Catechin-Mediated Restructuring of a Bacterial Toxin Inhibits Activity

Abstract

Background:

Catechins, polyphenols derived from tea leaves, have been shown to have antibacterial properties, through direct killing of bacteria as well as through inhibition of bacterial toxin activity. In particular, certain catechins have been shown to have bactericidal effects on the oral bacterium, Aggregatibacter actinomycetemcomitans, as well as the ability to inhibit a key virulence factor of this organism, leukotoxin (LtxA). The mechanism of catechin-mediated inhibition of LtxA has not been shown.

Methods:

In this work, we studied the ability of six catechins to inhibit LtxA-mediated cytotoxicity in human white blood cells, using Trypan blue staining, and investigated the mechanism of action using a combination of techniques, including fluorescence and circular dichroism spectroscopy, confocal microscopy, and surface plasmon resonance.

Results:

We found that all the catechins except (−)-catechin inhibited the activity of this protein, with the galloylated catechins having the strongest effect. Pre-incubation of the toxin with the catechins increased the inhibitory action, indicating that the catechins act on the protein, rather than the cell. The secondary structure of LtxA was dramatically altered in the presence of catechin, which resulted in an inhibition of toxin binding to cholesterol, an important initial step in the cytotoxic mechanism of the toxin.

Conclusions:

These results demonstrate that the catechins inhibit LtxA activity by altering its structure to prevent interaction with specific molecules present on the host cell surface.

General Significance:

Galloylated catechins modify protein toxin structure, inhibiting the toxin from binding to the requisite molecules on the host cell surface.

1. Introduction

Since the introduction of antibiotics in the 1940’s, a variety of molecules have been developed to effectively treat many bacterial diseases. With the recent rise in antibiotic resistance, the interest in alternative strategies, including anti-virulence approaches, has increased [1-3]. The goal of anti-virulence strategies is to specifically inhibit the virulence of pathogenic bacteria; thus, anti-virulence molecules have limited bacteriostatic or bactericidal activities. Some evidence has suggested that the development of resistance to these antivirulence strategies should be reduced, as there is less selective pressure than with traditional antibiotics [3, 4]. A majority of antibiotic molecules originated from naturally produced compounds [5, 6], and it is therefore likely that natural anti-virulence compounds exist as well [7].

One class of natural compounds with demonstrated antibacterial activity is the flavonoids, a group of benzo-γ-pyrone molecules derived from plants. Over 4,000 derivatives have been described, many with demonstrated biological activity [8]. One group of flavonoids that has shown strong antibacterial activity is flavan-3-ol, also known as catechins [9, 10]. Catechin and its derivatives are abundant in many tea leaves[11], especially green tea leaves, and are classified as generally recognized as safe (GRAS) by the United States Food and Drug Administration (FDA). The major constituents of green tea extracts are (−)-catechin (C), (−)-epicatechin (EC), (−)-epigallocatechin (EGC), (−)-epicatechin gallate (ECg), (−)-epigallocatechin gallate (EGCg), and (−)-gallocatechin gallate (GCg) [11] (Fig. 1). Each compound is named according to its stereoisomeric structure and location of the hydroxyl and galloyl groups [12].

Figure 1. Structure of Catechin Derivatives.

Some of the catechin derivatives as well as different tea extracts have been shown to have antibacterial effects on both Gram-positive and Gram-negative bacteria. For example GCg, EGCg, ECg, and catechin gallate (Cg) have been shown to inhibit the growth of Bacillus cereus, a Gram positive bacterium, at concentrations lower than required for vancomycin or tetracycline [13], and EGCg inhibits the growth of multidrug-resistant clinical isolates of Acinetobacter baumannii [14] and several strains of Staphylococcus [15]. In addition, tea extracts have demonstrated activity against Campylobacter jejuni [16], Clostridium perfringens [17], Escherichia coli O157:H7, and Salmonella typhimurium [18]. Besides their reported bactericidal and bacteristatic effects, catechins have demonstrated anti-resistance and anti-virulence properties as well. For example, several tea extracts showed inhibition of penicillin resistance in β-lactamase-producing strains of Staphylococcus aureus as well as an ability to alter the methicillin resistance activity in methicillin-resistant S. aureus (MRSA) [19]. Tea extracts have been used to disrupt quorum sensing of Pseudomonas aeruginosa [20], and EGCg was shown to disrupt biofilms of two oral bacteria, Streptococcus mutans [21] and Porphorymonas gingivalis [22].

In periodontal infections, catechins have demonstrated multiple activities, including inhibiting matrix metalloproteases (MMPs), which play a role in the bone and tissue destruction that is characteristic of periodontitis [23] and inhibiting the collagenase activity of gingival crevicular fluid from aggressive periodontitis patients [24]. Application of catechins along with scaling and root planing decreased the number of oral pathogen cells, including Aggregatibacter actinomycetemcomitans, detected within the pocket [25].

A. actinomycetemcomitans is a Gram negative bacterium associated with localized aggressive periodontitis (LAP) and endocarditis [26, 27]. LAP is generally treated with debridement and systemic antibiotics, usually tetracyclines; however, resistance in A. actinomycetemcomitans to tetracyclines has been increasing in recent years [28, 29], leading to a reduction in the efficacy of this approach [30-32]. During its colonization of the host, A. actinomycetemcomitans produces multiple virulence factors [33], including a leukotoxin, (LtxA). LtxA is a repeats-in-toxin (RTX) protein [34] that has been reported to be a “key” virulence factor of the organism, as it acts by killing immune cells [35], thus allowing the bacteria to colonize the host more readily. The most virulent strains of A. actinomycetemcomitans produce the most LtxA [36]; thus, inhibition of the activity of LtxA could represent a novel approach to inhibit A. actinomycetemcomitans pathogenicity.

There have been some reports on the ability of particular catechins to prevent LtxA-mediated cytotoxicity in white blood cells. Secreted LtxA exists in two forms, as a free, soluble protein [37-39] and associated with outer membrane vesicles (OMVs) [40, 41], and catechins have demonstrated inhibitory action against both forms. Specifically, galloylated catechins were shown to inhibit the activity of LtxA against HL-60 cells [42], and several catechins inhibited white blood cell lysis by A. actinomycetemcomitans OMVs [43], but the mechanism of this inhibition has not yet been demonstrated. In this work, we sought to understand how catechins inhibit the activity of the soluble form of LtxA. Because LtxA has strong membrane-interacting abilities [44-46] including an affinity for cholesterol/lipid rafts [44, 47, 48], we focused on this aspect of the toxin’s activity. Our results point toward a mechanism in which catechins alter LtxA structure, resulting in a reduced affinity for cholesterol.

2. Materials and Methods

2.1. Chemicals

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) was purchased from Avanti Polar Lipids (Alabaster, AL). Cholesterol (Chol), (−)-catechin (C), (−)-epicatechin (EC), (−)-epigallocatechin (EGC), (−)-epicatechin gallate (ECg), (−)-epigallocatechin gallate (EGCg), (−)-gallocatechin gallate (GCg), and poly-L-lysine (PLL) were purchased from Sigma-Aldrich (St. Louis, MO). N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (NBD-PE), Alexa Fluor^® 647 (AF647) and 6-dodecanoyl-2-dimethylaminonaphthalene (Laurdan) were purchased from Molecular Probes (Eugene, OR).

2.2. Cell Culture

THP-1 cells (ATCC), were grown in RPMI 1640 medium supplemented with 10% FBS, 0.05 mM 2-mercaptoethanol, and 1% Pen-Strep (100 U/mL penicillin and 100 μg/mL streptomycin).

2.3. LtxA Purification

LtxA was purified from the A. actinomycetemcomitans JP2 supernatant following a published protocol [49]. The purity of the purified toxin was determined using sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), and the identity was confirmed by Western blot (Fig. S1 A, B). The activity of the toxin was determined using a cytotoxicity experiment (Fig. S1C).

2.4. Preparation of Liposomes

Large unilamellar vesicles (LUVs) were prepared using the film deposition technique. Lipid stocks in chloroform were added to a glass vial in the required proportions. The mixture was dried under nitrogen, and residual solvent was evaporated by exposing the sample to vacuum overnight. The lipid films were then hydrated in buffer, vortexed to create multilamellar vesicles (MLVs) and then extruded through a polycarbonate membrane with 100-nm pores (GE Healthcare BioSciences, Pittsburgh, PA) using a LiposoFast extruder (Avestin Inc., Ottawa ON) to make LUVs.

For SPR analysis, the molar composition of the liposomes was 90% POPC and 10% Chol, and the liposomes were hydrated in phosphate buffered saline (PBS) at a pH of 7.4. For membrane rigidity experiments, the molar composition of the liposomes was 79% POPC, 20% Chol, and 1% Laurdan. These laurdan-labeled LUVs were hydrated in liposome buffer (150 mM NaCl, 5 mM CaCl₂, 5 mM HEPES, 3 mM NaN₃, pH = 7.4)

Giant unilamellar vesicles (GUVs) were created by mixing POPC, Chol, and NBD-PE in a 79/20/1 molar ratio. The lipids, dissolved in 250 μL of chloroform and 30 μL of acetonitrile, were gently placed at the bottom of 3 mL of PBS (pH 7.4) in a round bottom flask. Using a rotary evaporator, the flask was rotated under vacuum in a 40 °C water bath for approximately 5 s to evaporate the solvents to form GUVs.

2.5. Labeling THP-1 Cells with Laurdan

THP-1 cells at a density of 2 × 10⁶ cells/mL were resuspended in Laurdan-containing media without phenol red (1% (by mass) laurdan in THP-1 cell media without phenol red) and incubated for 15 mins. The cells were washed with PBS and resuspended in phenol red-free media before being used.

2.6. LtxA Labeling

The AF647 dye (Molecular Probes, Eugene, OR) was used to label LtxA. LtxA at a concentration of 5 μg/μL in 0.1 M NaHCO₃ buffer was incubated with 1 μL of AF647 dye at a concentration of 484 μM for 15 min. The mixture was run through a Zeba™ spin desalting column (ThermoFisher Scientific) with a 40k molecular weight cut-off (MWCO) resin to remove any excess dyes.

2.7. Membrane Rigidity Measurements

The GP value of Laurdan was used to quantify the change in the rigidity of LUVs and THP-1 cell membrane. In the LUV experiments, 100 μg of each catechin derivative was incubated with the liposomes for 30 mins. In THP-1 cell experiments, 10 μg of each catechin derivative was incubated with cells for 30 mins at 37 °C under 5% CO₂. The fluorescence spectra of all samples were recorded from 400 nm to 500 nm, with an excitation wavelength of 340 nm, using a Quantamaster 400 spectrofluorometer (PTI Horiba, Edison, NJ). The intensities of the spectra at 440 nm (I₄₄₀) and 490 nm (I₄₉₀) were used to calculate the GP value as follows:[50]

$$\mathit{GP}=\frac{I_{440}-I_{490}}{I_{440}+I_{490}}$$ (1)

2.8. Cell Viability

THP-1 cells were resuspended at a density of 2 × 10⁶ cells in fresh media and transferred to tubes that had previously been blocked with PBS/0.1% BSA. The cells were then treated with 10 μg of catechin and 4 μg of LtxA. The dose-dependent inhibitory ratio of catechin to LtxA was determined from a ladder experiment. The experiment was conducted using one of three incubation methods: 1) THP-1 cells, catechins, and LtxA were simultaneously incubated for 3 hrs. 2) THP-1 cells were pre-incubated with catechin for 30 mins, followed by 3 hrs of incubation with LtxA. 3) LtxA was pre-incubated with catechin for 30 minutes, followed by incubation with THP-1 cells for 3 hrs. All pre-incubations were performed at room temperature, and the 3-hr incubation with cells was conducted at 37 °C under 5% CO₂. The cell viability was determined using a Trypan blue assay.

2.9. Confocal Microscopy

At room temperature, 2 μg of labelled LtxA was pretreated with 5 μg of each catechin derivative for 30 mins, then added to 100 μL of GUVs. After 30 mins, the GUVs were transferred to a glass-bottom dish that was pretreated with PLL (Sigma-Aldrich, St. Louis, MO).

The fluorescently labelled GUVs and LtxA were imaged on a Zeiss LSM880 scanning confocal microscope using a 63 × oil objective with 488 nm and 633 nm laser sources. To quantify the co-occurrence of two channels, the channel noise was eliminated by setting an auto-threshold using ImageJ. Then the regions of interest were set for each GUV image to analyze the co-occurrence of LtxA with the labeled GUV membrane. The Manders’ coefficient was calculated using ImageJ to quantify the co-occurrence.

2.10. Surface Plasmon Resonance

A sensor chip, functionalized with a hydrophobic coating (Nicoya Lifesciences, Kitchener, ON, Canada), was coated with 200 μL of POPC/Chol (90/10) liposomes using a flowrate of 20 μL/min on OpenSPR system (Nicoya Lifesciences). Next, 2.4 mL of 100 nM LtxA, pretreated with catechin at a mass ratio of 4:10, was injected. Additionally, untreated LtxA and each catechin were injected separately over the POPC/Chol-coated chip.

The data were analyzed using TraceDrawer. The catechin-only curve was subtracted from the corresponding LtxA+catechin curve to obtain the affinity of LtxA for POPC/Chol in the presence of that catechin. Each sample was then compared to the curve corresponding to untreated LtxA.

2.11. Circular Dichroism (CD) Spectroscopy

The effect of the catechins on the secondary structure of LtxA was studied using CD spectroscopy using a J-815 spectrometer (JASCO, Easton, MD). LtxA at a concentration of 0.4 mg/mL in 0.1 M potassium phosphate buffer was incubated with EGCg at a mass ratio of 4:10 for 30 mins before the sample was placed in a quartz cuvette with a path length of 0.01 cm. The sample was scanned from 240 nm to 190 nm at a scanning speed of 20 nm/min with a bandwidth of 1.0 nm. Buffer-only, EGCg-only and C-only samples were run under similar conditions. The spectra from buffer-only, EGCg-only and C-only samples were subtracted from the corresponding LtxA-containing spectra, and the resulting spectra were deconvoluted to obtain the secondary structures of the toxin using DICHROWEB with the CONTINLL program and reference set 7 [51].

2.12. Statistical Analysis

All data are presented as the mean ± standard deviation. Statistical analysis was performed using an unpaired two-tailed student’s t-test, where p values less than 0.01 were considered to be statistically significant.

3. Results and Discussion

3.1. ECg and EGCg Increase the Rigidity of THP-1 Cell Membranes

Previous work has shown that galloylated catechins are able to inhibit the activity of LtxA in HL-60 cells [42], and these same catechins have been demonstrated to decrease cell membrane fluidity [52-54]. We have previously found that compounds that decrease membrane fluidity can inhibit LtxA activity by preventing the toxin from partitioning into the membrane [55]. For these reasons, we investigated whether catechin-mediated changes in target cell membrane fluidity might be responsible for their inhibitory effects against LtxA.

The laurdan probe has been used to measure membrane fluidity, due to its sensitivity to the presence of water in the membrane interior. This probe partitions to the hydrophobic core of the membrane. When the membrane is in the gel phase, the emission maximum of the probe occurs at a wavelength of 440 nm. However, as the membrane becomes more fluid, the local concentration of water increases, and the emission maximum shifts to a wavelength of 490 nm due to dipolar relaxation of the probe [56]. The generalized polarizability (GP) value, essentially a ratio of the emission intensity at these two wavelengths, can thus be used as a measure of membrane fluidity, with a low GP value indicating a more gel-like membrane and a high GP value indicating a more fluid-like membrane.

We first investigated the ability of six catechins, C, EC, EGC, ECg, EGCg, and GCg, to alter membrane fluidity in model membranes. As shown in Fig. S2, we observe that in model membranes composed of 100% POPC (Fig. S2A) the galloylated catechins, ECg, EGCg, and GCg, increase the GP of the membrane corresponding to a decrease in membrane fluidity. The non-galloylated catechins, C, EC, and EGC, also cause a decrease in membrane fluidity, but to a lesser extent. In 80% POPC/20% Chol membranes (Fig. S2B), only the galloylated catechins induce a decrease in the fluidity. These findings are consistent with other published observations [52-54].

In both types of membranes, with or without cholesterol, the galloylated catechins increased the GP value to approximately 0.45, regardless of the initial GP. The GP value in the absence of catechin depends on the presence of cholesterol, as cholesterol-containing membranes are less fluid than membranes lacking cholesterol [57]. Our results are consistent with a published report in which fluorescence polarization was used to determine that galloylated catechins decrease the fluidity of model membranes to a greater extent than do regular catechins [53]. This enhanced effect may be related to membrane partitioning, as galloylated catechins have been shown to partition more strongly into the membrane than regular catechins [58].

While model membrane studies can be quite useful, we wanted to ensure that the effect of the catechins on target cell membrane fluidity is similar to what is observed in model membranes. We therefore next investigated the effect of these six catechins on the membrane fluidity of THP-1 cells, a human monocytic cell line that is particularly susceptible to LtxA [59]. To investigate the effect of catechins on the rigidity of the THP-1 cell membrane, each of the catechins was incubated with laurdan-labeled THP-1 cells. The change in GP value of the membrane due to the addition of each catechin was calculated relative to an untreated control by subtracting the average GP value of the untreated cells from the GP value of each catechin-treated sample and dividing by average GP value of the untreated cells. As shown in Fig. 2, the regular catechins (C, EC, and EGC) decreased the GP value of the probe, corresponding to an increase in the membrane fluidity. However, only the decrease induced by EC was statistically significant. The galloylated catechins, EGCg and ECg, on the other hand, decreased the fluidity of the THP-1 cell membrane, which is consistent with what we (Fig. S2) and others [52-54] have observed in model membranes. Surprisingly, GCg, which has likewise been shown to decrease fluidity in model membranes [60], had the opposite effect on THP-1 cell membranes, resulting in a statistically significant increase in cell membrane fluidity. This result contrasts with our studies in model membranes where we observed that these three galloylated catechins mediate a significant decrease in membrane fluidity. We next conducted a series of cytotoxicity experiments to determine if these observed catechin-mediated changes in cell membrane fluidity result in inhibition of LtxA activity.

Figure 2. Catechin-Mediated Effects on THP-1 Cell Plasma Membrane Fluidity.

The laurdan probe was used as a measure of membrane fluidity, with a decrease in GP indicating an increased fluidity and an increased GP indicating a decreased fluidity. THP-1 cells were cultured in laurdan-containing media for 15 min. The GP value was measured initially and then after incubation with one of six catechins, and the change in GP value was calculated by subtracting the average GP value of untreated samples from the GP value after catechin incubation and dividing by the initial GP value. C, EC, EGC, and GCg induced small increases in the THP-1 cell membrane fluidity (decreased GP), while ECg and EGCg induced decreases in membrane fluidity (increased GP). The data are presented as the mean (N=3) ± standard deviation. The level of significance was determined using a two-tailed student’s t-test. N.S., not significant; *** p < 0.0001, relative to untreated control.

3.2. Catechins Inhibit LtxA Activity in THP-1 Cells

To investigate whether these six catechins can inhibit LtxA-mediated cytotoxicity in THP-1 cells, we conducted a cytotoxicity assay. In the initial experiment, LtxA and one of the six catechins, at an LtxA to catechin (L:C) mass ratio of 4:10, was added to the THP-1 cells, and the cell viability was measured after a three-hr incubation. As shown in Fig. 3A, C, EC, ECg, EGCg, and GCg inhibited LtxA-mediated cytotoxicity, but EGC did not. The galloylated catechins, ECg and EGCg, had the greatest inhibitory action.

Figure 3. Inhibitory Effect of Catechins on LtxA Activity.

A. LtxA and one of the six catechins were added to THP-1 cells simultaneously, and the cell viability was measured after 3 hr. B. THP-1 cells were pretreated with one of the six catechins for 30 mins before the addition of LtxA. Cell viability was measured after 3 hr. C. LtxA was pretreated with one of the six catechins for 30 mins before being added to THP-1 cells, and cell viability was measured after 3 hr. The cell viability was measured using Trypan blue assay. D. The cytotoxicity of catechins against THP-1 cells was measured. One of the six catechins was added to THP-1 cells, and the cell viability was measured after 3 hr using Trypan blue assay. Data are presented as the mean (N=3) ± standard deviation. The level of significance was determined using a two-tailed student’s t-test. N.S., not significant; *, p < 0.01; **, p < 0.001; *** p < 0.0001, relative to LtxA-treated cells for A-C and relative to untreated cells for D.

We next investigated whether the catechin-mediated change in cell membrane fluidity is the cause of the observed inhibitory action by pretreating the THP-1 cells with each of the six catechins for 30 min before adding LtxA. We expected that if the catechin-mediated membrane fluidity change is responsible for the inhibition of LtxA activity, we would see an enhanced inhibitory action of the catechins by preincubating them with the cells. However, in this case, we observed similar inhibitory action as in the case where catechin and LtxA were added to the cells simultaneously (Fig. 3B), suggesting that the catechins are not acting on a cellular target. We then repeated the experiment, instead pretreating LtxA with each of the catechins for 30 mins before adding the mixture to the THP-1 cells to investigate whether the catechins might be acting on the toxin directly. Fig. 3C shows that, except for C, pretreatment of the toxin with the catechins greatly improves their inhibitory action, indicating that the catechins act on the toxin rather than on the membrane to induce a protective effect. All three galloylated catechins had the greatest inhibitory effect, almost completely blocking LtxA activity.

At the concentrations used in these experiments, most of the catechins exhibited no significant toxicity in THP-1 cells; however, EC exhibited a slight, but statistically significant, toxic effect on the cells (Fig. 3D). Because EGCg was particularly effective in inhibiting LtxA, we next investigated whether decreased concentrations of this catechin were likewise effective in inhibiting LtxA-mediated cytotoxicity. As shown in Fig. 4, at 10 and 100 times lower concentrations (L:C mass ratio of 4:1 and 4:0.1, respectively), EGCg retained its inhibitory action, although to a reduced extent compared to the ratio of 4:10 used in the previous cytotoxicity experiment (Fig. 3A-C), demonstrating the effectiveness of this catechin in preventing LtxA-mediated cytotoxicity.

Figure 4. Minimum Inhibitory Ratio Between Catechin and LtxA.

LtxA was pretreated with EGCg at four different ratios (mass LtxA:mass EGCg = 4:0.1, 4:1, 4:10, and 4:100) and incubated for 30 mins before being added to THP-1 cells. After 3 hr, the cell viability was measured using a Trypan blue assay. Data are presented as the mean (N=3) ± standard deviation. N.S., not significant, *, p < 0.01, ***, p < 0.0001, relative to LtxA only sample.

Previous studies of the ability of catechins to prevent A. actinomycetemcomitans-mediated cell death have looked at the ability of these molecules to protect host cells against A. actinomycetemcomitans lipopolysaccharide (LPS) [61] and LtxA-containing OMVs [62]. In both cases, some of the catechins had a protective effect, but pretreatment of the cells with catechin did not improve this effect, suggesting that the catechins act, not on the cell, but on the bacterial components, as we have observed here. Additionally, the ability of galloylated catechins to inhibit LtxA-mediated cytotoxicity has been shown [63], but the enhancement of this effect by pretreating the toxin with catechins has not yet been demonstrated. With this result, we determined that our initial hypothesis that catechin-mediated effects on membrane fluidity are responsible for their anti-LtxA activity is incorrect. Rather, the results of our cytotoxicity experiment instead suggest that the catechins affect LtxA directly, as pretreatment with LtxA increased their inhibitory action. We therefore undertook a series of experiments to specifically investigate the mechanism by which the catechins interact with LtxA.

3.3. Catechins Inhibit Binding of LtxA to Lipid Membranes

In its interaction with host cells, LtxA employs several strong interactions with the plasma membrane lipids [45], including binding to cholesterol [44, 47]. We have previously demonstrated that disruption of this binding event abolishes the ability of the toxin to kill host cells [48, 64, 65]. Therefore, we hypothesized that the catechins, in their direct interaction with the toxin, may be disrupting the ability of the toxin to bind to cholesterol, leading to the observed inhibition of toxin activity.

To visualize the binding of LtxA to cholesterol-containing membranes in the presence of catechins, we imaged the interaction of fluorescently labelled LtxA and cholesterol-containing GUVs using confocal microscopy. LtxA was labeled with AF647 (red) and GUVs with NBD-PE (green). As shown in Fig. 5A, in the absence of catechin, LtxA associates readily with the GUV membrane. Non-galloylated catechins had little effect on this association, although EC did exhibit some inhibition. The galloylated catechins significantly inhibited the ability of LtxA to bind to the GUV membranes. Fig. 5B shows the quantification of this association using a Manders’ coefficient, a measure of the co-occurrence of red and green fluorescence [66]. This calculation was performed using eight confocal scans, each containing two to eight GUVs. Consistent with the representative images in Fig. 5A, the non-galloylated catechins have less of an effect LtxA association with the GUVs, while the galloylated catechins significantly inhibited association. Additional GUV images are shown in Fig. S3.

Figure 5. Binding of LtxA to GUV membranes in absence and presence of catechins.

LtxA was pretreated with one of six catechins or PBS for 30 mins before being incubated with cholesterol-containing GUVs. LtxA was labeled with AF647 (red) and the GUV membranes contained a small amount of NBD (green). A. Representative confocal images of LtxA association with GUVs. All of the galloylated catechins inhibited the ability of LtxA to bind to these cholesterol-containing membranes. Regular catechins had little effect on inhibition the binding, with the exception of EC which had a little effect. The length of the scale bar is 5 μm. B. Manders’ coefficient analysis of the interaction shown in panel A. Data are presented as the mean (N=8) ± standard deviation. The level of significance was determined using a two-tailed student’s t-test. N.S., not significant; ***, p < 0.0001, relative to LtxA-treated GUVs.

We further investigated the catechin-mediated inhibition of LtxA binding to membrane lipids using SPR. A hydrophobic sensor chip was coated with liposomes composed of 90 mol% POPC/10 mol% Chol, and LtxA alone or pretreated with one of the six catechins was flowed over the chip. Because the catechins also have a detectable membrane affinity, each catechin by itself was also flowed over the liposome coated chip, and this signal was subtracted from the LtxA-catechin signal to obtain a measure of LtxA binding.

Because of the need to subtract the catechin signal from the LtxA-catechin signal, we were unable to conduct the experiment in a manner to obtain an equilibrium dissociation constant (K_D). Instead, we investigated the association of LtxA to the membrane after steady-state binding was reached. These results are quantified in Fig. S4, where the intensity over the last 100 s was averaged and normalized relative to the LtxA-only sample. In this experiment, C increased the affinity of LtxA for cholesterol (although this increase was not statistically significant), while the remaining five catechins decreased the affinity of LtxA for cholesterol, consistent with the results from the GUV experiment and our cytotoxicity experiments. As observed in the cytotoxicity experiments, the galloylated catechins were most effective in inhibiting LtxA binding to cholesterol.

These findings indicate that the galloylated catechins affect the ability of LtxA to bind to cholesterol on the host cell membrane. Because this interaction is an essential step in the toxin’s mechanism, activity is reduced. The ability of LtxA to bind cholesterol is regulated by a cholesterol recognition amino acid consensus (CRAC) motif, located in the N-terminal region of LtxA [44]. We therefore hypothesized that the catechins may alter the conformation of LtxA, changing the exposure of this motif, and leading to the observed decrease in cholesterol binding by the toxin.

3.4. EGCg Induces a Conformational Change in LtxA

To determine if the catechins induce a restructuring of the toxin, CD spectroscopy was used. We limited our structural studies to EGCg, which was found in the previous experiments to have a large effect on the activity of LtxA, and C, which was found to have little effect on the activity of LtxA. The buffer-subtracted spectrum of LtxA is shown in Fig. 6A, along with the EGCg-subtracted spectrum of EGCg-treated LtxA and C-subtracted spectrum of C-treated LtxA as a negative control. In the absence of any catechin, the LtxA spectrum consists of a broad negative peak between 208 nm and 222 nm, and a large positive peak at 190 nm. In the EGCg-treated sample, the CD spectrum instead exhibits a positive peak at 212 nm and a negative peak at 200 nm, demonstrating that the secondary structure of LtxA has been greatly changed by the EGCg treatment. In contrast, the spectrum of C-treated LtxA is quite similar to that of untreated toxin. Deconvolution of the spectra shows that EGCg decreases the helical content of LtxA by approximately 70%; while C only reduces the helical content of LtxA by approximately 35% (Fig. 6B). Additionally, the peaks in the CD spectrum of EGCg-treated LtxA are characteristic peaks of an unordered protein, indicating that the effect of this catechin on LtxA is a substantial change in its secondary structure.

Figure 6. Structural Analysis of LtxA in the Presence of EGCg or C.

A. CD spectra of untreated LtxA (solid line), EGCg-treated LtxA from which the CD spectrum of EGCg alone has been subtracted (dotted line) and C-treated LtxA from which the CD spectrum of C alone has been subtracted (dashed line). The data represent the average of three runs. B. Percent change in helical structure compared to untreated LtxA, determined from deconvolution of the CD spectra.

To confirm the effect of EGCg on altering the secondary structure of protein, an additional CD experiment was performed using bovine serum albumin (BSA) (Fig. S5). All of the experimental conditions were kept the same, including the mass ratio of BSA to EGCg (4:10). However, the conformational change induced by EGCg on BSA was not as great as the effect on LtxA, suggesting that there is some specificity in the interaction of EGCg with LtxA.

Together, our results point toward a mechanism in which catechins, particularly galloylated catechins induce a restructuring in LtxA that prevents the toxin from interacting with requisite molecules on the host cell surface. This finding is consistent with work demonstrating that grape extract, which includes C, EC, ECg, and EGCg, among other molecules, inhibits the activity of Shiga toxin [67], heat-labile toxin (LT), and cholera toxin (CT) [68]. In these studies, pretreatment of the cells had no effect, suggesting that the extracts are acting directly on the toxins themselves. In fact, the extract was found to alter the structure of CT and inhibit its binding to its cellular receptor, ganglioside GM1 [68], and three additional polyphenol-containing extracts inhibited LT binding to GM1 [69]. Similarly, a number of polyphenols, including EC, EGC, and EGCg were found to inhibit the activity of the vacuolating cytotoxin, VacA, produced by Helicobacter pylori, with EGCg having a stronger effect than EC, as we have observed here [70]. EGCg was also shown to inhibit the activity of listeriolysin O (LLO), produced by Listeria monocytogenes.

The broad range of activities ascribed to these molecules, including non-specific lipid and protein interactions, highlight an important caveat in their use as anti-toxin agents. However, we propose that their topical use in the oral cavity during treatment of periodontitis could enhance, though not replace, current treatment options for this disease. The oral cavity is unique in that it is an accessible location, thus delivery of these molecules to the infected site is straight-forward and does not require targeted delivery devices. We expect that off-target interactions could be limited for this reason.

4. Conclusions

Our results indicate that the inhibition of LtxA activity by galloylated catechins is caused by a transient change in the toxin conformation, which results in a decrease in the necessary membrane interactions by the toxin. We propose that these molecules could be applied topically in the treatment of periodontitis to enhance current treatment regimens and improve patient outcomes.

Highlights.

• Galloylated catechins alter the structure of A. actinomycetemcomitans leukotoxin.

• Galloylated catechins prevent leukotoxin from binding to cholesterol.

• Leukotoxin activity is thus greatly inhibited by galloylated catechins.

ABBREVIATIONS

AF 647

Alexa Fluor^® 647

BSA

bovine serum albumin

C

(−)-catechin

CD

Circular dichroism

Cg

catechin gallate

Chol

Cholesterol

CRAC

cholesterol recognition amino acid consensus

CT

cholera toxin

EC

(−)-epicatechin

ECg

(−)-epicatechin gallate

EGC

(−)-epigallocatechin

EGCg

(−)-epigallocatechin gallate

GCg

(−)-gallocatechin gallate

GP

generalized polarizability

GUV

giant unilamellar vesicle

K_D

equilibrium dissociation constant

LAP

localized aggressive periodontitis

LLO

listeriolysin O

LPS

lipopolysaccharide

LT

heat-labile toxin

LtxA

leukotoxin

LUV

large unilamellar vesicle

MLV

multilamellar vesicle

MMP

matrix metalloprotease

MRSA

methicillin-resistant S. aureus

MWCO

molecular weight cutoff

NBD-PE

N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine

OMV

outer membrane vesicle

PBS

phosphate buffered saline

PLL

poly-L-lysine

POPC

1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine

RTX

repeats-in-toxin

SDS-PAGE

Sodium dodecyl sulphate-polyacrylamide gel electrophoresis

SPR

surface plasmon resonance

VacA

vaculoating cytotoxin