Inhibition of Fungal {beta}-1,3-Glucan Synthase and Cell Growth by HM-1 Killer Toxin Single-Chain Anti-Idiotypic Antibodies

Single-chain variable-fragment (scFv) anti-idiotypic antibodiesof an HM-1 killer toxin (HM-1) from the yeast Williopsis saturnusvar. mrakii IFO 0895 have been produced by recombinant DNA technologyfrom the splenic lymphocytes of mice immunized by idiotypicvaccination with a neutralizing monoclonal antibody (nMAb-KT).The fungicidal activity of scFv anti-idiotypic antibodies againstthe isolates of four Candida species was assessed by MIC analysis.scFv antibodies were fungicidal at concentrations of 1.56 to12.5 µg/ml in vitro against four Candida species. ThescFv antibodies exerted a strong candidacidal activity in vitro,with 50% inhibitory concentration (IC₅₀) values ranging from7.3 x 10^–8 to 16.0 x 10^–8 M, and were neutralizedby adsorption with nMAb-KT. Furthermore, all scFv antibodieseffectively inhibited fungal ß-1,3-glucan synthaseactivity in vitro, with IC₅₀ values ranging from 2.0 x 10^–8to 22.7 x 10^–8 M, values which almost coincide with thevalues that are inhibitory to the growth of fungal cells. Bindingassays showed that the scFv antibodies specifically bind tonMAb-KT, and this binding pattern was confirmed by surface plasmonresonance analysis. The binding ability was further demonstratedby the competition observed between scFv antibodies and HM-1to bind nMAb-KT. To the best of our knowledge, this is the firststudy to show that an antifungal anti-idiotypic antibody, inthe form of recombinant scFv, potentially inhibits ß-1,3-glucansynthase activity.

INTRODUCTION

Top

Introduction

Many yeast strains secrete proteins called "killer toxins" toinhibit the growth of other strains of yeast. HM-1 killer toxin(HM-1) is one such protein produced by Williopsis saturnus var.mrakii IFO 0895 (previously known as Hansenula mrakii) and isstrongly cytocidal against Saccharomyces cerevisiae (55, 56).HM-1 is a small protein consisting of 88 amino acids and fivedisulfide bridges, and its three-dimensional structure has beendetermined by using nuclear magnetic resonance analysis (1,56). It affects sensitive yeast cells primarily in the growthstage, but it is not toxic to yeast cells in the resting stageor to mammalian cells (21). The mechanism of cytocidal activityof HM-1 has been studied extensively, and the accumulated dataindicate that HM-1 kills yeast cells by extracellularly inhibitingß-glucan synthase, a transmembrane enzyme participatingin cell wall synthesis of yeasts and fungi (19, 52, 55). Thisinhibition by HM-1 results in the formation of a pore at thedistal tip of the developing bud and the protruding conjugationtube where cell wall synthesis is active, and cells treatedwith HM-1 die by discharging cellular materials from pores becauseof osmotic pressure (21).

The incidence of fungal infections is increasing worldwide becauseof increasing numbers of immunocompromised patients who areof advanced age, have AIDS or cancer, or are undergoing organtransplantation (18, 33). Human fungal pathogens are a highlydivergent group of fungal species. Candida albicans especiallyis a most dangerous pathogenic fungus, causing severe systemicinfections in immunocompromised populations (14). C. albicansis still the species most frequently isolated from patientswith bloodstream infections (58), while for other groups ofpatients non-C. albicans species have surpassed C. albicansas a cause of candidemia. Candida parapsilosis and Candida tropicalisare isolated more frequently than C. albicans in some Europeanand Latin American centers (6). The proportion of C. albicansinfections has decreased, whereas infections due to other species,such as C. parapsilosis, C. tropicalis, and Candida glabrata,have increased. Such an increase in non-C. albicans specieshas also been seen in retrospective reviews of the epidemiologyof candidemia (6, 35). An urgent need to develop new strategiesfor novel antifungal agents exists. ß-Glucan synthasehas been the target of antimycotic drug development to controlpathogenic fungi because it is common to all pathogenic andnonpathogenic fungi for cell wall biosynthesis (4, 46, 13).To inhibit fungal growth, various efficacious antibiotics havebeen developed to interfere with cell wall synthesis by targetingß-1,3-glucan synthase (9, 11, 12, 37, 53). However,no antifungal antibody that can inhibit ß-1,3-glucansynthase activity has ever been reported. A monoclonal antibody(MAb) that neutralizes the yeast killing activity of HM-1 (nMAb-KT)was produced and classified as immunoglobulin G1( ${kappa}$ ) [IgG1( ${kappa}$ )](49, 56). To apply the excellent biochemical properties of HM-1to the development of new antifungal drugs, we tested whetheranti-idiotypic antibodies having the internal image of HM-1can be raised from nMAb-KT and if such anti-idiotypic antibodiesinhibit ß-1,3-glucan synthase and the growth of yeastsand pathogenic fungi.

Anti-idiotypic antibodies can compete with external antigensfor the binding sites of specific antibodies by mimicking thestructures of the relative epitopes (38). Immunoglobulin variabledomains of heavy chains (V_H) and light chains (V_L) that formthe antigen-binding sites are expressed either as heterodimericFab fragments or as monomeric single-chain variable-fragment(scFv) regions in which the V_H and V_L domains are connectedby a short peptide linker (54). This simplified structure makesthem useful in the assessment and development of immunotherapeuticand immunodiagnostic applications (16, 30). In this context,anti-idiotypic antibodies with antifungal activity have previouslybeen reported for a killer toxin (KT) from Pichia anomala ATCC96603 (23, 44) which is apparently different from HM-1, basedon its large molecular mass (115 kDa) and also on its specificinteraction with cell wall receptors mainly consisting of ß-glucans(15). P. anomala KT is strongly cytocidal against Candida albicans(39). On the other hand, HM-1 has a strong cytocidal effecton S. cerevisiae and is also capable of inhibiting the growthof various Candida species. C. albicans is less sensitive toHM-1 (MIC, 300 µg/ml), while non-C. albicans species suchas C. glabrata (MIC, 0.4 µg/ml) and C. parapsilosis (MIC,12.5 µg/ml) are more sensitive to HM-1 (57). Thus, killertoxins differ between species or strains and demonstrate diversityin biochemical structure, immunity, and mechanism of killingsensitive cells (24, 27).

In this study, we used the nMAb-KT and phage display technologyto produce scFv anti-idiotypic antibodies specific to nMAb-KT.We generated four scFv anti-idiotypic antibodies and found thatthey inhibit ß-1,3-glucan synthase activity, resultingin a strong cytocidal effect on the growth of S. cerevisiaeand the pathogenic isolates of four Candida species, namely,C. albicans, C. tropicalis, C. parapsilosis, and C. glabrata.This study also shows that anti-idiotypic antibody medicinesbased on natural fungicidal proteins have a potential to bedeveloped as new antimycotic drugs.

MATERIALS AND METHODS

Top

Materials and Methods

Materials. HM-1 was extracted from the culture broth of the yeast W. saturnusvar. mrakii IFO 0895 and purified as described previously (52).C. albicans (ATCC 10231) and S. cerevisiae (A451) were obtainedfrom Japan Roche Research Center. C. albicans (NBRC 0197, NBRC0759, NBRC 1389, NBRC 1390, NBRC 1392, NBRC 1397, NBRC 1856,and NBRC 1974), C. tropicalis (NBRC 1400), C. parapsilosis (NBRC1396), and C. glabrata (NBRC 0622) were purchased from NITEBiological Resource Center. C. albicans (IFM 40213) was a generousgift from Koji Yokoyama, Research Center for Pathogenic Fungiand Microbial Toxicoses, Chiba University, Japan. Saccharomycesbayanus (AKU 4103) was obtained from Kyoto University. The hybridomaclone that produces mouse nMAb-KT was a gift from Tadashi Mayumi,Jichi Medical School, whose group initially reported HM-1-specificnMAb-KT (56). The purified nMAb-KT was prepared at TechnologyIncubation & Transfer Ltd. (Saitama, Japan). Mouse monoclonalantibodies 1F1 and 4A2, which bind to the HM-1 molecule buthave no neutralizing activity, were prepared and classifiedas IgG1( ${kappa}$ ) (22). The purified anti-human c-myc MAb and the antihistidinemouse MAb were purchased from Chemicon International (Temecula,CA) and Calbiochem (San Diego, CA), respectively.

Preparation of anti-idiotypic antibodies. The scFv anti-idiotypic antibodies used in this study were producedaccording to a procedure using the recombinant phage antibodysystem (Amersham Biosciences, Piscataway, NJ). Briefly, 50 µgof nMAb-KT was used to immunize female BALB/c mice (3 weeksold; 10 to 12 g) subcutaneously and intraperitoneally (boosterinjection). Three days after the final injection, the mice werekilled and their spleens were removed. mRNA was isolated fromthe splenic lymphocytes and reverse transcribed with randomhexamer primers. The genes encoding antibody variable regionsof heavy and light chains were amplified and assembled intoa single gene using a linker fragment and cloned into a specificphagemid vector, pCANTAB 5E. Recombinant phages produced intransformed Escherichia coli TG1 were repeatedly panned againstnMAb-KT and screened using a conventional enzyme-linked immunosorbentassay (ELISA) with the nMAb-KT. The nonsuppressor E. coli HB2151strain was infected with the selected recombinant phages toproduce soluble recombinant scFv antibodies that were purifiedby using affinity chromatography with an anti-E tag Sepharosecolumn (Amersham Biosciences). The protein concentration ofthe purified scFv antibodies was measured using the bicinchoninicacid method. The yield of the purified scFv antibodies rangedbetween 1.0 and 3.0 mg/liter. Four selected antibodies assayedby sodium dodecyl sulfate-polyacrylamide gel electrophoresisgave a single band corresponding to a molecular mass of around30 kDa, which agreed well with the calculated masses based onthe amino acid sequences of the antibodies. Surface plasmonresonance (SPR) analysis confirmed that purified scFv antibodiesbound to nMAb-KT but did not bind to the 1F1 or 4A2 MAb. Usingthe ELISA, it was confirmed that scFv antibodies specificallybound to nMAb-KT, whereas scFv antibodies did not bind to MAb1F1, MAb 4A2, or the irrelevant isotype-matched anti-human c-mycand antihistidine mouse MAbs. A control scFv antibody clonewhich did not bind to any of the MAbs was selected before panningand produced as described above.

MIC determination. MICs were determined by the standardized protocol for yeastsdeveloped by the National Committee for Clinical LaboratoryStandards (34). Briefly, cells of the Candida species were suspendedin sterile normal saline and diluted to a concentration of 5x 10⁵ cells/ml. The suspensions were diluted 1:1,000 in RPMI1640 medium with L-glutamine, without bicarbonate, which hadbeen buffered to pH 7.0 with 20 mM HEPES (Sigma, St. Louis,MO). Tubes containing 0.1-ml aliquots of scFv antibodies at10 times the final drug concentration were inoculated with 0.9ml of the diluted suspensions. The tubes were incubated at 30°Cfor 72 h with shaking. The MIC endpoints were read visuallyas the lowest concentration at which there was an absence ofgrowth.

In vitro antifungal activity by scFv antibodies. The antifungal activity of the purified scFv clones was testedin vitro by the CFU assay described previously (23). Approximately5 x 10² cells of each of the four Candida species were suspendedin 10 µl of phosphate-buffered saline, and 90 µlof various concentrations of purified scFv antibodies were added.An scFv antibody that did not bind to nMAb-KT was used as acontrol. In further control experiments, the same number offungal cells was also added to scFv antibodies that were previouslyadsorbed with 20 µl nMAb-KT. After overnight incubationat 37°C with the respective reagents to allow cell replication,fungal cells were dispensed into Sabouraud dextrose agar petridishes (three plates for each experiment). The plates were thenincubated at 30°C and observed after 48 h for fungal CFUenumeration. The 50% inhibitory concentration (IC₅₀) valueswere measured using semilogarithmic graphs.

Preparation of membrane fraction. Membrane fractions of Candida species were prepared using amethod described by Cabib and Kang (3) with some modifications.Cells in the mid-exponential phase were collected by centrifugation,washed with 1 mM EDTA, and then suspended in breaking bufferconsisting of 50 mM Tris-HCl (pH 7.5), 0.5 M NaCl, 1 mM EDTA,and 1 mM phenylmethylsulfonyl fluoride. Cells were disruptedby vortexing them with glass beads and were centrifuged for5 min at 1,000 x g at 4°C. The supernatant was centrifugedat 100,000 x g for 30 min at 4°C. The membrane fractionthus obtained was homogenized in a membrane buffer consistingof 50 mM Tris-HCl (pH 7.5), 10 mM EDTA, 1 mM ß-mercaptoethanol,and 33% glycerol and was stored at –80°C.

Measurement of ß-1,3-glucan synthase activity and IC₅₀ values. The ß-1,3-glucan synthase assay was performed usingthe method described by Cabib and Kang (3). The reaction mixturesconsisted of 5 mM UDP-D-[U-¹⁴C]glucose, 75 mM Tris-HCl (pH 7.5),0.75% bovine serum albumin, 25 mM KF, 0.75 mM EDTA, 20 µMguanosine 5'-[ ${gamma}$ -thio]triphosphate, and a 10-µl membranefraction of either one of the four Candida species or S. cerevisiaein a total volume of 40 µl. The reaction was started byadding the membrane fraction, and the mixture was incubatedat 30°C for 60 min. The scFv antibodies and the controlscFv were added to the reaction mixture, and in a positive controlexperiment, 10 and 25 µg/ml of purified HM-1 were addedto the reaction mixture of S. cerevisiae and C. albicans, respectively.The reaction was stopped by adding 250 µl cold 10% trichloroaceticacid, and the mixture was filtered through glass microfiberfilters. The filters were washed with 10% trichloroacetic acid,followed by 95% ethanol. The radioactivities retained on thefilters were determined using a liquid scintillation counter.To measure the IC₅₀s of scFv antibodies against ß-1,3-glucansynthase activity, various concentrations of scFv antibodieswere added to the reaction mixture and the IC₅₀ values weremeasured using semilogarithmic graphs. Each experiment was donethree times.

Pore formation by scFv antibodies. Pore formation was determined by observing the morphologicalchanges of cells treated with scFv antibodies or HM-1, usingphase-contrast microscopy as previously described (21). S. bayanuscells (4 x 10⁶ cells/ml) were incubated with 4 µg/ml ofHM-1 or 25 µg/ml of scFv anti-idiotypic antibodies inYPD (yeast extract-peptone-dextrose) medium containing 0.8 Msorbitol, and the mixture was shaken at 175 rpm at 30°Cfor 3 h. The cells were stained with 0.1% methylene blue in0.8 M sorbitol and were photographed under a phase-contrastmicroscope using immersion oil.

SPR analysis. The kinetics of scFvs binding to immobilized nMAb-KT were measuredusing SPR analysis with a Biacore X (Biacore AB, Uppsala, Sweden).The nMAb-KT or MAb 1F1 or 4A2 was diluted in 10 mM sodium acetate(pH 5.0) at a concentration of 35 µg/ml and immobilizedon a CM5 sensor chip by use of an amine coupling kit, and theunreacted moieties of the surface were blocked with ethanolamine.One channel of each sensor chip, prepared in the same way butwithout monoclonal antibody, was used to monitor the nonspecificbinding of scFvs. All measurements were done with HBS-EP bufferconsisting of 10 mM HEPES (pH 7.4), 150 mM NaCl, 3.4 mM EDTA,and 0.005% surfactant P20 at a flow rate of 10 µl/minat 25°C. After each measurement, the chip surface was regeneratedwith 10 µl of 1 mM HCl. The binding of HM-1 and scFv antibodieswas analyzed at concentrations of 15.8 to 250 nM. The equilibriumdissociation constant (K_D) was evaluated from the kinetic sensorgramcurves using BIAevaluation software.

Competitive binding of scFv antibodies with HM-1 to nMAb-KT. ELISA plates were coated with 10 µg/ml of nMAb-KT andblocked with 3% dry milk in phosphate-buffered saline for 1h at 37°C. The scFv anti-idiotypic antibodies were appliedin ratios of 1:5 to 1:20, followed by the addition of HM-1 (50ng/ml). Horseradish peroxidase-conjugated anti-E tag antibodywas to detect scFv antibodies. Anti-HM-1 rabbit serum and horseradishperoxidase-conjugated anti-rabbit IgG goat antibody were usedto detect HM-1. Using 0.022% of 2',2'-azino-bis(3-ethylbenzthiazoline-6-sulfonicacid)-diammonium salt in citric acid, the absorption was measuredat 405 nm.

RESULTS

Top

Results

In vitro antifungal activity of scFv antibodies against Candida species. The antifungal activity of recombinant scFv antibodies was examinedfor pathogenic strains of four Candida species, namely, C. albicans,C. tropicalis, C. parapsilosis, and C. glabrata. First, thein vitro susceptibilities of Candida species to scFv antibodieswere examined; the results are summarized in Table 1. The MICsof scFv antibodies ranged from 1.56 to 12.5 µg/ml. Amongthe scFv antibodies, scFv-A2 was found to be the most activeantifungal agent for the isolates of all the Candida species.Next, to measure the potency of scFv antibodies against Candidaspecies, the numbers of CFUs were determined after incubationwith increasing concentrations of scFv antibodies. Figure 1shows the killing and neutralization assay of a representativeCandida isolate after overnight incubation with scFv antibodyand appropriate controls. Overall, the IC₅₀ values of scFv antibodiesranged from 2.2 to 4.8 µg/ml (7.3 x 10^–8 to 16.0x 10^–8 M) for all Candida species tested (Table 2), indicatingthe potent antifungal activity of scFv antibodies. Preincubationwith nMAb-KT eliminated the inhibition of Candida growth byscFv antibodies, suggesting that the observed cytocidal effectof scFv antibodies was due to their structural resemblance toHM-1 (Fig. 1C).

View this table:
[in this window]
[in a new window]
TABLE 1. MICs for in vitro susceptibilities of four Candida species to recombinant scFv anti-idiotypic antibodies^a

View larger version (118K):
[in this window]
[in a new window]
FIG. 1. Candidacidal activity of an scFv antibody. The effect of scFv-A2 on the growth of C. albicans ATCC 10231 cells was measured by a CFU assay as described in Materials and Methods. (A) Fungal cells treated with the control scFv antibody; (B) fungal cells treated with 4 µg/ml of scFv-A2; (C) fungal cells treated with scFv-A2 neutralized with nMAb-KT. One of three plates used for each sample is shown.

View this table:
[in this window]
[in a new window]
TABLE 2. IC₅₀ values for recombinant scFv anti-idiotypic antibodies against the cell growth of four Candida species^a

Inhibition of yeasts and fungal ß-1,3-glucan synthase by scFv antibodies. HM-1 strongly inhibits ß-1,3-glucan synthase activityin vitro and kills sensitive yeast cells (21, 19, 52, 55). Toexamine whether scFvs obtained from nMAb-KT affect ß-1,3-glucansynthesis of yeasts and other fungal cells, we used an in vitroreaction system consisting of the ß-1,3-glucan synthaseof the membrane fractions of growing S. cerevisiae and C. albicans.All scFvs inhibited ß-1,3-glucan synthase in S. cerevisiaeand C. albicans by 75% at a concentration of 25 µg/ml,whereas the control scFv antibody did not (Fig. 2A and B). Underthe same conditions, HM-1, included in this study as a positivecontrol, showed almost the same levels of inhibition of ß-1,3-glucansynthase activity.

View larger version (16K):
[in this window]
[in a new window]
FIG. 2. Inhibition of yeast and fungal ß-1,3-glucan synthase activity by purified scFvs. The membrane fractions were obtained from S. cerevisiae A451 (A) and C. albicans ATCC 10231 (B). The scFvs were added to the reaction mixture at a concentration of 25 µg/ml, as described in Materials and Methods. The activity was expressed in comparison to the activity of the sample without scFv and HM-1 as 100%. Data represent the mean of results for triplicate experiments ± standard error of the mean.

Table 3 shows the IC₅₀ values of scFv anti-idiotypic antibodiesagainst ß-1,3-glucan synthase in four Candida species.The IC₅₀ values of the scFv anti-idiotypic antibodies rangedfrom 0.62 to 6.8 µg/ml (2.0 x 10^–8 to 22.7 x 10^–8M) and corresponded well with the IC₅₀ values obtained fromthe CFU assay.

View this table:
[in this window]
[in a new window]
TABLE 3. IC₅₀ values of recombinant scFv anti-idiotypic antibodies against the ß-1,3-glucan synthase activity of four Candida species^a

Mechanism of antiyeast effect of scFv antibodies. HM-1 inhibits the growth of yeast cells by forming a pore atthe growing tip of the daughter cell, resulting in the formationof a protruding structure and eventual cell death (21). To examinewhether scFv also forms protruding structures in growing cells,we added purified scFv antibodies to an S. bayanus cell cultureand analyzed the change in morphology of the yeast cells. Themicroscopic study showed that most cultured cells treated withscFvs (3 h) had a pearlike structure with protruding materials,characteristic of pore formation and similar to the morphologychange after treatment with HM-1 (Fig. 3B and C). This morphologicalchange was caused by all four scFvs tested and was clearly distinguishedfrom the smooth, round shape of untreated control cells (Fig.3A). These data clearly indicate that the scFvs appear to havethe same effect as HM-1 on sensitive yeast cells.

View larger version (65K):
[in this window]
[in a new window]
FIG. 3. Pore formation in yeast cells by scFv antibodies shown by phase-contrast microscopy of S. bayanus AKU 4103 cells treated with scFv-A2 and HM-1. The sample preparation is described in Materials and Methods. (A) Control yeast cells; (B) HM-1-treated cells; (C) scFv-A2-treated cells. Bar, 5 µm.

Kinetic parameters of selected scFv antibodies. The binding specificities and kinetic parameters of purifiedscFv antibodies were determined by using SPR analysis. The Biacoresensorgram curve shows the interactions of scFv-A1, scFv-A2,scFv-A3, and scFv-A4 at different concentrations with immobilizednMAb-KT. Table 4 shows the calculated values of the associationrate constant (k_on), dissociation rate constant (k_off), andequilibrium dissociation constant (K_D), indicating that theinteraction of scFv-A2 with nMAb-KT was the most effective ofthe four scFvs. Overall, the results of SPR analysis showedthat all four scFvs have kinetic affinities similar to thatof nMAb-KT, despite differences in their amino acid sequences.None of the four scFvs showed a signal curve with immobilizedMAb 1F1 or 4A2 (data not shown). These results indicate thatthe scFv antibodies bind specifically to nMAb-KT but not toMAb 1F1 or 4A2, which binds to HM-1 but is unable to neutralizeHM-1 antifungal activity. HM-1, included in this study as apositive control, showed the greatest affinity (K_D = 5.48 x10^–9 M) for immobilized nMAb-KT.

View this table:
[in this window]
[in a new window]
TABLE 4. Kinetic parameters for the binding of four scFv antibodies to nMAb-KT as measured by SPR analysis^a

Competitive binding of scFv antibodies with HM-1 to nMAb-KT. To verify whether scFv antibodies compete with HM-1 to bindnMAb-KT, a competitive binding ELISA was performed (Fig. 4).When the scFv antibody concentration was increased, the bindingof scFv antibody to nMAb-KT was also increased, but the HM-1binding to nMAb-KT was concomitantly decreased. In contrast,the control scFv antibody did not affect the binding of HM-1to nMAb-KT. These results suggest that all four scFv antibodyclones were homologous in competition with HM-1 for bindingto nMAb-KT.

View larger version (19K):
[in this window]
[in a new window]
FIG. 4. Competitive binding of scFv antibodies with HM-1 to nMAb-KT. ELISA plates coated with nMAb-KT were incubated with increasing concentrations of the selected scFv antibodies and the control scFv antibody as described in Materials and Methods. The symbols represent the respective binding of scFv antibodies (solid symbols) and HM-1 (open symbols) with nMAb-KT. •, scFv-A1; ${blacktriangleup}$ , scFv-A2; ${blacksquare}$ , scFv-A3; ${diamondsuit}$ , scFv-A4; ${blacktriangledown}$ , control scFv antibody. The maximum binding activities (A405nm, absorption at 405 nm) of scFv-A1, -A2, -A3, and -A4 were 0.928, 0.968, 0.891 and 0.852, respectively.

DISCUSSION

Top

Discussion

Jerne proposed a biological significance for idiotypic determinantsin the network theory of immune regulation (17). Idiotypes areepitopes that are unique to an antibody, associated with theheavy and light chains of the antibody in which the participationby variable regions that is usually needed for immune activityoccurs (10, 38). Anti-idiotypic antibodies representing theinternal image of some antigenic determinants have been proposedas surrogate vaccines (36), and antibodies conjugated with toxinshave been proposed in the immunotherapy of cancer (30). In somecases, amino acid sequence homology between the protein antigenand the anti-idiotypic antibody variable region exists (2).Antigenic mimicry is usually more functional than biochemical,and the functional mimicry of ligands of biological receptorsby anti-idiotypic antibodies has been extensively studied (10).

The ability of the phage display system to express antibodyfragments offers several important advantages over hybridomatechnology in identifying functional immunoglobulin domains(5, 54). Previously, anti-idiotypic antibodies with antimicrobialactivity have been made with the neutralizing antibodies ofKT from P. anomala ATCC 96603 (23, 44). Each killer toxin hasunique properties, different targets, and an intrinsic strategyto attack the host organism, and these profiles also dependon the species and strains (20, 31). HM-1 inhibits ß-1,3-glucansynthesis but does not inhibit the synthesis of protein, chitin,or mannan in S. cerevisiae protoplast cells (55). ß-1,3-Glucansynthase from the S. cerevisiae membrane is inhibited in vitroby HM-1 at a concentration that coincides well with the cellgrowth inhibitory concentration, implying that inhibition ofß-1,3-glucan synthase is responsible for the cytocidalactivity of HM-1 (52). This inhibition is partly reduced byadding ß-1,3-glucan fragments, suggesting that HM-1binds both ß-1,3-glucan and its synthase (19). HM-1kills only growing cells and does not affect cells in the restingstage (19, 21). These excellent biochemical properties of HM-1are important for antifungal drug development and encouragedus to generate recombinant immunoglobulin that has an internalimage of the HM-1 active site responsible for ß-1,3-glucansynthase inhibition and antifungal activity.

To generate recombinant single-chain antibodies, we used mRNAfrom the splenocytes of mice immunized by parenteral idiotypicvaccination with an nMAb-KT and a phagemid that enables expressionof scFv recombinant antibodies as fusion proteins with phagegene III protein. This allowed the selection of phages thatdisplay anti-idiotypic antibodies and the production of solublescFv anti-idiotypic antibodies. We selected four clones, scFv-A1,scFv-A2, scFv-A3, and scFv-A4, based on the strength of theirbinding activity to nMAb-KT and on their nucleotide sequences.The binding properties of scFv antibodies demonstrated by SPRanalysis indicated that scFv antibodies react to nMAb-KT butnot to MAb 1F1 or 4A2.

The wide spectrum of the antifungal activity of scFv antibodiesagainst the Candida species was investigated because Candidaspecies are pathogenic to humans. C. albicans especially hasbeen the species most often associated with neonatal infections.Recent reports, however, have suggested an increasing numberof infections attributable to C. parapsilosis associated witha common source (47). In terms of Candida species, there hasrecently been a shift towards more reports of non-C. albicansinfections by some authors, especially with patients with hematologicaldiseases and transplant patients (35, 45). In this study, wehave determined the in vitro antifungal activity of scFv antibodiesagainst pathogenic isolates of four Candida species. When theantifungal activity was evaluated by the CFU assay, the scFvantibodies showed strong antifungal activity against C. albicans,C. tropicalis, C. parapsilosis, and C. glabrata in vitro, withIC₅₀ values ranging from 7.3 x 10^–8 to 16.0 x 10^–8M. The fungal susceptibilities were also determined by MICs,and the scFv antibodies were found to be fungicidal at concentrationsof 1.56 to 12.5 µg/ml in vitro against the Candida isolates.

All four scFv antibodies inhibited in vitro glucan synthesiscatalyzed by ß-1,3-glucan synthase in the membraneof S. cerevisiae as well as that of C. albicans (Fig. 2A and B).They also inhibited ß-1,3-glucan synthase in fourCandida species in vitro, with IC₅₀ values ranging from 2.0x 10^–8 to 22.7 x 10^–8 M. The good correlation ofIC₅₀ values in the inhibition of glucan synthesis and of cellgrowth support the speculation that scFv antibodies and HM-1share a common target molecule(s) that is most likely to bethe ß-1,3-glucan synthase on the cell surface. Indeed,the binding of scFv antibodies to nMAb-KT was competitive withthat of HM-1 (Fig. 4), and nMAb-KT that neutralized the killingactivity of HM-1 eliminated the cytocidal activity of scFv (Fig.1). In our established colorimetric ß-1,3-glucan synthaseassay system, HM-1 and scFv antibodies inhibit the enzyme reactionin a noncompetitive manner (unpublished results). These resultssuggest that the killing action of scFvs is the same as thatof HM-1, which distorts the growing end of the cell wall (Fig.3). The echinocandin analogues such as caspofungin, micafungin,and anidulafungin antifungals are the clinically useful antibioticsthat inhibit fungal growth through the inhibition of ß-1,3-glucansynthase (9, 11, 12, 37, 53). Although echinocandins are reportedto be active against several medically important fungi, theyare relatively inactive against Cryptococcus species (7, 8,51). In this context, it should be noted that in addition toinhibiting Candida species, HM-1-derived scFv antibodies arealso able to inhibit the cell growth and ß-1,3-glucansynthase of Cryptococcus species (50).

The cloning system used in this study was based on a heterogeneouspopulation of lymphocytes that does not necessarily ensure thatthe original pairing of the heavy and light chains will be conserved.Most pairings with binding activity are likely to be fortuitous.Nevertheless, the scFv clones were selected solely because oftheir high binding affinity to the idiotype nMAb-KT. We expectedthat the functional mimicking of HM-1 (i.e., ß-1,3-glucansynthase inhibition, pore formation in the growing yeast, andthe consequent antifungal activity) might reflect amino acidsequence similarity to HM-1. To our surprise, however, the selectedscFv antibodies share no apparent homology with HM-1 in theiramino acid sequences. Perhaps the three-dimensional structure,particularly that formed by the domain of V_H shared by all fourscFvs, is most responsible for the functional mimicry of HM-1by the scFv antibodies.

Molecules able to selectively interact with microbial cell wall-relatedcomponents that are not present in mammalian cells should rationallybe considered as antimicrobial agents (48). From this pointof view, some other promising candidacidal antibodies have beendescribed. Magliani et al. (23, 25, 26) and Polonelli et al.(40-42, 44) demonstrated that several models of antibodies againstthe killer toxin of P. anomala were therapeutically active inmurine models of invasive candidiasis. Moragues et al. (32)described how a MAb (MAb C7) directed against a protein epitopeof a cell wall stress mannoprotein expressed in different agentsexerted a direct in vitro candidacidal activity. Matthews etal. (28) reported that different antibodies against fungal heatshock protein 90 (HSP90) were also therapeutically active inmouse models. The preclinical assessment of a humanized recombinantantibody, Mycograb, against an epitope of HSP90 showed cytocidalactivity against a wide range of yeast species. Mycograb isintrinsically fungicidal in vitro, with an MIC of 128 to 256µg/ml and a mechanism of action that participates in theinhibition of HSP90 (29). In this study, we showed that scFvanti-idiotypic antibodies have potential candidacidal activityin vitro through inhibition of ß-1,3-glucan synthase.We believe that the recombinant scFv antibodies obtained inthis study are also excellent candidates for antimycotic drugsand should support the concept of a family of fungicidal antibodiessuch as the natural monoclonal as well as recombinant killerantibodies of P. anomala KT (23, 43, 44).

On the basis of our recent observations, peptides reproducingthe complementarity-determining regions of both heavy and lightchains of scFv antibodies were synthesized and their effectivenessfor candidacidal activity in vitro was tested (unpublished observations).There is a possibility that scFv antibody-derived peptides couldprovide a unique approach for the development of a new classof antibiotic, which would be deliverable directly to the mucosalsite. They could also be active against pathogenic microorganismsthat are currently resistant to conventional drugs.

ACKNOWLEDGMENTS

We thank Tadashi Mayumi and his colleagues at Jichi MedicalSchool for kindly providing nMAb-KT-producing hybridoma clones.We also thank Koji Yokoyama of Chiba University for providingthe C. albicans strain. Special thanks go to Toshinori Nakayamaand Masaru Taniguchi of the Graduate School of Medicine, ChibaUniversity, for valuable discussions.

This work was supported by a grant from the Ministry of Education,Science, Sports, and Culture of Japan.

FOOTNOTES

* Corresponding author. Mailing address: Department of Biochemistry, Faculty of Pharmaceutical Sciences, Niigata University of Pharmacy and Applied Life Sciences, 265-1 Higashijima, Niigata 956-8603, Japan. Phone: 81-250-25-5273. Fax: 81-250-25-5021. E-mail: tkomiyam{at}niigata-pharm.ac.jp.

REFERENCES

Top