In Vitro Inhibition of Streptococcus mutans Biofilm Formation on Hydroxyapatite by Subinhibitory Concentrations of Anthraquinones

Abstract

We report that certain anthraquinones (AQs) reduce Streptococcus mutans biofilm formation on hydroxyapatite at concentrations below the MIC. Although AQs are known to generate reactive oxygen species, the latter do not underlie the observed effect. Our results suggest that AQs inhibit S. mutans biofilm formation by causing membrane perturbation.

Dental plaque displays several properties typical of biofilms, including reduced susceptibility to antimicrobial agents (9, 13). Streptococcus mutans is considered to be the primary cariogen within dental plaque (12), and prevention or reduction of biofilm formation by S. mutans could thus contribute to the prevention of caries. In the present study we evaluated the abilities of anthraquinones (AQs) to inhibit S. mutans biofilm formation at concentrations below the MIC, as well as their mechanism of action.

S. mutans LMG 14558^T, Micrococcus luteus NRRL B-2618, and Vibrio harveyi strains were routinely grown in brain heart infusion (BHI) broth (Becton Dickinson[BD], Franklin Lakes, NJ) at 37°C, on tryptic soy agar (BD) at 30°C, or on Difco marine agar (BD) at 37°C, respectively. The AQs tested are listed in Table 1. To determine the MIC of each compound, a microdilution assay in 96-well microtiter plates (TPP, Trasadingen, Switzerland) was used (8). S. mutans LMG 14558^T biofilms were grown on hydroxyapatite disks in modified Robbins' devices, and the biofilm biomass on each disk was estimated using fluorescent staining with SYTO9 (Invitrogen, Carlsbad, CA) (8). Inhibition of glucosyltransferase by AQs was assessed by an enzymatic assay (16). Induction of reactive oxygen species (ROS) by AQs was measured by two separate assays. In the first assay, MICs were determined in the presence or absence of 1.5 mM glutathione (GSH), 0.025% (wt/vol) cysteine, and 10 mM mannitol (all from Sigma) by using the modified microdilution assay described above. In the second assay, we used the fluorescent probe 2′,7′-dichlorofluorescein diacetate (DCF-DA) to quantitate the amount of ROS produced (5, 6). The inhibition of mutacin production can be used as an indirect assay to detect competence-stimulating peptide-based quorum sensing (QS) (22). Mutacin production was determined using M. luteus NRRL B-2618 as an indicator strain. Inhibition of autoinducer-2-based QS was studied using V. harveyi biosensor strains, as described previously (7). Lateral diffusion of fatty acids in the cell membrane was measured by the intermolecular excimerization of the fluorescent probe pyrene (2). To study the rotational diffusion of the fatty acid acyl chains in the interior of the membrane, fluorescence anisotropy was measured using 1,6-diphenyl 1,3,5-hexatriene (DPH) (2). Cellular fatty acid analysis was performed as described previously (21).

TABLE 1.

MICs, concentrations used in biofilm experiments, and effects of AQs^a

Compound	MIC (μg/ml)	Concn in MRDb (μg/ml)	Biofilm formation (%)c (avg ± SD)	Significanced
Emodin	>250	5	10.9 ± 0.9	<0.001
		2	52.1 ± 7.2	<0.001
		1	79.0 ± 10.9	0.005
		0.5	85.7 ± 12.7	NS
		0.1	100.8 ± 13.9	NS
Hypericin	250	5	47.8 ± 16.7	<0.001
Carminic acid	>250	5	66.5 ± 23.15	0.005
Chrysophanic acid	250	5	75.0 ± 10.0	<0.001
Rhein	10	5	75.2 ± 25.0	NS
Quinizarin	250	5	80.0 ± 17.2	0.01
Sennidin A	>250	5	83.7 ± 17.8	NS
Chrysazin	250	5	91.3 ± 9.3	NS
Anthraflavic acid	250	5	98.2 ± 12.5	NS
Aloe-emodin	>250	50	99.2 ± 22.2	NS
Physcion	>250	5	119.9 ± 17.7	NS
Chlorhexidine	2	0.1	103.5 ± 8.0	NS
		1200	34.5 ± 21.3	<0.001
Triclosan	16	5	78.13 ± 5.5	0.005
		1333	10.9 ± 10.8	<0.001

^aFor comparison, we also determined the inhibition of biofilm formation by the antibacterial agents triclosan and chlorhexidine, compounds commonly included as antibacterial agents in toothpaste and mouthwash.

^bMRD, modified Robbins' device.

^cRelative to biofilm formation with the BHIS control.

^dP value for difference between biofilm formation in BHIS plus the indicated compound and that in BHIS alone (by a one-tailed independent-sample t test). NS, not significant (P > 0.01).

The overall mean fluorescence response for S. mutans biofilms grown on hydroxyapatite in BHI supplemented with 1% sucrose (BHIS) (positive controls) (386 disks) after staining with SYTO9 was 13.22 × 10⁵ ± 2.83 × 10⁵ relative fluorescence units, which correlates with approximately 5 × 10⁸ to 6 × 10⁸ cells per disk (8). The MICs for various AQs are shown in Table 1. Most of the AQs tested in the present study exhibited no activity against planktonic S. mutans LMG 14558^T cells, with MICs of ≥250 μg/ml (Table 1). Biofilms grown in the presence of emodin, hypericin, carminic acid, chrysophanic acid, or quinizarin (5 μg/ml) revealed significantly lower fluorescence responses (P ≤ 0.01) than biofilms grown in BHIS without AQs (Table 1). For emodin, the most active compound, there was a quasilinear relationship between its concentration and relative biofilm formation (Table 1). None of the AQs investigated showed significant inhibition of glucosyltransferase (data not shown). The addition of 1.5 mM GSH, 10 mM mannitol, or 0.025% cysteine did not result in an altered MIC for emodin in M1 medium (17) (Fig. 1). Similarly, supplementation of BHIS medium containing 5 μg/ml emodin with 0.025% cysteine did not result in increased biofilm formation, i.e., cysteine had no protective effect against the action of emodin. This strongly suggested that ROS generation is not the mechanism by which emodin affects S. mutans cells. This was confirmed by using the oxidative-stress-specific fluorescent probe DCF-DA (data not shown). No effect of AQs on competence-stimulating peptide- or autoinducer-2-based QS was observed (data not shown). Incubation of S. mutans LMG 14558^T with emodin or hypericin resulted in a decreased membrane lateral and rotational diffusion (Table 2; Fig. 2), indicating reduced membrane fluidity. Although the average difference in anisotropy between DPH-labeled emodin-treated cells (0.1528) and untreated cells (0.1327) was low (15.13%) and not significant (P = 0.512), these differences were observed consistently. In addition, small changes (approximately 10%) in fluorescence anisotropy may reflect marked changes (of about 25%) in membrane microviscosity (2, 11). Together with the significant decrease in membrane lateral diffusion, the observed differences in fluorescence anisotropy strongly suggest an effect of emodin on membrane microviscosity.

FIG. 1.

Susceptibility of S. mutans LMG 14558^T to increasing emodin concentrations in various media. MICs were determined using a modified microdilution assay in 96-well microtiter plates, as previously described (8), in the presence and absence of 1.5 mM GSH (Sigma), 0.025% (wt/vol) cysteine (Sigma), and 10 mM mannitol (Sigma). These compounds are ROS scavengers, protect cells from oxidative damage (18, 23), and will increase the MIC of emodin if its antibacterial effect is due to ROS production. M1 is the chemically defined medium described in reference 17.

TABLE 2.

Relative fluidity of membranes of S. mutans grown under different conditions, as measured using pyrene and DPH

Conditions	Pyrene		DPH
	Fluidity (%)a (avg ± SD)	P value	Anisotropyb (avg ± SD)	% Differencec
BHI	100 ± 27.8		0.1327 ± 0.0433
BHI + 5 μg/ml emodin	21.2 ± 47.4	<0.01	0.1528 ± 0.0107	15.13
BHI + 5 μg/ml hypericin	35.6 ± 28.2	<0.05	ND	ND
BHI + 1% green tead	23.4 ± 39.2	<0.05	ND	ND

^aRelative to fluidity with BHI alone. At a constant absolute temperature T, the relative fluidity of the membranes of cells treated with AQs (F) compared to the fluidity of membranes of untreated cells (F_r) is calculated as follows: F = (I_e/I_m)/(I/I_m)_r × F_r, where I_e/I_m is the pyrene excimer-to-monomer fluorescence intensity at temperature T (14, 15). For convenience, F_r (the relative fluidity of untreated S. mutans cells grown in BHI) was set to 1.

^bFluorescence anisotropy (A) is defined as I_‖ − I_⊤/I_‖ + 2I_⊤, where I_‖ and I_⊤ are the fluorescence intensities parallel and perpendicular to the direction of the excitation beam, respectively (2). ND, not determined.

^cCalculated by the formula (A_BHI − A_emodin)/A_emodin × 100.

^dThe green tea extract was selected as a positive control because it was previously shown that green tea catechins affect membrane fluidity (20).

FIG. 2.

Fluorescence emission spectra of DPH-labeled S. mutans LMG 14558^T cells grown in BHI in the presence or absence of 5 μg/ml emodin. DPH is allowed to insert into the membrane, the sample is subsequently excited with polarized light, and the extent of polarization of the emitted light (which depends on the rotational Brownian motion) is measured. The magnitude of the rotational diffusion of DPH depends on the temperature and the microviscosity (fluidity) of the surrounding membrane. Cell suspensions were incubated with 5 × 10⁻⁶ M DPH for 1 h at 37°C, and subsequently steady-state fluorescence anisotropy was measured with a Photon Technology International spectrofluorimeter (excitation with vertically polarized light of 360 nm; emission at 430 nm).

It has been reported previously that membrane fluidity has an influence on many cellular processes, including permeability, cold adaptation, and growth and survival at suboptimal temperatures (3, 4, 10, 19). Bacterial membrane fluidity is most often modulated by altering the fatty acid composition, but there were no significant differences in membrane fatty acid composition between S. mutans cells grown in the presence or absence of emodin (data not shown). It has been reported that emodin becomes inserted into the phospholipid bilayer, strongly affects van der Waals interactions between hydrocarbon chains of phospholipids, and destabilizes membrane bilayers by promoting nonbilayer phases (1). Based on this, we suggest that the antibiofilm effect of emodin is caused by insertion of the planar molecule into the cell membrane and/or by binding of the same molecule to membrane-embedded molecules, including proteins. To our knowledge, there are at present no data on the effect of changes in bacterial membrane fluidity on biofilm formation, although an effect on adhesion potential appears plausible. The observation that the formation of an S. mutans biofilm can be significantly reduced by AQs at subinhibitory concentrations is unusual and may lead to novel strategies to prevent dental plaque and caries.