Nisin-Triggered Activity of Lys44, the Secreted Endolysin from Oenococcus oeni Phage fOg44

Abstract

The intrinsic resistance of Oenococcus oeni cells to the secreted endolysin from oenophage fOg44 (Lys44) was investigated. Experiments with several antimicrobials support the hypothesis that the full activity of Lys44 requires sudden ion-nonspecific dissipation of the proton motive force, an event undertaken by the fOg44 holin in the phage infection context.

Double-stranded DNA bacteriophages use an endolysin-holin strategy to achieve lysis of their hosts (17). Holins are small hydrophobic proteins with one to three transmembrane domains that regulate lysis timing by means of programmed cytoplasmic membrane disruption (17). Endolysins are peptidoglycan-degrading enzymes that are usually accumulated in the cytosol until access to the cell wall substrate is provided by the holin membrane lesion (10, 15). It has recently come to light that some endolysins are endowed with secretion signals, engaging the Sec system of the host for their translocation into the extracytoplasmic environment. One example is the endolysin from phage fOg44 (11), which possess a typical cleavable signal peptide. fOg44 is a temperate bacteriophage of the gram-positive lactic acid bacterium Oenococcus oeni. Another example is the Escherichia coli phage P1 endolysin, which exhibits an atypical signal sequence designated SAR that mediates the translocation of the enzyme without cleavage (16). Clearly, in both cases the holin is not necessary for endolysin export, and the holin's function therefore remains elusive. In the case of Lys44, the fOg44 endolysin, we could envisage that the endolysin activity would be restrained in the cell wall environment until the holin lesion acted as a signal for its activation through the dissipation of the membrane proton motive force (PMF). Alternatively, the activity of Lys44 could be counteracted by repair synthesis of the cell wall. Such synthetic activity would come to a halt upon holin-mediated cell metabolic arrest, and only then would lysis prevail. To test these models, we have investigated here the activity of Lys44 added exogenously to cell suspensions or produced inside cells with its signal peptide. The Lys44 activity in the presence of cell wall synthesis inhibitors was compared to the activity in the presence of antimicrobials which target the membrane.

Purification of Lys44 derivatives.

In order to study the enzymatic activity of the fOg44 endolysin, a signal peptide-less derivative of Lys44 endowed with a C-terminal hexahistidine tail (Lys44*) was overproduced in E. coli and purified from crude extracts. The construct was generated as follows: a version of the Lys44 gene devoid of the signal peptide coding region was amplified by PCR, and the corresponding 1,218-bp DNA fragment was purified, digested with NcoI (Takara Bio) and Cfr9I (Fermentas), and ligated into the similarly digested overexpression vector pIVEX2.3d (Roche Applied Science), yielding plasmid pIVEXlys44*. In this recombinant vector, the first 27 codons of the Lys44 gene were replaced by an ATG codon and the 3′ end of the gene was fused to a hexahistidine coding sequence. The plasmid pIVEXlys44E₁₂₄A*, harboring the Lys44(E₁₂₄A)* mutant gene in which the codon corresponding to essential glutamate residue E₁₂₄ of the enzyme catalytic site (4) was replaced by an alanine codon, was constructed by following a QuikChange XL site-directed mutagenesis kit (Stratagene) protocol. Primer sequences are presented in Table 1 posted at http://dbv.fc.ul.pt/documentos+temporarios/Table1-JournalOfBacteriology.pdf. Overnight cultures of E. coli BL21 carrying pGP1-2 (10) and pIVEX2.3d were reinoculated at 1:100 into 10 ml of Luria-Bertani broth supplemented with the appropriate antibiotics, and the suspensions were incubated with shaking at 28°C to an A₆₀₀ of 0.3 and subjected to thermal shock at 42°C for 4 h in order to induce protein overexpression. Ten-milliliter aliquots of thermoinduced E. coli cultures were pelleted, and Lys44* and Lys44(E₁₂₄A)* proteins were purified under denaturing conditions by His tag affinity chromatography according to the instructions of the manufacturer of the nickel-nitrilotriacetic acid spin column (QIAGEN). The column eluate was subjected to dialysis in renaturation buffer (50 mM Na₂HPO₄/NaH₂PO₄ buffer, 250 mM NaCl, 1 mM CaCl₂, 1 mM MgCl₂, and 0.1% Triton X-100, pH 5) at room temperature. After dialysis, insoluble material was removed by centrifugation (10,000 × g for 30 min at 4°C). The protein concentration was determined by the method of Bradford (Bio-Rad Laboratories) by using bovine serum albumin as a standard. Protein stock solutions with protein concentrations in the range of 100 to 600 μg/ml were routinely obtained and stored at 4°C.

Lysis of oenococcal cells from without: effect of nisin.

The ability of Lys44* to induce cell lysis when added from without was tested at 30°C using O. oeni ML34-C10 cells as the substrate. Lysis assays were performed in 1-ml cuvettes, and the absorbance at 600 nm was periodically measured. To determine the kinetic profile of Lys44*-mediated lysis, early-exponential-growth-phase cultures of O. oeni (A₆₀₀, 0.3 to 0.4) growing in Man-Rogosa-Sharpe tomato juice (MTJ) medium at pH 5.5 (14) were centrifuged at 7,000 × g for 5 min and the pellets were suspended in fresh MTJ broth adjusted to pH 5 to a final A₆₀₀ of 0.6. Cells of O. oeni were initially challenged with different concentrations of Lys44*, up to 10 μg/ml (Fig. 1 and data not shown), and lysis was never observed within the tested incubation period (2 to 3 h). This resistance of growing cells to lysis was expected from our model, in which the dissipation of the membrane PMF would be a prerequisite for Lys44 activity. Accordingly, we decided to test whether membrane-depolarizing agents could trigger Lys44*-mediated lysis of oenococcal cells. The lantibiotic nisin (Sigma-Aldrich) proved to be a most effective agent in sensitizing cells to the action of Lys44*. As seen in Fig. 1, the addition of 500 ng of nisin/ml (ca. 150 nM; MIC) to cells preincubated with 10 μg of Lys44*/ml resulted in instantaneous lysis of the culture. Likewise, a short preincubation of cells with nisin followed by the addition of Lys44* to different final concentrations resulted in dose-dependent lysis rates, and as little as 25 ng of Lys44*/ml was sufficient to yield significant culture lysis (Fig. 1). When added alone, nisin did not induce the lysis of oenococcal cells during the time course of the experiment (Fig. 1).

FIG. 1.

Kinetics of nisin-dependent Lys44*-mediated lysis of O. oeni. Early-log-phase cultures of O. oeni were concentrated to an A₆₀₀ of 0.6 in fresh MTJ broth, pH 5, at 30°C and the absorbance of the suspensions over time was monitored. At time point zero, purified Lys44* was not added (×) or was added at 25 ng/ml (⧫), 50 ng/ml (▴), 100 ng/ml (▪), or 200 ng/ml (•) to cells previously incubated for 15 min with 500 ng of nisin/ml. In reciprocal experiments, cells were incubated in the presence of Lys44* at 200 ng/ml (○) or 10 μg/ml (□) and 500 ng of nisin/ml was added 90 min after endolysin addition. Results from a control assay of cells without any additions are also shown (+). Plotted data are the means of results from three independent experiments, and standard deviations are shown as error bars. For purposes of clarity, lysis curves are truncated to avoid superposition.

Lys44 is composed of a catalytic N-terminal domain with homology to N-acetyl muramidases of glycosyl hydrolase family 25 (4) and a C-terminal region comprising two LysM modules mediating its binding to peptidoglycan (8, 13). To ascertain if nisin was triggering lysis by favoring the attachment of the enzyme to cell walls, binding experiments were performed in its presence or absence. For this purpose, a Lys44 derivative devoid of lytic activity (Lys44E124*) was used (4) so that binding could be measured in the absence of lysis. For the production of radioactive Lys44(E₁₂₄A)*, E. coli cultures were incubated in minimal medium M9 instead of Luria-Bertani broth (9) and, 40 min after thermal induction of expression, a 10-μCi/ml concentration of a ¹⁴C-radiolabeled amino acid mixture (Amersham Biosciences) was added in order to produce radioactive proteins. Purification of the labeled protein by affinity chromatography was as described above. O. oeni cells prepared as for the lysis assay were incubated with 400 ng of radiolabeled Lys44(E₁₂₄A)*/ml in the presence or absence of 500 ng of nisin/ml. Every minute after protein addition, 1.1-ml aliquots of the cell suspensions were centrifuged at 14,000 × g for 5 min and the radioactivity of the supernatant (a 1-ml sample mixed with 8 ml of Ready Safe cocktail from Beckman Coulter) was quantified using fresh MTJ broth as a blank. As shown in Fig. 2, the kinetics of binding of Lys44(E₁₂₄A)* to nisin-treated oenococcal cells and untreated cells were very similar, indicating that nisin does not operate by significantly increasing the binding affinity of the endolysin for its substrate.

FIG. 2.

Kinetics of binding of Lys44(E₁₂₄A)* to O. oeni cells. Early-log-phase cultures of O. oeni were concentrated to an A₆₀₀ of 0.6 in fresh MTJ broth at pH 5 and incubated with (•) or without (○) 500 ng of nisin/ml for 15 min before the addition of radiolabeled Lys44(E₁₂₄A)* (in an amount corresponding to ca. 600 cpm; a 400-ng/ml final concentration). At the indicated times, cells were collected by centrifugation and the amount of radiolabeled Lys44(E₁₂₄A)* remaining in the supernatant was determined. A cell-free control of radiolabeled protein added to MTJ was evaluated (□). The error bars represent the standard deviations of means of results from three independent experiments.

Dissecting nisin-induced Lys44-mediated lysis.

The high potency and microbial specificity of nisin derive from its affinity for murein precursor lipid II and the formation of a 2.5-nm pore responsible for cytoplasmic membrane PMF dissipation. The pore structure is not selective towards ions or other small metabolites, such as amino acids and ATP, thus allowing their free flow across the membrane (reference 2 and references therein). In order to clarify which component of the dual targeting mechanism of nisin was triggering Lys44-mediated lysis, we performed assays in the presence of other drugs selectively affecting cell wall synthesis or membrane permeabilization. From several drugs assayed, six (all purchased from Sigma-Aldrich) were retained for further testing given their inhibitory activity on exponentially growing Oenococcus cultures: enduracidin, which binds to lipid II but does not form membrane pores (MIC = 30 μg/ml, or 12.55 μM); ampicillin, acting as a competitive inhibitor of transpeptidase (MIC = 500 μg/ml, or 1.35 mM); bacitracin, which interferes with the dephosphorylation of C₅₅-isoprenyl pyrophosphate (MIC = 250 U/ml, or 2.5 mM); the ionophores nigericin (MIC = 10 μM) and 2,4-dinitrophenol (DNP; MIC = 5 mM), which dissipate the proton gradient across the cytoplasmic membrane without pore formation; and gramicidin D, which forms ion-selective (H⁺ > K⁺ > Na⁺) 0.25-nm-wide transient pores throughout the cytoplasmic membrane (MIC = 10 μg/ml, or 5.32 μM). The addition of any of these drugs to cells incubated with Lys44* did not elicit an immediate lysis response as observed for nisin. Therefore, cells were exposed to the antimicrobials for the equivalent of one generation (ca. 6 h for O. oeni under the reported growth conditions) and then challenged with Lys44*. Exposure to the antibiotics led to growth arrest, as determined by absorbance measurements, but in no case was antibiotic-mediated lysis observed. Under these conditions, preincubation with the cell wall synthesis inhibitor bacitracin, ampicillin, or enduracidin did not render cells sensitive to Lys44* (data not shown). In contrast, Lys44-mediated lysis was observed with all tested PMF-dissipating agents (Fig. 3), although to different extents: nigericin induced complete lysis but at a lower rate than nisin, DNP incubation resulted in relatively fast but partial lysis of the culture, and gramicidin D led to a much slower progressive decrease in culture absorbance (Fig. 3). Although currently we have no experimental explanation for these differences in lysis profiles, the ensemble of the data clearly points to a role of the energized membrane in regulating the lytic activity of Lys44*. Enduracidin was also used in assays in combination with the membrane depolarizers in order to test a synergistic effect between lipid II binding and PMF dissipation, but nisin-associated lysis rates were not successfully duplicated (data not shown).

FIG. 3.

Lys44*-mediated lysis of O. oeni is triggered only by antibiotics affecting the PMF. Early-log-phase cultures of O. oeni ML34-C10 (A₆₀₀, 0.3 to 0.35) were incubated for the equivalent of one generation (ca. 6 h) in the presence of a 500-ng/ml concentration of nisin (• and ○), 10 μM nigericin (▪ and □), 5 mM DNP (⧫ and ⋄), or a 20-μg/ml concentration of gramicidin D (▴ and ▵) or without any antibiotics (× and +). At time point zero, 200 ng of Lys44*/ml was added (closed symbols and ×) or not (open symbols and +), and the absorbance of the cultures over time was monitored. Plotted data are the means of results from three independent experiments. For visualization purposes, standard deviations are not shown. Standard errors varied from 0.01 to 0.07.

Cationic peptides such as nisin have been proposed to play a role in autolysis in staphylococci by means of competition with an amidase autolysin in binding to negatively charged cell wall teichoic and teichuronic acids. In this model, anionic polymers are inhibitors of autolysis (1). To test whether nisin could be acting through its cationic nature, we have tested the activity of Lys44* in cells treated with enduracidin to block lipid II, followed by nisin addition. Under these circumstances, no decrease in optical density was observed (data not shown), indicating that pore formation (coupled to lipid II binding) and not the cationic nature of nisin is the decisive factor for triggering Lys44* activity. The use of 15- to 100-μg/ml concentrations of streptomycin, a cationic molecule also capable of inducing staphylococcus autolysis (1), did not trigger Lys44-mediated lysis.

Recently, nisin has been described as also inducing morphological changes derived from septum deregulation in Lactococcus lactis, Bacillus subtilis, and other members of the Bacillus genus (5, 6). These nisin-induced cellular responses may contribute to the full sensitization of cells to Lys44, but such a contribution is unlikely because these phenomena were generally observed at higher nisin concentrations, on the order of several micrograms per milliliter.

Lysis of lactococcal cells from within.

Besides its effect when added from without, the activity of the endolysin when produced with its signal peptide from within was studied. Due to the lack of genetic tools for performing such a study with Oenococcus, a previously described L. lactis-based gene expression system was employed. In this case, the full-length Lys44 gene in pCSJ28 is under the control of the chloride-inducible Pgad promoter, and detectable amounts of the secreted enzyme are apparent in cell extracts 2 to 4 h after induction with 0.5 M NaCl (reference 12 and unpublished results). As seen in Fig. 4, the expression of Lys44 in this system did not result in lysis, at least during the test period of 5 h postinduction. Nevertheless, the addition of nisin at 4 h postinduction did result in an immediate decrease in culture absorbance, indicating that nisin is able to trigger Lys44-mediated lysis from within in L. lactis. The other PMF-dissipating drugs tested in this study failed to elicit a similar response, probably due to the requirement for prolonged incubation referred to above (Fig. 4). Control assays performed under the same rationale described above in order to elucidate lysis triggering by nisin were duplicated in this system with identical results (data not shown). This finding indicates that the same regulatory mechanism appears to operate on Lys44-mediated lysis whether Lys44 is produced from within or is added from without.

FIG. 4.

Lys44-mediated lysis of L. lactis is triggered by nisin only. Early-log-phase cultures (A₆₀₀, 0.3 to 0.4) of L. lactis MG1363 carrying pCSJ28 (closed symbols and ×) or the control plasmid pGKV259 (open symbols and +) were supplemented with 0.5 M NaCl and incubated for 4 h. At this time (time zero), cultures were supplemented with a 500-ng/ml concentration of nisin (• and ○), 5 μM nigericin (▪ and □), 5 mM DNP (⧫ and ⋄), or an 80-ng/ml concentration of gramicidin D (▴ and ▵) or incubated without any antibiotic additions (× and +). Plotted data are the means of results from three independent experiments. For purposes of clarity, only error bars representing the standard deviations for the strain harboring pCSJ28 in the presence of nisin are shown. Standard errors for the remainder of the assays oscillated between 0.02 and 0.1.

Effect of chloroform and UV treatment on the sensitization of O. oeni cells to Lys44*.

The results of the experiments described above suggested that the dissipation of the electrochemical gradient(s) across the membrane is required to render cells sensitive to Lys44. To further test this hypothesis, lysis induction by chloroform, a commonly used cell permeabilization reagent, was also attempted. The addition of chloroform (1% final concentration) to high-cell-density suspensions such as those of Lys44-expressing Lactococcus cells or concentrated Oenococcus cells to which Lys44* had been exogenously added produced only a weak lysis response (data not shown). However, significant and fast chloroform-induced lysis in the presence of Lys44* could be reproducibly obtained using oenococcal cells incubated to early log phase (A₆₀₀ of 0.3 to 0.35) without further concentration (Fig. 5). As a control, UV-killed cells of O. oeni were also assayed for sensitivity to Lys44*. In this case, a 10-ml sample of cells grown in MTJ was first resuspended in a less complex medium, FT80 (3), poured into a 9-cm-diameter petri dish, and exposed under a 254-nm UV lamp (Camag) for 60 min (ca. 3.6 mJ/cm²). These cells were again resuspended in MTJ medium and used for cell viability determination and lysis experiments. A reduction of 95.8% ± 3.7% (mean ± standard deviation of results from three experiments) in viable CFU counts was achieved following this procedure. This value is probably an underestimation, as O. oeni ML34-C10 tends to grow in rather long chains and a single surviving cell in a chain may be sufficient to produce a colony. As shown in Fig. 5, UV-killed cells were not lysed by exogenously added endolysin. Overall, these results further support that Lys44 activity is not the result of cell metabolic arrest (leading to loss of viability) and requires permeabilization of the cytoplasmic membrane.

FIG. 5.

Lys44-mediated lysis is dependent on membrane permeabilization. Early-log-phase cultures of O. oeni ML34-C10 (A₆₀₀, 0.3 to 0.35) were incubated in the presence (closed symbols) or absence (open symbols) of 100 ng of Lys44*/ml, and the absorbance of the suspensions over time was monitored. At time point zero, Lys44* was added (▴) or not (▵) to UV-treated cells, whereas Lys44* was added at the 15-min time point to the remainder of the cell suspensions. At time point zero, a 500-ng/ml concentration of nisin (• and ○) or 1% chloroform (▪ and □) was added to nonirradiated cultures. Control assays of nonirradiated cells without further additions were also performed (⧫ and ⋄). Plotted data are the means of results from three independent experiments, and standard deviations are shown as error bars.

Concluding remarks.

The data presented herein indicate that bacterial membrane PMF regulates the lytic activity of the secreted endolysin Lys44 from O. oeni phage fOg44. To our knowledge, this is the first time that the activity of a soluble endolysin has been shown to be hindered by the cytoplasmic membrane electrochemical gradient. However, the high cell lysis rate observed in the presence of nisin (or chloroform) and Lys44 could not be duplicated with other PMF-dissipating agents, so we conclude that cytoplasmic membrane voltage dissipation is necessary but not sufficient for the full sensitization of cells to Lys44.

We propose that nisin mimics the holin disruption of the cytoplasmic membrane electrical and chemical gradients observed during phage infection, which triggers the event(s) that leads to the onset of Lys44 activity in the cell wall. Unlike nisin, chloroform, or holins, the PMF dissipators used in this study were essentially selective towards the proton gradient of PMF and probably did not induce the rapid equilibration of the remaining chemical gradients (electrogenic or otherwise) across the cytoplasmic membrane. This effect may account for their inability to induce endolysin-mediated lysis as effectively as nisin (or chloroform). The work here presented does not elucidate how the cytoplasmic membrane electrochemical composition is transduced as a negative regulator of Lys44. Membrane-bound lipoteichoic acids are good candidates for the role of response regulators because they are abundant cell wall polyanions capable of conformational changes depending on their ionization state and are known to regulate bacterial autolysins (reference 7 and references therein).

It will be of interest to investigate the activity of other endolysins as a function of the PMF since a putative role of holins in triggering endolysin-mediated lysis may have been obscured in those cases in which the holin plays a decisive role in endolysin export.