Ebook Current topics in medical mycology (Vol.4): Part 2

(BQ) Part 2 book “Current topics in medical mycology” has contents: Killer system interactions, allylamine antifungal drugs, teaching medical mycology in latin america, the'' need for a national mycoses reporting system,… and other contents.

5-Killer System Interactions

L POLONELLI, G MORACE, S CONTI, M GERLONI,

W MAGLIANI, AND C CHEZZI

Viruses and Fungi

Over the last few years, our concept of yeasts has changed vastly Once thought of as "E coli with a nucleus,"l these organisms currently represent

cells of universal use by modern molecular biologists Retroviral elements, ubiquitin, calmodulin, actin, and tubulin are only some of the many biologi-cal elements that are being investigated in yeast cells Ras-related genes, strongly implicated in the transformation of normal mammalian to cancer cells,2 have also been discovered in yeasts, opening the way for exciting new cytological, biochemical, and experimental genetic strategies that were im-possible to carry out in animal cells

Cytoplasmic viruslike particles (VLPs) were first observed in diseased mushrooms.3 Since then, viruses have been detected in more than 100 differ-ent fungal species In most of these species, the persistent viral infection produces no discernible effect on the fungal host's phenotype

However, the finding that the antiviral and interferon-inducing activities

of extracts (statolon, hellenin) from Penicillium species were due to the

pre-sence of double-stranded RNA (dsRNA) has sparked an explosion of interest

in what has now become a new area for research in mycology.4 The virulence of certain forms of Endothia parasitica, which may cause chestnut

hypo-blight disease, has also been found to be related to the presence oflipid-rich cytoplasmic vesicles containing dsRNA Reduction of cytochrome oxidase and respiratory deficiency resulting in abnormal growth and morphology have been linked to specific dsRNA segments in Ophiostoma ulmi (which

causes Dutch elm disease)

The biological properties of the dsRNA genomes that are in capsid form in noninfectious VLPs within the fungal cytoplasm have proved to be unique These mycoviruses are apparently incapable of extracellular transmission through lysis of the host cell They are normally maintained at a relatively stable copy number and transmitted predominantly by the intracellular route In hyphomycetes, for example, transmission occurs through the How

137

138 Poionelli et aI

of protoplasm toward the growing hyphal tip, whereas cytoplasmic mixing that occurs during budding, mating, or other means of cell fusion is responsi-ble for the spread of these viruses in yeasts Another unusual characteristic

of the dsRNA VLPs in fungi is that they generally produce no adverse effect

on their hosts and, in some cases, may even be beneficial In this respect they may be compared with bacterial plasm ids, in that they act as nonessen-tial extrachromosomal elements which can, however, be quite useful to their hosts

In addition to conjugation behavior, enterotoxin production, and tic resistance, bacterial plasmids encode the production of bacteriocins by certain strains that exert lethal effects on closely related strains The killer toxins produced by certain yeasts are extremely similar to the bacteriocins The killer characteristics of these yeasts are often compared with those of bacterial species associated with C factors The resemblance is even more

antibio-striking between killer yeasts and Paramecium species carrying K particles

In the latter case, a dominant nuclear gene has been found to be responsible for the organism's ability to maintain these particles.5 The occurrence of similar phenomena in organisms with such widely divergent evolutionary histories is fascinating

The Killer Phenomenon in Yeast

The yeast killer phenomenon was first observed among Saccharomyces

cere-visiae strains.6 These strains secrete glycoproteins that have toxic effects on other, sensitive strains of this species as well as closely related ones The yeast's ability to produce a killer toxin, its immunity to the effects of that protein, and its resistance to those produced by other species are all encoded

by satellite dsRNAs

In S cerevisiae there are two main recognized killer systems (Kl and K2)

which do not elicit cross-immunity Kl is found in laboratory strains and is the most deeply investigated yeast killer system, whereas K2 is exclusively related to wine yeasts When no distinction is made, reference to the Kl killer system is generally intended In the type I system, cells may have one offour possible phenotypes: K+ R+ (killer), K- R- (sensitive), K- R+ (neut-ral), and K+ R- (suicidal)

Two types of dsRNA have been (ound inside icosahedral VLPs of

approx-imately 40 nm in diameter within wild killer strains of S cerevisiae: L

dsRNA (4.9 kb) is present in approximately 1,000 copies per haploid cell, while the M dsRNA appears in about 100 copies Deletion of certain se-quences of the latter results in suppressive S dsRNA.7 Both L and M dsRNA interact with mak (maintenance killer) genes, which are widely scattered over the chromosomal map of the yeast, and these genes are necessary for maintaining autoreplication of M dsRNA VLPs In fact, mutants that have

lost certain mak genes also lose M dsRNA, though L dsRNA, which is also present in nonkiller yeast cells, is maintained Recessive mutations of other

chromosomal genes, such as ski (superkiller), result in increased toxin

pro-duction by the yeast, presumably because of the roles they play in the plication of M dsRNA.8

re-Other chromosomal genes are essential for the secretion (sec) and killer expression (kex) of the toxin as well as for the virus-mediated host immunity

to the toxin (vpl [vacuolar protein localization], end [endocytosis], and rex

[resistance expression] genes) although not necessarily for replication of the

cytoplasmic killer genome The kex 2 gene product is, however, also

re-quired for mating functions and meiotic sporulation

The relationship between the L and M dsRNA genomes would be gous to that between a helper and a defective virus rather than that of com-ponents of an interdependent segmented mycovirus genome.9 In the S

analo-cerevisiae Kl and K2 killer systems, L dsRNA encodes the major capsid polypeptides of the VLPs The M and L types of dsRNA probably do not exist within the same capsid, and although the polypeptide composition of the VLPs has not been completely described yet, particles containing L dsRNA reportedly include a large polypeptide of 75,000 Da and two smaller ones of 55,000 and 37,000 Da Whether or not the M dsRNA-containing particles are structurally identical to the L particles cannot be confirmed at this time However, in light of the cross-antigenicity observed between the two VLPs, it is likely that they share at least one polypeptide

The M dsRNA is related to toxin activity, by encoding the killer as well as the immune phenotype Killer toxin-coding cDNA copies of the MdsRNA from

S cerevisiae have been cloned, and a toxin precursor gene sequence has

been identified 10.11 This sequence encodes a 35-kDa protein (preprotoxin), which contains a signal sequence-encoding region Either this molecule or

a processed product (43-kDa protoxin) of glycosylation in the endoplasmic reticulum is responsible for immunity of the toxin-producing strain After digestion by specific proteolytic enzymes, the protoxin is processed in the Golgi apparatus or in secretory vesicles into the mature killer toxin, which in

the S cerevisiae Kl killer system is composed of ~vo disulfide-linked a

(9.5-kDa) and {3 (9.0-(9.5-kDa) polypeptide components 12.13

Mutation of various nuclear genes may drastically affect the yeast's ability

to maintain killer dsRNA VLPs and its killer phenotype The study

ofkiller-resistant (kre) mutants has shed light on the mechanism by which killer toxins destroy sensitive yeast cells Strains of S cerevisiae that had been rendered resistant to S cerevisiae Kl killer toxin by mutation of the nu-

clear genes krel and kre2 were found to bind 35S-labeled killer toxin more

weakly than wild, sensitive strains 14 Although killing activity is not yet fully understood, ATe do know that rapid, energy-independent binding of the toxin to a (1,6)-{3-n-glucan-linked component of the cell wall occurs during

the initial phase and that the products of either the krel or the kre2 nuclear gene are necessary to this process Mutations in the nuclear loci krel and

140 Poionelli et al

kre2 result in a reduction and modification of (1,6)-J3-n-glucan content of the cell wall

The initial binding to the wall, presumptively by the J3 subunit, might

make the a subunit of the toxin somehow accessible by an

energy-dependent process at a plasma membrane site where the toxic effect is ifested by ion leakage and cell death 15 When constant concentrations of sensitive cells were treated with sub saturating concentrations of killer toxin, linear rates of killing were observed, thus suggesting a single-hit process 16 Since krel and kre2 mutant spheroplasts are sensitive to the toxin, while those with mutation of the kre3 nuclear gene are resistant, even though normal cell wall binding occurs, it may be that the latter gene encodes for or

man-is in some other way involved with the cytoplasmic membrane receptor site Two different hypotheses have been proposed for killer toxin resistance in the immune cell The immunity determinant (22-kDa) might alter or mask the plasma membrane receptor site, rendering its interaction with the a

domains derived from exogenous killer toxin impossible Alternatively, the immunity determinant might mediate the relocation or removal of the re-ceptor from the cytoplasmic membrane, a process in which vp2 and end

genes might be involved,17 though protease production by immune strains for cleaving killer toxin should not be excluded

Studies using artificial phospholipid bilayer membranes have revealed that the purified toxin from Pichia kluyveri, like the K1 and K2 S cerevisiae

toxins, causes ion-permeable channels to form in the bilayer 18 The tion of pores and proton pumping is not part of the killing effect of a P mrakii

forma-toxin This basic polypeptide, composed of 88 amino acid residues, is devoid

of mannosides and has an isoelectric point of 9.1 and a molecular size of 10,721 Da 19 It selectively inhibits the synthesis of J3-glucan in the cell wall

of sensitive S cerevisiae cells 20 Cell wall synthesis of proteins, mannan, chitin, and the alkali-insoluble, acid-soluble polysaccharides is not affected

by the toxin

Like S cerevisiae, the corn smut pathogen Ustilago maydis secretes protein toxins 21 There are three different, though closely related, killer systems (PI, P4, and P6) which are associated with cytoplasmic dsRNAs in VLPS.22-24 The strains in one system kill those of the other two systems, though they are immune to their own toxins

glyco-All three toxins consist of a 12.5-kDa and a 1O-kDa peptide chain linked by disulfide bonds only, and both polypeptides are essential for toxic activity Temperature-sensitive, nonkiller mutants secrete an inactive toxin which lacks the 1O-kDa polypeptide; when the missing peptide is added, the killer effects of the toxin are restored The two polypeptides appear to interact sequentially: the 1O-kDa component initiates the toxic effect by acting as a recognition element It interacts with a cell wall receptor to render the cell accessible to the catalytic effects of the 12.5-kDa polypeptide, which apparently induces endonucleolytic cleavage of nucleic acids The U maydis

killer strains are lethal only for members of their own and very closely

re-lated species and have no effect on yeast isolates that are sensitive to other killer yeasts

The plasmids of both prokaryotic and eukaryotic microorganisms are usually covalently closed circular DNA molecules, though linear forms do

exist.25 The killer system of Kluyveromyces lactis, for example, is mediated

by two linear dsDNA plasmids: pGKI-l (8.9 kb) and pGKI-2 (13.4 kb) There are approximately 100 copies of each in each haploid cell, and they are cyto-plasmically inherited in a non-Mendelian fashion.26 The pGKI-l plasmid confers upon the host cell either the killer (gene located in the central part)

or the immunity (gene located in the terminal part of the plasmid) type Replication and maintenance of the pGKI-l plasmid are probably con-trolled by the pGKI-2 plasmid

pheno-The K lactis killer toxin consists of three subunits: a glycosylated polypeptide with an approximate molecular size of 100,000 Da and two smaller nonglycosylated components of 30,000 and 27,500 Da These three proteins are produced by two distinct RNAs, each of which includes a sig-nal sequence.27 The first two subunits are derived from a larger precursor The toxin inhibits adenylate cyclase in sensitive strains, causing them to arrest in the G 1 phase of growth This arrest can be reversed by the addition

of cyclic AMP, which is recognized as necessary for the initiation of a new mitotic cycle in normal cells

While many aspects of the killer phenomenon in K lactis (such as pH and temperature range for toxin activity, spectrum of activity, and even mode of

action) are different from those of S cerevisiae, killer cells of both species can

be deprived of their toxic properties by physical and chemical curing cesses (growth in cycloheximide or ethidium bromide or at elevated tem-peratures) Both dsRNA and dsDNA plasmids may be transferred, by protoplast fusion and transformation techniques, to (heterologous) nonkiller (sensitive) strains to confer the killer (and resistant) phenotype to the reci-pient cells K lactis dsDNA plasm ids can replicate autonomously and stably

pro-in the S cerevisiae new host and can coexist with the resident dsRNA,

although they are incompatible with mitochondrial DNA.28 The resultant killer strains may produce larger amounts of killer toxin than the K lactis did Extracellular transmission of the dsRNA VLPs of S cerevisiae has also

been demonstrated 29

The toxin of Pichia anomala (Fig 5-1a) has proven to be a potent killer of

many different genera of yeasts, including many pathogenic species.3O The genetic basis and the mechanism of action have not yet been identified

Studies of the saturation kinetics of this toxin in Candida albicans cells

sug-gest the presence of a toxin receptor, probably located on the cell wa1l3l

(Fig 5-1b) Killing activity against C albicans is greatest during the early exponential growth phase of P anomala, while the highest activity against Saccharomycodes ludwigii occurs during the late phase This would suggest

that P anomala produces more than one active component

Like the toxin of S cerevisiae, the P anomala toxin acts by binding to a

142

INDIRECT

IMMUNOFLUORESCE 'CE

OCCUI\IIE CE or

~II.I.ER PHE OME ON

KILLER SYSTEM INTERACTIONS

1 1on

YEAST KII.I.ER TOXI THERAPEUTIC EFFECT

Polonelli et aI

IOIOTYPIC VACCI ATION

FIG 5-1 Killer system interactions For detailed descriptions, see text

receptor site, but its range of activity is much broader, including organisms of various genera Unlike the toxin of K lactis, this toxin is not

micro-counteracted by cyclic AMP These phenomena suggest that the P anomala

killer toxin has a unique mode of action 32

Hat-spored species that were formerly classified in the genus Hansenula have now been included in the genus Pichia, since identity of the two genera

was demonstrated by DNA comparisons.33 Segregation of the recombinant killer phenotype from the meiotic tetrads of crosses between killer and non-

killer strains of P anomala have shown that one or more nuclear genes must

be involved in expression of the killer character.34 The search for RNA or

DNA plasmids in many other Pichia (former Hansenula) species have also

failed to yield positive results (N Gunge, personal communication)

Killer toxin-producing yeast isolates belonging to the pathogenic genera

Cryptococcus and Torulopsis have been isolated from natural habitats,

although the genetic determinants for their toxinogenesis and toxin bioaction modalities are still unknown 35

The field has been extensively reviewed in whole or in part by several authors.36- 41 These reviews have been very useful to the authors, and read-ers will find in them further references and sometimes alternative view-points

The different susceptibilities of potential sensitive yeasts to the activity of

potential killer yeasts must be attributed to different mechanisms of

immu-nity In the S cerevisiae Kl system, binding of the killer toxin to the cell wall receptor is substantially reduced in krel and kre2 mutants,42 which are con-

sidered resistant Binding would thus seem to be a prerequisite for the tive phenotype.43 The Kl killer toxin is lethal for spheroplasts of Candida, Kluyveromyces, and Schwanniomyces isolates, though intact cells of the

sensi-same microorganisms are toxin resistant Killer spheroplasts themselves also remain immune to their own toxin The Kl toxin is poorly retained by the cell wall of K iacUs, but the C albicans wall binds the toxin to more or less the same extent as do that of sensitive S cerevisiae species Thus, yeast

killer toxin specificity is defined by cell wall receptors which are necessary for binding but not sufficient for toxin action at the plasma membrane of the intact cells.44 It is possible that there are additional unidentified cell wall

components for killer toxin activity that are missing in the wall of C cans Alternatively, there could be some structural differences in the wall of this species which prevent the glucan receptor-bound toxin from reaching the plasma membrane Such observations emphasize the need for further study of the structure and protein transport systems of the yeast cell wall

albi-Saccharomyces cerevisiae strains carrying a 1.8-kb M dsRNA, which codes

for polypeptide toxin and a resistance function, are resistant to S cerevisiae

Kl killer toxin, though they do not present a significant reduction in the number of (1,6)-J3-D-glucan cell wall receptors Remarkable amounts of killer toxin are retained by binding to the killer cell's own cell wall The chromo-

somal rexl gene is also involved in this form of resistance 45

Whether this phenomenon represents the mechanism of resistance or not, all resistant strains bind a certain amount of toxin to the cell walls, indicating that they contain receptors similar to those that exist in wild-type sensitive strains.42 Killer cells themselves present cell wall receptors for their own toxin, which act as a barrier to the toxin after it passes through the plasma membrane on its outward path Binding of the toxin by isolates that contain sterically compatible cell wall receptors might also reduce toxin activity

against sensitive cells (unpublished data) In krel mutants, which lack such

receptors, a superkiller phenotype results when M dsRNA is introduced Phenylmethylsulfonyl fluoride-sensitive protease in the cell wall may also degrade the toxin before it can be secreted In fact, mutation of the super-

killer (skiS) gene, which controls this enzyme, results in the production of an

active killer toxin more concentrated than that produced by the nonmutant strain 46

The various mechanisms of immunity invariably result in differential ceptibility of the yeasts to the toxins of defined killer strains In a reciprocal killer assay, yeast isolates were alternatively used as killer or sensitive strains Many strains exhibited both killer and susceptible behavior, de-pending on the strain they were matched against and the conditions under which the assay was performed 47

sus-The production of inhibitory factors (killer toxins) can be assumed to play

144 Poionelli et al

an important role in the modification of the ecosystems of natural habitats34 and infected organisms (amensalism) Studies in animals have clearly shown that the killer yeast P anomala is able to secrete toxin in vivo in both im-munosuppressed and normal mice after experimental infection 48

The use of selected killer yeasts during brewing processes has been sidered to prevent the growth of contaminating strains It has been specu-lated, moreover, that the potential capacity of U maydis killer proteins to specifically inhibit U maydis sensitive strains could be used in the biological control of cereal smuts, if informational molecules for the production of tox-

con-in, introduced into the plant cell cytoplasm, could replicate and express their killing function 49

The Killer System as an Epidemiological Marker

The impact of hospital-acquired yeast infections has been dramatically demonstrated over the last decade, and the need for a simple, reliable, and sensitive method for differentiating fungal strains beyond the species level has become increasingly pressing Several different methods have been used for this purpose: serotyping, morphologic differentiation, study of mating behavior, enzyme profiles, chemical analyses, and chemical assimilation or resistance patterns 50 The ultimate method for biotyping fungal isolates would be complete DNA base sequencing, but even if this approach was technologically feasible, it would probably result in the complete differentia-tion of all isolates tested, thus invalidating its use as an epidemiological marker

First reported among S cerevisiae strains,6 the killer yeast phenomenon has subsequently been observed among many other yeast genera 51-53 Killer yeasts have been grouped according to their specificity for killing sensitive yeasts,54 and conversely, sensitive isolates of the same species have been differentiated on the basis of their differential susceptibility to the activity of various killer yeasts The killer typing system, using zone assays similar to those used in phage typing of Salmonella species, was first used to differenti-ate isolates of the pathogenic yeast species C albicans 55 and subsequently applied to the epidemiological study of nosocomial infections caused by this yeast species 56 The adoption of simple test conditions allowed investigators

to evaluate the susceptibility of these isolates to the activity of selected killer yeasts, primarily those of Pichia spp Under the same test conditions it was possible to extend the typing to other opportunistic species of yeasts (C ryp- tococcus neoformans, C glabrata, C parapsilosis, C pseudotropicalis, and

C tropicalis) 30 On the basis of their susceptibilities to the panel of selected killer strains, these opportunistic yeasts could be grouped into reproducible categories that contained strains that were serotypically heterogeneous The ability of the killer yeasts to exert their toxic effects was naturally

affected by temperature, pH, and composition of the growth medium, tions that varied according to the requirements of the target species being studied To avoid some of these restrictions, isolated and partially purified toxins were used instead of streaked whole yeasts 57

condi-The use of killer toxins in place of killer yeasts, although more laborious, improved the standardization of the system Killer toxins proved to be stable when partially purified, concentrated, and stored at 4°C, thus ensuring a high degree of test reproducibility A computer program automatically divided the C albicans isolates studied into groups according to their suscep-tibility to the killer toxins The program was designed to allow a maximum error of 5% in strain differentiation The computer-aided program permitted the biotyping of the investigated yeast isolates in terms of the group percen-tage of probable affinity Computer interpretation of the results eliminated subjective interpretation The storage of the data in the computer allowed a rapid comparison of any new result with the results of the groups coded previously, thus simplifying its application in epidemiological studies The original test conditions were also used to evaluate the occurrence of sensitivity of hyphomycetes, bacteria, and achlorophyllous microorganisms

to the activity of recognized killer yeasts 58 Killer toxins appeared to be hibitory to a wide variety of prokaryotic and eukaryotic microorganisms other than yeasts (Fig 5-1c) As expected, the highest activity was displayed

in-by the isolates of Pichia spp tested All of the bacteria and fungi that were able to grow under the experimental conditions proved to be sensitive to at least one killer yeast tested The killing effect was expressed differently against the various species of bacteria and hyphomycetes and strains within the same species The inhibitory effect observed might not necessarily be caused by the killer toxins themselves but rather by other metabolic prod-ucts If the killer phenomenon among yeasts, bacteria, aerobic actinomy-cetes, hyphomycetes, and achlorophyllous microorganisms could be con-firmed with purified killer toxins, this would imply common cell wall binding receptors and bioaction modalities

The differential susceptibility of bacteria and hyphomycetes to the killer strains could be used for epidemiological differentiation of strains within the same species in much the same way that bacteriocin typing of gram-negative and gram-positive bacteria is used 59,60

Gross morphology and microscopic features are inconsistent criteria for distinguishing intraspecific mycelial cultures Pleomorphism and lack of sporulation pose major difficulties to mycologists attempting the biotyping of mycelial fungi Unfortunately, with the possible exception of C albicans 61

and C neoformans, 62 serologic differentiation of fungi has proven to be less discriminatory than serotyping of pathogenic bacteria

The exoantigen technique has revealed two serotypes in Blastomyces dermatitidis,63 and antigenic variability among isolates of Sporothrix schenc- kii has been detected by indirect immunofluorescence.64 Monoclonal anti-bodies have proven to be very useful for serotyping yeastlike and mycelial

146 Polonelli et aI

fungi by using the Western blot technique.65•66 This immunological approach, however, requires reagents and technology that are not often available to small laboratories

The toxic effect of numerous selected killer yeasts has been studied on

Penicillium camemberti, S schenckii, Aspergillus niger, Pseudoallescheria boydii,67 and A fumigatus and related taxa 68 The killer system proved to be

a reliable tool for the biotyping of these mycelial cultures

Although different procedures have been developed for differentiating lates of gram-positive and gram-negative bacterial species, fewer possibili-ties have been reported for distinguishing strains of slowly growing bacteria such as aerobic actinomycetes and mycobacteria

iso-The killer system, preViously standardized for yeasts and hyphomycetes, has been adapted to the specific growth conditions of the bacterial isolates The (modified) killer system proved to be a convenient and flexible biotyping method for strain differentiation of Nocardia asteroides, N brasiliensis, N otitidis caviarum, and Actinomadura madurae69 as well as Acinetobacter calcoaceticus, Escherichia coli, Pseudomonas aeruginosa, Haemophilus injluenzae, Neisseria meningitidis (beyond the conventional serotype level), staphylococcus aureus, group A /3-hemolytic streptococci, Myco- bacterium tubercolosis, M fortuitum, and M smegmatis 70

The specific growth conditions of each bacterial species under which killer yeasts could still exert their potential killer activity were identified, and there were remarkable variations in pH, temperature, and oxygen concen-tration It is likely, however, that more than one killer toxin is produced by the same yeast, each of which is active under different conditions

Several typing systems, in addition to serotyping and antibiotic ceptibility, have, of course, been used for such bacterial species: phage susceptibility,71-73 enzyme production,74,75 bacteriocin production or susceptibility, 76, 77 DNA hybridization,78 and R plasmid analysis 79 Most of these methods, however, are too laborious to be reproduced with a large number of isolates in small microbiological laboratories The killer system, properly adapted to the growth requirements of the sensitive species, may

sus-be a convenient biotyping method for a large numsus-ber of prokaryotic and eukaryotic microorganisms It requires no specific technical expertise and can be carried out using commercially available media and a set of suitable killer yeasts

Antibiotic Potential of Yeast Killer Toxins

Many of the etiologic agents of systemic mycoses have a parasitic form in infected tissues that is structurally and morphologically different from the one they present in the cultural phase The in vitro susceptibility to selected killer toxins of both the mycelial and yeast forms of S schenckii isolates were therefore evaluated comparatively80 (Fig 5-1d)

Temperature, pH, aeration, and carbon source are recognized to play portant and specific roles in the development of the mycelial or yeast form of this species, making it impossible to establish the same cultural conditions for exclusive development of specific morphological types of growth For this reason, the original killer system conditions had to be modified to test pure cultures of mycelial- or yeast-form isolates

im-The pure conidial and yeast suspensions were challenged with the same amounts of killer toxins, and the numbers of colony-forming units (CFU) were compared with those obtained with control challenges using Sabouraud broth Both the yeast and mycelial forms of S schenckii showed significant susceptibility to the activity of yeast killer toxins, and this finding, regardless

of the specific mechanism of action, suggests that these toxins or their rivatives might be used in the treatment of systemic mycoses

de-In E coli KB TOOl, the outer membrane receptors for colicins E and phage BF 23 are also involved in vitamin B12 transport 81 If the cell wall receptors for killer toxins were found to have other functions as well, agents that inhibit the latter might be developed for use as antibiotics 82 The prim-

ary target of the P mrakii killer toxin on sensitive yeast and mold cells is the synthesis of J3-glucan 2o,83 Certain antifungal antibiotics, such as aculeacin A, echinocandin B, and papulocandin B, have been found to specifically inhibit the cell wall J3-glucan synthesis in many ascomycetous and deuteromycetous yeasts These compounds selectively inhibit J3-glucan synthesis in growing yeast cells and also the in vitro activity of J3-(1,3)-glucan synthetase obtained from the yeasts Thus, the P mrakii toxin appears to act against sensitive yeast cells in much the same way that these cell wall-active antibiotic com-pounds act against their target cells J3-glucan, especially J3-(1,3)-glucan, which is recognized as the most important structural component of the yeast cell wall, might be rendered osmotically fragile and defective by killer toxin inhibition of J3-glucan synthesis, thus resulting in lytic cell death

Because of the killer toxins' lability during purification processes and the instability of the purified molecules, there is little information on the effec-tive antimicrobial spectrum of these toxins 84 We therefore attempted to investigate the therapeutic potential of the toxin produced by a P anomala

isolate (UCSC 25F) This isolate was selected as a reference candidate cause it had exhibited the largest sppdrum of activity against prokaryotic and eukaryotic microorganisms and had proven to be relatively stable at 37°C and in unbuffered conditions

be-Pityriasis versicolor-like lesions were experimentally induced in guinea pigs and rabbits with a strain of Malassezia furfur, which had previously been recognized to be sensitive in vitro to the activity of the reference toxin The animals recovered well after topical application of the crude toxic extract (Fig 5-le) The same behavior was obtained in dogs with otitis media ex-perimentally induced with M pachydermatis 85 It was not possible to ascer-tain the potential therapeutic effect of parenterally administered toxin in experimental systemic mycoses because toxic effects occurred that were probably related to the impurity of the substance

148 Polonelli et aI Monoclonal Antibodies That Neutralize Yeast Killer Toxin

Since conventional chemical procedures (gel exclusion chromatography, chromatofocusing, anion-exchange chromatography) were not efficient enough for the production of large amounts of purified killer toxin, mono-clonal antibodies (MAbs) were produced for the one-step immunochemical purification and characterization of yeast killer toxin by affinity chromato-graphy19 (Fig 5-1£) Monoclonal antibodies might also be used for analyzing biologically active epitopes of the toxin in order to chemically synthesize small peptides with killer activity and poor immunogenicity

Monoclonal antibodies were obtained after fusion of mouse myeloma cells with spleen cells isolated from mice primed with a crude extract of yeast

killer toxin produced by the P anomala UCSC 25F strain.86 A comparative enzyme-linked immunosorbent assay, used for selecting the antibody-producing hybrids, was adapted to determine the reactivity of the antibodies

in the culture fluid Hybrids that also reacted with the growth medium of the killer yeast were discarded The hybrids that produced antibodies which reacted only with the crude toxic extract were expanded and cloned Mono-

clonal antibodies of the IgG class were produced When the crude P anomala

killer toxin reference antigen was electrophoretically separated in ing gels and then immobilized on nitrocellulose strips (Western blot tech-nique), only slight diversity among the MAbs to the killer toxin proteins was detected

denatur-One of these MAbs (designated KTl) showed a greater avidity than another (designated KT4), although both reacted with the antigenic determi-nants that had molecular sizes of 92 and 115 kDa Antibody-rich fluids pro-duced from the two expanded clones reacted differently to the reference antigen in immunodiffusion tests (Fig 5-1g) Monoclonal antibody KT4 pro-duced a clear precipitin band, whereas KTI failed to precipitate with the antigen No reactions were observed in immunodiffusion with the growth medium When graduated amounts of KT4 (ammonium sulfate-purified

ascitic fluid) were added to the P anomala UCSC 25F crude killer toxin, killer activity against a recognized sensitive yeast (C albicans UCSC 10) was

gradually neutralized (Fig 5-1h) No neutralization was observed when equal amounts of a nonspecific MAb (ascitic fluid) were used as a control Monoclonal antibody KT4 was also observed to react with toxins produced

by other Pichia species, expecially that of P mrakii UCSC 255

Cross-re-activity between the killer toxins of these species implies the presence of common antigenic determinants However, several yeast killer toxins from

P mrakii with different chemical and biological properties have been

defined 87,88 The toxin that has been most fully characterized (as having high heat and pH stability and a molecular weight of 10,700) is probably not the one reacting with MAb KT4

Various killer toxins produced by Saccharomyces, Kluyveromyces, Pichia,

and Candida species were therefore investigated for reactivity with MAb KT4 The genetic determinants of some of these toxins have yet to be iden-tified Double immunodiffusion using the killer toxins as antigens (Fig 5-1g) and indirect immunofluorescence on whole killer cells (Fig 5-li) revealed that MAb KT4 reacted only with the killer toxins and the whole cells of yeasts belonging the genus Pichia 89 This finding suggests that the toxins coded by different genetic systems are antigenically heterogeneous

High magnification of yeast killer cells of P anomala UCSC 25F under indirect immunofluorescence revealed differential staining depending on the growth phase of the cells Apparently, mature blastospores fluoresced more than younger daughter cells, suggesting that killer toxin secretion is associ-ated with the later phases of growth

The genus Pichia was formerly limited to a heterogeneous group of species characterized by hat-shaped, spheroidal, smooth or rough ascospores and certain common coenzyme properties.90 On the basis of DNA base sequenc-ing studies,33, many species of the genus Hansenula have recently been transferred to 'the genus Pichia, eliminating the earlier distinction based on nitrate assimilation Species and genus differentiation cannot be based on the killer toxin phenotype, which is quite variable In fact, we have observed

a great deal of intrageneric variation in the reactions of 25 Pichia isolates to MAb KT4 Some of these strains, whose killer activity had been confirmed previously when tested against sensitive strains, showed no reactivity what-soever to the MAb.91 For this reason, the significance of killer interactions must be interpreted with caution

Observation of KT4 reactivity with at least one killer toxin produced by most of the Pichia spp tested prompted us to attempt the one-step purifica-tion of correspondent killer toxin with affinity chromatography We found that the resulting toxin was still somewhat toxic and immunogenic to mice,

as might be expected of a large foreign protein (unpublished data) This finding, together with the toxin's lability at neutral pH and at elevated temperatures, makes it unlikely that this protein can be used therapeutically

in systemic mycoses Perhaps the solution to these problems lies in the use

of toxin derivatives as antifungal agents

Yeast Killer Toxin Mimicking Anti-Idiotypic

Antibodies

The steric interaction that occurs between a yeast killer toxin and the cell wall receptor might be similar to that which takes place between antigens and antibodies The variable regions of antibodies are known to be im-munogenic, and the antigenic determinants of these regions are termed idiotypes Antibodies produced against these immunogenic regions of the antibody are called anti-idiotypes or anti-idiotypic antibodies (antilds) Anti-idotypic antibodies have been described in a variety of clinical conditions

In light of these findings, we attempted to clarify the nature of killer toxin cell wall receptors in sensitive cells, using antilds raised in New Zealand rabbits These animals had been immunized with MAb KT4-secreting hybri-doma cells on the assumption that lymphocytes bearing the immunoglobulin idiotype as receptors might act as a more efficient immunogenlOO (Fig 5-1j) The specificity of the antilds raised in the rabbit antiserum was evaluated by

a double-immunodiffusion procedure The reference system consisted of ammonium sulfate (50% of saturation)-purified MAb KT4 and the P anomala

UCSC 25F killer toxin used as an antigen

The rabbit anti-idiotypic antiserum reacted in immunodiffusion with MAb KT4, producing a single precipitin band that was homologous to the one of the reference system (Fig 5-1g) No reaction was observed when the rabbit anti-idiotypic antiserum was tested with another MAb produced against an exoantigen of C albicans lOl The results suggested that the anti-idiotypic antiserum was highly specific for the variable region (idiotype) of MAb KT4 The precipitin band, moreover, disappeared after adsorption of MAb KT4 with the yeast killer toxin This observation implies that the antigen blocked the combining site of MAb KT4 The rabbit antiserum thus appeared to contain subpopulations of antilds that shared structural similarities with the portion of the killer toxin that binds with MAb KT4, while the combining site of MAb KT4 can be considered as the cell wall receptor-like idiotype

Detection of Yeast Killer Toxin Microbial Receptors

Ideally, MAb KT4 could be used to visualize yeast killer toxin bound to sensitive cells and to ascertain the presence of common cell wall receptors on the surfaces of taxonomically unrelated microorganisms The cell wall of C

albicans has been recognized to contain a (1,6)-f3-D-glucan, 102 which is the likely site of binding for the toxin However, previous studies using a con-ventional sandwich constituted of sensitive cell, yeast killer toxin, yeast killer toxin MAb, anti-mouse fluorescein-conjugated immunoglobulins (or anti-mouse biotinylated antibodies and fluorescein-streptavidin) have failed

to detect killer toxin receptors in sensitive C albicans cells (unpublished data) This negative finding may have been caused by the dissociation of the bound toxin during washings, to the masking of specific killer toxin epitopes for the MAb by cell wall receptors themselves, or other reasons

In contrast, the use of antilds that mimicked the activity of yeast killer

toxin made it possible to visualize directly the interaction with the cell wall receptors on sensitive C albicans yeast cells by indirect immunofluores-cence (Fig 5-1k) Reactivity was clearly detectable on the outer cell wall of the yeast cells The degree of fluorescence varied according to the growth phase of each cell Immunofluorescence was mainly detectable in budding cells and germ tubes and never inside yeast cell No fluorescence was observed when antilds, previously adsorbed with MAb KT4, were used which attests to the specificity of the reaction103 (Fig 5-11) Interestingly, cell wall receptors were also revealed by the antilds in killer cells of P ano- mala, which were obviously resistant to their own toxin, and in K lactis

cells, which proved to be sensitive to the P anomala killer toxin (Fig 5-1m) Preliminary results have suggested that taxonomically unrelated eukaryo-tic and prokaryotic isolates, such as S schenckii (yeast and mold form of growth) and gram-positive (S aureus) and gram-negative (E coli, P aerugi- nosa) bacteria, which had previously proven to be susceptible to killer yeasts, might show the presence of compatible yeast killer toxin cell wall receptors under some circumstances (Fig 5-1n) The therapeutic implica-tions of such a finding become even more evident when we consider that no fluorescence was detectable on animal cells (HeLa, Vero, HEp2) in vitro (unpublished data) (Fig 5-10) Theoretically, it should be possible to de-velop new antibiotics that react with a specific physiological target (the yeast killer toxin cell wall receptor) It is reasonable to hope that these antilds antibiotics will create fewer adverse side effects than killer toxins themselves

or other more conventional drugs adopted as antifungal agents in clinical use

Antibiotic Activity of Anti-Idiotypic Antibodies

Anti-idiotypic antibodies purified by affinity chromatography against mobilized MAb KT4 were used to investigate their in vitro killing activity on sensitive C albicans (CDC B385) cells (Fig 5-1p) After overnight adsorp-tion, CFUs were compared in inocula incubated with antilds and with phosphate-buffered saline used as a control The former repeatedly showed statistically significant reduction of CFU when compared with the control

im-It was also interesting to observe that in the same experimental design, the antilds were able to kill P anomala cells secreting UCSC 25F killer toxin that had previously shown resistance to the action of their own toxin (Fig 5-1q) The cell's normal mechanisms of immunity to its own toxins (pro-teases, reduction in cell wall receptors) are apparently ineffective against the toxin-mimicking antilds

The neutralization of the killer activity of antilds after adsorption with the complementary MAb KT4 should attest to the specificity of their action104 (Fig 5-1r) The activity of antilds against toxin-resistant killer cells such as P

anomala UCSC 25F implies that antilds may have wider ranges of activities

152 Polonelli et aI than the toxins themselves It may be, however, that binding of the antiIds

to a cell wall receptor alone, as visualized in prokaryotic microorganisms, is not sufficient for killing activity

Idiotypic Vaccination

An important aspect of the therapeutic potential of antilds is the fact that they can be elicited in vivo by immunization with the complementary MAb (idiotypic vaccination) Administration of MAb KT4 to syngeneic mice according to standardized schedules of immunization elicited high titers of antilds which were associated with a significant degree of protection in vivo

against lethal infections with C albicans cells (Fig 5-1s)

Increased survival rates were seen with lower infecting doses and higher MAb immunization doses The specificity of this immunoprotection was demonstrated by high antild levels detected during the entire course of immunization by neutralization enzyme-linked immunosorbent immuno-metric assay exploiting the competition of antilds with yeast killer toxin for the binding site of MAb KT4 as well as by the in vitro killer activity of mouse antilds purified by affinity chromatography against MAb KT4 tested

in the CFU assay on the yeast cells used for infection (manuscript in tion) (Fig 5-lt)

publica-Specific immunity may be evoked in animals against an infectious agent by the whole antigen (or its fractions), as occurs in conventional vaccination, by its derivatives, as occurs in recombinant vaccination, or through the media-tion of internal images of the antigen, as demonstrated in anti-idiotypic vac-cination

Our approach, in which the immunizing agent was represented by MAb KT4, constitutes a unique type of vaccination in that the protective im-munoglobulins that it elicits (antilds) act more like antibiotics ("antibiobo-

dies") than like antibodies against the target organism (C albicans)

Perspectives

New techniques that are capable of generating human or chimeric MAbs could provide autologous antibodies for the induction of antilds in humans Alternatively, the hypervariable regions of murine MAbs could be se-quenced to allow synthesis of artificial haptens that could be carried on hu-man proteins These approaches would theoretically eliminate the problem

of toxicity associated with the therapeutic use of either killer toxins selves or heterologous immunoglobulin derivatives while maintaining the high levels of antimicrobial activity The network theory implies that prop-erly engineered antiIds could be used not only prohylactically but also therapeutically in animals and humans already affected by systemic mycoses

1 Herskowitz I Yeast as the universal cell Nature (London) 1985; 316: 678-679

2 Nurse P Yeast aids cancer research Nature (London) 1985; 313:631-632

3 Hollings M Viruses associated with a die-back disease of cultivated mushroom

6 Bevan EA, Makower M The physiological basis of the killer character in yeast,

in Geerts SJ (ed), Genetics Today, XIth Int Congr Genet (Vol 1) Pergamon

toxin-12 Bussey H, Boone C, Dmochowska A, et al Secretion and action of yeast K1 killer toxin, in Koltin Y, Leibowitz MJ (eds), Viruses of Fungi and Simple Eukaryotes Marcel Dekker Inc, New York 1988; pp 161-178

13 Sturley SL, EI-Sherbeini M, Kho S-H, et al Acquisition and expression of the killer character in yeast, in Koltin Y, Leibowitz MJ (eds) Viruses of Fungi and Simple Eukaryotes Marcel Dekker Inc, New York 1988; pp 179-208

14 Bussey H, Saville D, Hutchins K, et al Binding of yeast killer toxin to a cell wall receptor on sensitive Saccharomyces cerevisiae ] Bacteriol 1979;

154 Polonelli et al

19 Yamamoto T, Imai M, Tachibana K et al Application of monoclonal antibodies

to the isolation and characterization of a killer toxin secreted by H ansenula mrakii FEBS Lett 1986; 195:253-257

20 Yamamoto T, Hiratani T, Hirata H, et al Killer toxin from Hansenula mrakii

selectively inhibits cell wall synthesis in a sensitive yeast FEBS Lett 1986;

197:50-54

21 Koltin Y, Day PRo Inheritance of killer phenotypes and double-stranded RNA

22 Koltin Y Virus like particles in Ustilago maydis: mutants with partial genomes Genetics 1977; 86:527-534

23 Koltin Y, Kandel JS Killer phenomenon in Ustilago maydis: the organization of

the viral genome Genetics 1978; 88:267-276

24 Koltin Y The killer system of Ustilago maydis: secreted polypeptides encoded

by viruses, in Koltin Y, Leibowitz MJ (eds), Viruses of Fungi and Simple Eukaryotes Marcel Dekker Inc, New York 1988; pp 209-242

25 Gunge N Some structural aspects oflinear DNA plasmids in yeast, in Wickner

RB, Hinnebusch A, Lainbowitz AM, et al (eds), Extrachromosomal Elements

in Lower Eukaryotes Plenum Press, New York 1986; pp 449-461

26 Gunge N, Tamaru A, Ozawa F, et al Isolation and characterization of linear deoxyribonucleic acid plasmids from Kluyveromyces lactis and the plasmid

associated killer character J Bacteriol1981; 145:382-390

27 Stark MJ, Boyd A The killer toxin of Kluyveromyces lactis: characterization of

the toxin subunits and identification of the genes which encode them EMBO J

1988; 5: 1995-2002

28 Gunge N, Yamane C Incompatibility of linear DNA killer plasmids pGKIl and pGK12 from Kluyveromyces lactis with mitochondrial DNA from Sacchar-

29 El-Sherbeini M, Bostian KA Viruses in fungi: infection of yeast with Kl and K2 killer viruses Proc Natl Acad Sci USA 1987; 84:4923-4927

30 Morace G, Archibusacci C, Sestito M, et al Strain differentiation of pathogenic yeasts by the killer system Mycopathologia 1984; 84: 81-85

31 Sawant AD, Abdelal AT, Ahearn DG Anti-Candida albicans activity of Pichia anomala as determined by a growth rate reduction assay Appl Environ Micro- bioll988; 54:1099-1103

32 Sawant AD, Abdelal AT, Ahearn DG Purification and characterization of the

anti-Candida toxin of Pichia anomala WC 65 Antimicrob Agents Chemother

1989;33:48-52

33 Kurtzman CPo Synonymy of the yeast genera Hansenula and Pichia

demon-strated through comparison of deoxyribonucleic acid relatedness Antonie van LeeuwenhoekJ Microbiol SeroI1984; 50:209-217

34 Starmer WT, Ganter PF, Aberdeen V, et al The ecological role of killer yeasts

in natural communities of yeasts Can J Microbiol1987; 33: 783-796

35 Middelbeek EJ, Peters GJWH, Stumm C, et al Properties of a Cryptococcus laurentii killer toxin and conditional killing effect of the toxin on Cryptococcus albidus FEMS Microbiol Lett 1980; 9:81-84

36 Pietras DF, Bruenn JD The molecular biology of yeast killer factor Int J

Biochem 1976; 7: 173-179

37 Bruenn JD Virus-like particles of yeast Annu Rev Microbiol1980; 34:49-68

38 Bussey H Physiology of killer factor in yeast Adv Microb Physiol 1981;

41 Fujimura T, Esteban R, Wiclmer RB Double-stranded DNA replication in

Saccharomyces cerevisiae: genetic control and in vitro replication, in Koltin Y,

Leibowitz MJ (eds), Viruses of Fungi and Simple Eukaryotes Marcel Dekker

Inc, New York 1988; pp 9l-115

42 Hutchins K, Bussey H Cell wall receptor for yeast killer toxin: involvement of

(1,6)-J3-n-glucan ] Bacteriol1983; 154: 161-169

43 Al-Aidroos K, Bussey H Chromosomal mutants of Saccharomyces cerevisiae

affecting the cell wall binding site for killer factor Can] Microbiol 1978;

24:228-237

44 Zhu H, Bussey H The K1 toxin of Saccharomyces cerevisiae kills spheroplasts

of many yeast species Appl Environ Microbiol1989; 55: 2105-2107

45 Wiclmer RB Chromosomal and non chromosomal mutations affecting the killer character of Saccharomyces cerevisiae Genetics 1974; 76:423-432

46 Bussey H, Steinmetz 0, Saville D Protein secretion in yeast: two chromosomal mutants that oversecrete killer toxin in Saccharomyces cerevisiae Curr Genet

1984: 7: 449-456

47 Rogers D, Bevan EA Group classification of killer yeasts based on cross

reac-tions between strains of dilferent species and origin ] Gen Microbiol 1978;

52 Stumm C, Hermans JMH, Middelbeek EJ, et al Killer-sensitive relationships

in yeasts from natural habitats Antonie van Leeuwenhoek] Microbiol Serol

56 Polonelli L, Morace C Yeast nosocomial infections Ig Mod 1983; 79:315-334

57 Polonelli L, Castagnola M, Rossetti DV, et al Use of killer toxins for aided dilferentiation of Candida albicans strains Mycopathologia 1985;

computer-91:175-179

58 Polonelli L, Morace C Reevaluation of the yeast killer phenomenon ] Clin Microbiol1986; 24: 866-869

59 Holland lB Physiology of colicin action Adv Microb Physiol1975; 12:56-139

60 Tagg JR, Dajani AS, Mannamaker LW Bacteriocins of gram-positive bacteria

Bacteriol Rev 1976; 40: 722-756

61 Hasenclever HR, Mitchell WOo Antigenic studies of Candida I Observation

of two antigenic groups in Candida albicans ] Bacteriol1961; 82: 570-573

62 Wilson DE, Bennett JE, Bailey JW Serologic grouping of Cryptococcus neoformans Proc Soc Exp Bioi Med 1968; 127: 820-823

63 Kaufman L, Standard PC, Weeks RJ, et al Detection of two Blastomyces matitidis serotypes by exoantigen analysis ] Clin Microbiol1983; 18: 110-114

der-64 Nishikawa T, Harada T, Harada S, et al Serologic dilferences in strains of othrix schenckii Sabouraudia 1975; 13:285-290

Spor-65 Polonelli L, Morace C Specific and common antigenic determinants of

68 PoloneIIi L, Conti S, Campani L, et al Biotyping of Aspergillus fumigatus and

related taxa by the yeast killer system, in Samson RA, Pitt JI (eds), Modern Concepts in Penicillium and Aspergillus Classification Plenum Press, New

York, 1990; pp 225-233

69 Morace G, Dettori G, Sanguinetti M, et al Biotyping of aerobic actinomycetes

by modified killer system Eur] Epidemiol1988; 4:99-103

70 Morace G, Manzara S, Dettori G, et al Biotyping of bacterial isolates using the yeast killer system Eur] Epidemiol1989; 5:303-310

71 Marsik FJ, Paris is JT Bacteriophage types and 0 antigen groups of Escherichia coli from urine Appl MicrobioI1971; 22: 26-31

72 Smith PB Bacteriophage typing of Staphylococcus aureus, in Cohen JO (ed), The Staphylococci Wiley Interscience, New York 1972; pp 431-441

73 Lindberg RB, Latta RI Phage typing of Pseudomonas aeruginosa: clinical and

epidemiologic considerations ] Infect Dis 1974; 130(Suppl): S33-S42

74 Aber RC, Mackel DC Epidemiologic typing of nosocomial microorganisms

Am] Med 1981; 70:899-905

75 Janda JM, Bot,tone EJ Pseudomonas aeruginosa enzyme profiling: predictor of

potential invasiveness and use as an epidemiological tool.] Clin Microbiol1981;

81 Di Masi DR, Klute JC, Schnaitman CA, et aI Transport of vitamin Bl2 in

Escherichia coli: common receptor sites for vitamin Bl2 and the E colicins on

the outer membrane of the cell envelope.] Bacteriol1973; 115:506-513

82 Bussey H, Skipper N Killing of Torulopsis glabrata by Saccharomyces visiae killer factor Antimicrob Agents Chemother 1976; 9:352-354

cere-83 Yamamoto T, Uchida K, Hiratani T In vitro activity of the killer toxin from

yeast Hansenula mrakii against yeasts and molds.] Antibiot 1988; 41 :398-403

84 Otha Y, Tsukada Y, Sugimori T Production, purification and characterization of HMI, an anti-yeast substance produced by Hansenula saturnus Agric BioI Chem 1984; 48:903-908

85 Polonelli L, Lorenzini R, De Bernardis F, et al Potential therapeutic effect of yeast killer toxin Mycopathologia 1986; 96: 103-107

86 Polonelli L, Morace G Production and characterization of yeast killer toxin monoclonal antibodies.] Clin Microbiol1987; 25:460-462

87 O'Leary E Secretion of multiple killer toxins by Williopsis mrakii Abstr Annu Meet Am Soc Microbiol1988; p 394

88 Ashida S, Shimazaki T, Kitano K, et aI New killer toxin of Hansenula mrakii Agric BioI Chem 1983; 47: 2953-2955

89 Polonelli L, Manzara S, Conti S, et al Serological study of yeast killer toxins by monoclonal antibodies Mycopathologia 1989; 108:211-215

90 Yamada Y, Okada T, Ueshima 0, et al Coenzyme Q system in the classification

of the ascosporogenous yeast genera Hansenula and Pichia ] Gen Appl

Micro-bioi 1973; 19: 189-208

91 Polonelli L, Conti S, Campani L, Gerloni M, Morace G, Chezzi C Differential toxinogenesis in the genus Pichia detected by an anti-yeast killer toxin mono-

clonal antibody Antonie van Leeuwenhoek 1991; 59: 139-145

92 Burdette S, Schwartz RS Idiotypes and idiotypic networks N EnglJ Med 1987;

hemagglutinin-97 Sacks DL, Esser KM, Sher A Immunization of mice against African somiasis using anti-idiotypic antibodies ] Exp Med 1982; 155: 1108-1119

trypano-98 Sacks DL, Sher A Evidence that anti-idiotypic induced immunity to perimental African trypanosomiasis is genetically restricted and requires recog- nition of combining sites-related idiotopes ] Immunol1983; 131: 1511-1515

ex-99 Sege K, Peterson PA Use of anti-idiotypic antibodies as cell-surface receptor probes Proc Natl Acad Sci USA 1978; 75:2443-2447

100 Polonelli L, Morace G Yeast killer toxin-like anti-idiotypic antibodies ]

Clin Microbiol1988; 26: 602-604

101 Polonelli L, Morace G Specific and common antigenic determinants of

Candi-da albicans isolates detected by monoclonal antibodies ] Clin Microbiol1986;

23: 366-368

102 Gopal PK, Sheperd MG, Sullivan PA Analysis of wall glucans from yeast, hyphal and germ-tube forming cells of Candida albicans ] Gen MicrobioI1984;

130: 3295-3301

103 Polonelli L, Fanti F, Conti S, Campani L, Gerloni M, Castagnola M, Morace

G, Chezzi C Detection by immunofluorescent anti-idiotypic antibodies of yeast killer toxin cell wall receptors of Candida albicans ] Immunol Methods

1990; 132:205-209

104 Polonelli L, Conti S, Gerloni M, Campani L, Castagnola M, Morace G, Chezzi

C "Antibiobodies": antibiotic-like anti-idiotypic antibodies ] Med Vet ogy, in press

Mycol-6-Allylamine Antifungal Drugs

The allylamines constitute a recently developed class of synthetic mycotics characterized functionally by their action as squalene epoxidase inhibitors 1 Figure 6-1 shows the structures of three representative allyla-mines Naftifine, the first of these compounds to be discovered, was first synthesized in 1974,2 and its antifungal properties were identified during routine screening The potent antifungal activity of naftifine in vitr03 and in viv04 led to its clinical development, and this drug has been marketed since

anti-1985 as a topical antimycotic Naftifine provided the basis for an extensive program of chemical derivatization5 - 8 aimed at improving the antimycotic efficacy, especially with regard to oral administration This goal was achieved

in the form of terbinafine (SF 86-327),1,6-13 the efficacy of which has now been confirmed in numerous clinical studies involving both topical and oral application Parallel to this development, detailed investigations were car-ried out concerning the mechanism of action of the allylamines, 14- 26 includ-ing much basic research on the biochemistry of ergosterol biosynthesis in pathogenic fungi

A considerable body of literature on the experimental and clinical erties of the· allylamines has now arisen in the decade since the first presentation27 of naftifine In this chapter, we aim to present a comprehen-sive overview of the available data on this new class of antimycotics, with emphasis on the underlying biological and biochemical aspects of relevance

prop-to their clinical application

Experimental Antimycotic Activity

Spectrum of Antifungal Action In Vitro

In in vitro tests, both naftifine and terbinafine are active against a broad spectrum of pathogenic fungi.3,9,10,12,13,28 Activity is extremely high against dermatophytes, with minimum inhibitory concentrations (MICs) in the

158

FIG 6-1 Structures of the allylamine antimycotics naftifine (1), terbinafine (2Land SDZ 87-469 (3)

nanogram/milliliter range in the case of terbinafine and generally very good against the majority of filamentous and dimorphic fungi The 3-chloro-7-benzo[b]thienyl derivative SDZ 87-469 is even more active than terbinafine against most fungi tested 29.30 Terbinafine has also been reported to have activity against a wide range of unusual fungal pathogens 31- 34 Table 6-1 summarizes the published data on the spectrum of action of these three allylamines against filamentous and dimorphic fungi, and Table 6-2 shows data for yeasts In addition to its effects on human and animal pathogens, terbinafine has also been reported to inhibit several fungi of agricultural importance as plant pathogens 42-45

Activity of the allylamines against yeasts is quite variable between ent species and strains For example, Candida parapsilosis is very suscepti-ble whereas C glabrata shows little response (Table 6-2) Susceptibility of

differ-C albicans varies considerably among strains SDZ 87-469 is markedly more active than terbinafine against pathogenic yeasts in generaJ.29 Of particular interest is the much higher susceptibility to allylamines of the filamentous form ofC albicans,22.46.47 which plays an important role in pathogenicity of this organism This is probably a significant factor in the clinical efficacy of allylamines against Candida infections

It should be noted that the data in Tables 6-1 and 6-2 are derived from several workers using a variety of methods and are intended only as a guide

to the spectrum of activity of the allylamines In vitro antifungal data are notoriously variable and can be strongly influenced by factors such as medium composition, pH, inoculation size, and endpoint interpretation 48

162 Neil S Ryder and Hubert Mieth TABLE 6-2 Spectrum of action in vitro of naftifine, terbinafine, and SDZ 87-469 against pathogenic yeasts

Range of MIC values (J.tglml) Organism Naftifine Terbinafine SDZ 87-469 a Reference(s)

"All data for SDZ 84-469 are from references 29 and 30

Activity of a compound in vitro is no guarantee of in vivo efficacy, while compounds virtually inactive in vitro may show high activity in animal mod-els In our experience, the in vitro activity of allylamines against der-matophytes correlates quite well with potential for in vivo efficacy and ,also with biochemical activity (see later sections) In contrast, MIC values for other fungi, especially yeasts, are difficult to correlate with experimental or clinical antimycotic efficacy

Fungicidal Action

One of the decisive factors in the selection of naftifine for further ment was the early recognition of its primary fungicidal action (Le., the MIC equals the minimum fungicidal concentration).3,12 Fungicidal action is an important attribute of an antimycotic, contributing to rapid therapeutic ac-tion, and is of particular significance for treatment of immunocompromised patients The specific mechanism of action of the allylamines appears to be responsible for the characteristic fungicidal action of this class of compounds,

develop-as discussed below Terbinafine is fungicidal in vitro against dermatophytes

and against dimorphic and filamentous fungi,9,10 including Aspergillus

species37 (Fig 6-2) SDZ 87-469 and other allylamines, as far as they have been tested, have similar properties The type of activity against yeasts is

variable, being for example fungicidal against C parapsilosis and in general fungistatic against C albicans (Fig 6-3) Terbinafine is secondarily fungicid-

al against C albicans at higher concentrations 10,12

The hair root invasion test for measuring fungal skin infections was veloped with the specific aim of evaluating the fungicidal action of test com-

de-Terblnaflne: Fungicidal activity against T.mentagrophvtes t: 158, t.t.canls t: 846,

A.fumlgatus I: 159, Sc.brevlcaulls t: 901

cJ.u./mI and S.schenckll t: 177 respectively

FIG 6-2 Primary fungicidal action of terbinafine against Trichophyton rophytes, Microsporum canis, Aspergillus fumigatus, Scopulariopsis brevicaulis, and Sporothrix schenckii From reference 11

mentag-Terblnaflne: Fungistatic and fungicidal activity against

C.alblcans t: 9 and C.parapsllosls t: 39 respectively

and C parapsilosis From reference 11

164

100 a<' 80 5

~60 '"

days starting 48 h after inoculation • , Clinical efficacy; , mycological efficacy

From reference 49

pounds in experimental animals.49 To fulfill the criteria of efficacy, the test drug must penetrate to the depth of the hair follicle in sufficient concentra-tion to exert a fungicidal action Both naftifine (Fig 6-4) and terbinafine exhibit high efficacy in this discriminating model 11,49 An autoradiographic study using radiolabeled naftifine has demonstrated that the drug penetrates rapidly into both the stratum corneum and the hair follicles after a single application to guinea pig skin.50,51 The experimental evidence thus indicates that the allylamines have the potential to exert a fungicidal action during therapeutic application

Antimycotic Activity in Experimental Animals

Naftifine proved to be highly effective as a topical treatment in several guinea pig models of dermatophytosis using either clinical or mycological criteria of efficacy (summarized in Table 6-3) A novel experimental approach involving the establishment of dermatophyte infections on the ears of guinea pigs has also been used (Fig 6-5) The resulting inBammation is quantified

by measuring the increase in skin surface temperature, application of tive antimycotic therapy leading to a reduction of both inBammation and skin temperature.13,52 Treatment of such animals with naftifine (1% cream, 2 days) led to a rapid drop in skin temperature, which then remained in the range of normal untreated animals 52 This model is, however, more difficult

effec-to interpret than the standard mycological models, as efficacy may also be inBuenced by direct anti-inBammatory action of the test drug There is clinical evidence (discussed below) that naftifine does indeed possess such activity

Terbinafine is considerably more potent than naftifine in topical treatment

of experimental dermatomycoses (summarized in Table 6-4) More

sig-TABLE 6-3 Antimycotic activity of naftifine (topical administration) in experimental animals in vivo: summary of published studies

Guinea pig Guinea pig Guinea pig (hair root invasion test) Guinea pig

(skin temperature model)

Drug concn (%) for 100%

mycological cure 0.5

Trichophyton mentagrophytes, and the animals were treated orally once daily for 10

consecutive days Severity of inflammation was assessed by measurement of skin temperature with a sensor From reference 51

166 Neil S Ryder and Hubert Mieth

TABLE 6-4 Antimycotic activity ofterbinafine in experimental animals in vivo: summary of published studies

• In one of three strains tested Activity was strain dependent: 1% gel gave 95, 90, and 50% cure rates, respectively, with the three strains

nificantly, terbinafine also shows outstanding efficacy after oral application (Fig 6-5) with 100% mycological cure of Trichophyton mentagrophytes in-fections at a dose of6 mglkglday,ll Naftifine has only borderline oral activity

at high doses (ISO mglkg).6 Terbinafine is also active topically against cutaneous infections of C albicans in guinea pigs and in a rat vaginal candi-dosis model.u SDZ 87-469 possesses oral and topical antimycotic activity similar to that of terbinafine in the models described here 29,30

Despite the excellent oral activity of terbinafine in dermatophytoses, there is at present no indication that the compound is effective in ex-perimental models of systemic mycoses Negative results were obtained in murine models of sporotrichosis, 55 phaeohyphomycosis,56 and cryptococco-sis57 and in aspergillosis in guinea pigs 58 The reason for this is not clear but may be related to the pharmacokinetic properties of this drug (discussed below)

Antiprotozoal Activity

Certain parasitic protozoa biosynthesize sterols in a manner similar to that of fungi, thus rendering them potentially suceptible to antifungal agents in-hibiting ergosterol biosynthesis Terbinafine has been shown to inhibit both growth and sterol biosynthesis in Leishmania mexicana promastigotes,59 while naftifine and terbinafine inhibit growth of L major amastigotes in macrophage cultures.60 Terbinafine also inhibits growth in vitro of the im-portant pathogen Trypanosoma cruzi (the agent of Chagas disease) and, in-terestingly, shows strong synergy with the ergosterol biosynthesis inhibi-tor ketoconazole 61

Squalene Squalene epoxide Lanosterol 14-Demethyllanosterol

FIG 6-6 Simplified pathway of ergosterol biosynthesis Modified from reference 21

Biochemical Mechanism of Action

Inhibition of Ergosterol Biosynthesis

Initial investigations14,15 showed that naftifine-treated fungal cells become

deficient in ergosterol while accumulating the intermediate squalene

Sub-sequently, extensive studies were carried out to characterize the effects,

qualitative and quantitative, of allylamines on ergosterol biosynthesis in

var-ious pathogenic fungi Ergosterol biosynthesis can be readily evaluated by

incorporation of radiolabeled acetate in whole cells or of mevalonate in

cell-free extracts (Fig 6-6) Untreated cells of yeasts and filamentous fungi

incorporate acetate primarily into the ergosterol fraction of the

nonsaponifi-able lipids, with traces of activity in precursors such as squalene and the

4-methylsterols.16,18,24,25 Treatment with an allylamine leads to a

dose-dependent inhibition of incorporation into ergosterol accompanied by a

parallel accumulation of labeled squalene Dermatophytes are exquisitely

sensitive to this inhibition, showing reproducible effects at terbinafine

con-centrations below 1 ng/ml (Table 6-5) Qualitatively similar results have been

obtained with cells of several filamentous fungi and yeasts treated with

naf-tifine, terbinafine, or SDZ 87-469 Examining the results from a large

num-ber of experiments (Table 6-6), it is clear that inhibition of fungal growth is

related to potency of ergosterol biosynthesis inhibition Both the respective

antifungal activities of naftifine and terbinafine and the high susceptibility of

168 Neil S Ryder and Hubert Mieth TABLE 6-5 Effect ofterbinafine on incorporation of[14C]acetate into sterols and squalene in

Terbinafine Total % Total incorporation by:

a Data are from reference 25

TABLE 6-6 Concentrations ofnaftifine and terbinafine causing inhibition of fungal growth (MICs) and of sterol biosynthesis in cells of various pathogenic fungi

MIC

(p,g/ml)

0.05 0.05 1.6

50

100 0.003 0.003 0.8 0.4 3.1

100

Inhibitory concn a

(JLg/ml)

50% 95% 0.005

0.006

0.23 0.35 0.34 0.0005 0.002 0.07

0.006

0.008 0.040

0.11

0.12 4.0 7.5 11.3 0.02 0.04 1.2 0.3 0.2 0.9 aConcentration of drug causing 50 and 95% inhibition of sterol biosynthesis compared with untreated controls (mean of three separate experiments) Sterol biosynthesis was measured by (14C), acetate incorporation MIC values are in medium l6 buffered at pH 6.5 Data are from reference 23

dermatophytes in comparison with other fungi correlate with extent of sterol biosynthesis inhibition

ergo-A qualitatively and quantitatively similar inhibition by allylamines is found

in cell-free ergosterol biosynthesis preparations from C albicans, using

mevalonate as a substrate (Table 6-7) The agreement between inhibitory concentrations in whole cells and cell-free extracts indicates that the Candi-

da cell envelope does not represent a significant barrier to uptake of the

drugs Up to now it has not proved possible to develop a cell-free ergosterol biosynthesis assay from dermatophytes, leaving open the question as to whether their high sensitivity is due to drug accumulation in the cells or to greater sensitivity of the target enzyme

TABLE 6-7 Inhibition by allylamines of ergosterol biosynthesis in cell-free

prepara-tions and whole cells of Candida albicans

Test system

Strain 124 cell-free prepn

Strain 124 cells (yeast)

Strain 126 cells (yeast)

Strain 126 cells (mycelial)

Concn (/-tg/ml) for 50% inhibition a

Naftifine 0.19 0.35 1.21 0.17

Terbinafine 0.008 0.008 0.014 0.013

SDZ 87-469 0.003 0.011 0.020 0.010

a Mean of three separate experiments, each performed in triplicate

In view of the much greater susceptibility of the filamentous form of C

albicans to allylamines,46,47 the inhibition of ergosterol biosynthesis was

compared in pure yeast and mycelial cultures of a C albicans strain selected

for its ability to grow as mycelium in a defined medium As shown in Table

6-7, slightly lower drug concentrations are required to inhibit ergosterol biosynthesis in the filamentous form than in the yeast form (significantly so only in the case of naftifine) These differences do not account for the much greater susceptibility of the filamentous form to the drugs 46,47 It would seem that the filamentous growth form is inherently more susceptible to inhibition of ergosterol biosynthesis, in agreement with the data for other fungi in Table 6-6 A similar situation regarding C albicans occurs in the

case of the azoles,62,63 which inhibit ergosterol biosynthesis at a later stage of the pathway (lanosteroI14a-demethylation)

Further information concerning effects of allylamines on cellular

ergoster-ol biosynthesis was obtained by using the technique of sterergoster-ol side chain methylation with [methyP4C]methionine as a substrate (Fig 6-6).19,24

In Candida, side chain methylation normally occurs distal to nuclear

demethylation,I9 as in Fig 6-6, while in some other fungi, such as

Aspergil-lus, methylation of 4-methylsterols also takes place In the case of a full

inhibition of squalene epoxidase, the side chain methylation technique can measure any residual ergosterol biosynthesis occurring distal to the point of inhibition and utilizing endogenous sterol precursors The extent of this re-sidual biosynthesis varies considerably among different fungi treated with terbinafine and is inversely proportional to the susceptibility of the fungus to the drug (Table 6-8) The effectively total and immediate blockade of ergo-sterol biosynthesis in dermatophytes may be a significant factor in the high efficacy of allylamines against these fungi

The occurrence of a significant degree of residual side chain methylation, which is not further reduced by allylamine concentrations greater than those required for full inhibition of squalene epoxidation, indicates that the com-pounds do not inhibit any distal stages of the ergosterol pathway 19 Candida

cell-free systems using acetyl coenzyme A, squalene epoxide, and lanosterol

as substrates were additionally used to confirm that naftifine and terbinafine

170 Neil S Ryder and Hubert Mieth TABLE 6-8 Residual ergosterol biosynthesis measured by sterol side

chain methylation in fungal cells inhibited with terbinafine

10 1.0 1.0

Residual biosynthesis (% of control)a 0.0

0.5± 0.4 9.7± 2.1 3.6 ± 1.0 39.1 ± 10.8

(2) (6) (6) (5) (10)

a Degree of residual sterol biosynthesis occurring over 3 h in cells treated with terbinaflne at the concentrations shown, which block squalene epoxidation by more than 95% Biosynthesis was measured by incorporation from [methyl-

14Cjmethionine.19 Results are in the form mean ± standard deviation The number of experiments is given in parentheses Data are from reference 23

inhibit only squalene epoxidation.18,25 More than 50 compounds of the lamine class have been examined in one or more of the ergosterol biosyn-thesis assays described above and found to inhibit squalene epoxidation to varying degrees There is thus a large body of evidence to show that the antifungal activity of the allylamines is due to specific inhibition of ergosterol biosynthesis at the level of squalene epoxidation

ally-Mechanism of Fungicidal Action

Ergosterol is an essential component of the cell membrane in virtually all fungi Inhibition of ergosterol biosynthesis can therefore easily explain the growth-inhibiting activities of the allylamines on the basis of the evidence summarized previously However, this does not obviously account for the primary fungicidal action of these drugs, especially in view of the fact that the other main group of antifungal ergosterol biosynthesis inhibitors (the azoles) is fungistatic in action The rather slow onset (over days) of cell death

in allylamine-treated fungi (Figs 6-2 and 6-3) suggests that cidal action is secondary to ergosterol biosynthesis inhibition rather than due to direct cell-damaging effects of the drug This was confirmed by studies which failed to show primary effects of naftifine, terbinafine, or SDZ 87-469 (up to 100 JLg/

ml) on membrane integrity, cellular uptake processes, respiration, or nucleic acid and protein biosynthesis in Trichophyton and Candida At high concen-trations (100 JLg/ml), naftifine affects several cellular processes in C albicans,

including interference with uptake of adenosine and methionine17 and ulation of endogenous respiration.23 These effects are specific to naftifine and may contribute to its topical efficacy against C albicans

stim-A clue to the mechanism of cidal action may be obtained from examination

of the data in Table 6-6, from which fungi may be classified into two groups: (1) those in which complete growth inhibition is achieved with drug concen-trations that cause only partial inhibition of ergosterol biosynthesis and (2)

., ~2

t::

., u;

<0 g10 0 ~

0.01~

FIG 6-7 Effect ofterbinafine (0.003 ~g/ml) on viability and lipids in Trichophyton mentagrophytes cells • , Viable cell count; f::", squalene content; 0, ergosterol con- tent From reference 64

those in which complete growth inhibition requires total inhibition of sterol biosynthesis Fungi subject to cidal action by the allylamines are in the first group, suggesting that rather than ergosterol deficiency, the accumula-tion of squalene might be the critical effect leading to fungal cell death Detailed studies of the correlations between cell death, growth inhibition, and lipid content in several fungi treated with allylamines have confirmed that cidal action is indeed associated with squalene accumulation in both e

ergo-parapsilosis and T mentagrophytes 21 64 In Trichophyton cells treated with terbinafine at its MIC, cell death coincides with a massive rise in intracellu-lar squalene concentration, while ergosterol is only slightly depleted (Fig 6-7) The possible mechanism of toxicity of squalene in fungal cells is not clear, although it might be expected to increase membrane fluidity.65 Ultra-structural studies of fungi treated with naftifine66•67 or terbinafine68•69 re-vealed the presence of large numbers of lipid bodies in the cytoplasm and also in the cell wall, probably representing the location of the accumulating squalene It has recently been shown that in yeast cells treated with naf-tifine, the accumulated squalene was present in the cytosol rather than in the membrane fraction 70

In contrast, the fungistatic action of allylamines in e albicans correlates well with depletion of ergosterol Maximum levels of squalene are reached at concentrations of naftifine16 or terbinafine17.26 causing only partial inhibition

of growth At higher drug concentrations, sterol precursors disappear, sterol is depleted to about 40% of the control level, and growth ceases Inhibition of Squalene Epoxidase

ergo-The experimental evidence reviewed above indicates that squalene idase is the primary target of the allylamines in fungi This enzyme was characterized in microsomes ofe albicans 71 and e parapsilosis20 and found

epox-172 Neil S Ryder and Hubert Mieth

INHIBITION OF SQUALENE EPOXIDASE BY NAFTIFINE: DIXON PLOT

FIG.6-B Inhibition of Candida albicans squalene epoxidase by naftifine in the

pre-sence of increasing substrate concentrations (Dixon analysis) Squalene tions: 0,14.7 p.M;., 32.2 p.M; 0,116.9 p.M From reference 20

concentra-to be potently inhibited by naftifine and terbinafine.20 The enzymology of squalene epoxidase and its inhibitors is the subject of a comprehensive re-cent review72 and will therefore not be discussed in detail here Allylamines are specific reversible inhibitors of the Candida epoxidase, with apparently noncompetitive kinetics (Fig 6-8), and also inhibit the enzyme in solubilized form.22 All antifungally active allylamines that have been tested were found

to inhibit the squalene epoxidase The extent of epoxidase inhibition is ficient to account for the observed inhibition of ergosterol biosynthesis in

suf-Candida cells Comparison of naftifine, terbinafine, and SDZ 87-469 shows that their potency as epoxidase inhibitors correlates well with their respec-tive antifungal activities (Table 6-9) Clearly then, epoxidase inhibition is a requirement for activity of the allylamines However, some effective epox-idase inhibitors were found to have little or no antifungal activity, presum-ably because of instability or lack of uptake into fungal cells Also, intrinsic physiological differences among different fungi must play an important role

in determining susceptibility, as in the case of the two Candida species which have widely differing susceptibility to allylamines (Table 6-6) but very similar sensitivity in terms of squalene epoxidase inhibition (Table 6-9) Despite extensive biochemical studies, the precise mechanism of inhibi-tion of the epoxidase by allylamines is still not clear One possibility is that

TABLE 6-9 Inhibition offungal and mammalian microsomal squalene

144

Terbinafine 0.03 0.04

93

SDZ 87-469 0.01 0.02

One of the clinically relevant implications of the mechanism of action of the allylamines is with regard to fungal resistance Although development of resistance to clinically used antimycotics has been rare up to now, this might become a significant problem in the future, especially in the context oflong-term treatment of immunocompromised patients Pathogens that have de-veloped resistant to antimycotics such as azoles or polyenes would not be expected to be cross-resistant to allylamines, which have a completely differ-ent mode of action Studies with several different fungi have demonstrated that cross-resistance is indeed absent at least in the case of experimentally produced mutants.44,45,77 Further implications of epoxidase inhibition for therapeutic application are discussed in the following section

Effects on Mammalian Systems

Cholesterol Biosynthesis

Squalene epoxidase is an essential enzyme in the biosynthesis of cholesterol

in mammals as well as of ergosterol in fungi, and the mammalian and fungal enzymes have quite similar properties 72 Unlike fungi, mammals obtain much of their cholesterol from dietary sources and would thus not be ex-pected to be as sensitive to inhibition of sterol biosynthesis; however, severe

174 Neil S Ryder and Hubert Mieth

inhibition might be toxic to certain tissues The effects of allylamines on cholesterol biosynthesis have therefore been investigated in detail, particu-larly in the case of systemically acting compounds such as terbinafine In

preliminary studies, a rat liver cell-free system similar to that from Candida

was used, with labeled mevalonate as a substrate.l8,24 Naftifine has only a very weak nonspecific inhibitory effect in this system, while terbinafine at high concentrations shows a specific inhibition of squalene epoxidation (50% inhibition at 30 J.Lglml) There is no evidence for inhibition of any other steps

in the cholesterol pathway Cholesterol biosynthesis is thus several orders of magnitude less sensitive to terbinafine than is fungal ergosterol biosynthesis

in similar experimental systems Later studies have confirmed that na6ne has no significant effect on cholesterol levels either in experimental animals or in patients undergoing oral therapy with the drug Terbina6ne was recently reported78 to inhibit platelet-derived growth factor-induced mitogenesis in smooth muscle cells in vitro, and also neointimal proliferation

terbi-in the rat carotid artery terbi-in vivo after oral admterbi-inistration (200 mglkglday) This phenomenon might be due to inhibition of cholesterol biosynthesis (which is important in cell proliferation) in view of the high dose used, but it cannot be ruled out that the drug has some other antiproliferative effect at high concentrations

Following discovery of the surprising degree of selectivity of the mines for fungal but not mammalian sterol biosynthesis, comparative studies were performed with the respective squalene epoxidases in order to clarifY the enzymatic basis of selectivity 20,22 There are both qualitative and quan-titative (several orders of magnitude) differences in the effects of terbina6ne

allyla-on the Candida and rat liver epoxidases.2o,72 The very low sensitivity of the mammalian enzyme (Table 6-9) explains the lack of effect of allylamines on cholesterol biosynthesis in vitro and also in vivo, since sufficiently high drug concentrations would not be reached Detailed investigations of mammalian squalene epoxidases22 revealed that soluble cytosolic factors in liver partially protect the epoxidase from inhibition by terbina6ne However, selectivity is primarily due to intrinsic differences between the respective epoxidase en-zymes, although the nature of these differences at the molecular level is still unknown.72 The occurrence of such a high degree of inhibitor selectivity between two very similar enzymes, which could not have been predicted from our knowledge of these enzymes, has interesting implications for the development of pharmacological agents in general

Cytochrome P-450

Squalene epoxidase from either fungal or mammalian sources is not an zyme of the cytochrome P-450 superfamily, 71, 72 so that allylamines have no inherent tendency to inhibit this important class of enzymes Many im-portant physiological functions, including drug metabolism and steroid

en-hormone biosynthesis, involve cytochrome P-450-dependent reactions Sterol 14a-demethylation, the target of azole antifungals, is catalyzed by a cytochrome P-450, thus providing the potential for a variety of side effects For example, ketoconazole has been reported to inhibit steroid hormone biosynthesis,79 hepatic drug metabolism,80 and cholesterol biosynthesis,81 although the more recently developed azoles have greatly improved selec-tivity In contrast, spectral binding studies with naftifine and terbinafine indicate that these drugs do not interfere with cytochromes P-450 from a range of tissues involved in biosynthesis of steroid hormones and prostaglandins 82,33 No significant binding was found in the case of human placenta (microsomes), bovine testis (microsomes), bovine adrenals (mito-chondria and microsomes), and bovine aorta (microsomes) The allylamines also had no effect on binding of the endogenous substrates or of carbon mon-oxide to cytochrome P-450 from these tissues In liver microsomes, terbina-fine is degraded predominantly by cytochrome P-450-mediated oxidation and shows a typical substrate (type I) binding spectrum 84 Terbinafine binds with moderate affinity to about 25% of the cytochrome P-450 in liver microsomes from rats and guinea pigs but to less than 5% of cytochrome P-450 in human liver.84 In agreement with these binding studies, investiga-tions with human liver microsomes show terbinafine to have no effect on cytochrome P-450-mediated drug metabolism in vitro 85,86 In addition, ter-binafine shows no potential for induction of drug-metabolizing enzymes, as indicated by experiments in which rats were treated for up to 3 months with the compound 13

These in vitro studies indicating lack of inhibition of cytochrome P-450 by allylamines have been confirmed in clinical investigations with human volunteers Administered orally at therapeutic doses, terbinafine had no effect on either volume of distribution or clearance of antipyrine 85,87,88 Ter-binafine (500 mg orally) reduced clearance of intravenous caffeine,89 suggest-ing potential for competition between terbinafine and other drugs showing type I binding Two clinical studies90,91 have shown that terbinafine (in con-trast to ketoconazole) has no effect on plasma testosterone levels in healthy males

Pharmacokinetics

The development of naftifine and terbinafine has necessitated numerous studies on the pharmacokinetics, distribution, and biotransformation of the compounds in vivo, of which only a brief overview will be presented here The clinical pharmacokinetic properties of terbinafine have been recently summarized by Jensen, 88 and detailed studies have been published concern-ing the pharmacokinetics13,84,92 and biotransformation92,93 of allylamines in various animal species Absorption of orally administered terbinafine is high

in all species tested88,93 and at least 70% in humans (after a single 250-mg

176 Neil S Ryder and Hubert Mieth

oral dose) The bulk of the absorbed drug is transported by the portal route

to the liver, thereby undergoing a fairly heavy first-pass metabolism before reaching the systemic circulation The metabolism of naftifine in the liver appears to be much more rapid (about lO-fold) than that of terbinafine,13 which probably explains the poor oral activity of naftifine Lymphatic trans-port also plays a significant role in the systemic distribution of both naftifine94 and terbinafine,13 as a result of association of the lipophilic drug with the chylomicrons As expected for such a lipophilic compound, terbi-nafine is highly bound to plasma proteins 88 This binding is nonsaturable at therapeutic concentrations and not specific, involving all plasma fractions, including albumin and the lipoproteins In humans, peak plasma levels of terbinafine (0.8 to 1.5 JLg/ml) are attained about 2 h after oral administration

of a 25O-mg dose A bioassay has been developed95 to enable determination

of terbinafine concentrations in serum, the results agreeing well with those obtained by high-pressure liquid chromatography analysis 96

Terbinafine is extensively metabolized to more polar derivatives, which lack antifungal activity The main biotransformation routes of the allylamines are N-dealkylation (at all three possible carbon atoms), oxidation at both aliphatic and aromatic carbon atoms, and conjugation with glucuronic acid and glycine The main metabolites of naftifine, terbinafine, and SDZ 87-469

have been identified in a number of species 92,93 The predominant lites found in humans have also been identified in one or more of the animal species investigated (rat, mouse, dog, rabbit, and guinea pig) The polar metabolites are excreted by both renal and biliary routes, urinary excretion being predominant in humans After topical application of naftifine in humans or animals, the drug is only very slowly released from the skin into the blood, but the pattern of metabolites remains qualitatively similar to that observed after oral administration.93 As a result of its lipophilicity, terbinafine accumulates preferentially in the skin and adipose tissue of ex-perimental animals,13 and probably also in humans,88 leading to relatively slow elimination of the drug This property is clearly of importance for the oral efficacy of terbinafine in dermatomycoses The apparent lack of activity

metabo-of terbinafine in animal models metabo-of systemic mycoses is difficult to explain, since the drug was found to be well distributed to the various organs 13 One possibility could be lack ofbioavailability, since terbinafine has a high affinity for intracellular components (higher than for plasma proteins) and is aVidly taken up by mammalian cells 13

Clinical Applications

Clinical therapeutic investigation of the allylamines was a logical step in view

of the high experimental chemotherapeutic activity of the drugs as described previously This decision was supported by the favorable preclinical toxico-