zinc-bound HP0892 were collected at a resolution of 1.80 Å at about −173°C using the ADSC Quantum 210 CCD detector system at the NW12A beamline of Photon Factory, Japan. One hundred and eighty images were collected using an oscillation of 1°. The program suite HKL2000 was used for raw data processing and scaling.⁶² Crystals of zinc-bound HP0892 belong to the space group P2₁ with unit cell parameters of a = 36.01, b = 47.03, and c = 52.60, and α = γ = 90.0°, and β = 110.0°. Two monomers of zinc-bound HP0892 are present in each asymmetric unit, with a calculated crystal volume per protein weight (V_M) of 2.01 ųDa⁻¹ and a solvent content of 38.8%.⁶³ Initially, molecular replacement was attempted using the NMR structure of HP0892 (PDB code 2OTR),⁴⁵ but was unsuccessful. Then, SeMet-L66S HP0894 model (PDB code 4LS4)⁴⁶ was used successfully to determine the structure of zinc-bound HP0892 by molecular replacement. A cross-rotational search followed by a translational search was performed using the program PHENIX. Subsequent model building was performed manually using the program COOT,⁴⁸ and the model was refined using the program REFMAC⁶⁴ in the CCP4 suite⁶⁵ including bulk solvent correction. The test data for the calculation of R_free, which was 5% of the data, were randomly set aside.⁶⁶ The stereochemistry of the refined model was evaluated by MolProbity.⁶⁷ The data collection and refinement statistics are summarized in Table I.

Table I.

Data Collection and Refinement Statistics

Zinc-bound HP0892
Data collection
Beamline	NW12A beamline of PF, Japan
Wavelength (Å)	1.0000
Space group	P21
Cell dimension (Å)
	a = 36.01
	b = 47.03
	c = 52.60
	β = 110.0°
Resolution (Å)	50–1.8 (1.83–1.80)a
Rmerge (%)b	4.7 (10.7)a
I/σ(I)	48.2 (18.6)a
Redundancyc	3.6 (3.7)a
Completeness (%)	91.0 (98.7)a
Unique reflections	14,014 (741)a
Refinement
Rworkd(%)	18.22
Rfree e(%)	22.04
No. Atoms
Protein	1472
Water	140
B factor(Å2)
Protein	36.21
Water	32.47
RMSDf
Bond lengths (Å)	0.02
Bond angles (°)	2.25
Molprobity score	2.52 (22nd percentile)
Ramachandran plot (%)
Favorable region	97.2
Allowed region	2.8
Disallowed region	0.0
PDB accession code	4NRN

^aValues in parentheses indicate the highest resolution shell.

^bR_merge = ∑_hkl ∑_i|I_i(hkl) − <I(hkl)>|/∑_hkl ∑_i I_i (hkl), where I(hkl) is the intensity of reflection hkl, ∑_hkl is the sum over all reflections and ∑_i is the sum over i measurements of reflection hkl.

^cN_obs/N_unique.

^dR_work = ∑_hkl||F_obs| − k |F_calc||/∑_hkl|F_obs| was calculated with the reflections used for refinement.

^eR_free was calculated by same way as R_work, but with the 5% of the reflections excluded from the refinement.

^fRMSD was calculated with REFMAC.⁶⁴

PDB accession code

The atomic coordinates and structure factors have been deposited in the Protein Data Bank under accession code 4NRN for zinc-bound HP0892.

Supporting Information

Additional Supporting Information may be found in the online version of this article.

Supporting Information Figure 1.

Supporting Information Figure 2.

Supporting Information Table 1.

Supporting Information Table 2.