Khafrefungin, a Novel Inhibitor of Sphingolipid Synthesis*

In the course of screening for antifungal agents we have discovered a novel compound isolated from an endophytic fungus that inhibits fungal sphingolipid synthesis. Khafrefungin, which is composed of aldonic acid linked via an ester to a C22 modified alkyl chain, has fungicidal activity against Candida albicans,Cryptococcus neoformans, and Saccharomyces cerevisiae. Sphingolipid synthesis is inhibited in these organisms at the step in which phosphoinositol is transferred to ceramide, resulting in accumulation of ceramide and loss of all of the complex sphingolipids. In vitro, khafrefungin inhibits the inositol phosphoceramide synthase of C. albicans with an IC50 of 0.6 nm. Khafrefungin does not inhibit the synthesis of mammalian sphingolipids thus making this the first reported compound that is specific for the fungal pathway.

The incidence of human systemic fungal infections has increased dramatically in the last decade due to a rise in immunocompromised patients, including those receiving cancer chemotherapy and immunosuppressants, and in the human immunodeficiency virus infected population. Therapies are currently limited to a small number of compounds for the treatment of a rather diverse array of pathogenic fungi, which include Candida albicans and other Candida species, Cryptococcus neoformans, Aspergillus sp., and Histoplasma capsulatum. Each drug has limitations; toxicity is an issue with treatments based on amphotericin B, and resistance, which precluded the use of flucytosine as a stand-alone antifungal soon after its introduction, is now beginning to emerge as a problem with the azole and triazole class of inhibitors. Even with the most aggressive therapy that is available, the rate of mortality from aspergillosis is extremely high in some patient populations thus highlighting the need for the development of new treatments (1).
The sphingolipid biosynthetic pathway has been suggested as a good target for antifungal therapy (2). Although sphingolipids comprise a relatively small proportion of fungal phospholipids, they are essential. Saccharomyces mutants that do not make sphingolipids are not viable (3), and pathogenic fungi treated with inhibitors of sphingolipid synthesis are killed (4,5). The initial steps in sphingolipid biosynthesis, from ketodihydrosphingosine synthesis through ceramide formation, are conserved in fungi and mammals except that fungi make phytosphingosine as their predominant sphingoid base with lesser amounts of dihydrosphingosine. Mammalian sphingolip-ids are composed of dihydrosphingosine and sphingosine. Fungi do not make sphingomyelin and instead transfer phosphoinositol to the C1 hydroxyl of ceramide to make inositol phosphoceramide (IPC) 1 . IPC is further modified by the addition of mannose to make mannosyl inositol phosphoceramide, and the addition of a second inositol phosphate group to make mannosyl diinositol diphosphoceramide (6). Some fungi, including pathogenic species of Aspergillus, have been reported to make glucosylceramide and lactosylceramide (7), but the major sphingolipids of Candida, Cryptococcus, and Histoplasma are based on IPC (6,8).
A number of inhibitors of the sphingolipid pathway that also have antifungal activity have been discovered, all from natural product sources. Three structurally distinct classes of inhibitors of serine palmitoyltransferase have been found: 1) the sphingofungin family, which includes sphingofungins A through F (4,9) and myriocin/ISP1, which has immunosuppressant activity in addition to its antifungal activity (10); 2) the lipoxamycins, which are produced by actinomycetes (11); and 3) a newly described family of inhibitors called the viridiofungins (12). All of these compounds are very potent inhibitors (nanomolar to picomolar) of serine palmitoyltransferase, and though they kill a broad array of pathogenic fungi, they also inhibit the mammalian enzyme. Two types of inhibitors of the ceramide synthase have been described: the fumonisins, which are inhibitors of mammalian ceramide synthase (13) and have poor antifungal activity although they do inhibit the fungal enzyme; and australifungin, which has very potent antifungal activity (5). The fumonisins are associated with severe toxicities in animals and possibly humans. Toxicity has been attributed not only to the loss of ceramide and complex sphingolipids, but also to the accumulation of the sphingoid base precursors of ceramide, which are components of signal transduction pathways (14,15).
In this report we describe a novel inhibitor that shows specificity for fungal sphingolipids. Khafrefungin inhibits the IPC synthase of Saccharomyces cerevisiae and pathogenic fungi at picomolar to nanomolar concentrations but does not inhibit mammalian sphingolipid synthesis. Ceramide accumulates in response to khafrefungin treatment, and fungi are killed. sphingosine was exposed to tritium gas and palladium on carbon, deacetylated, and purified by HPLC to isolate [4, H]dihydrosphingosine. N-stearoyl-D-sphingosine was incubated with NaBT 4 to synthesize N-stearoyl-D- [4, H]dihydrosphingosine ( 3 H-ceramide) that was purified by HPLC.
Antifungal Assays-Growth inhibition was determined by microtiter broth dilution assay in Difco yeast nitrogen base medium containing 2% glucose (YNBD) and 0.078% complete supplement mixture (CSM, Bio101). Cells were inoculated at A 600 nm ϭ 0.002 (about 2 ϫ 10 4 yeast cells/ml), and serial 2-fold dilutions of inhibitors were made from 125 g/ml (250 g/ml for A. fumigatus); the minimum inhibitory concentration value was the lowest inhibitor concentration, which prevented visible growth after 48 h at 30°C. Absorbance readings were obtained using a SLT 340 ATTC (Tecan Instruments) after cell resuspension. Minimum fungicidal concentrations were determined by diluting 0.1 ml of the drug-treated cell suspension and spreading aliquots onto agar plates to monitor colony formation. The minimum fungicidal concentrations value was the lowest inhibitor concentration, which reduced the viable cell count by 99% of the inoculum level.
Microtiter Assay for Sphingolipid Synthesis-Cells were grown in YNBD medium with 1% casamino acids buffered to pH 5.2 with 40 mM sodium succinate. Logarithmic phase cells at 2.5 ϫ 10 6 cells/ml were mixed with 2.5 Ci/ml [ 3 H]inositol and dispensed into 96-well plates containing inhibitor. The plates were incubated at 30°C (S. cerevisiae and C. albicans) and 37°C (C. neoformans) for 120 min (S. cerevisiae) or 180 min (C. albicans and C. neoformans). The assay was terminated with the addition of trichloroacetic acid to 5%, and the plates were chilled at 4°C for 20 min. Precipitated cells were harvested onto filtermats and total [ 3 H]inositol incorporated into both phosphatidylinositol (PI) and sphingolipid was quantitated in an LKB BetaPlate liquid scintillation counter (Wallac). The filtermats were removed from scintillant and treated with 0.1 N KOH in methanol/toluene (1:1) at room temperature for 60 -120 min and washed sequentially in methanol, 5% trichloroacetic acid and twice more in methanol. The remaining radiolabeled material (alkali stable sphingolipids) was measured by counting the filtermats again, and the counts incorporated into PI were calculated by subtracting the sphingolipid counts from the total.
Lipid Synthesis and TLC Analysis-1-ml cultures of C. albicans were grown to 0.25 A 600 nm in YNBD medium at 30°C and labeled with 2 Ci/ml [ 3 H]dihydrosphingosine for 30 min. Lipids were extracted and resolved by TLC in CHCl 3 , methanol, 4.2 N NH 4 OH (9:7:2) as described (17). To isolate the lipid that accumulated with khafrefungin treatment, 5-ml cultures of C. albicans were labeled with 0.5 Ci/ml [ 14 C]acetate for 2 h in the presence and absence of 0.4 g/ml khafrefungin, and extracted lipids were resolved by TLC in CHCl 3 , methanol, 2 N NH 4 OH (40:10:1). Radiolabeled bands were visualized by overnight exposure to x-ray film (Kodak XAR5), and the material that accumulated in response to khafrefungin treatment was scraped and eluted three times in 1.5 ml chloroform/methanol/H 2 O (16:16:5).
HepG2 cells (ATCC HB-8065) maintained in minimum essential medium containing streptomycin, penicillin, 2 mM glutamine, and 10% fetal bovine serum were seeded into 6-well plates for labeling. After 2-3 days growth to about 70% confluence, the cells were washed two times with 2 ml of phosphate-buffered saline to remove residual serum before 1 ml of serine-free Dulbecco's minimum essential medium (Life Technologies, Inc.) was added. Inhibitors or vehicle were added to cells at a final concentration of 1% solvent (methanol or Me 2 SO), and the cells were incubated at 37°C for 1 h before labeling with 20 Ci/ml of [ 3 H]serine. After a 5 h incubation period, the medium was removed; the cells were washed two times with cold phosphate-buffered saline and scraped into 1 ml of phosphate-buffered saline. Lipids were extracted by adding sequentially 1.5 ml of chloroform:methanol (1:2), 1 ml of chloroform, 1 ml of H 2 O. The chloroform layer was collected and washed two times with 2 ml of H 2 O. Extracts were dried and subjected to alkaline methanolysis using monomethylamine reagent as described (17). Half of the deacylated sample was applied to TLC plates and resolved as described above.
Fatty Acid Analysis-To identify the fatty acid component of the material that accumulated with khafrefungin treatment, heptadecanoic acid (50 g), tricosanoic acid (50 g), and [ 3 H]palmitate (250,000 dpm) were added to the eluted material as internal standards. The samples were dried under N 2 and hydrolyzed with 1 ml 2 N KOH in methanol/ H 2 O (2:1, v/v) at 80°C for 18 h. The hydrolysates were cooled to room temperature, acidified with 1 ml 3 N HCl and 1 ml of H 2 O, and extracted three times with 1.5 ml of petroleum ether. The ether extracts were combined, washed with 1.5 ml of 0.5 N HCl, and evaporated to dryness. Fatty acids were converted to phenacyl derivatives as described (18) and chromatographed on a YMC analytical C8 column (4.6 ϫ 250 mm).
Elution was carried out at 40°C with a methanol-acetonitrile-H 2 O gradient (75:25:25 to 95:2.5:2.5 over 60 min) at 2 ml/min. Absorbance was monitored at 242 nm, and 0.4 min fractions were collected across the gradient and counted by liquid scintillation counter.
Sphingoid Base Analysis-To analyze the sphingoid base component of the material that accumulated during khafrefungin treatment, type IV ceramides (Sigma) and [ 3 H]dihydrosphingosine (250,000 dpm) were added as internal standards and dried under N 2 . The sample was acid hydrolyzed, extracted, and the long chain bases were converted to biphenylcarbonyl derivatives as described (19). The derivatives were resolved with methanol-H 2 O (90:10) at 1 ml/min on a Beckman Ultrasphere 5 m ODS column (250 ϫ 4.6 mm) fitted with a Phenomenex Spherex 5 m C18 guard (30 ϫ 4.6 mm). Absorbance was monitored at 254 nm, fractions were collected every 0.5 min, and radioactivity was quantitated by liquid scintillation counter.
In Vitro IPC Synthase-Microsomal membranes were prepared from C. albicans, S. cerevisiae, and C. neoformans (Cap64) cultures grown to A 600 of 1.0 as described (5). In vitro IPC synthase reactions (100 l) contained 50 mM Tris-HCl, pH 7.0, 50 mM KCl, 0.25% sodium cholate, 5 g of membrane protein, 25 M PI, and traces of 3 H-ceramide (stearoyl [ 3 H]dihydrosphingosine, 0.18 mCi/ml) with or without inhibitor. Khafrefungin was added from an ethanol stock; the concentration of ethanol in the assay was less than 2% and did not affect IPC synthase activity. The substrates were combined with 0.1% ␤-octylglucoside and sonicated until clear (ϳ10 s in a special ultrasonic cleaner, model G112SP1G, Laboratories Supplies Co.) and diluted 10-fold into the assay to initiate the reaction. The reaction was conducted at 22-25°C for 1 h and quenched with 20 l of a 5% (w/v) deoxycholate solution in water. 3 H-IPC was separated from 3 H-ceramide on small (0.5 ml bed volume) anion exhange columns using Bio-Rad AG 4-X4 resin in 95% ethanol. The columns were equilibrated with formic acid and rinsed with water prior to use. The assay solution was applied to the column and 3 H-ceramide was removed by washing the columns with 20 ml of 95% ethanol. 3 H-IPC was eluted with 1 M K-formate in 95% ethanol, mixed with 5 ml of Aquasol, and counted in a Beckman liquid scintillation counter. Controls were carried out with the medium devoid of enzyme, and all data were corrected for the radioactive blank.

RESULTS
Antifungal Activity of Khafrefungin-In the course of screening for new antifungal agents, an unidentified sterile fungus cultured from a Costa Rican plant sample was found to produce a potent inhibitor of sphingolipid synthesis. Isolation of the active component resulted in identification of the novel inhibitor khafrefungin. 2 As shown in Fig. 1, khafrefungin is composed of aldonic acid ester linked to a C22 modified alkyl chain.
Khafrefungin inhibited the growth of many species of yeast and filamentous fungi in agar diffusion assays and in liquid broth. The growth of C. albicans, C. neoformans, and S. cerevisiae in liquid culture was inhibited at khafrefungin concentrations of 1 g/ml and higher, with minimum inhibitory concentrations of 2, 2, and 15.6 g/ml for the three organisms, respectively (Fig. 2). Khafrefungin killed the fungi with minimum fungicidal concentrations of 4, 4, and 15.6 g/ml for C. albicans, C. neoformans, and S. cerevisiae, respectively. Examination of C. albicans and S. cerevisiae cells after drug treatment did not reveal any gross changes in morphology or unusual distribution of budded cells indicative of cell cycle arrest. A. fumigatus was not sensitive to growth inhibition by the inhibitor in liquid or in agar and the hyphal morphology was not affected even at high concentrations (250 g/ml).
Khafrefungin Inhibits Sphingolipid Synthesis-The effect of khafrefungin on [ 3 H]inositol incorporation into lipids of C. al-bicans using a new microtiter-format assay, is shown in Fig. 3. Inositol is first incorporated into PI and then transferred to sphingolipids. These two fractions were distinguished by degrading the ester-linked PI with mild alkaline methanolysis; the remaining counts represent the amide-linked sphingolipids, which are resistant to saponification. Khafrefungin inhibited inositol incorporation into the sphingolipid fraction with an IC 50 of 0.09 g/ml (150 nM) but did not block inositol incorporation into PI. Compounds that inhibit sphingolipid synthesis at serine palmitoyltransferase (viridiofungin A) (12) or ceramide synthase (australifungin) (5) also specifically inhibited inositol incorporation into the sphingolipid fraction but were less potent than khafrefungin in this assay (Fig. 4).
Sphingolipid synthesis was tested in other organisms that were sensitive to growth inhibition by khafrefungin. Fig. 5 shows that khafrefungin inhibits inositol incorporation into the sphingolipids of S. cerevisiae and C. neoformans. These fungi were 4 -5-fold less sensitive than C. albicans with IC 50 values of 0.5 and 0.35 g/ml for S. cerevisiae and C. neoformans, respectively. In all organisms, the deacylation resistant fraction was confirmed to be composed of sphingolipids by TLC analysis of lipid extracts, and the inositol containing sphingolipids were the only lipids that were found to be inhibited by khafrefungin (data not shown).
Ceramide Accumulates with Khafrefungin Treatment-To identify the step in sphingolipid synthesis that was blocked by khafrefungin, C. albicans cells were labeled with precursors to the pathway ( Fig. 6 is a fluorograph of lipids labeled with [ 3 H]dihydrosphingosine, which is incorporated most readily into the sphingolipid fraction. At a concentration that inhibited most of the label incorporation into IPC and mannosyl inositol phosphoceramide (200 ng/ml), khafrefungin did not affect the sphingoid base intermediates, which accumulate when ceramide synthesis is blocked, but did cause accumulation of a very nonpolar lipid that migrated slightly below the stearoyl-dihydrosphingosine standard. In the presence of khafrefungin, all of the sphingolipid precursors except [ 3 H]inositol labeled this lipid, and simultaneous treatment with australifungin, the ceramide synthase inhibitor, prevented accumulation of this lipid (Fig. 6). Thus, by virtue of its mobility, sensitivity to inhibitors, and substrate composition, the accumulating lipid was most likely to be hydroxyceramide, the predominant species of ceramide made by fungi.
We wanted to verify that the accumulating intermediate was ceramide and identify the sphingoid base and fatty acid components. [ 14 C]acetate was used to label both components of ceramide, and the khafrefungin-dependent lipid that accumulated was isolated and hydrolyzed. Fractions were extracted, derivatized, and chromatographed by HPLC using methods designed to analyze long chain fatty acids (as their phenacyl derivatives) and sphingoid bases (as their biphenylcarbonyl derivatives). The accumulating intermediate was found to contain primarily a hydroxylated C 24 fatty acid with small amounts of hydroxy C 26 and nonhydroxy C 24 (Fig. 7A). In the sphingoid base analysis, the major component detected was C 18 phytosphingosine with a second broad peak that comigrated with both C 18 dihydrosphingosine and C 20 phytosphingosine, which are only partially resolved by this system (20) (Fig. 7B). By TLC analysis, which resolves dihydrosphingosine from phytosphingosine but does not separate species with different chain lengths, phytosphingosine was almost exclusively found (data not shown). Thus, khafrefungin causes the accumulation of hydroxyceramide with a composition that is consistent with a recent description of sphingolipids present in the hyphal form of C. albicans (20).
Khafrefungin Inhibits IPC Synthase-The accumulation of ceramide suggested that IPC synthesis might be the biosynthetic step that is blocked by khafrefungin. Inhibition of IPC synthase was confirmed in an in vitro enzyme assay. Previous assays for IPC synthase have relied on deacylation of [ 3 H]PI substrate and chromatographic methods (21,22) or differential organic extraction (2). An improved assay for IPC synthase was developed that uses a simple ion exchange procedure to separate [ 3 H]ceramide from [ 3 H]IPC. The assay has a precision of Ϯ3% standard error with background radioactivity less than 0.4%, and it is suitable for large scale screening. IPC synthase, being membrane bound and operating on membrane components as substrates, is not amenable to standard kinetic analysis, primarily because the concentration of substrates in crude membranes is difficult to control. Although IPC synthase can be assayed by adding trace [ 3 H]ceramide to a membrane containing endogenous PI, the rates of incorporation are slow and exogenous PI has little influence. Cholate, almost alone among a number of detergents tested, markedly enhanced the cata-lytic activity and extent of incorporation of [ 3 H]ceramide at concentrations below or approaching its critical micelle concentration and was adopted as a component in subsequent assays. Nonetheless, the reaction remained a complex function of PI, ceramide, and detergent concentrations. In the standard assay, khafrefungin was found to be a very potent inhibitor of the C. albicans enzyme with an IC 50 of 0.6 nM as shown in Fig. 8. The IPC synthase enzymes from S. cerevisiae and C. neoformans were 10-and 50-fold less sensitive with IC 50 values of 7 and 31 nM, respectively. These values appeared relatively insensitive to changes in substrate or detergent concentration, but increased linearly with increasing enzyme-membrane concentration.
Khafrefungin Does Not Inhibit Mammalian Sphingolipid Synthesis-Known inhibitors of serine palmitoyltransferase and ceramide synthase have approximately the same potency against fungal and mammalian enzymes. These compounds also inhibit mammalian sphingolipid synthesis in intact cells, as described below. We sought to determine whether khafrefungin inhibits mammalian sphingolipid synthesis given that fungal IPC synthase differs significantly from its mammalian counterpart in substrate specificity. Sphingomyelin synthase is the analogous enzyme in the mammalian pathway; it transfers phosphocholine to ceramide to make sphingomyelin. HepG2 cells were labeled with 3 H-serine and lipids extracted, deacylated, and separated by TLC (Fig. 9). Viridiofungin (5 M) prevented serine incorporation into sphingomyelin and ceramide, but did not affect serine incorporation into phosphatidylserine or phosphatidylethanolamine, which are converted to glycerophosphoserine and glycerophosphoethanolamine by alkaline transacylation. With australifungin treatment (10 M), ceramide and sphingomyelin synthesis were partially inhibited, and 3 H-serine incorporation was enhanced into several lipids, including dihydrosphingosine, glycerophosphoethanolamine, and a lipid comigrating with the nonpolar species of sphingomyelin which may be sphinganine 1-phosphate. These results with australifungin are the expected consequence of ceramide synthase inhibition, based on observations with fumonisin B 1 which has been shown to promote 3 H-serine incorporation into sphingoid bases and their degradation products (23). In contrast to these compounds that inhibit enzymes in the initial stages of sphingolipid synthesis, khafrefungin did not inhibit serine incorporation into sphingomyelin or any mammalian lipid at a concentration that was more than 300-fold above that required to inhibit IPC synthesis. Thus, khafrefungin is the first inhibitor described that is specific for fungal sphingolipid synthesis.

DISCUSSION
S. cerevisiae has been found to be surprisingly flexible in terms of its requirements for phospholipids. The phospholipid composition varies dramatically in different growth conditions and mutant studies have been particularly revealing about which lipids are essential (24,25). For instance, studies on mutants that lack phosphatidylserine synthase have resulted in the discovery that phosphatidylserine is dispensable for growth (although other processes are impaired), and analysis of mutants defective in the methylated phospholipids, phosphatidylethanolamine, phosphatidylmonomethylethanolamine, phosphatidyldimethylethanolamine, and phosphatidylcholine, indicate that these lipids can largely substitute for one another (25). In contrast, there is an essential requirement for PI, and cells that lack the ability to synthesize PI die rapidly in a process that is thought to be due to a lack of coordination between PI synthesis and other metabolic processes (26). PI has several important roles in Saccharomyces that include the regulatory aspects of the phosphorylated species, the synthesis of glycosylphosphatidylinositol anchors on mannoproteins that constitute a major component of the yeast cell wall and sphingolipid synthesis.
Sphingolipids, although present in relatively low abundance, are also essential, and inhibition of sphingolipid synthesis is lethal. Like PI, sphingolipids play multiple roles in yeast, any one of which could serve their essential function. Sphingolipids and their intermediates appear to be lipid-signaling molecules in mammalian cells where they regulate key enzymes involved in growth and differentiation (15,(27)(28)(29). They may play a similar role in Saccharomyces; some of the components of a ceramide signal transduction pathway have been identified and ceramide has been shown to activate a protein phosphatase and mediate G1 arrest (30). Sphingoid bases also have a regulatory role in yeast where they have been shown to inhibit several key enzymes in phospholipid biosynthesis when induced to accumulate by fumonisin treatment (31). Sphingolipids are involved in two ways with the synthesis of glycosylphosphatidylinositol anchors in Saccharomyces; a substantial proportion of mature glycosylphosphatidylinositol-anchored proteins are composed of IPC instead of PI in what appears to be a remodeling step (32), and inhibition of sphingolipid synthesis impedes processing of glycosylphosphatidylinositol-anchored proteins through the secretory pathway (33,34). Finally, sphingolipids appear to be the major, and perhaps sole repository for very long chain fatty acids (C 24  paired could explain the pleiotrophic nature of the fatty acid elongation mutants. It is intriguing that suppressor mutants selected for their ability to grow in the absence of sphingolipids make novel lipids that mimic sphingolipid structure, not only in their phosphoinositol and mannose-containing head groups, but also in the presence of a very long chain fatty acid in the SN-2 position of the diacylglycerol backbone (18).
Our understanding of the function of sphingolipids in fungi, albeit limited, comes entirely from work done in Saccharomyces. Other fungi, including pathogenic organisms, contain sphingolipids with structures that are very similar to those found in Saccharomyces (6,8), and through the use of specific inhibitors, it is now known that sphingolipids are also essential in the pathogens. Furthermore, there are significant differences between the mammalian and fungal sphingolipid biosynthetic pathways, thus making sphingolipid synthesis an attractive target for antifungal therapy. Fermentation extracts of fungi and bacteria appear to provide a diverse source of inhibitors to this pathway and several structural classes of inhibitors to serine palmitoyltransferase and ceramide synthase have been discovered (4, 5, 10 -13). Very recently, it has been reported that aureobasidin A1, a cyclic depsipeptide antifungal whose mechanism of action was unknown (37), is in fact an inhibitor of IPC synthase (2). In this paper we describe a novel compound isolated from an endophytic fungus that is structurally unrelated to the aureobasidins but inhibits IPC synthase.
Khafrefungin specifically inhibited sphingolipid synthesis in fungi that were sensitive to growth inhibition. The compound caused the accumulation of ceramide and inhibited IPC synthase in vitro at very low concentrations. Of the organisms tested, C. albicans was the most susceptible to whole cell sphingolipid inhibition and in vitro inhibition by khafrefungin, and in all likelihood, the antifungal activity is due to inhibition of IPC synthase. The toxicity of khafrefungin to fungi may be imparted not only by loss in biosynthesis of the mature sphingolipids, but also by the accumulation of ceramide, which mediates G1 arrest in Saccharomyces (30) and promotes cell death in an IPC synthase mutant that accumulates ceramide in response to exogenously added phytosphingosine (2). Khafrefungin does not have antifungal activity against an important pathogen, A. fumigatus, a limitation that is also shared by aureobasidin. Sphingolipids are thought to be essential in Aspergillus, as inhibitors at earlier steps in the pathway inhibit the growth of this pathogen (5,38). We have found that A. fumigatus does synthesize alkali-stable inositol lipids, and their synthesis is inhibited by khafrefungin but at much higher concentrations than those required for the other fungi. Lack of growth inhibition may be due to a resistant enzyme or poor uptake of the drug. Alternatively, the inositol-containing sphingolipids may not be essential in this organism, which is also reported to contain glycosylated sphingolipids (7). Future studies with inhibitors and mutants may help to determine the role of sphingolipids in Aspergillus.
Unlike reported inhibitors of serine palmitoyltransferase and ceramide synthase which affect fungal and mammalian enzymes with equal potency, khafrefungin did not inhibit mammalian sphingolipid synthesis. Since mammalian sphingomyelin synthase and fungal IPC synthase both use ceramide as a substrate, it is tempting to speculate that the highly hydroxylated acidic polar headgroup on khafrefungin (Fig. 1) confers fungal specificity by virtue of its resemblance to phosphoinositol. Unfortunately, we have been unable to test whether khafrefungin is a competitive inhibitor due to the complexities of the in vitro enzyme reaction that employs a crude membrane preparation, detergent, two lipid substrates, and a hydrophobic inhibitor that partitions into the lipid fraction. More rigorous kinetic analysis awaits purification of the enzyme, which may be facilitated by the recent identification of the IPC synthase gene in Saccharomyces (2). Khafrefungin should provide a useful tool to manipulate ceramide levels and probe the function of ceramide in fungal cell growth, differentiation, and stress response pathways.