Candidacidal Activity of Salivary Histatins

Candida albicans is the predominant species of yeast isolated from patients with oral candidiasis, which is frequently a symptom of human immunodeficiency virus infection and is a criterion for staging and progression of AIDS. Salivary histatins (Hsts) are potent in vitro antifungal agents and have great promise as therapeutic agents in humans with oral candidiasis. The molecular mechanisms by which Hsts kill yeast cells are not known. We report here, that unlike other antimicrobial proteins, Hsts do not display lytic activities to lipid membranes, measured by release and dequenching of the fluorescent dye calcein. Analysis of the magnitude and time course of Hst-induced calcein release from C. albicans cells further showed that loss of cell integrity was a secondary effect following cell death, rather than the result of primary disruption of the yeast cell membrane.125I-Hst 5 binding studies indicated that C. albicans expressed a class of saturable binding sites (K D = 1 μm), numbering 8.6 × 105 sites/cell. Both Hst 3 and Hst 4 competed for these binding sites with similar affinities, which is consistent with the micromolar concentration of Hsts required for candidacidal activity. Specific 125I-Hst 5 binding was not detected to C. albicans spheroplasts, which were 14-fold less susceptible to Hst 5 killing, compared with intact cells in candidacidal assays. In overlay experiments, 125I-Hst 5 bound to a 67-kDa protein detected in C. albicans whole cell lysates and crude membrane fractions, but not in the yeast cell wall fraction. Consistent with the overlay data, cross-linking of 125I-Hst 5 toC. albicans resulted in the appearance of a specific 73-kDa125I-Hst 5-containing complex that was not detected in the cell wall. 125I-Hst 5-binding protein of similar size was also observed in susceptible S. cerevisiae strain TI#20. This is the first description of Hst 5 binding sites on C. albicans which mediate cell killing and identification of a 67-kDa yeast Hst 5-binding protein. The binding characteristics of Hst 5 are in agreement with the observed potency of its biological effect and provide crucial information to the use of Hst 5 as a therapeutic agent. The presence of a specific C. albicans Hst 5-binding protein provides further insight into the potential mechanism of yeast killing and suggests a basis for differential activity between yeast killing and the nontoxic nature of Hsts to humans.

Oral candidiasis is a frequent superficial infection in humans associated with mechanical or traumatic factors or with an immunocompromised host (1). Candida albicans is the predominant species of yeast isolated from oral tissues of patients with oral candidiasis (2). Oral candidiasis is commonly associated with HIV 1 infection and is a criterion for the development and progression of AIDS (3). Relatively few antifungal drugs are available for clinical treatment of oral or systemic candidiasis. Increased use of these antifungal agents to treat candidiasis in late stage AIDS and cancer patients has resulted in a dramatic increase in emergence of candidal species with antifungal drug resistance, especially to azole-based drugs.
Growth or adhesion of pathogenic yeasts as well as bacteria in the oral cavity are regulated by glandular secretion of an array of salivary proteins (4,5). Histatins (Hsts) are structurally related basic proteins of acinar cell origin, which possess in vitro candidacidal and candidastatic (6 -8) activities and, to a lesser degree, bactericidal properties (9) and participate in nonimmune host defense of the oral cavity (10). Hsts have potential as therapeutic agents against oral candidiasis, being potent antifungal molecules with no toxicity to humans.
Histatin 1 (Hst 1), histatin 3 (Hst 3), and histatin 5 (Hst 5) are the major Hsts products found in saliva from human submandibular-sublingual and parotid glands (11,12) and are present in healthy adults at concentrations of 50 -425 g/ml (13). Hst 1 and Hst 3 are the full-length precursor molecules, which are cleaved by specific proteolytic processing events during secretion to produce smaller histatins (14 -16). In vitro, Hst 5 is the most potent candidacidal member of the family that kills pathogenic Candida species from 90% to 100% at physiological concentrations (15-30 M) (10,17). We have shown that Hst M, the middle portion of Hst 3 (residues 9 -24), has equivalent candidacidal activity with the full-length molecule (17). Candidacidal activity of this fragment has been confirmed (8), and duplication of these residues (residues 13-24) as a tandem repeat within Hst 3 has produced a peptide with enhanced candidacidal activity at low concentrations (18).
Histatins should be added to the described families of natural antibiotics produced by host tissues in contact with indigenous microorganisms (19). These antimicrobial proteins are produced by a variety of tissues and organisms as a means of nonimmune host defense. Although Hsts have similar size and net positive charge as other naturally occurring antimicrobial peptides, they possess structural features unique from other described low molecular weight cationic proteins. The high histidine content, lack of disulfide bonds, and weak amphipathic character of the ␣-helical structures distinguish Hsts from the known natural antimicrobial proteins such as defensins, bactenecins, and cecropins (19).
The mode of action of a number of cationic host defense polypeptides is related to increased membrane permeability and disruption of microbial cell structure. Studies with artificial lipid bilayers have demonstrated a range of effects including channel formation by defensins (20 -22) and disruption of lipid packing by bactenecins (23), dermaseptin (24), nisin (25), cecropin (26), and tachyplesin (27), which results in permeabilization of the membrane. Lytic properties have also been described for the polyene antimycotics, which are currently used drugs in treatment of candidiasis. Polyene antimycotics form pores in the plasma membrane by complex formation with ergosterol of the plasma membrane (28,29), resulting in loss of potassium and leakage of other cell products. These drugs have additional secondary effects on inhibition of plasma membrane ATPase leading to proton efflux as well as reduction of activity of many biosynthetic enzymes (29). The synthetic azole derivatives are the other major class of antifungal drugs. The azolebased drugs act on yeast mitochondria and through a complex multistep pathway inhibit the biosynthesis of ergosterol, the major sterol of yeast plasma membranes (30,31).
The mechanisms by which salivary Hsts exert candidacidal activity are not well understood. Electron micrographs of C. albicans cells following Hst treatment revealed damage at the cell membrane, as well as the presence of cytoplasmic vacuoles, suggesting loss of intracellular materials (32). In addition, C. albicans cells were found to release potassium in the presence of a fraction of histidine-rich polypeptides purified from human parotid saliva, indicating a loss of yeast membrane integrity (33). Further, Hsts were effective in killing azole-resistant Candida species, thus implicating a different mechanism of Hst action from that of the azole-based antifungal drugs (34,35). Studies using Hst 5 variants with one or two amino acid substitutions showed that His 18 , Lys 13 , Arg 22 , and the dipeptide sequence Phe 14 /His 15 may be important for optimal activity (36). Moreover, Hst 5 variants with reduced killing ability exhibited similar helical content to Hst 5, suggesting that the ␣-helical conformation alone is not solely responsible for optimal candidacidal activity (36).
Although earlier potassium release results suggested that Hsts' biological effects may be related to direct peptide-membrane interactions, recent structural studies of the active fragment of Hst (37) and two-dimensional NMR studies of Hst 5 (38) revealed that the weak amphipathic character of the helical structure precludes spontaneous insertion into the membrane. The size of the active fragment (16 residues) also argues against its ability to function as a channel-forming peptide. Moreover, the yeast cell wall is rich in mannoproteins and glucans containing anionic sites, which may form electrostatic complexes with Hsts and prevent interaction with the cell membrane.
Hsts' potent antifungal activity, lack of toxicity to humans, and ability to kill azole-resistant yeast strains underscore the importance of detailed understanding of their mechanism of action. In this study, we examined whether Hsts can induce direct physicochemical damage to yeast cell membranes (cationic peptide model) or whether binding to a specific yeast component is required for Hst killing by performing biological and fluorescent dye release assays, Hst binding, and crosslinking studies on C. albicans intact cells and spheroplasts.

EXPERIMENTAL PROCEDURES
Materials-C. albicans strain DS1 was isolated from the palate of a denture stomatitis patient (39) and Saccharomyces cerevisiae strains TI#20 and BJ2169 were obtained from Dr. J. Bruenn (Department of Biological Sciences, SUNY, Buffalo, NY). Sabouraud dextrose agar and YPD media were from Difco. Wang resins and Fmoc amino acid derivatives were purchased from Calbiochem (NovaBiochem Co., La Jolla, CA) and Sigma. Na 125 I was purchased from Amersham Pharmacia Biotech. Zymolyase-20T was from ICN Biomedicals (Costa Mesa, CA). Calcein-AM and calcein were obtained from Molecular Probes (Eugene, OR). Egg yolk phosphatidylcholine (ePC) and phosphatidylserine (PS) were obtained from Avanti Polar Lipids (Birmingham, AL); cholesterol and ergosterol were purchased from Sigma. Sephadex G-25 and Sephadex G-10 were from Amersham Pharmacia Biotech; disuccinimidyl suberate (DSS), bis(sulfosuccinimidyl)suberate (BS), and BCA protein reagent were purchased from Pierce, and the electrophoretic reagents were from Bio-Rad.
Purification of Hsts was performed by reversed phase high performance liquid chromatography using a Rainin Dynamax-60A C 18 column and a gradient of acetonitrile and water (both containing 0.1% trifluoroacetic acid) as the solvent system. Purity of Hsts was assessed by amino acid analysis, mass spectroscopy and confirmed by amino acid sequencing using an Applied Biosystems (model 471A) protein sequencer. Bactenecin 5 (Bac 5) was synthesized and purified as described previously (23,40). MUC7-derived synthetic peptides (Table II) were provided by Dr. M. J. Levine (Department of Oral Biology, SUNY, Buffalo, NY).
Yeast Culture Conditions-C. albicans was maintained on Sabouraud dextrose agar plates as described previously (17), and S. cerevisiae was maintained on YPD agar plates. C. albicans was inoculated into 10 ml of sucrose-salts-biotin yeast synthetic medium (SSB) (17,41), and S. cerevisiae was inoculated into 10 ml of YPD medium. Cells were grown to stationary phase for 48 h at 25°C with rotary agitation at 200 rpm. Blastoconidial cell morphology and cell numbers were determined by phase contrast microscopy using a cell counting chamber.
Conversion of Yeast Cells to Spheroplasts-C. albicans cells were grown for 48 h at 25°C in SSB medium and converted to spheroplasts with Zymolyase 20T (ICN Biomedicals, Inc.) as described (42). Briefly, C. albicans cells from 10-ml cultures were harvested by centrifugation at 600 ϫ g, washed twice with TE buffer (10 mM Tris, pH 7.4, 1 mM EDTA), and resuspended in 900 l/g wet weight cells spheroplasting buffer (10 mM Tris, pH 7.4, 1 mM EDTA, and 1 M sorbitol as an osmotic protector). After the addition of 2-mercaptoethanol to a final concentration of 30 mM and 150 units Zymolyase-20T/g wet weight yeast cells, cells were incubated for 1-1.5 h at 30°C with gently shaking. Spheroplast formation was monitored microscopically by lysis of osmotically sensitive cells in 5% SDS; under these conditions, more than 90% of the cells were converted to spheroplasts. Spheroplasts were washed twice and resuspended in buffer containing 1 M sorbitol for further use. In some cases, spheroplasts were solubilized in Laemmli sample buffer and subjected to SDS-PAGE.
Candidacidal Assay-Antifungal activity of Hsts was examined by microdilution plate assay as described previously (17) with the following modifications. Fungicidal assays were performed on stationary C. albicans cells in the presence or absence of Hst 5, Hst 3, Hst 4, Hst M, Hst 4a, or Hst 3 C (0.98 -125 M) or 500 M amylase, cytochrome c, and insulin-␤-chain. Briefly, C. albicans cells were washed twice with 10 mM sodium phosphate buffer (Na 2 HPO 4 /NaH 2 PO 4 ), pH 7.4, and resuspended at 1.8 ϫ 10 5 cells/ml. Cell suspensions (20 l) were mixed with 20 l of 10 mM phosphate buffer, pH 7.4, containing the indicated proteins and incubated for various times at 37°C with shaking. Control cultures were incubated with 20 l of 10 mM phosphate buffer alone. The reaction was stopped by addition of 360 l of yeast nitrogen base (YNB); 40 l of the suspension (360 cells) were spread onto Sabouraud dextrose agar plates and incubated for 48 h at 37°C. In some experiments, candidacidal assays were performed in 10 mM phosphate buffer, pH 7.4, with 1 M sorbitol. In preliminary experiments, Hst 5 candidacidal activity was tested in 10 mM phosphate buffer containing either 1 M sorbitol, 2 mg/ml BSA, 0.6 M KCl, 0.5 M NaCl, or in 40 mM HEPES. Hst 5 activity was unaffected in 10 mM phosphate buffer containing 1 M sorbitol or 2 mg/ml BSA; however, it was completely inhibited in buffers containing high salt concentration or in HEPES buffer and therefore these reagents were not used throughout the experiments. Candidacidal assays were performed in duplicate or triplicate. Loss of viability was calculated as [1 Ϫ (colonies from suspension with peptide/ colonies from suspension with no protein)] ϫ 100. The Hst concentration required to kill 50% of colony-forming unit yeast cells (LD 50 ) was calculated from the dose-response curves. Statistical significance of the results was calculated using Student's t test.
Cell Permeability Assay-Cell permeability was monitored by release of the intracellular dye calcein. Following cellular uptake of calcein acetoxymethyl ester (calcein-AM), intracellular esterases convert the molecule into a nonpermeable acid form of calcein. Intracellularly, high concentration of dye leads to energy exchange between calcein molecules resulting in quenching. Leakage of calcein into the extracellular media was detected by measuring characteristic fluorescence emission spectra. Calcein-AM loading of C. albicans cells was performed on early stationary phase yeast cell cultures. Cells were washed twice with 10 mM phosphate buffer and resuspended at 10 7 cells/ml. C. albicans (10 7 cells) were loaded with 5 l of 1 mM calcein-AM at a final concentration of 5 M for 2 h at room temperature. Cells were washed four times to remove unincorporated dye, and 100 l (10 6 ) of cells were transferred to a quartz microcuvette filled with 300 l of 10 mM phosphate buffer. In some experiments, whole cells were loaded with calcein-AM and subsequently converted to spheroplasts as described above and measurements were performed in 10 mM phosphate buffer containing 1 M sorbitol. Hst 5 (50 or 200 M) was then added to calcein-loaded cells, and the fluorescence intensity of the induced calcein release (I r ) was recorded every 5 min at excitation and emission wavelengths of 485 and 530 nm, respectively, in a Hitachi F-2000 fluorescent spectrophotometer. Experiments were run for 60 or 90 min. The optimal number of cells, incubation time, and calcein-AM concentration were determined by preliminary experiments to reduce potential artifacts from overloading, dye leakage, and toxic effects. Fluorescence intensity corresponding to 100% of potentially available calcein (I int ) was determined by boiling the cells for 10 min. To confirm that all intracellular calcein was counted, calcein-loaded whole cells were converted to spheroplasts, lysed in boiling water, and total fluorescence measured. Total fluorescence released by lysis of osmotically sensitive cells was very similar to that released following boiling of whole cells. Total fluorescent intensity of the cell population following boiling was assumed to be equivalent to total potentially available intracellular calcein (I int ). The fluorescence intensity of calcein released from loaded cells before the addition of peptides was measured as background (I bkg ). Only cell preparations that showed less than 0.1% changes in the background over 10 min were used for further assays. Calcein efflux was calculated as percent fluorescent release ϭ [(I r Ϫ I bkg )/(I int Ϫ I bkg )] ϫ 100.
Preparation of Phospholipid Vesicles-The reverse-phase evaporation method of vesicle formation (43,44) was used to prepare small unilamellar vesicles using ePC or PS in liposome buffer (150 mM NaCl, 10 mM TES, pH 7.5) and 25 mM calcein. For liposomes containing sterols, cholesterol or ergosterol were recrystallized twice from ethanol and were incorporated into ePC and PS preparations at a lipid/sterol ratio of 2:1. Liposomes were extruded through 0.1-m pore polycarbonate membranes using a high pressure extruder to produce uniformly sized small unilamellar vesicles. Liposomes were then separated from unencapsulated calcein by gel chromatography with a Sephadex G-25 column by eluting with liposome buffer. The phospholipid concentration of collected fractions containing liposomes was determined by phosphorus assay (45). The lipid concentration of each liposome preparation was 1-2 mM. Liposome preparations were adjusted to 0.1 mM phospholipid in liposome buffer, and background fluorescence monitored for 30 min to ensure stability of each preparation. Hst 5, Hst M, and Bac 5 were added to the liposome preparations at varying lipid to peptide concentration ratios and fluorescence of the released dye was recorded as described above. Liposomes were then lysed with 0.5% Triton X-100 to establish total fluorescent content of each preparation. Percent fluorescent release was calculated as [(I r Ϫ I bkg )/(I int Ϫ I bkg )] ϫ 100.
Iodination of Histatin 5-Synthetic histatin 5 (125 g) was dissolved in 10 mM sodium phosphate buffer, pH 7.8, and radioiodinated using a chloramine T method as described previously (46). Free iodine was separated from 125 I-Hst 5 on Sephadex G-10 column that had been pre-equilibrated with 1 mg/ml BSA. The specific activity of the radioiodinated Hst 5 was typically around 20 cpm/fmol. The purity of radiolabeled Hst 5 was assessed by 15% SDS-PAGE using a Tris-Tricine electrode buffer and appeared as a single band. Based on the specific radioactivities of Na 125 I (ϳ15.5 mCi of 125 I/g of iodine) and 125 I-Hst 5 (ϳ20 cpm/fmol), the iodotyrosines were estimated to be less than 2% of the tyrosine residues in the final product. Nonradioactive iodination of Hst 5 was performed parallel to the radioactive labeling, except that 900 ng of Na 127 I was used in place of Na 125 I. Fractions collected after Sephadex G10 chromatography were concentrated by lyophilization and dissolved in 10 mM phosphate buffer, pH 7.4. Protein concentration was determined by amino acid analysis. 127 I-Hst 5 dissolved in either 10 mM phosphate buffer, pH 7.4, or 10 mM phosphate buffer with 2 mg/ml BSA was further tested for biological activity on candidacidal bioassay.
Binding Studies-125 I-Hst 5 binding was performed on C. albicans cells that were grown for 48 h in SSB medium. Cells were washed twice with 10 mM sodium phosphate buffer, pH 7.4, and resuspended at 4 ϫ 10 7 cells/ml in binding buffer (10 mM phosphate buffer, pH 7.4, 2 mg/ml BSA). Suspension assays were performed at cell densities of approximately 10 6 cells (25 l) for each point in a final volume of 50 l binding buffer containing 125 I-Hst 5 for 30 min at room temperature. Following centrifugation (5000 ϫ g, 3 min) and two rinses with binding buffer, the ligand-bound cell pellet was counted in a ␥ counter at 55% efficiency. Specific binding was calculated as total minus nonspecific, where nonspecific binding (typically 15-28% of the total ligand bound) was determined in the presence of 100 -1000-fold excess of unlabeled Hst 5. Competitive inhibition binding studies were carried out in binding buffer containing various concentrations (100 nM to 500 M) of Hst 5, Hst 3, or Hst 4; plus 500 M amounts of cytochrome c, insulin-␤-chain, salivary amylase, or MUC7-derived peptides. Optimal cell number, incubation time, and temperature were determined by preliminary experiments. Cells were assayed for each experimental point in triplicate. The competition curves were fitted to the experimental points using the equation for one binding site using curve fit program LIGAND (Munson, NIH) and Kaleidagraph (Synergy System, Reading, PA).
Cell Fractionation-C. albicans cells were grown for 48 h in SSB, harvested by centrifugation, and washed twice with 10 mM phosphate buffer, pH 7.4. Cells (0.7-1.0 g wet weight) were resuspended in 15 ml of cold homogenization buffer (10 mM phosphate buffer, pH 7.4, 1 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, 1 g/ml aprotinin, 1 g/ml pepstatin A, 1 g/ml leupeptin, and 1 g/ml benzamidine) and added to a 50-ml Bead Beater Chamber (Biospec, Bartlesville, OK) containing 15 ml w/v of prechilled 0.5-mm glass beads. Homogenization was carried out in five 1-min bursts at 4°C. The beads were rinsed twice with 5 ml of homogenization buffer, and the lysate was centrifuged for 5 min at 1000 ϫ g at 4°C to remove unbroken cells and cell debris. Crude membrane fraction was separated from the soluble proteins by centrifugation for 1 h at 170,000 ϫ g. The proteins from whole cell lysate (the low speed supernatant from the first centrifugation) were precipitated overnight at Ϫ20°C with six volumes of ice-chilled acetone and centrifuged for 30 min at 14,000 rpm, and the dried pellets were solubilized in Laemmli sample buffer. The solubilized proteins from whole cell lysates or membrane fractions were separated on 7.5% or 10% SDS-PAGE and transferred to nitrocellulose membrane for overlay assay.
Overlay Assays-Nitrocellulose membranes were blocked for 2 h in 10 mM Tris-HCl, pH 7.4, with 1% BSA and then incubated for 4 h with 50 nM 125 I-Hst 5 in binding buffer. Blots were washed with 10 mM Tris-HCl, pH 7.5, containing 0.1% Tween 20, and autoradiography was performed by exposing dried blots at Ϫ80°C for 1-5 days on X-Omat film (Eastman Kodak Co.).
Cross-linking Studies-Washed C. albicans cells were resuspended in binding buffer at 1.8 ϫ 10 9 cells/ml, and 150 l (3 ϫ 10 8 cells) were incubated for 20 min at room temperature with 100 nM 125 I-Hst 5 in a final incubation volume of 200 l. Where indicated, a 1000-fold excess of Hst 5 was added as an unlabeled competitor. Cells were then placed on ice and subsequently incubated for 20 min with 0.9 mM non-cleavable membrane-permeable DSS or membrane-impermeable BS cross-linking reagents (47), followed by three washes with binding buffer. Yeast protein extract was prepared from the 125 I-Hst 5 cross-linked cells by enzymatic digestion of the cell wall and subsequent lysis of the osmotically sensitive wall-free cells. Cross-linked C. albicans cells were resuspended in 200 l of spheroplasting buffer (10 mM Tris, pH 7.4, 1 mM EDTA, 1 M sorbitol, 480 g of Zymolyase-20T, 30 mM 2-mercaptoethanol) and incubated for 1 h at 30°C with gentle shaking. Wall-free cells were separated from the released cell wall material by centrifugation, solubilized in Laemmli sample buffer, boiled, and subjected to SDS-PAGE. Autoradiography was performed by exposing dried gels at Ϫ80°C for 5-15 days.

Histatin-induced Loss of Yeast Cell Viability-
The antifungal activity of the major members of the Hst family has been characterized and active concentrations for candidacidal activity determined (6,11,17). We have confirmed and extended these results by testing the concentration dependence of synthetic Hsts representing various regions of the major Hsts (Table I) (Table I). Hst 4a, which included three additional amino acids at its N terminus (GYK), had equivalent candidacidal activity with Hst 4. Hst C, which represents the eight-amino acid C terminus of Hst 3, did not produce loss of yeast cell viability even at high concentrations (500 M). These data demonstrate that the middle and N-terminal regions of full-length Hsts are required for optimal candidacidal activity and that the C-terminal region may not be essential for yeast killing and may even reduce Hsts' fungicidal activity.
We next tested killing ability of the most potent Hst, Hst 5, on two strains of S. cerevisiae (BJ 2169 and TI#20). Both strains responded to Hst 5, although these yeast cells were less susceptible when compared with C. albicans. Concentrations of Hst 5 (31.25 M) required to kill more than 95% of C. albicans induced only 34% and 67% loss of viability of the BJ 2169 and TI#20 cells, respectively (Fig. 1). Higher concentrations (125 M) of Hst 5 did not further increase cell killing, suggesting that a subpopulation of S. cerevisiae cells was unresponsive to Hst 5.
Previous studies on Hst structure-function have indicated that the charge and weak amphipathicity of the molecule do not favor spontaneous insertion into yeast cells, thus implying that Hsts may require interactions with C. albicans cell surface components to mediate killing (37). We converted C. albicans cells to spheroplasts to test whether cell wall-free cells would still respond to Hst 5. Removal of the cell wall with Zymolase in the presence of 2-mercaptoethanol drastically reduced cell susceptibility to Hst 5. Spheroplasts were 14-fold less susceptible to Hst killing compared with whole cells (LD 50 ϭ 25 M versus 1.8 M, respectively) ( Fig. 1, Table I). These results suggest that either a cell wall component(s) is required for maximal Hst killing, or that spheroplasts are protected from Hst 5 lethal effects due to removal and/or alteration of membrane components during cell wall digestion with Zymolyase and 2-mercaptoethanol.
To test the specificity of Hsts as candidacidal molecules, three previously characterized proteins were selected for use in candidacidal assays. These molecules were salivary amylase, a 56-kDa protein that is found in C. albicans pellicles formed in saliva (18); insulin ␤-chain, a 3.5-kDa protein with size similarity to Hst 5 (3.0 kDa); and hemoprotein cytochrome c (12.4 kDa), a highly basic protein (pI ϭ 10.2) like Hst 5 (pIϾ10). These proteins were tested at 500 M, which is 16-fold greater than the concentration of Hst 5 required for complete killing of C. albicans cells (Table II). Neither salivary amylase nor insulin had any effect on cell viability; however, cytochrome c killed approximately 50% of cells. This is consistent with previous reports that cytochrome c caused release of nucleotide fractions and amino acid pools from Candida utilis leading to cell death (48).
Hsts Induce Loss of Intracellular Dye from Yeast Cells-Preincubation of C. albicans with a highly enriched fraction of histidine-rich polypeptides (HRPs) purified from human parotid saliva has been shown to produce rapid release of potassium from cells, thus implicating damage of the cell membrane as an early event in the HRP-induced killing (33). We examined the ability of synthetic Hsts to induce yeast cell membrane destabilization by measuring the release of the fluorescent dye calcein from yeast cells. C. albicans cells were loaded with calcein-AM, and fluorescent emission of the released free dye in response to Hsts was recorded over a period of 90 min.
Preliminary experiments were carried out to determine dye retention profiles in these cells. In contrast to the previous study (33), which reported 30% release of potassium from untreated cells, less than 1% of spontaneous dye release was detected over a period of 1 h. Evaluation of longer time periods showed a maximum base line of 2% dye released from cells after 6 h of incubation in phosphate buffer. This is in agreement with studies on human fibroblasts, in which calcein was well retained and not exported by multidrug transporters (49). Under conditions of the candidacidal assays employed here and by others (36), the minimum concentration of Hst 5 required for complete killing of cells is 31.25 M. Time course experiments of Hst 5-induced killing of C. albicans showed that 50% of the cells were killed following 10 min of Hst 5 exposure and complete killing was achieved following 90 min of incubation with Hst 5 (Fig. 2). Calcein release from cells following 10 min of incubation with 50 M Hst 5 was about 2%. Approximately 10% of calcein release was detected over a period of 60 min, which corresponded to 86% cell killing in the candidacidal assay; the maximum dye efflux from loaded cells was 12% during the 90-min exposure to Hst 5 (Fig. 2). Incubation of calcein-loaded cells with higher concentration of Hst (200 M) resulted in proportionally increased dye efflux. Hst 3, Hst 4, and Hst M induced dye efflux similar to that detected in response to Hst 5. This effect was not observed in response to Hst C, which also lacked candidacidal activity. Although our dye release results suggest that cell membrane destabilization is a consequence of Hst activity, the time course and magnitude of calcein efflux indicate cell permeabilization and dye efflux are secondary effects associated with cell death rather than the primary cause of cell death. Consistent with findings that Hstinduced killing of C. albicans was significantly reduced when the yeast cell wall was removed, maximum calcein release in response to Hst was also dependent on the cell envelope integrity. C. albicans whole cells were loaded with calcein-AM, converted to spheroplasts, treated with Hst 5, and the release of the free dye recorded. As shown in Fig. 2, no significant release of calcein from spheroplasts was detected in response to Hst 5 over a 60-min period.
Hsts Do Not Induce Leakage from Phospholipid Vesicles-Candidacidal and dye release assays suggested that C. albicans components in the cell envelope are required for optimal Hst activity. To determine whether Hsts can disturb a lipid bilayer that does not contain proteins, model membranes of differing compositions were constructed. The lysis of lipid vesicles was monitored by the dequenching of fluorescence caused by leakage of encapsulated calcein, which is self-quenched inside liposomes. Bac 5, a proline-and arginine-rich polypeptide of bovine neutrophil granules with bactericidal activity against Gramnegative bacteria, was also tested in the dye release assay. We have previously shown that Bac 5 is a potent candidacidal molecule and exhibits lytic effects on negatively charged liposomes (23). Increasing concentrations of Hsts and Bac 5 were mixed with calcein-containing small unilamellar vesicles of varying phospholipid composition. Membrane disruption or perturbation was measured by fluorescent intensity of dye release. Bac 5 caused rapid dye release from ePC liposomes; however, neither Hst 5 nor Hst M elicited significant increase in fluorescence when mixed with ePC liposomes at molar ratios examined (Fig. 3). No release of dye was detected with any Hst when mixed with ePC liposomes containing ergosterol or cholesterol, or PSϩcholesterol. These data point to differing mechanisms of candidacidal action between Hst and Bac 5. Bactenecins have been shown to effect Gram-negative bacteria by increasing the permeability of inner and outer membranes and disrupting membrane potential (50,51). In contrast, salivary Hsts do not interact spontaneously with the model membranes selected, in agreement with previous conformational studies of Hst M (37), and suggest that a protein or other cellular components may be crucial for Hst killing ability.
Analysis of Hsts Binding to C. albicans Cells-To test whether C. albicans express binding sites for Hsts, binding experiments were performed using 125 I-Hst 5. Preliminary results were carried out to determine whether the iodinated Hst 5 retained biological activity. In a candidacidal bioassay, 127 I-Hst 5 in 10 mM phosphate buffer containing 2 mg/ml BSA had similar biological activity as the non-iodinated protein (data TABLE II Specificity of total 125 I-Hst 5 binding and candidacidal activity C. albicans cells were incubated with 100 nM 125 I-Hst 5 in the presence or absence of Hst 5, Hst 3, and Hst 4 (all at 100 M) or 500 M ␣-amylase, insulin-␤-chain, cytochrome c, or MUC7-derived peptides. Total binding is expressed as a percentage of control (average control was 1.9 Ϯ 0.7 pmol/10 6 cells) and values are means Ϯ S.D. from triplicates from five (cytochrome c) and three (Hst 5, Hst 3, Hst 4, ␣-amylase, insulin ␤-chain) and two (MUC7-derived peptides a and b) independent experiments. Candidacidal activity of Hst 5, Hst 3, and Hst 4 (125 M), or ␣-amylase, insulin ␤-chain, and cytochrome c (500 M), was assayed as described for Table I. Candidacidal results represent means Ϯ S.D. from duplicates from 11 (Hst 5), 3 (␣-amylase, insulin ␤-chain, cytochrome c), and 2 (Hst 3, Hst 4) independent experiments.

not shown).
Binding experiments were carried out on C. albicans cells with various concentrations of 125 I-labeled Hst 5. Scatchard analysis of the results indicated a class of saturable binding sites on C. albicans detectable between 800 pM and 2.4 M 125 I-Hst 5 with an average equilibrium dissociation constant (K D ) of 9.5 Ϯ 4.9 ϫ 10 Ϫ7 M (n ϭ 5) and B max of 8.4 Ϯ 3.8 pmol (Fig. 4). Binding assays were routinely performed on 10 6 C. albicans cells, and the detected binding level corresponded to 8.6 Ϯ 3.9 ϫ 10 5 binding sites/cell.
Specificity of 125 I-Hst 5 Binding-Binding specificity was investigated using unlabeled Hst 5 and two related family members possessing candidacidal activities, Hst 3 and Hst 4, to compete with 125 I-Hst 5 for the identified sites on C. albicans. Hst 3 and 4 competed for the same binding sites as 125 I-Hst 5 with affinities (K i ϭ 8.4 Ϯ 3.6 ϫ 10 Ϫ7 M, n ϭ 4 and 1.1 Ϯ 0.28 ϫ 10 Ϫ6 M, n ϭ 2, respectively) which were similar to the affinity of 125 I-Hst 5 (K D ϭ 9.5 Ϯ 4.9 ϫ 10 Ϫ7 M) (Fig. 5). Moreover, the affinity of unlabeled Hst 5 (K i ϭ 7.8 Ϯ 1.4 ϫ 10 Ϫ7 M, n ϭ 2) was very similar to the affinity of 125 I-Hst 5, verifying that radiolabeled Hst 5 retained full binding ability. Although candidacidal assays showed differences in the potency of Hst 5, Hst 3, and Hst 4, data from competition binding experiments indicated these three Hsts have approximately the same affinities to the 125 I-Hst 5 sites on C. albicans. We next tested whether salivary amylase, insulin, and cytochrome c, which were used in candidacidal assays, interact with binding sites on C. albicans (Table II). Even at high concentrations, salivary amylase and insulin did not produce a significant decrease in the total 125 I-Hst 5 binding. In addition, cytochrome c, which induced 50% loss of C. albicans cell viability, did not inhibit 125 I-Hst 5 binding.
Low molecular weight mucin (MUC7) is a prominent salivary glycoprotein, which contains in its protein core an N-terminal domain with 53% sequence homology to Hst 5. 2 Although MUC7 is not a candidacidal molecule, a synthetic 15-amino acid histidine-rich N-terminal fragment of MUC7 has been recently shown to possess candidacidal activity, 2 albeit significantly lower than Hst 5 (LD 50 ϭ 39.1 M versus 1.5 M, respectively, Tables I and II). This Hst 5-like domain of MUC7 and its N-succinyl derivative (LD 50 ϭ 305.3 M, Table II), were tested in competition binding assays. Hst 5-like domain of MUC7 decreased 125 I-Hst 5 binding (38% inhibition of total binding), which is consistent with its reduced candidacidal activity compared with Hst 5. Moreover, the N-succinyl derivative, which is 170-fold less potent than Hst 5, 2 did not produce significant inhibition of 125 I-Hst 5 binding (Table II). The results from the competition binding experiments showed correlation between binding ability and killing activity and confirmed the specificity of 125 I-Hst 5 binding to C. albicans, thus implying that the identified 125 I-Hst 5 binding sites on C. albicans mediate its biological effect.
Because C. albicans spheroplasts were less sensitive to Hst 5 in killing assays, we tested whether 125 I-Hst 5 binding sites were altered in the cell wall-free cells. In contrast to whole cells, specific binding of 125 I-Hst 5 (800 pM to 2.4 M) was not detected and binding was not saturable at the highest concentration of unlabeled Hst 5 used (800 M). It is unlikely that the Small unilamellar vesicles containing calcein were prepared using reverse-phase evaporation procedures. Hst 5, Hst M, or Bac 5 were added to the liposomes at increasing peptide to lipid ratios, and the fluorescent intensity of the release dye (I r ) was recorded at excitation and emission wavelengths of 485 and 530 nm, respectively. Liposome preparation was lysed with 0.5% Triton X-100 to establish total fluorescent content, and percent fluorescent recovery was calculated as described in Fig. 2. spheroplast formation, inasmuch as spheroplasts, although less susceptible, were still killed by Hst 5. It is possible that Zymolase treatment of the cells in the presence of a reducing agent decreased the binding affinity of 125 I-Hst 5 sites or that a high affinity binding component was removed, leaving a large number of low affinity sites that were not detected with the concentrations of 125 I-Hst 5 used in the whole cell binding assay.
Identification of a C. albicans Hst 5-binding Protein-We next examined whether Hst 5 binds to C. albicans proteins using an overlay assay. C. albicans cells were disrupted with glass beads and proteins from whole cell lysates and crude membrane fractions were separated by gel electrophoresis and analyzed on nitrocellulose membranes with radiolabeled Hst 5. The results revealed that 125 I-Hst 5 bound to a C. albicans protein with an apparent molecular mass of 67,000 Ϯ 4000 (n ϭ 4). This protein was detected in C. albicans whole cell lysate and crude membrane fraction and was the only protein consistently observed to bind Hst 5 (Fig. 6, lanes 1 and 2). In some experiments, a similar sized protein was found in the high speed supernatant containing the soluble post-membrane fraction (data not shown). 125 I-Hst 5-binding proteins were not detected from Zymolase C. albicans cell wall digest or lysates prepared from human embryonic kidney epithelial (HEK 293) cells or neutrophil plasma membranes, even in longer exposures (data not shown). Interestingly, when proteins from Hst 5-susceptible yeast S. cerevisiae strain TI#20 were incubated with 125 I-Hst 5 on overlay assay, a doublet with relative molecular masses of 70,000 and 87,000 was observed (Fig. 6, lane  3). The faster migrating component was of similar size and may be related to the C. albicans Hst 5 binding component, or alternatively, may be a degradation fragment of the slower migrating protein in S. cerevisiae.
Cross-linking of iodinated Hst 5 was next employed to study the functional interaction of Hst with C. albicans leading to cell killing. C. albicans cells were incubated with 100 nM 125 I-Hst 5 and proteins were covalently linked to the iodinated factor with a membrane-permeable, non-cleavable cross-linking reagent, DSS. In order to extract yeast cell proteins, cell wall from 125 I-Hst 5-cross-linked cells was removed and separated from the wall-free cells by centrifugation. Proteins from osmotically sensitive cells were then solubilized in SDS-containing buffer and a 125 I-Hst 5-containing complex with a relative molecular mass of 73,000 Ϯ 1000 (n ϭ 2) was resolved after SDS-PAGE and autoradiography (Fig. 7, lane 1). The labeling of this component was specific, as the complex was not detected when cells were incubated with an excess of unlabeled Hst 5 during the binding (Fig. 7, lane 2). Consistent with the overlay data, 125 I-Hst 5 cross-linked proteins were not detected in the cell wall material (Fig. 7, lane 3), thus implying that Hst 5-binding proteins are not likely to be located on the yeast cell wall. Together, these results suggest Hst 5 interacts with at least one C. albicans-binding protein.

DISCUSSION
The family of salivary Hsts consists of structurally related, low molecular weight histidine rich proteins, which are part of the non-immune host defense system of the oral-esophageal area. The finding that Hsts are potent in vitro antifungal agents, while nontoxic to human cells, provided promise for their therapeutic potential in patients with oral candidiasis. Despite the characterization of Hsts at the structural and conformational level (38), at present little is known about their fungicidal mechanism of action and the basis for selectivity between eukaryotic fungal and host mammalian cells. This study represents the first description of Hst 5 binding sites on C. albicans that mediate cell killing, and identifies a 67-kDa yeast Hst 5-binding protein. We report here that Hst-induced killing is initiated upon binding to a candidal receptor rather than spontaneous permeabilization and destabilization of cell membranes based on biological and fluorescent dye release assays as well as Hst 5 binding and cross-linking studies. Hsts binding to a specific component on target yeast cells may be the basis for their selectivity and lack of toxicity to human host cells.
This study used three major salivary Hsts and the middle active region (Hst M), which represent truncated variants of full-length salivary Hst 3. The candidacidal activity of the synthetic peptides, shown here, is consistent with the reported activities of naturally occurring (6,11,15), synthetic (17), and recombinant (18,36) salivary Hsts. Presence of the C-terminal portion of the molecule reduced the fungicidal activity, as demonstrated by the lower activity of Hst 3 and Hst 4 when compared with structurally equivalent Hst 5 and Hst M, which lack the C-terminal region (Table I). Thus, full-length Hst 3 may be a precursor form of Hsts and proteolytic cleavage of the Cterminal region represents a post-translational processing step for activation of the protein. This concept is in agreement with the findings of Perinpanayagam et al. (14) that Arg 25 is a primary cleavage site of Hsts in vivo. This cleavage site resides at the junction of M and C regions defined in the present study, and may have functional significance in proteolytic processing to produce a functional molecule. In contrast, proteolytic cleavage at other sites in the defined M and N-terminal regions may produce a less active Hst. We have shown that C. albicans cells have the ability to degrade 125 I-Hst 5 (45). In this respect, less active or inactive Hst fragments detected in vivo (14) and generated by proteolytic cleavage at sites Lys 5 , Arg 6 , Lys 12 , His 18 , and Arg 22 may represent a microbial protective mechanism against the cytotoxic Hsts.
Many polypeptide antibiotics exert their activity through spontaneous insertion into microbial membranes forming ion channels or pores, thereby disrupting cell structure and integrity. If Hsts function primarily as membrane lytic agents, direct exposure to the plasma membrane of wall-free spheroplasts should enhance Hst candidacidal activity. However, treatment of C. albicans cells with Zymolyase in the presence of a reducing agent to generate wall-free cells resulted in a 14-fold reduction of cellular sensitivity to Hst 5 compared with intact cells (Table I and Fig. 1). Zymolyase-20T preparation contains a mixture of hydrolytic enzymes including ␤-1,3-laminaripentaohyhydrolase, ␤-1,3-glucanase, protease, and mannase (52). Therefore, either the yeast cell wall is necessary for Hst killing, or an enzymatic cleavage and/or alteration of specific yeast membrane component critical for optimal killing accounts for decreased sensitivity of spheroplasts to Hst 5. Similarly, Lactococcus lactis spheroplasts or trypsin-pretreated whole cells were less sensitive to the bacteriocin Lactostrepcin 5, thus implicating a putative membrane receptor mediating Lactostrepcin 5 and Lactococcin A effects (53). Driscoll and colleagues (54) found that lyticase treatment of C. albicans in the presence of 2-mercaptoethanol resulted in spheroplasts equally sensitive to Hst 5 as the intact cells. However, these lyticaseproduced spheroplasts survived a 30-min incubation in hypotonic 10 mM potassium phosphate buffer, which should lyse osmotically sensitive cells. Different methodologies used for generating spheroplasts may explain apparent differences in the sensitivity of Zymolyase-and lyticase-produced spheroplasts to Hst 5 in this study.
A wide variety of naturally occurring peptide antibiotics have been identified and isolated, including defensins and bactenecins from neutrophils, maganins from frog skin, tachyplesins and polyphemusins from hemocytes of horseshoe crabs, and melittin from bee venom (19). Despite structural variations, virtually all carry a net positive charge and their antimicrobial activities are related to the ability to disrupt microbial membranes. Bactenecins, defensins, maganins, and tachyplesins have been shown to adopt amphiphilic ␣-helical, poly-L-proline II, or ␤-sheet structures leading to spontaneous insertion into membrane and disruption of cell structure (23,55). In contrast, structural and conformational analysis of Hst 5 revealed that the weak amphipathic character of the helical structure precludes spontaneous insertion into microbial mem-branes and formation of pores or ion channels across the membrane (37,38). Hsts' inability to disrupt lipid membranes, measured here by the release and dequenching of calcein, further supports these observations. Unlike Bac 5, which was effective in disrupting artificial membranes (Fig. 3) (23), Hst 5 and Hst M do not display membrane lytic activities to any of the membranes selected. Furthermore, analysis of the magnitude and time course of Hst 5-induced calcein release from C. albicans cells (Fig. 2) showed that loss of cell integrity is a secondary effect following cell death rather than the result of primary disruption of the membrane by channel or pore formation. Together, these results suggest a selectivity of Hst fungicidal activity that is exerted through interaction with a specific yeast component(s). An early study utilizing a partially purified fraction of HRPs from human parotid saliva reported 90% release of potassium from C. albicans cells after a 30-min exposure to HRPs (33). Although this work correlated HRPinduced loss of cell viability with loss of cell membrane integrity, our calcein-release experiments utilizing pure synthetic Hst 5 indicate that cell death is not correlated with leakage of the dye, e.g. with Hst 5-induced yeast plasma membrane disruption. The observed difference in the response of C. albicans cells to a mixture of purified HRPs reported previously (33) as compared with pure Hst 5 in the present study is difficult to explain. The use of purified mixtures of HRPs in the earlier studies suggests the possibility of the presence of additional salivary components in the mixture, which could be responsible for potassium leakage from the cells. Moreover, the correlation of candidacidal activity and the leakage of potassium ions from yeast cells using a mixture of Hsts that exhibit variations in their candidacidal potency and multifunctional properties may not be unequivocal. 125 I-Hst 5 equilibrium binding results indicated that C. albicans cells express a class of saturable binding sites for Hst 5, Hst 3, and Hst 4, numbering 8.6 ϫ 10 5 /cell (Fig. 4). The affinity binding constant (K d ϳ1 M) is in agreement with the active concentrations on the candidacidal assay (LD 50 ϭ 1.8 M) (Table I) and is similar to the K d for 125 I-Hst 5 in equilibrium binding studies on Porphyromonas gingivalis (K d ϭ 1.5 M) (56). Although Hst 5 bactericidal activity to P. gingivalis has not been reported, Hst 5 has been shown to be effective in inhibition of coaggregation of these cells with other bacteria by binding to specific sites numbering 3600/cell. Analysis of Hst binding and killing of C. albicans spheroplasts further suggested a correlation between binding and killing ability. Specific 125 I-Hst 5 binding was not detected to spheroplasts, which were 14-fold less susceptible to killing, compared with intact cells (Table I). The appearance of lower affinity and/or a larger number of binding sites for Hst 5 not detected with the concentrations of 125 I-Hst 5 used in the whole cell binding assays could explain our inability to measure specific Hst 5 binding to spheroplasts. Hst 3 and Hst 4 recognized 125 I-Hst 5 sites on C. albicans with very similar affinities (Fig.  5), although they were 5-and 2.5-fold, respectively, less potent than Hst 5 on the candidacidal assays (Table I). Observed differences in candidacidal activities of Hst 3, Hst 4, and Hst 5 may reflect different biological effects including killing, inhibition of germination, or increased proteolysis. We selected three proteins for use in our killing and binding assays to address the specificity of Hst 5 candidacidal and binding activities: ␣-amylase, a salivary protein; insulin-␤-chain with a similar size; and cytochrome c, which has a high pI. Amylase and insulin neither induced cell killing nor competed for Hst 5 binding sites (Table II). Cytochrome c (500 M) did not inhibit 125 I-Hst 5 binding; however, it induced about 50% loss of C. albicans cell viability, thus suggesting a different mechanism for Hst and cytochrome c yeast cell killing. Earlier experiments have shown that cytochrome c exhibits lytic activity to C. utilis, perhaps due to electrostatic interactions with the yeast cell membrane (47). Recent studies have indicated an involvement of cytochrome c in apoptosis of eukaryotic cells by its release from the mitochondrial intermembrane space into the cytosol (57).
Overlay assays demonstrated the existence of a candidal 67-kDa 125 I-Hst 5-binding protein (Fig. 6). This protein was detected in whole cell lysates and crude membrane fractions prepared from glass-bead disrupted C. albicans cells, but not in the yeast cell wall fraction. Moreover, cross-linking experiments revealed the presence of a specific 73-kDa 125 I-Hst 5-containing complex (Fig. 7). This complex may contain one or two molecules of 125 I-Hst 5 covalently linked to the 67-kDa binding protein, detected in the overlay assays. Because of the small size of Hst 5 (ϳ3 kDa), our experiments did not clarify the exact stoichiometry of the binding event. The absence of cross-linked products in the cell wall material further confirmed overlay results that the Hst 5-binding protein is not located on the cell wall. 125 I-Hst 5-binding protein of similar size was also observed in susceptible S. cerevisiae strain TI#20 (Fig. 6), but was not detected in extracts from human kidney epithelial cells and neutrophils, suggesting that the presence of a yeast receptor for Hst may be the basis for its lack of toxicity to human cells. The presence of an additional higher molecular mass binding protein in S. cerevisiae TI#20 cells may account for the differences in the sensitivity of C. albicans and S. cerevisiae TI#20 to Hst 5. Interestingly, preliminary overlay assay data showed that Bac 5 partially competed with labeled Hst 5 for the C. albicansbinding protein. 3 Although Bac 5 differs from Hst 5 in its ability to lyse lipid membranes, it is possible that its killing mechanism may overlap that of Hst 5. Additional experiments are needed to support this assertion.
Although the exact location of the Hst 5-binding protein is yet to be determined, our data show that it is not located within the yeast cell wall. Removal of the cell wall with 2-mercaptoethanol and Zymolyase, a preparation containing a variety of hydrolytic enzymes including a protease, produced cells with decreased Hst 5 sensitivity, suggesting alteration of a cell membrane protein as the source of reduced Hst 5 effect. Preliminary cross-linking experiments that employed membraneimpermeable cross-linking reagent BS to covalently link 125 I-Hst 5 to C. albicans cells detected a 125 I-Hst 5-containing complex of similar size (data not shown) to the one formed with its membrane-permeable analog DSS (Fig. 7). These results, together with the detection of Hst 5-binding protein in the C. albicans crude membrane fraction (Fig. 6, lane 2), provide evidence that Hst 5-binding protein may be located on the plasma membrane.
C. albicans has been shown to express several adhesins that are receptors for fibronectin, fibrinogen, laminin, collagen, and complement receptors for C 3 d and iC 3 b (1, 58 -62). These receptors uniformly have a yeast cell wall location and are associated with pathogenicity of Candida species as related to adherence to epithelial and endothelial tissues. It has been suggested that similarities between these receptors may in fact reflect a single C. albicans surface mannoprotein with a number of binding activities (59). The cell wall location and binding characteristics of these candidal mannoprotein receptors (K d ϳ1 nM, numbering about 5000/cell) differ from the presently characterized Hst 5-binding protein. Hst binding characteristics reported here are more similar to those described for the pore-forming colicins (63). These bacterial toxins kill susceptible Escherichia coli by formation of ion channels in the inner membrane of the target cells via binding to specific outer membrane receptors. The density of OmpF, the receptor for colicin N, has been evaluated at approximately 4 ϫ 10 5 copies/E. coli with a binding affinity K d ϳ1 M (64,65). Following binding, colicins are translocated across the cell envelope by helper proteins, which belong to nutrient uptake systems.
It is currently unclear how association of Hsts with the yeast-binding protein leads to cell death. Hsts may utilize an existing yeast membrane transport system for vital nutrients to gain illicit access into the cell where secondary effects occur. Membrane-bound transport proteins have been described for both C. albicans and S. cerevisiae, which are involved in uptake of amino acids, peptides, and sugar molecules (66,67). Yeasts also express drug efflux pumps which have been divided into two major classes: membrane proteins belonging to the ABC (ATP binding cassette) or MSF (major facilitator superfamily) families, which share structural and presumably functional homology to bacterial components involved in the active transport of a wide range of molecules including oligopeptides, histidine, maltose, ribose, phosphate, vitamin B 12 , and hemolysin) (68,69). Intracellularly, Hsts may effect sites such as mitochondria to inhibit energy production, induce release of cytochrome c, a process characteristic for apoptosis, or inhibit DNA or RNA synthesis. Finally, interaction with the yeast receptor may be required for correct orientation or oligomerization of Hst following insertion into the membrane. In this respect, ␣-hemolysin, a toxin secreted from the human pathogen Staphylococcus aureus binds to the membrane as a monomer, forms a heptamer and then inserts and lyses human platelets, erythrocytes, and endothelial cells (70).
In conclusion, major disadvantages of currently used antifungal drugs are their toxicity and the development of resistant yeast strains (71). In contrast, salivary Hsts are nontoxic to humans and yet potent candidacidal agents even with drugresistant strains. Although the therapeutic potential of Hsts is becoming apparent, particularly against azole-resistant strains, the exact mechanism of cell killing must be elucidated to fully utilize them as therapeutic agents. The identification of a specific Hst 5-binding protein on C. albicans in this study provides insight into the potential mechanism of yeast killing and suggests a basis for the nontoxic nature of Hsts. Finally, the binding characteristics of Hst 5, described here, are in agreement with the observed potency of its biological effect and further provide crucial information to the use of Hst 5 as a therapeutic agent.