Antifungal activity of local anesthetics against Candida species.

OBJECTIVE
To evaluate the activity of benzydamine, lidocaine, and bupivacaine, three drugs with local anesthetic activity, against Candida albicans and non-albicans strains and to clarify their mechanism of activity.


METHODS
The minimal inhibitory concentration (MIC) was determined for 20 Candida strains (18 clinical isolates and two American Type Culture Collection strains). The fungistatic activity was studied with the fluorescent probe FUN-1 and observation under epifluorescence microscopy and flow cytometry. The fungicidal activity of the three drugs was assayed by viability counts. Membrane alterations induced in the yeast cells were evaluated by staining with propidium iodide, by quantitation of intracellular K+ leakage and by transmission electron microscopy of intact yeast cells and prepared spheroplasts.


RESULTS
The MIC ranged from 12.5-50.0 microg/mL, 5.0-40.0 mg/mL, and 2.5-10.0 mg/mL for benzydamine, lidocaine, and bupivacaine, respectively. The inhibitory activity of these concentrations could be detected with the fluorescent probe FUN-1 after incubation for 60 minutes. A very fast fungicidal activity was shown by 0.2, 50, and 30 mg/mL of benzydamine, lidocaine, and bupivacaine, respectively.


CONCLUSIONS
At lower concentrations, the tested drugs have a fungistatic activity, due to yeast metabolic impairment, while at higher concentrations they are fungicidal, due to direct damage to the cytoplasmic membrane.

Many nonantibiotic drugs including antidiuretic, antidiabetic, [3-blockers, psychotherapeutic, and nonsteroidal anti-inflammatory molecules possess an antimicrobial action, which has generally been regarded as a side effect 6 and therefore neglected for potential clinical use. Benzydamine, lidocaine, and bupivacaine are known nonantibiotic antimicrobials. Topical use of these anesthetic drugs may be useful in the management of cutaneous and vaginal candidosis. We studied the activity and mechanism of action of benzydamine, lidocaine, and bupivacaine against C. albicans and non-albicans strains.

Candida Strains
Twenty Candida strains were used: 18 clinical isolates and two American Type Culture Collection (ATCC) strains ( Table 1). The yeasts were kept at -70C in Brain-Heart broth (Difco Laboratories, Detroit, MI) with 5% glycerol until tested. For each experiment, the strains were subcultured twice on Sabouraud agar (Difco) for 24 hours at 35C and either resuspended in saline (stationary growth phase cells) or subcultured in Sabouraud broth to the middle of the exponential growth phase.

Incubation of Yeast Cells With Drugs
Yeast cells in stationary phase were resuspended in 10 mmol/L sodium N-2-hydroxyethylpiperazine-N-2-ethanesulfonic buffer (HEPES, pH 7.2), supplemented with 2% glucose (GH solution), at a density of x 106-5 x 106 cells/mL, with or without serial concentrations of the drugs (see legends to figures). Incubations were carried out at 35C, with shaking at 200 strokes/min. At the end of the incubation, the cells were centrifuged for 10 minutes at 1800g, and the antifungal activity of the drugs was assayed by viability counts and by staining with the fluorescent probes Propidium Iodide (PI) and FUN-I, as described below.

Yeast Cell Counts
The total number of yeasts in the suspensions was determined in a Neubauer hemocytometer (Agar Scientific Ltd, Stansted, UK). Enumeration of viable yeast cells in the untreated control suspensions and in those exposed to local anesthetics or sodium azide was carried out by counting colonyforming units (CFU) after plating serial dilutions (in saline) of the suspensions on Sabouraud agar plates. The number of colonies was counted after 48 hours of incubation at 35C.

Studies of Membrane Damaging
Two independent procedures were used to assess the capacity of benzydamine, lidocaine, and bupivacaine to damage the fungal cytoplasmic membrane. One relied on the use of the membraneimpermeable fluorescent dye PI. Previous experiments have been carried out to optimize the flow cytometric conditions that were used in the current study (Pina-Vaz et al, in press). That is, optimal results were obtained when using 106 yeast cells/mL, stained with lpg/mL of PI, for 30 minutes in 0.05 mol/L sodium HEPES buffer, pH 7.2, at room temperature in the dark. Incubation with lpg/mL of PI under the above-mentioned conditions had no toxicity to Candida cells (as determined by viability counts) and stained 100% of Candida cells killed by boiling for 30 minutes (Pina-Vaz et al., in press). For each sample, the percentage of PI-positive cells was determined by flow cytometry as described in detail below. From these values, the concentration of the assayed drugs resulting in 50% PI-positive cells was calculated according to a linear regression equation. The PI staining was found to be an adequate indicator of cell death (see "Results"). Therefore, we considered those drug concentrations causing half of the cells to be stained as representing median lethal concentration (LC50).
The second method we used to study cytoplasmic membrane damage estimated the leakage of intracellular K from the yeast cells. Because the intracellular accumulation of K is higher in yeasts in the exponential growth phase, 8 Candida cells grown at 35C in Sabouraud broth supplemented with 0.5% KzHPO4 were harvested at the middle of the exponential growth phase. The yeasts were washed twice with saline and exposed at 35C for 10 minutes to the assayed drugs dissolved in saline at indicated concentrations. After 5 and 10 minutes, treated and control suspensions were filtered through 0.45 pm Millipore filters (Millipore, MA). The filtrates were assayed for K using a K +sensitive glass electrode connected to a Spotlyte analyzer (Menarini Diagnostics). The values are presented as the percentage of K leaked in comparison to that from cells boiled for 30 minutes. 9'1 Viability counts and the percentage of cells stained by PI were also determined in the suspensions used for the K leakage assays. The leakage of K was also analyzed for Candida cells treated with 20 mM sodium azide, or with 2pg/mL of amphotericin B, for 10 minutes.

Assays of Metabolic Vitality
The processing of the fluorescent probe FUN-1 by the yeast cells was used to detect nonlethal metabolic alterations. Two methods to assess FUN-1    yeast cells grown to middle of exponential phase (about x 107 cells/mL). The cells were collected by centrifugation at 1800g for 10 min, washed once with water and once with 1.4 mol/L sorbitol before the pellets were resuspended at a concentration of x 107-5 x 107 cells/mL in 0.04 M HEPES buffer (pH 7.4), 0.5 M MgC1 z, and 0.5% mercaptoethanol (Sigma) in 1.4 M sorbitol. Lyticase was added at a concentration of 10 units/107 cells, and the suspension was incubated at 30C with gentle, occasional shaking. Spheroplast production was monitored by phase contrast microscopy, assessing the lysis of yeast cells exposed to 5% sodium dodecyl sulfate (SDS). Yeasts unexposed to Lyticase did not lyse with SDS. When most Candida cells were converted into spheroplasts, the suspension was can-  Fig. 4. Exponential phase C. albicans, ATCC strain 10231 cells exposed for I0 min to 0.3 mg/mL of benzydamine (Benz), 50 mg/mL of lidocaine (LID), 30 mg/mL of bupivacaine (BUP) or 20 mM sodium azide (Na azide). In the same suspensions, three parameters were determined, i.e., the percentage of nonviable cells (determined by CFU counts) (A), the percentage of PI-positive cells (B), and K efflux (as percent of the total intracellular K/) (C). trifuged at 1800g for 10 minutes and the pellet resuspended in GH medium supplemented with 1.4 M sorbitol with indicated concentrations of the assayed antifungal drugs.
Transmission Electron Microscopy To analyze the ultrastructural alterations induced by the tested drugs, C. albicans blastoconidia and spheroplasts were studied. Blastoconidia of untreated and treated cells of C. albicans strain ATCC 10231, were prefixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2, followed by washing in the same buffer. The cells were then fixed with 1.5 % potassium permanganate in water for hour, 14 followed by washing with water and postfixation with aqueous 1% uranyl acetate for 30 minutes. 1S Samples of control and treated spheroplasts were prefixed with 2.5% glutaraldehyde (1 volume of 25% glutaraldehyde stock solution added to 9 volumes of spheroplast suspensions).
After at least 4 hours at room temperature, the samples were washed with 0.1 M cacodylate buffer, pH 7.2, and fixed overnight at room temperature with 1% OsO4 in 0.1 M acetate-veronal buffer, pH 7.0, supplemented with 10 mM calcium chloride. 16 After washing with water, the cells were postfixed with aqueous 1% uranyl acetate for 30 minutes at room temperature, is Fixed intact cells and spheroplasts were dehydrated in ethanol and embedded in Epon (TAAB Aldermaston, Berks, UK). 7 Ultrathin sections were cut with an LKB Ultratome III microtome (LKB-Produkter AB, Stockolm-Bromma, Sweden) and contrasted with uranyl acetate followed by lead citrate. 16 To improve visualization of ribosomes in the spheroplasts, sections were treated with 3% hydrogen peroxide for 10 minutes before lead citrate staining. 7 Observations and micrographs were done with a Zeiss EM 10C electron microscope (Carl Zeiss, Oberkochen, Germany).

Statistical Analysis
Correlation coefficients (r) were calculated using the Anova program (program Statistica for Windows, Stat Soft, CA).

RESULTS
The three drugs tested inhibited growth of all Candida strains studied, with MICs ranging from 6.25 to 50.0 lag/mL for benzydamine, from 1.25 to 40.0 mg/mL for lidocaine, and from 2.5 to 10 mg/mL for bupivacaine ( Table 1). The correlation coefficients between the MIC of each drug for the 20 strains studied were good when comparing lidocaine with bupivacaine (r 0.905), but not for benzydamine vs. lidocaine (r 0.212) and not for benzydamine vs. bupivacaine (r 0.328). Short exposure of C. albicans strain ATCC 10231 to the MIC of each of the three drugs did not result in any cell death and did not make the cells permeable to PI or induce any significant K leakage (not shown). However, under these conditions, the drugs significantly impaired the vitality of the Candida cells, as was shown by FUN-1 experiments. Thus, yeast cells exposed for 1 hour to the drugs at the MIC did not form CIVS (not shown), while an increase in the intracellular fluorescence was detected by flow cytometry (Fig. 1). Treatment of C. albicans, ATCC strain 10231, with increasing concentrations of the anesthetics resulted in a dose-dependent fungicidal effect ( Fig. 2A, C, and E). Staining with PI showed that the number of PI-positive cells correlated well with the number of nonviable cells (Fig. 2B, D and F).  Table 1). The correlation coefficient between the LCs0 of each drug for the 12 strains tested was r 0.073 when comparing benzydamine with lidocaine, r 0.001 for benzydamine vs. bupivacaine, and r 0.955 for lidocaine vs. bupivacaine.
The concentrations of the three drugs causing an extensive rate of killing (Fig. 4A) and permeability to PI (Fig. 4B) in C. albicans, ATCC 10231, cells induced a quick and extensive leakage of intracellular K (Fig. 4C). A similar pattern of K leakage was seen with the membrane-active antifungal amphotericin B, with 85.7% and 94.1% of intracellular K being lost after treatment of C. albicans ATCC 10231 cells with 2 lag/mL of the antibiotic for 5 and 10 minutes, respectively. Treatment of C. albicans, ATCC strain 10231 cells, with 20 mM sodium azide for 10 minutes resulted in extensive cell death (Fig. 4A) but did not make the cells permeable to PI (Fig. 4B), nor did it induce a significant K leakage (Fig. 4C). Candida glabrata strain H30 was the strain most resistant to the three drugs assayed. When it was treated with high concentrations of lidocaine, no K leakage was seen (not shown).
Transmission electron microscopy showed that exposure of C. albicans, ATCC strain 10231, to cidal concentrations of the anesthetics induced severe alterations of the cytoplasmic organelles (Fig. 5).
Exposure of blastoconidia of C. albicans, ATCC strain 10231, to Lyticase--under the described conditions--resulted in detergent-sensitive spheroplasts, which often had loose cell wall remnants (Fig. 6A). Only occasionally completely wall-free protoplasts were found. Spheroplasts not exposed to the drugs were intact and showed numerous ribosomes and normal intracellular organelles, i.e., the nucleus and the mitochondria (Fig. 6, A and B) and intact cell membranes with a continuous triplelayered profile (Fig. 7A). On the contrary, when  (Fig. 7, B and C). Ultrastructural images of plasma membrane splitting were seen in some drug-exposed yeast cells (Fig. 7C).

DISCUSSION
Local anesthetic drugs possess antibacterial and antifungal properties. 9,18-22 However, their antifungal mechanisms have not yet been elucidated. Using a combination of complementary methodologies, we found the drugs to have a marked antifungal activity on C. albicans that ranged from growth inhibition to complete loss of viability. Growth inhibition can bc detected by conventional protocols for MIC determination, as demonstrated in our study. We found that the fungistatic activity of the anesthetics could also be detected, with advantages, by use of FUN-1. Yeast cells exposed to MIC of the drugs showed both inhibition of conversion of the FUN-1 monomer into CIVS and an increase in diffuse intracellular fluorescence, as demonstrated by flow cytomctry. These alterations were already detectable after an incubation period of hour, thereby allowing for a more rapid assay as compared with the conventional protocol for MIC determination requiring 24 hours.
The following results point to the mechanism of the fungicidal action of the drugs being due to direct damage to the yeast cytoplasmic membrane. Thus, Candida cells exposed to fungicidal concentrations of the drugs quickly become permeable to the membrane-impermeable fluorochrome PI. This probe has previously been used to evaluate membrane permeability in yeasts and is considered a good marker for cell death associated with membrane alterations, e,e4 Quantitation of K leakage from microorganisms, both from bacteria 9,es and yeasts, l,e6-e9 has also been used to evaluate membrane damage by various compounds. Yeast cells in the exponential growth phase are known to accumulate K intracellularly, with concentrations as high as 220 mM. s This cation is quickly lost to the extracellular milieu when the selective membrane permeability is lost. ,e6-e9 A very quick and extensive K leakage was induced by fungicidal concentrations of the local anesthetics, with more than 90% of the intracellular cation being lost during the initial 10 minutes after exposure to 0.3, 50, or 30 mg/mL of benzydamine, lidocaine, or bupivacaine, respectively. This indicates that the fungicidal effect results from a direct damage to the cell membrane, rather than from a metabolic impairment leading to secondary membrane damages. In support of this interpretation is our observation that exposure of the Candida cells to fungicidal concentrations of the metabolic inhibitor sodium azide induced K leakage at a much slower rate as compared to the three local anesthetics. As expected, the membrane-active antimicrobial agent amphotericin B also induced an extensive and rapid K leakage. The extensive loss of intracellular K may not bc the sole factor responsible for the fungicidal activity of the drugs, as the membrane disorganization resulted in multiple perturbations that eventually could be lethal.
Additional support for a direct membrane damaging action of fungicidal concentrations of the anesthetics was evident from the quick lysis of sphe- roplasts in an osmotically protective medium, when exposed to fungicidal concentrations of the drugs. Bacterial protoplasts are quickly lysed by treatment with phenethyl alcohol, tetrazolium salts, or local anesthetics, molecules that are known to directly disorganize bacterial cell membranes. 9'es The ultrastructural studies revealed a second support for the assumption, which was the severe membrane alterations, with fracturing and solubilization, seen after 10 minutes of exposure to fungicidal concentrations of the drugs.
The poor correlation found between the MIC or LC.s0 of bcnzydamine and those of lidocaine and bupivacaine suggests that although the drugs kill Candida by acting on a common target (the cell membrane), the mechanisms for the membrane damage would differ in regards to benzydaminc on the one hand and lidocainc and bupivacaine on the other. Spheroplasts from the same preparation as in Fig. 6A The proposed mechanism for the fungicidal activity of the drugs is in conformity with the lipophilic properties of membrane-active molecules due to their lipophilic character. It is therefore expected that the antifungal activity of bupivacaine was higher than that of lidocaine because the former anesthetic is more lipophilic. The same difference in the activity between the two local anesthetics was also found in a previous study on the inhibition of germ tube formation by C. albicans. 19'22 A good correlation between the lipophilic and antibacterial activity of local anesthetics has also been described. 9,18 These observations agree with the accepted interpretation that anestheticmembrane interaction is of a hydrophobic character, whereby the anesthetic molecule ultimately penetrates the membrane bilayer and accommodates in its hydrophobic interior. 1,3z Membrane splitting resulting from insertion of several lipophilic molecules, including the local anesthetic tetracaine, into the hydrophobic core of bacterial membranes was described. 3 This ultrastructural alteration was seen in the present study in lidocaine-treated Candida plasma membranes (Fig.  7C).
The results showing that the three local an,esthetics assayed possess an antifungal activity but exposed to 20 mg/mL of bupivacaine. Notice the altered ultrastructure of the yeast protoplast, with abnormal mitochondria (M), and with absence of ribosomes and nucleus.
Cell walls are partially digested. Cellular remnants are present in the much enlarged periplasmic space. Section stained with uranyl-lead. (x 35,000). through a membrane-damaging action are in keeping with what has been reported for other lipophilic molecules. This is the case, among others, with amphotericin B, lipophilic azoles, 4 butenafine, "5 and phenothiazines. "6,37 The observed good correlation coefficients between the MIC (indicator of fungistatic activity) and LCso (indicating a fungicidal activity) of each of the drugs for the 12 Candida strains tested would suggest that the mechanisms for the fungistatic and fungicidal activities of the drugs are similar, the plasma membrane being a common target. However, when exposing yeast cells to the fungistatic concentrations of the three drugs, no indication of membrane damage was found, since there was no permeability to PI nor production of K leakage. This may be explained by the short incubation time (10 minutes) used in the incubations prior to the above two membrane integrity tests.
The present results show that the three drugs now studied could be used topically to treat mucosal or cutaneous candidosis because concentrations with antifungal activity can be obtained with the commercial formulations available, mainly with benzydamine, the most active of the three drugs we tested. In fact, benzydamine is available in gels with 3% and 5% of the drug, "8 and these concen- In C, the membrane remnants show zones with increased membrane thickness, indicating membrane splitting (arrow heads). W, partially digested cell wall. Sections stained with uranyl-lead. (x 72,000).
trations are 600 and 1000 times higher, respectively, than the MIC for the least susceptible Candida strain we studied. Benzydamine solutions for oral and vaginal applications contain 0.15%, 38 a concentration that is 30 times higher than the MIC of the least susceptible Candida strain. Commercial formulations of lidocaine and bupivacaine contain drug concentrations 3s that are up to 40 times above the MIC of most strains we studied. It should be emphasized, additionally, that the analgesic properties of the three drugs we studied represent an additional advantage for their topical use in the management of Candida infections.