Molecular Structure and Functional Analysis of Pyocin S8 from Pseudomonas aeruginosa Reveals the Essential Requirement of a Glutamate Residue in the H-N-H Motif for DNase Activity

Pyocins are proteins produced by Pseudomonas aeruginosa strains that participate in intraspecific competition and host-pathogen interactions. They were first described in the 1950s and since then have gained attention as possible new antibiotics. However, there is still only scarce information about the molecular mechanisms by which these molecules induce cell death. Here, we show that the metal-dependent endonuclease activity of pyocin S8 is involved with its antimicrobial action against strain PAO1. We also describe that this killing activity is dependent on a conserved Glu residue within the H-N-H motif. The potency and selectivity of pyocin S8 toward a narrow spectrum of P. aeruginosa strains make this protein an attractive antimicrobial alternative for combatting MDR strains, while leaving commensal human microbiota intact.

ABSTRACT

Multidrug resistance (MDR) is a serious threat to public health, making the development of new antimicrobials an urgent necessity. Pyocins are protein antibiotics produced by Pseudomonas aeruginosa strains to kill closely related cells during intraspecific competition. Here, we report an in-depth biochemical, microbicidal, and structural characterization of a new S-type pyocin, named S8. Initially, we described the domain organization and secondary structure of S8. Subsequently, we observed that a recombinant S8 composed of the killing subunit in complex with the immunity (ImS8) protein killed the strain PAO1. Furthermore, mutation of a highly conserved glutamic acid to alanine (Glu100Ala) completely inhibited this antimicrobial activity. The integrity of the H-N-H motif is probably essential in the killing activity of S8, as Glu100 is a highly conserved residue of this motif. Next, we observed that S8 is a metal-dependent endonuclease, as EDTA treatment abolished its ability to cleave supercoiled pUC18 plasmid. Supplementation of apo S8 with Ni²⁺ strongly induced this DNase activity, whereas Mn²⁺ and Mg²⁺ exhibited moderate effects and Zn²⁺ was inhibitory. Additionally, S8 bound Zn²⁺ with a higher affinity than Ni²⁺ and the Glu100Ala mutation decreased the affinity of S8 for these metals, as shown by isothermal titration calorimetry (ITC). Finally, we describe the crystal structure of the Glu100Ala S8 DNase-ImS8 complex at 1.38 Å, which gave us new insights into the endonuclease activity of S8. Our results reinforce the possibility of using pyocin S8 as an alternative therapy for infections caused by MDR strains, while leaving commensal human microbiota intact.

IMPORTANCE Pyocins are proteins produced by Pseudomonas aeruginosa strains that participate in intraspecific competition and host-pathogen interactions. They were first described in the 1950s and since then have gained attention as possible new antibiotics. However, there is still only scarce information about the molecular mechanisms by which these molecules induce cell death. Here, we show that the metal-dependent endonuclease activity of pyocin S8 is involved with its antimicrobial action against strain PAO1. We also describe that this killing activity is dependent on a conserved Glu residue within the H-N-H motif. The potency and selectivity of pyocin S8 toward a narrow spectrum of P. aeruginosa strains make this protein an attractive antimicrobial alternative for combatting MDR strains, while leaving commensal human microbiota intact.

INTRODUCTION

Increasing rates of antibiotic resistance among Gram-negative pathogens such as Pseudomonas aeruginosa represent a serious risk to worldwide health (1, 2). The prevalence of multidrug-resistant (MDR) strains capable of inactivating a broad array of antibiotics has stimulated the development of alternative therapies (3–5). Bacteriocins produced by P. aeruginosa, termed pyocins, represent an emerging antimicrobial option for the treatment of infections caused by MDR strains (6, 7). These molecules are selective protein antibiotics produced by some bacterial species to kill their closely related strains during intraspecific competition (8, 9). Their potency and selectivity toward a narrow spectrum of P. aeruginosa strains make pyocins an attractive antimicrobial to combat MDR strains, while leaving commensal human microbiota intact (10).

Based on their structure, pyocins can be classified into three types, called R, F, and S. While R/F-type pyocins are high-molecular-weight complexes that resemble phage tails, S-type pyocins are binary protein complexes, comprising a large portion that harbors the killing function and a smaller immunity portion (Fig. 1A) (8, 11). The immunity protein protects the pyocin-producing strains by binding with very high affinity (equilibrium dissociation constant [K_d] ≈ 10⁻¹⁴ M) and specificity, thereby inactivating the killing protein (12).

FIG 1.

Structural organization of pyocin S8. (A) Proposed multidomain architecture of pyocin S8 based on similarities with others pyocins. The S8 pyocin is a binary protein complex consisting of a large protein that harbors the killing function and a smaller immunity protein. (B) Predicted secondary structure of pyocin S8 using JPred4 software (48). The amino acid sequences of the different domains are colored according to their respective colors shown in panel A. Arrows and cylinders represent strand and helices, respectively. The coiled-coil structures are indicated by black underlines. Residues highlighted by red boxes comprise the clusters of repeated R(Q/X)(A/X)E amino acid sequences present inside the coiled-coil structure. The H-N-H motif is highlighted by a bracket. (C) Amino acid identities among pyocins and colicins containing the conserved H-N-H-endonuclease motif.

Generally, both killing and immunity proteins are encoded by two separate genes organized in an operon. The transcription of pyocin genes is regulated by the coordinated action of the proteins PrtN, PrtR, and RecA (13). PrtN is a transcriptional activator that induces the expression of pyocin genes in response to stressful conditions that cause DNA damage. Under normal conditions, the transcription of PrtN is inhibited by the repressor PrtR. Under stressful conditions, the nucleoprotein filament formed by RecA binding to single-stranded DNA induces the autocleavage of PrtR, leading to the derepression of PrtN and subsequent pyocin production (13–15).

Following their synthesis, the S-type pyocins are released into the extracellular medium as stoichiometric killing-immunity protein complexes. To gain access into their intracellular targets, the S-type pyocins initially parasitize several nutrient uptake pathways that are important for cell viability under nutrient-limited conditions. This process is mediated by the so-called N-terminal receptor-binding domain that binds a specific receptor on the surface of the target cell (16–19). The N-terminal domain then interacts with several components of the proton motive force (PMF)-linked systems (Tol-Pal or Ton systems) in the inner membrane (IM) that span the periplasm to mediate pyocin transport across the outer membrane (16, 18, 20). The so-called translocation domain participates in this process, but the details still remain poorly understood (19). Some S-type pyocins also need to cross the inner membrane to interact with a cytoplasmic target. However, again, mechanisms underlying this process have been only poorly described. There is evidence that only the C-terminal cytotoxic domain enters the cytoplasm, with the immunity protein being released into extracellular medium (21–23). Once inside the cytoplasm, the cytotoxic domain exerts its lethal effect, promoting cell killing through a range of mechanisms.

Different S-type pyocin cytotoxic domains have been described, most of them displaying nuclease activities toward DNA, rRNA, or tRNA (reviewed in reference 8). DNase domains of pyocins S1, S2, and AP41 bear a conserved H-N-H-endonuclease motif as their catalytic core (24). This H-N-H motif is also found in colicins, which are well-studied DNases from Escherichia coli, as well as in other enzymes widely distributed in all kingdoms, serving a variety of functions, including homologous recombination, DNA repair, mobile intron homing, and apoptosis (25, 26). In the case of pyocins and colicins, the H-N-H motif coordinates a single divalent metal ion, and this group catalyzes random cleavages of the bacterial genome (12, 26–29).

The H-N-H motif is composed of 32 amino acids located at the C terminus of the cytotoxic domain, adopting a topology similar to that of a zinc finger, containing two antiparallel β-strands and one α-helix (ββα-Me) with a single metal ion sandwiched between them (30–32). Several conserved residues of the H-N-H motif play important roles in metal ion binding and DNA hydrolysis. Consequently, substitution of alanine for these residues abolishes the colicin activity in vivo (26). Curiously, some of these mutants retain their ability to bind transition metal ions in vitro and are able to catalyze DNA cleavage (26). Although the H-N-H motif has been extensively studied in colicins (27), structural information on this motif in pyocins is still scarce.

We have isolated a novel S-type pyocin displaying potent antibacterial activity against MDR P. aeruginosa isolates, demonstrating its potential as a protein antibiotic (33). This uncharacterized pyocin, previously designated S8 by in silico analysis (8), has a cytotoxic domain with metal-dependent DNase activity based on an H-N-H motif. Here, we report an in-depth characterization of pyocin S8 using multiple biophysical approaches, which include the elucidation of its crystal structure. We also demonstrate the essential role of the highly conserved glutamic acid (Glu100) of the H-N-H motif for pyocin activity.

(This article was submitted to an online preprint archive [34].)

RESULTS

Amino acid sequence identities and domain organization of pyocin S8.

We have previously shown that pyocin S8 displays a wide spectrum of killing activity against distinct MDR P. aeruginosa strains (33). Pyocin S8 is synthesized as a binary protein complex composed of a large killing subunit that contains 772 amino acids and its cognate smaller immunity protein with 85 amino acids. Similar to other S-type pyocins, the S8 large subunit has a multidomain organization (Fig. 1A).

The first 50 amino acids of pyocin S8 (domain I) are rich in proline and glycine residues, and secondary structure predictions suggest that this region lacks a regular secondary structure (Fig. 1A and B). This prediction is supported by the recently reported structures of pyocins S2 and S5 (16, 20). There is a proline-rich region (residues 25 to 32) within the N-terminal domain, which is predicted to adopt a conformation that mimics the natural ligands of iron transporters that pyocins utilize to enter cells (20). This N-terminal unstructured domain is followed by a helical domain with a coiled-coil structure (domain II) (Fig. 1A and B). Within this helical domain, there is a long cluster of α-helices that contain clusters of R(Q/X)(A/X)E amino acid repeats (Fig. 1B, red rectangles). The function of these sequences, which were also identified in pyocin AP41 (35), is still unknown. The central domain (domain III) is closely related to the N-terminal domain of many colicins and probably plays a function in pyocin translocation (8).

The C-terminal cytotoxic domain (domain IV) bears the H-N-H endonuclease motif that is widely conserved (see Fig. S1A in the supplemental material) and constitutes the core of pyocins’ catalytic sites. Although the H-N-H motif is conserved among H-N-H endonucleases, there is still a high level of sequence diversity within the remainder of the cytotoxic domain, especially in the region involved in the interaction with the immunity protein (immunity protein exosite [IPE]) (Fig. S1A and B). The cytotoxic domain of pyocin S8 shares about 60% amino acid identity with the cytotoxic domains of pyocins S2 and AP41 (Fig. 1C). The amino acid identities among the cytotoxic domains of pyocin S8 and colicins E2, E7, E8, and E9 are around 50% (Fig. 1C).

Interestingly, pyocin S8 shares a high amino acid sequence similarity with pyocin AP41, with the exception of the cytotoxic and immunity domains (Fig. 1C). The amino acid sequence identities in the first three domains are around 95%, suggesting that these two pyocins are translocated into the target cells by similar mechanisms.

Killing spectrum of pyocin S8.

The purification of pyocin S8 from P. aeruginosa cells is laborious, yielding small amounts of the protein. Therefore, we cloned the genes encoding the entire killing subunit and its cognate immunity protein (ImS8) (Fig. 1A) into an Escherichia coli expression vector. Since ImS8 binds to the killing subunit very tightly, these two subunits were copurified by nickel affinity chromatography due to the presence of a His₆ tag at the C-terminal end of the immunity protein (Fig. 2A).

FIG 2.

Killing activity of recombinant pyocin S8. (A) SDS-PAGE of purified pyocin S8-ImS8 complex. (B) Determination of pyocin S8 activity using an overlay spot plate method. A purified pyocin S8 (starting concentration, 5,000 μg/ml) was 10-fold serially diluted and spotted onto Mueller-Hinton agar plates containing a growing layer of P. aeruginosa strains. Clear zones indicate cell death. (C) Effect of pyocin S8 on growth of strain PAO1. The growth of the cultures in the presence of different concentrations of pyocin S8 was followed at OD₆₀₀.

Next, we analyzed the ability of recombinant pyocin S8 to kill P. aeruginosa cells. As the PAO1 strain does not produce pyocin S8 (8), it was employed as a sensitive strain in the killing activity assays. As expected, the S8-ImS8 complex was highly active against the PAO1 strain, as confirmed by clear zones of no growth around the spotting area (Fig. 2B, left panel). In contrast, the ET02 strain was completely resistant to pyocin S8, even at the highest concentrations tested (Fig. 2B, center panel). This finding was expected, as the ET02 strain contains a gene encoding ImS8 (33), which can bind to the cytotoxic domain, thereby neutralizing its endonuclease activity.

To confirm whether the cytotoxic effect is due exclusively to pyocin S8 DNase activity, we tested the pyocin activity of the Glu736Ala mutant protein. The residue at position 736 is localized in the H-N-H motif and is predicted to play an essential role in colicin activity (26). The Glu736Ala mutation rendered pyocin S8 inactive against the PAO1 strain (Fig. 2B, right panel), indicating that cell death is dependent on the pyocin DNase activity.

To measure pyocin S8 killing activity more precisely, liquid cultures of the PAO1 strain were incubated with different amounts of this recombinant protein and its effect on bacterial growth was monitored. The growth of bacterial cultures was not affected at concentrations of pyocin up to 1 μg/ml (Fig. 2C). At a 10-μg/ml concentration, growth was severely affected. At concentrations above 100 μg/ml, growth curves displayed an interesting feature: initially, the culture turbidity increased slightly, but after 90 min, it rapidly dropped to levels lower than those observed at the beginning of the curve. These results indicate that pyocin S8 induced cell lysis after a 90-min period.

Endonuclease activity of pyocin S8.

In order to test the endonuclease activity of pyocin S8, only the cytotoxic domain of the killing subunit (Fig. 1A, domain IV) was utilized, since it is predicted that only this domain penetrates the cytoplasm of the target cells to induce death. The sequence numbering of the cytotoxic domain (referred to here as the S8 DNase domain) was defined by alignment of the C-terminal amino acid sequence of pyocin S8 with the DNase domain of the well-characterized colicin E9 (30, 36) (Fig. S1A). Thus, to facilitate comparison with other H-N-H endonucleases, we refer here to sequence positions with respect to only the E9 DNase domain.

Since the S8 DNase domain cannot be overexpressed without ImS8 due to elevated toxicity for the host strain, the two proteins were copurified as a heterodimeric complex by nickel affinity chromatography, by employing the C-terminal His₆ tag on the immunity protein (Fig. 3A, lane 1). Subsequently, the S8 DNase domain was separated from ImS8 by treatment with guanidine hydrochloride, followed by another nickel affinity chromatography step to specifically remove ImS8, which remained immobilized in the column (Fig. 3, lanes 2 and 3, respectively). Finally, the S8 DNase domain was refolded after many cycles of dialysis to remove the denaturing agent (guanidine hydrochloride). In most cases, EDTA was added after these two rounds of nickel affinity chromatography and was removed together with guanidine hydrochloride by extensive dialysis procedures (Fig. 3B, +EDTA). In a few cases, no EDTA was added after the guanidine hydrochloride/nickel affinity chromatography procedures (in Fig. 3B, −EDTA).

FIG 3.

Metal-dependent endonuclease activity of the S8 DNase domain. (A) SDS-PAGE of the purified S8 DNase domain and its cognate immunity protein. The S8 DNase-ImS8 complex was initially purified by nickel affinity chromatography (lane 1). The S8 DNase domain (lane 2) was then separated from ImS8 by guanidine hydrochloride, followed by another nickel affinity chromatography step. ImS8 was then eluted by imidazole (lane 3). Finally, the S8 DNase domain (lane 2) and ImS8 (lane 3) were subjected to many cycles of dialysis. (B to D) Plasmid nicking assay to evaluate the endonuclease activity of different protein samples. DNA cleavage was monitored by the conversion of pUC18 plasmid DNA from the supercoiled (s) form to the open-circle (o) or linear (l) DNA forms. (B) In some samples (+EDTA), EDTA was added after these two rounds of nickel affinity chromatography and was afterwards removed by extensive dialysis to allow attempts to recover the endonuclease activity by specific metal supplementation shown in panel C. Open-circle or linear DNA forms of pUC18 plasmid were detected in the absence of this EDTA treatment, indicating metal contamination in buffers and reagents employed in the nicking assay. In other samples, the immunity protein was added (+ImS8) and completely abolished DNA cleavage. (C) Recovery of the endonuclease activity of the EDTA-treated wild-type S8 DNase domain by supplementation with different metal ions at 10 mM concentration. (D) Endonuclease activity of the EDTA-treated Glu100Ala S8 DNase as assayed in panel C.

The endonuclease activity of pyocin S8 was assessed by a plasmid-nicking assay using the supercoiled pUC18 plasmid as the DNA substrate. In our experimental conditions, all of the supercoiled (s) plasmid was converted into the open (o) or linear (l) forms (Fig. 3B, −EDTA). In contrast, this endonuclease activity was strongly inhibited by the pretreatment of the recombinant protein with EDTA, as the band corresponding to the supercoiled band remained intense even 20 min after the beginning of the reaction (Fig. 3B, +EDTA). DNase activity was totally blocked when a chelator was present at all times throughout the experimental procedures, such as at the zero time point of the +EDTA sample (Fig. 3B), indicating that contaminant metals in buffers and other reagents supported the endonuclease activity of the S8 DNase domain. Finally, as expected, addition of the immunity protein completely abolished the appearance of the open and linear forms of the plasmid (Fig. 3B, +ImS8), excluding the possibility that the observed endonuclease activity is coming from a contaminant protein.

Next, the metal specificity of this endonuclease activity was analyzed (Fig. 3C). The addition of Mg²⁺ and Mn²⁺ ions reactivated the endonuclease activity of the EDTA-pretreated S8 DNase domain. In contrast, addition of Zn²⁺ did not support the endonuclease activity. Indeed, the time course of supercoiled DNA disappearance in the Zn²⁺-supplemented sample (Fig. 3C, Zn²⁺) was similar to that of the EDTA-treated S8 DNase domain (Fig. 3B, +EDTA). On the other hand, Ni²⁺ induced a high endonuclease activity, degrading the pUC18 DNA to a much greater extent and precluding the detection of any bands whatsoever. Control experiments in the absence of the pyocin S8 DNase domain and containing the metal ion with supercoiled plasmid pUC18 did not result in any DNA degradation (Fig. S2). Therefore, the DNase activity is very dependent on the metal ion used, with the highest activity associated with Ni²⁺ and weaker activities associated with Mg²⁺ and Mn²⁺. The endonuclease activity of pyocin AP41 displayed similar metal dependencies (12).

The Glu100Ala S8 DNase mutant (corresponding to Glu736Ala considering the full killing subunit sequence) was not capable of converting supercoiled DNA into linear forms in the presence of Mg²⁺ (Fig. 3D), indicating the involvement of the H-N-H motif in this activity. In the presence of Ni²⁺, the mutant exhibited appreciable activity, although considerably lower than that of the wild-type S8 DNase. These results are in contrast with a previous work showing that the substitution of alanine for glutamic acid in colicin ColE9 from Escherichia coli abolishes its endonuclease activity in the presence of nickel (26).

Affinity of S8 DNase-metal complexes.

Since we observed distinct effects of metals on the DNase activity, we decided to investigate metal-protein interactions by isothermal titration calorimetry (ITC). The stoichiometry of Ni²⁺ and Zn²⁺ binding to the wild-type S8 DNase domain was 1:1 (Fig. 4, lower two panels) with a substantial enthalpic contribution (>20 kcal/mol), indicating that it was a highly favorable event. The dissociation constants for both Zn²⁺ and Ni²⁺ were calculated (Table 1) and indicated that S8 DNase binds Zn²⁺ with an affinity approximately 26-fold higher than Ni²⁺. No binding of Mn²⁺ and Mg²⁺ to the S8 DNase domain was detected (data not shown), which is agreement with previous work on the ColE9 DNase domain (37). Possibly, the presence of DNA is required for the S8 DNase domain to bind Mn²⁺ and Mg²⁺ (37).

FIG 4.

Isothermal titration calorimetry (ITC) for metal binding to the S8 DNase domain. Upper two panels, ITC thermogram responses for the titration of metals (1 mM final concentration) into 27 μM concentrations of the indicated proteins in 200 mM NaCl, 20 mM Tris-HCl, pH 7.5, buffer at 25°C. Data from two independent experiments were fitted to a simple noncooperative binding model with n independent binding sites. Lower two panels, heat profile from peak integration of the ITC thermograms of the wild-type (WT) and S8 DNase Glu100A mutant, respectively, with the fits to the model with n independent binding sites shown by black lines. The corresponding thermodynamic parameters are given in Table 1.

TABLE 1.

Thermodynamic parameters for metal ion binding to the S8 DNase domain determined by ITC experiments^a

Expt	Kd (nM)	n	ΔH (kcal/mol)
S8 + Zn2+	26.5 ± 0.2	0.91 ± 0.02	−20.8 ± 0.1
S8 + Ni2+	689 ± 50	0.92 ± 0.02	−22.0 ± 0.6
S8 mutant + Zn2+	508 ± 20	0.96 ± 0.02	−17.1 ± 0.5
S8 mutant + Ni2+	2,170 ± 100	0.92 ± 0.04	−15.0 ± 0.9

^aITC profiles were fitted using a set of sites with the same constant affinity model to obtain values for the equilibrium dissociation constant (K_d), ligand binding sites (n), and enthalpy change (ΔH). Standard errors from duplicate experiments are given.

The metal-protein affinity of the Glu100Ala S8 DNase mutant was also measured (Fig. 4). The mutant was still able to bind both Zn²⁺ and Ni²⁺ metal ions, even though with a reduced affinity in comparison to the wild-type protein. Possibly, the lower affinity of the S8 Glu100Ala mutant for metals is associated with the fact that this protein could not kill bacterial cells (Fig. 2B).

Crystal structure of the S8 DNase-immunity complex.

As described above, the Glu100Ala S8 DNase mutant binds transition metal ions (Fig. 4) and catalyze DNA hydrolysis in the presence of Ni²⁺ (Fig. 3), although with considerably lower efficiency than the wild-type protein. Furthermore, this mutant is completely inactive in vivo (Fig. 2B). To better understand the role of this highly conserved residue in the H-N-H motif in the context of pyocins, the crystal structure of the Glu100Ala S8 DNase domain in complex with ImS8 (here referred to as Glu100Ala S8 DNase-Im) was elucidated at 1.38-Å resolution (Table 2). After many attempts, we unfortunately failed to crystallize the wild-type protein. However, the crystal structure of wild-type pyocin AP41 (12) could be used as a template to identify the structural features caused by the Glu100Ala mutation because these two pyocins share a high degree of amino acid sequence identity (Fig. 1C).

TABLE 2.

Data collection and refinement statistics

Parameter	Value(s) for Glu100Ala S8 DNase-Im (PDB ID 6W0V)a
Data collection statistics
Space group	P212121
Cell dimensions
a, b, c (Å)	42.2, 47.1, 99.9
α, β, γ (°)	90.0, 90.0, 90.0
Resolution (Å)	29.96–1.38 (1.40–1.38)
When I/σ(I) = 2.00	(1.40)
Rpim	0.02 (0.30)
I/σ(I)	30.4 (1.7)
CC1/2	1.00 (0.8)
Completeness (%)	89.6 (32.1)
Multiplicity	9.0 (1.7)
Refinement statistics
Resolution (Å)	29.98–1.38
No. of reflections	35,452
Rwork/Rfree	0.18/0.21
No. of atoms
Protein	1,629
Water	213
Ramachandran plot (%)
Favored	98.5
Allowed	1.5
B factors
Protein	21.3
Ligand/ion
Water	27.6
RMSD
Bond length (Å)	0.010
Bond angle (°)	1.45

^aValues in parentheses are for the highest-resolution shell.

The asymmetric unit of the Glu100Ala S8 DNase structure is composed of one monomer of the cytotoxic domain bound to one monomer of the immunity protein (Fig. 5). The N-terminal methionine and the C-terminal six residues of the DNase domain were not observed in the structure, probably because they are unstructured. The overall structure of Glu100Ala S8 DNase is very similar to that of pyocin AP41 DNase (PDB ID 4UHP) and S2 DNase (PDB ID 4QKO) complexes, with their Cα atoms displaying root mean square deviation (RMSD) values of 0.72 Å and 0.80 Å, respectively (Fig. S3). However, there are notable differences related to the orientation of the amino acids in the active site (see below).

FIG 5.

Crystal structure of the Glu100Ala S8 DNase mutant domain in complex with its immunity protein in a cartoon representation. Glu100Ala S8 DNase and its cognate immunity protein are colored light orange/green and gray, respectively. The active site comprising the H-N-H motif is colored green. The S8 DNase domain adopts the typical mixed α/β DNase fold, while the immunity protein is a four-helix bundle. This illustration of the protein structure was created using PyMOL.

The structure of the S8 DNase domain is composed of a mixed α/β fold, with its 32-amino-acid H-N-H motif located at the extreme C terminus of the enzyme (Fig. 5, region colored green). This motif comprises the active site of the nuclease and adopts a V-shaped architecture, which binds in the minor groove of DNA. ImS8 adopts a four-helix bundle structure and binds to the DNase domain at a region distant from the active site. Although the DNase active site is still exposed in spite of its association with ImS8, somehow this interaction blocks its binding with DNA, probably by steric and electrostatic clashes (12, 30).

We detected an inadvertent mutation (Tyr9His) in the α-helix I of ImS8. This mutation probably does not interfere with the inhibitory properties of ImS8, since α-helix I is far away from the cytotoxic-immunity binding interface. Indeed, in the pUC18 cleavage assay, the recombinant immunity protein with the Tyr9His mutation inhibited the ability of S8 DNase to cleave DNA (data not shown). Furthermore, the interactions between the DNase and immunity proteins are also highly conserved among the complexes (12).

S8 DNase active site.

To further investigate the role of the conserved glutamic acid in the H-N-H motif, we compared the active site of the Glu100Ala S8 DNase mutant with other H-N-H DNase structures. Although the amino acid sequences of the structures analyzed are highly similar, they differ considerably in terms of metal ligation. While there are structures showing no metal (apo form) in the active sites, others are bound to different metals, such as Zn²⁺, Ni²⁺, and Mg²⁺ (12, 28–32, 38, 39) (Fig. 6C).

FIG 6.

Active site of the H-N-H endonucleases. (A) Multiple amino acid sequence alignment, with some residues described in this work highlighted at the bottom as conserved amino acids. The amino acids responsible for the name of the H-N-H motif are highlighted in bold. The numbers at the top refer to the four highly conserved histidine residues. Pyocin S8 is highlighted in red. The amino acids at positions 96 and 100 (predicted to interact with each other by a salt bridge) are indicated by black arrows. (B) Close view of the Glu100Ala S8 DNase active site, shown in cartoon representation with selected side chains shown as sticks. The Glu100Ala mutation is highlighted in red. (C) Superposition of the mutant S8 active site with different H-N-H DNase structures, with highlighting of key amino acids for DNase activity. The active site is shown in cartoon representation with selected side chains shown as sticks. The structures superposed with Glu100Ala S8 DNase were as follows: Mg²⁺-bound S2 DNase-ImS2 structure from reference 12 (PDB ID 4QKO), apo AP41 DNase-ImAP41 structure from reference 12 (PDB ID 4UHP), Ni²⁺-bound AP41 DNase structure from reference 12 (PDB ID 4UHQ), Ni²⁺-bound ColE9 DNase-ImColE9 from reference 30 (PDB ID 1BXI), apo ColE9 DNase-ImColE9 structure from reference 40 (PDB ID 1EMV), and Zn²⁺-bound ColE7 DNase-ImColE7 structure from reference 31 (PDB ID 7CEI). Figures of protein structures were created using PyMOL.

The H-N-H motif contains four histidine residues that directly participate in DNA hydrolysis (Fig. 6A, numbered residues). His131 is not observed in our structure due the absence of electron density in the last six C-terminal residues of the DNase domain (Fig. 6B). His103 is in a very similar conformation to that observed for the corresponding residues in other H-N-H DNase structures (Fig. 6C). In fact, this residue plays an important role in catalysis, acting as a general base that activates a water molecule for nucleophilic attack of the scissile phosphodiester bond (29, 32). The remaining two histidine residues (His102 and His127) showed slight changes in their conformations in comparison to the other structures (Fig. 6C). These histidine residues have been proposed to participate in metal ion coordination and seem to have considerable structural variability depending on the nature of the metal ion present in the structure (29, 32).

The orientation of the His102 and His127 side chains is compatible with metal binding being similar to that observed for the Ni²⁺-bound form of AP41 (12) (Fig. 6C). However, although weak electron density is visible at the expected metal site in difference maps, it was not possible to attribute this to any particular metal or to establish its occupancy. Therefore, the active site of the S8 DNase domain structure described here is probably predominantly apo (Fig. 6B). It is also possible that the weak electron density in the active site could be a polyethylene glycol (PEG) molecule (or other component of the crystallization solution), but it was not possible to model it due to steric hindrance effects with the histidine side chains. Thus, although the Glu100A S8 DNase domain is able to bind transition metal ions, the crystal structure presented here appears to be metal free.

Although Glu100 is not thought to have a direct role in DNA hydrolysis, this residue is highly conserved in the H-N-H motif (Fig. 6A, residue indicated by a black arrow). Indeed, most of the structures show that Glu100 forms a salt bridge with Arg96 (Fig. 6C) (12, 29–32, 40). However, there is an exception in the case of pyocin AP41, which contains a lysine in place of arginine. Importantly, in the AP41 DNase structure, Glu100 seems to not interact with Lys96, and consequently, these residues show high mobility (B factor) in the structure (Fig. S4).

The Glu100-Arg96 salt bridge has been proposed to play an important role in distorting the DNA substrate, approximating the scissile phosphate toward the metal ion (32). The substitution of alanine for Glu100 (Fig. 6B, residue highlighted in red) releases constraints on the movement of Arg96, which is consistent with its higher B factor (Fig. S4). Therefore, the loss of the Glu100-Arg96 salt bridge probably is the cause of pyocin S8 inactivation.

DISCUSSION

Here, we describe a biochemical and structural characterization of a new member of the S-type pyocin group. We show that recombinant pyocin S8 is highly effective against the PAO1 strain, inducing cell death at 10 μg/ml or higher concentrations. Similarly, other pyocins induced cell death at concentrations within the same range (4, 18, 41, 42). This result prompted us to investigate the molecular basis by which this protein shows potent bactericidal activity.

Pyocin S8 is an endonuclease with activity based on the highly conserved H-N-H motif that requires a divalent metal ion as a cofactor. There are some variations regarding the metal ion identity used by H-N-H endonucleases. The most commonly reported metal ion is Mg²⁺, although other divalent cations can also support the H-N-H DNase activity, including Mn²⁺, Zn²⁺, and Ni²⁺ (12, 26, 28, 29, 37, 43, 44). The metal requirements for ColE9 and ColE7 appear to be distinct. While both enzymes are activated by Ni²⁺ and Mg²⁺, Zn²⁺ activates only ColE7, having inhibitory effects for ColE9 (28, 37, 43, 44). In our case, pyocin S8 exhibited DNase activity when supplemented with Mg²⁺, Mn²⁺, and Ni²⁺, displaying no detectable activity in the presence of Zn²⁺ (Fig. 3C). Accordingly, pyocin AP41 DNase displays higher endonuclease activity in the presence of Ni²⁺ or Mg²⁺ ions, showing no detectable activity in the presence of Zn²⁺ (12). Given that a variety of metal ions can support catalysis by H-N-H DNases, the identity of the biologically active metal ion remains a challenge.

Possibly, metals bind H-N-H endonucleases in different ways, activating or inhibiting the histidine residues that are directly involved in the hydrolysis of phosphodiester bonds. The crystal structures of ColE9 in complex with Zn²⁺-DNA and Mg²⁺-DNA provided some structural insights into the distinct effects of metals on the endonuclease activity of this bacteriocin (32). In the ColE9 Zn²⁺-DNA structure, Zn²⁺ is bound to atoms of the three active site histidine residues (His102, His127, and His131), while in the ColE9 Mg²⁺-DNA structure, Mg²⁺ is bound to only two histidine residues (His102 and His127). Therefore, Zn²⁺ binds to the H-N-H motif with high affinity, using His131 as an additional metal-binding site and consequently trapping the enzyme in a catalytically inactive state. Thus, “active” metal ions are those that bind more weakly to the H-N-H motif, leaving His131 in a disengaged state, allowing a water molecule to complete the catalytic cycle (32). In agreement with this model, our ITC analyses showed that Zn²⁺ binds to S8 with higher affinity than Ni²⁺. However, the situation is complicated by the fact that the AP41 Ni²⁺-bound complex shows the metal bound to all three histidine side chains. Therefore, the crystal structures of pyocin S8 in complex with metals and with DNA are required to understand the mechanisms by which the endonuclease activity of this bacteriocin is modulated by metals.

Besides metals, this work provided new insights into the function of the H-N-H motif. Our structure of the S8 DNase-Im8 complex is only the third cognate pyocin DNase-Im protein complex solved, and this is the first displaying a mutation of the conserved Glu (Glu100) in the H-N-H motif. As this conserved Glu (Glu100) is essential to killing activity (Fig. 2), comparison of our structure with the others available provided valuable information. The superposition of H-N-H DNase-Im structures showed that the Glu100-Arg96 salt bridge is highly conserved (Fig. 6C). Noteworthy, Arg96 is positioned in a distinct rotamer in our DNase (Glu100Ala) structure, suggesting that Arg96 presents a high degree of freedom to assume distinct conformations.

The H-N-H motif constitutes the DNase active site of pyocins and closely related colicins. The current model of DNA hydrolysis by H-N-H endonucleases suggests that the double helix of DNA needs to undergo some distortion to approximate the scissile phosphate close to the metal center of the H-N-H motif (26, 29, 32). This is achieved by positioning the H-N-H motif into the minor groove of DNA, mainly due to the insertion of the Arg96-Glu100 salt bridge into the groove itself. As a result of these DNA-enzyme interactions, the scissile phosphate is positioned toward the metal ion embedded within the H-N-H motif (32).

In the Glu100Ala mutant, the lack of the salt bridge with Glu100 releases constraints on the Arg96 side chain movements that present a weak electron density. Moreover, Arg96 moved away from its canonical orientation found in the other structures. Possibly, this mutation influences the ability of the V-shaped H-N-H motif to interact with DNA and to catalyze DNA hydrolysis. Most likely, Glu100 orientates the guanidinium group of Arg96 by accepting hydrogen bonds, involving N_εH and one of the NH₂ groups. This leaves the remaining hydrogen atoms free to interact with hydrogen bond acceptors from both the DNA backbone and one of the bases (32). In the absence of such interactions in the Glu100Ala mutant, the distortion of DNA probably does not occur, which could explain the loss of killing activity. The elucidation of the crystal structure of pyocin S8 in complex with its DNA substrate is required to test this hypothesis.

As described before, the H-N-H motif is composed of four conserved histidine residues that play important roles in DNA hydrolysis. The role of His131 is still elusive, and it appears to be involved in the activation of a water molecule to protonate the 3′ oxygen-leaving group of a phosphodiester bond (29). In some cases, this residue can also coordinate metals, as showed in the Zn²⁺-bound DNase structures (28, 31, 32). His102 and His127 residues appear always engaged in metal ion coordination, regardless of which metal is bound to the enzyme. Finally, His103 activates a water molecule for the nucleophilic attack of the scissile phosphate bond (29, 32). With the exception of His131, which is disordered in our structure, the other three histidine residues showed orientations similar to those of equivalent residues in the other DNase-Im structures. Thus, although the putative metal binding sites (represented by the active site histidine residues) were only moderately affected in our structure, no metal ion was found in the active site. Joshi and coworkers (12) have reported the crystal structures for the wild-type pyocin AP41 DNase-Im and pyocin S2 DNase-ImS complexes. The former is in the apo form, while the latter, curiously, has been modeled to include a Mg²⁺ ion at the active site despite the absence of DNA in this structure. The diversity of the active site configurations observed in the current structures of the H-N-H endonucleases suggest that these enzymes have a dynamic catalytic center, able to adapt to different crystallization conditions.

Bacteriocins such as pyocins and colicins play important roles in shaping bacterial communities and also affect host-pathogen interactions (15, 45–47). The narrow specificity of these toxins toward closely related strains gives rise to an attractive possibility of using them in the treatment of bacterial infections with minimal collateral damage to the human microbiota. However, the great potential of these molecules might be reduced by the occurrence of the pyocin resistance phenotype. Therefore, understanding the molecular mechanisms by which pyocins kill target cells is an essential step in attacking this problem and minimizing the appearance of resistance to these proteinaceous toxins.

MATERIALS AND METHODS

Secondary structure prediction and sequence alignments.

The secondary structure elements of pyocin S8 were predicted by using the JPred4 server available in the Jalview program (48). Pyocin protein sequence alignments were performed using the Jalview program.

Plasmid construction.

The genes encoding either full-size pyocin S8 or its immunity protein were PCR amplified from ET02 genomic DNA using primers 5′-GGAATTCCATATGAGCGACGTTTTTGACCTTGGA-3′ and 5′-ATGGATCCTTGCCAGCCTTGAAGCCAGGGAG-3′. The PCR product was digested with NdeI and BamHI and ligated into the corresponding sites of the E. coli expression vector pET29b to give pET29b-PyoS8-ImS8, which encodes pyocin S8 with a C-terminal His₆ tag fused to the immunity protein. To create the substitution of alanine for the conserved glutamic acid 736 (glutamic acid at position 100 when considering only the cytotoxic domain) in the H-N-H motif, pET29b-PyoS8-ImS8 was used as a template for the site-directed mutagenesis (QuikChange XL site-directed mutagenesis kit; Agilent Technologies) to yield pET29b-PyoS8Glu736Ala-ImS8.

The coding sequence for the pyocin S8 cytotoxic domain starting with the aspartic acid residue at position 6 (residues 642 to 772 in terms of the full-length pyocin S8 killing subunit) and its cognate immunity protein were PCR amplified from ET02 genomic DNA by using primers 5′-ATTATATCATATGGATGAGCCGGGTGTTGCTACC-3′ and 5′-ATGGATCCTTGCCAGCCTTGAAGCCAGGGAG-3′. The PCR product was digested with NdeI and BamHI and ligated into the corresponding sites of pET29b to give pET29b-PyoS8DNase-ImS8. The coding sequence for the pyocin S8 cytotoxic domain containing the mutation of glutamic acid at position 100 to alanine and the immunity protein were PCR amplified from pET29b-PyoS8Glu736Ala-ImS8 by using primers 5′-ATTATATCATATGGATGAGCCGGGTGTTGCTACC-3′ and 5′-AAGGTACCGCCAGCCTTGAAGCCAGGGAG-3′. The PCR product was digested with NdeI and KpnI and ligated into the corresponding sites of pET29b to give pET29b-PyoS8DNaseE100A-ImS8.

Pyocin S8 expression and purification.

All constructs used for protein expression were routinely checked by DNA sequencing in order to verify the presence of mutations. The expression vector encoding full-size pyocin S8 and its immunity protein was transformed into E. coli BL21(DE3)/pLysS competent cells. Overnight cultures were diluted in 0.5 liter of LB broth to an optical density at 600 nm (OD₆₀₀) of 0.2, and cells were grown at 37°C in a shaking incubator to an OD₆₀₀ of 0.6 to 0.8. The culture was then shifted from 37°C to 30°C, and protein expression was induced by the addition of 0.5 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG) for a period of 6 h. Cells were then harvested and resuspended in 20 ml of start buffer (20 mM sodium phosphate, 500 mM NaCl, pH 8.0) containing 0.05 mg/ml lysozyme. After incubation at 4°C for 30 min, 500 μl of a 10× Complete EDTA-free protease inhibitor cocktail (Sigma-Aldrich) solution was added to the cell suspension. Cells were lysed by sonication, and the cell extract was kept on ice for 20 min during a 1% streptomycin sulfate treatment. Cell debris was removed by centrifugation at 15,000 × g at 4°C for 40 min, and the supernatant was further clarified by filtration using a 0.45-μm pore membrane. The cell-free lysate was applied to a 5-ml HiTrap chelating HP column (GE Healthcare) equilibrated in start buffer. After loading, the column was connected into an AKTA fast protein liquid chromatography (FPLC) system (Amersham Biosciences, GE Healthcare) and washed with 50 ml start buffer containing 50 mM imidazole to remove unbound protein. The bound protein was then eluted by an imidazole gradient (20 to 500 mM). The purified proteins were then dialyzed against TN₂₀₀ buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl) using a HiTrap desalting column (GE Healthcare), and the protein was further purified by gel filtration chromatography on a HiLoad 16/600 Superdex 75 or HiLoad 16/600 Superdex 200 column (GE Healthcare) equilibrated in the same buffer. The purity of the proteins was assessed by SDS-PAGE, and the fractions were pooled to purity and concentrated using an Amicon Ultra concentrator (Merck Millipore) with a 10-kDa cutoff. The proteins were divided into small aliquots and stored at −80°C. Prior to use, protein concentrations were determined by absorbance at 280 nm using the molar extinction coefficient (ε₂₈₀ = 77,240 M⁻¹cm⁻¹) obtained using the ProtParam tool.

Purification of DNase domain from ImS8.

Initially, the pyocin S8 DNase-ImS8 complex was overexpressed and purified essentially as described above. Then, the pyocin S8 DNase domain was isolated from the pyocin S8 DNase-ImS8 complex using 6 M guanidine hydrochloride as described previously (12). Briefly, the pyocin S8 DNase-ImS8 complex at 5.5 mg/ml was diluted 10-fold in a denaturing buffer containing 500 mM NaCl, 20 mM Tris-HCl (pH 8.0), and 6 M guanidine hydrochloride. The sample was then incubated for 1 h at room temperature with occasional agitation. The pyocin S8 DNase domain was separated from its cognate immunity protein by loading the sample onto a 5-ml HiTrap chelating HP column (GE Healthcare) equilibrated with denaturing buffer. In this step, the immunity protein was immobilized in nickel affinity resin due to a C-terminal His₆ tag. The DNase domain was then concentrated and incubated with 20 mM EDTA for 1 h at 25°C to remove adventitious metal ions. The chelating and denaturing agents were removed by extensive dialysis against 50 mM Tris (pH 7.5) buffer. Finally, after concentration and quantification, protein was divided into small aliquots and stored at −80°C. The pyocin S8 DNase E100A domain was purified in the same way described above.

Pyocin sensitivity assays.

The activity of recombinant pyocin S8 was assessed by the overlay spot plate method as described previously (12) with a slight modification. Four hundred microliters of a sensitive strain culture at 0.5 McFarland turbidity (1.5 × 10⁸ CFU/ml of bacteria) was added to 6 ml of 0.7% soft agar and poured onto Mueller-Hinton or G medium agar plates. Purified pyocin S8 at an initial concentration of 5 mg/ml was 8-fold serially diluted, and 3 μl of each dilution was spotted onto the plates and incubated for 16 to 18 h at 37°C.

Alternatively, bacterial cultures at an OD₆₀₀ of 0.1 were incubated with different concentrations of pyocin S8 and grown at 37°C with continuous shaking in a 96-well plate. The effect of pyocin S8 on growth was monitored by measuring the turbidity of the cultures every 10 min for 3 h using the automated Synergy H1 absorbance microplate reader (BioTek).

Endonuclease activity assay.

The endonuclease activity was evaluated by a plasmid nicking assay as described previously (12) with some modifications. Supercoiled pUC18 (1 μg) was used as the substrate for the plasmid nicking assay. Assays were performed at 25°C in a buffer containing 50 mM Tris-HCl (pH 7.5) in a final volume of 50 μl. Reactions were started by the addition of the pyocin S8 DNase domain to a final concentration of 0.4 μM. Most samples were pretreated with EDTA after removal of the immunity protein and after two rounds of affinity chromatography (see “Purification of DNase domain from ImS8”). Importantly, EDTA was removed afterwards by extensive dialysis to allow studies of endonuclease activity recovery by metal supplementation. In a few cases, however, no EDTA pretreatment was performed (Fig. 3B, −EDTA samples). Metal dependence was tested by supplementing these pyocin S8 DNase domain preparations with different metal ions at final concentrations of 10 mM. At different time points, aliquots of 10 μl were removed and the reaction was stopped by the addition of an equal volume of gel loading dye, purple (6×) (New England Biolabs), containing 60 mM EDTA. The plasmid nicking results were analyzed on 1% agarose gels.

ITC.

The isothermal titration calorimetry (ITC) experiments were carried out at 25°C in 20 mM Tris-HCl, 200 mM NaCl, pH 7.5, buffer using a MicroCal ITC200 (GE Healthcare). The protein solutions (27 μM) were gently loaded into the sample cell consisting of a 200-μl working volume. The titration involved 0.3- to 0.5-μl injections of ligands present in a syringe filled with 39 μl of metal ion solutions diluted in ultrapure water to a final concentration of 1 mM. The measurements were performed by 77 injections of 0.5 μl or 0.3 μl for Ni²⁺ and Zn²⁺, respectively, separated by a delay of 120 s between each injection. Control dilution experiments involving injections of ligands into buffer were performed, and results were subtracted from the integrated data before curve fitting. Data were fitted to a model with n sites with the same binding affinities by using MicroCal software (Oring package) to determine the dissociation constant, enthalpy of binding (ΔH), and stoichiometry. These parameters are reported as mean values obtained from two independent experiments performed on distinct protein samples and the corresponding sample standard deviations.

Crystallization and data collection.

Crystals were grown at 18°C using the hanging-drop vapor diffusion method. The Glu100Ala S8 DNase-Tyr9His ImS8 mutant crystals were grown by adding 10 mg/ml of 6×His-protein (diluted in 5 mM Tris-HCl, pH 7.4) to an equal volume of the reservoir solution containing 0.2 M sodium acetate, 30% (wt/vol) PEG 4000, 0.1 M Tris-HCl, pH 8.0. The crystals were all flash frozen in liquid N₂ and collected at beamline MX2 at the Brazilian Synchrotron Light Laboratory (LNLS).

Structure determination and refinement.

All data were indexed and integrated with XDS (49) and scaled using AIMLESS in the CCP4 program suite (50). Initial phases were obtained by molecular replacement with PHASER (51) using P. aeruginosa pyocin AP41 (PDB ID 4UHP) as the search model. One copy of a monomer of the 1cytotoxic:1immunity complex was found in the asymmetric unit. Models were built in COOT (52) and refined using REFMAC5 (53). Details of the refinement statistics are presented in Table 2.

Data availability.

Data for Glu100Ala S8 DNase-Im have been deposited in the Protein Data Bank (PDB) under accession number 6W0V.