Inhibition of Sphingolipid Biosynthesis by Fumonisins IMPLICATIONS FOR DISEASES ASSOCIATED WITH

Culture materials and grains contaminated with certain isolates of Fusarium moniliforme cause equine leucoencephalomalacia, porcine pulmonary edema syndrome, and liver cancer in rats. The causative agents are thought to be a family of compounds called fumonisins, which bear considerable structural similarity to the long-chain (sphingoid) base backbones of sphingolipids. Incubation of rat hepatocytes with fumonisins inhibited incorporation of [14C]serine into the sphingosine moiety of cellular sphingolipids with an IC50 of 0.1 microM for fumonisin B1. In contrast, fumonisin B1 increased the amount of the biosynthetic intermediate sphinganine, which suggests that fumonisins inhibit the conversion of [14C]sphinganine to N-acyl-[14C]sphinganines, a step that is thought to precede introduction of the 4,5-trans double bond of sphingosine (Merrill, A.H., Jr. and Wang, E. (1986) J. Biol. Chem. 261, 3764-3769). In agreement with this mechanism, fumonisin B1 inhibited the activity of sphingosine N-acyltransferase (ceramide synthase) in rat liver microsomes with 50% inhibition at approximately 0.1 microM and reduced the conversion of [3H]sphingosine to [3H]ceramide by intact hepatocytes. As far as we are aware, this is the first discovery of a naturally occurring inhibitor of this step of sphingolipid metabolism. These findings suggest that disruption of the de novo pathway of sphingolipid biosynthesis may be a critical event in the diseases that have been associated with consumption of fumonisins.


Inhibition of Sphingolipid Biosynthesis by Fumonisins IMPLICATIONS FOR DISEASES ASSOCIATED WITH
FUSARZUM MONZLZFORME* (Received for publication, April 23, 1991) Elaine WangS, William P. Norred §, Charles W. Bacon §, Ronald T. Rileygll Culture materials and grains contaminated with certain isolates of Fusarium moniliforme cause equine leucoencephalomalacia, porcine pulmonary edema syndrome, and liver cancer in rats. The causative agents are thought to be a family of compounds called fumonisins, which bear considerable structural similarity to the long-chain (sphingoid) base backbones of sphingolipids. Incubation of rat hepatocytes with fumonisins inhibited incorporation of [14C]serine into the sphingosine moiety of cellular sphingolipids with an ICso of 0.1 pM for fumonisin B1. In contrast, fumonisin B1 increased the amount of the biosynthetic intermediate sphinganine, which suggests that fumonisins inhibit the conversion of [14C]sphinganine to N-acyl- [14C] sphinganines, a step that is thought to precede introduction of the 4,5-trans double bond of sphingosine (Merrill, A. H., Jr. and Wang, E. (1986) J. Biol. Chem. 261, 3764-3769). In agreement with this mechanism, fumonisin B1 inhibited the activity of sphingosine Nacyltransferase (ceramide synthase) in rat liver microsomes with 50% inhibition at approximately 0.1 p~ and reduced the conversion of [3H]sphingosine to [3H] ceramide by intact hepatocytes. As far as we are aware, this is the first discovery of a naturally occurring inhibitor of this step of sphingolipid metabolism. These findings suggest that disruption of the de novo pathway of sphingolipid biosynthesis may be a critical event in the diseases that have been associated with consumption of fumonisins.
Fusarium moniliforme (Sheldon) is one of the most prevalent fungi on maize, other grains, and agricultural commodities in the United States and throughout the world (1). Culture materials from certain isolates of (and grains naturally contaminated with) F. moniliforme have been shown to be toxic and carcinogenic for animals (2)(3)(4)(5); furthermore, consumption of contaminated maize has been correlated with esophageal cancer in areas of southern Africa, China, and other countries (6)(7)(8). Several mycotoxins, termed fumonisins, have been * This work was supported by National Institutes of Health Grant GM33369. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. isolated from extracts of F. moniliforme (9) and naturally contaminated corn (10,11). Fumonisin B1 has been shown to cause equine leucoencephalomalacia (12), porcine pulmonary edema (13), and promotion of liver tumors in rats (14). Recent surveys indicate that high levels of fumonisin B1 are present in United States feeds associated with field cases of these animal diseases (15).
The molecular mechanism of action of the fumonisins is not known; however, these compounds bear a remarkable structural similarity to sphingosine (Fig. l), the long-chain (sphingoid) base backbone of sphingomyelin, cerebrosides, sulfatides, gangliosides, and other sphingolipids. Sphingolipids are thought to be involved with the regulation of cell growth, differentiation, and neoplastic transformation through participation in cell-cell communication and cellsubstratum interactions and possible interactions with cell receptors and signaling systems (16)(17)(18)(19)(20). Hence, we hypothesized that disruption of sphingosine metabolism might be a target of fumonisins. This manuscript reports that fumonisins inhibit de nouo sphingolipid biosynthesis by rat liver hepatocytes and identifies an important site of inhibition as the reaction catalyzed by sphingosine N-acyltransferase (ceramide synthase).

EXPERIMENTAL PROCEDURES
Materials-Fumonisin B1 and BZ were purchased from Division of Food Sciences and Technology, Council for Scientific and Industrial Research, Pretoria, South Africa. Tissue culture media were purchased from GIBCO, and collagenase (Type IV), collagen (Type III), and palmitoyl-CoA were from Sigma. Matrigel (from Engelbreth-Holm-Swarm mouse tumor) were obtained from Collaborative Research Inc. (Bedford, MA). The [2-"C]serine and the [3H]sphingosine (prepared by catalytic exchange) were from Du Pont-New England Nuclear. The sphingolipid standards were purchased from Sigma or prepared synthetically (21).
Hepatocyte Culture-Hepatocytes were prepared by a collagenase perfusion method (22) from male Sprague-Dawley rats (125-200 g) fed a chow diet (Ralston Purina, St. Louis, MO) ad libitum. The hepatocytes were isolated using aseptic procedures and plated in 60mm tissue culture dishes coated with collagen for short term ( t 2 4 h ) or with Matrigel for longer term (3-4 days) experiments. After plating for 4 h, the medium was changed to 2 ml of Dulbecco's modified Eagle's medium (with 3.7 g/l' NaHC03) containing 10 rg/ml insulin, 6.1 g/1 penicillin, 10 g/1 streptomycin sulfate, and varying concentrations of fumonisins (added from a 1 mM solution in Dulbecco's phosphate-buffered saline). The cells were maintained in a tissue culture incubator a t 37 "C and an atmosphere of 5% CO,; the medium was changed every 24 h. Cell viabilities were assessed by the ability of the cells to exclude 0.1% trypan blue and were greater than 90%.
Lipid Analyses-The de nouo biosynthesis of sphingosine was measured as described before (23,24). Briefly, cells that had been The abbreviations used are: 1, liter; HPLC, high performance liquid chromatography. were added, and the lipids were extracted and acid-hydrolyzed to liberate the free long-chain bases using conditions that give optimal hydrolysis with minimal decomposition of the labile long-chain bases (23)(24)(25). The sphinganine and sphingosine were separated by thinlayer chromatography on silica gel H (Brinkman) plates developed with CHC13:methanol:2 N NH,OH (40:10:1, v/v/v), and the radiolabeling profile (shown in Fig. 2) was determined by radiometric scanning of the plates with a Bioscan System 200 Imaging Scanner. Sphingosine and sphinganine were visualized using ninhydrin, which also revealed any acid decomposition products, which migrate as a ninhydrin-positive material near the solvent front (when seen, the data were not considered reliable). The appropriate regions of the chromatoplate were scraped, placed in scintillation vials, and 0.2 ml of water and 4 ml of a detergent-based scintillation mixture were added. After several hours, the samples were counted with the appropriate corrections for quenching. In our experience (23,24,26,27), these procedures give a good representation of the total amount of label in the long-chain base moieties of cellular sphingolipids.
To measure the relative activity of sphingosine N-acyltransferase in intact hepatocytes, the cells were incubated with 1 FM fumonisin B, for 1 h, then 1 WCi of ['Hlsphingosine was added, and the cells were incubated for varying times. The ['Hlsphingosine was freshly purified because it decomposes rapidly to a material with an R, similar to that of ceramides on TLC. The [3H]sphingosine was diluted to 1 Ci/mmol and applied to a small silica gel column (Unisil, Clarkston, IL) in chloroform. The column was washed with 10-column volumes of chloroform, the ['Hlsphingosine was eluted with methanol, and the purity confirmed by TLC. The ['Hlsphingosine was transferred to a small test tube, the solvent was evaporated under N,, a small volume of culture medium was added immediately, and the tube was sonicated for a few minutes using a bath-type sonicator. Aliquots of this solution were added to the cells, and at the end of the specified incubation period the cells were scraped from the dish, 25 fig of unlabeled ceramide carrier was added, and the lipids were extracted as described above. The ['Hlceramides were separated from ['HI sphingosine using silica gel H TLC plates developed with diethyl ether:methanol (99:1, v/v), visualized with I2 vapor, and the amount of radiolabel was determined by scintillation counting.
Mass measurements of the long-chain bases were conducted by HPLC as previously described (28) with C2O-sphinganine as an internal standard. To quantitate free long-chain bases, the extracts were base-treated to remove glycerolipids before analysis (28); whereas, to determine total sphingolipids, the extracts were acidhydrolyzed before preparing the samples for HPLC and corrected, if necessary, for any losses during hydrolysis or extraction by spiking replicate samples with a known amount of ceramide or sphingomyelin.
In Vitro Assays of Serine Palmitoyltransferase and Sphingosine N-Acyltransferase-A microsomal fraction was isolated from rat liver and assayed for serine palmitoyltransferase (24, 29) and sphingosine N-acyltransferase (30) in the presence of varying concentrations of fumonisin Bl. The assay mixture for sphingosine N-acyltransferase contained 1 /AM ['Hlsphingosine, 25 mM potassium phosphate buffer (pH 7.4), 0.5 mM dithiothreitol, 200 PM palmitoyl-CoA, and approx-imately 0.2 mg of microsomal protein in a total volume of 0.1 ml, which were found to be optimal assay conditions as reported previously (30). The reaction was initiated by adding palmitoyl-CoA, and after incubation at 37 "C for 15 min, the products were extracted, resolved by TLC, and quantitated as described above. Background counts were subtracted using data from identical assays that omitted palmitoyl-CoA.
Other Methods-The incorporation of radiolabel into other lipids was determined by separating the phospholipids by TLC using silica gel H plates developed with CHC13:methanol:formic acidwater (56:30:2:1, v/v/v/v), visualization of the spots and standards with I, vapor, and quantitation by scintillation counting. The amount of label in fatty acids was estimated using samples that had been incubated with 0.1 M KOH in methanol for 1 h at 37 "C and separated using the cermaide TLC system because the methyl esters migrate near the solvent front under these conditions.

RESULTS AND DISCUSSION
Rat hepatocytes were selected as a model for the effects of fumonisins because F. moniliforme culture materials, now known to contain the cancer promoting fumonisin B, (9,15), are hepatotoxic and hepatocarcinogenic in the rat (4, 10, 31). Furthermore, sphingosine biosynthesis de nouo by rat hepatocytes is relatively easy to follow using [14C]serine and appears to proceed via the reactions shown in   (Table I) were reduced significantly. Fatty acid biosynthesis from [14C]acetic acid was not altered (Table I). Hence, the inhibition of sphingolipid biosynthesis does not appear to be due to the inability of the cells to take up [14C]serine and to incorporate it into lipids in general nor to affect the ability of the cell to form the other biosynthetic precursors ( i e . fatty acids).
Although de novo sphingolipid biosynthesis was completely blocked by 10 p~ fumonisin B,, there was only a slight reduction in the mass of total sphingolipids after 1 day (Table  I). This is probably due to the generally slow turnover of sphingolipids (32). A greater effect was seen when the cells were plated on Matrigel to allow incubation with 1 ~L M fumonisin B1 for 4 days, which reduced the level of total sphingolipids by about half (i.e. There are several potential sites at which fumonisins might affect sphingosine metabolism (Fig.  2), hence, these were investigated further. Fumonisins do not appear to act at the first step of this pathway, because even a fairly high concentration (25 p~) of fumonisin B1 did not reduce the activity of serine palmitoyltransferase in vitro (i.e. the activities in the presence and absence of 25 p~ fumonisin B, were 88 & 3 and 61 f 15 pmol 3-ketosphinganine formed per min/mg of microsomal protein, respectively). Furthermore, treatment of hepatocytes with fumonisin B1 followed by assays of this enzyme in disrupted cells revealed no inhibition (i.e. the activities were 7.7 f 0.2 and 8.2 f 0.8 pmol/min/mg of cellular protein in the presence and absence of 2.5 p~ fumonisin B1, respectively). Inhibition at the second step of the pathway is also unlikely, because 3-ketosphinganine was not seen to accumulate (Fig. 2), and there was no reduction in the formation of [14C]sphinganine (which will be discussed below).
Fumonisins apparently inhibit at the step where [14C]sphinganine is converted to N-acyl-[14C]sphinganine, because the amount of radiolabel in [14C]sphinganine increased when hepatocytes were treated with fumonisin B1 (Fig. 4A). This was accompanied by an increase in the mass of free sphinganine (Fig. 4B); when hepatocytes were incubated with 1 p~ fumonisin B, for 4 days, sphinganine increased 110-fold (i.e. to 1499 & 18 pmol/dish compared with 13.6 & 0.4 pmol/dish for the control). There was only a small reduction in the amount of free sphingosine within the first 2 h (Fig. 4B); however, free sphingosine decreased significantly after 4 days (i.e. to 52 k 1 pmol/dish compared with 233 f 14 pmol/dish for the control). These findings are consistent with the view that the free sphingosine found in hepatocytes is not an intermediate of the de novo biosynthetic pathway (23,24) (hence, will be affected little by short term treatment with fumonisin B1) but arises from the turnover of complex sphingolipids (27,32) and would, therefore, be most affected when there is a decrease in total sphingolipids due to longer term exposure to these compounds.
Inhibition at this step of the pathway was demonstrated directly by in vitro assays of sphingosine N-acyltransferase, which has been reported to acylate both sphinganine and sphingosine (30) (Fig. 5 B ) and by following the conversion of ['Hlsphingosine to [3H]ceramide by intact cells (Fig. 5A). The

TABLE I Effects of fumonisin B1 on selected lipids of rat hepatocytes
Hepatocytes were incubated for 16 h with [14C]serine and the concentrations of fumonisin B1 shown using the conditions described in Fig. 2. The lipids were extracted, separated by thin-layer chromatography, and quantitated as described in the text. The groups that are significantly different ( p < 0.05) from the control (no fumonisin) group are designated by asterisks.  [3H]sphingosine was added, and the cells were incubated for the times shown. The lipids were extracted and analyzed by TLC as described in the text. Asterisks designate the groups that are different from control (no fumonisin treatment) with p < 0. 05. apparent IC5,, for inhibition of this activity i n uitro was approximately 0.1 p~ for fumonisin B1, but the inhibition may be biphasic (Fig. 5 B ) . All together, these findings suggest strongly that the inhibition of sphingosine N-acyltransferase accounts for the disruption of de nouo sphingolipid biosynthesis by fumonisins, although these compounds may have additional effects on other enzymes of sphingolipid metabolism.
The structural basis for this inhibition is unknown; however, one can speculate that similarities between the fumonisins and long-chain (sphingoid) bases (Fig. 1) allow them to be recognized as substrate (or transition state or product) analogs by sphingosine N-acyltransferase. The absence of a hydroxyl group a t carbon 1 may alter their orientation in the active site of this enzyme and preclude acylation or, if acylated, result in an inhibitory ceramide that cannot be removed by addition of a sphingolipid headgroup at that position. The lack of a hydroxymethyl group at position 1 is shared by a number of other fumonisin-like compounds, such as the hostspecific phytotoxins produced by another fungus Alternaria alternata (33); hence, more naturally occurring inhibitors of this pathway may exist. These compounds are also unable to follow the usual pathway of long-chain base catabolism, which proceeds via phosphorylation at position 1 (32), and this may contribute to the persistence of the inhibition.
These findings with rat hepatocytes provide the first identification of a biochemical target for the action of fumonisins and imply that inhibition of de nouo sphingolipid biosynthesis in uitro may underlie the hepatotoxicity and hepatocarcinogenicity of this mycotoxin i n uiuo. The hepatic levels of fumonisins under these conditions are not known; however, if all the fumonisin given in feeding studies with rats (1-10 mg) (3)(4)(5)31) was distributed uniformly in the body, the concentrations could be between 10 and 100 p~. This is surely an over-(or under-) estimation of the amount that reaches liver i n uiuo, but certainly considerably higher than the ICs0 for inhibition of sphingolipid biosynthesis by hepatocytes (0.1 p~) . In preliminary studies, we have noted a small (10 and 35%) reduction in [14C]sphingosine biosynthesis by hepatocytes isolated from two rats given fumonisin B1 (5 mg/200 g body weight) by gavage for just 2 days. Perhaps more interestingly, there was a pronounced reduction in the ratio of free ['4C]sphingosine to [14C]sphinganine from 2.6 * 0.14 (a typical ratio for control hepatocytes) to 0.5 & 0.1.
Disruption of this pathway is an attractive mechanism for at least some of the pathological effects of fumonisins, since sphingolipids have been hypothesized to regulate various aspects of cell growth, differentiation, and transformation (18- 20). For example, the degeneration of neuronal cells seen in equine leucoencephalomalacia may be due to inhibition of sphingolipid biosynthesis because brain contains high levels of sphingolipids (17). One can also speculate that an accumulation of sphinganine in cells exposed to fumonisins might lead to cell death since long-chain bases can be highly cytotoxic (34,35) or to cell proliferation since these compounds are mitogenic to some cell types (36) and affect diverse cell systems (20), including protein kinase C (37), the epidermal growth factor receptor (38), the Na+/K+-ATPase (39), and phosphatidic acid phosphohydrolase (40), inter alia. Although toxic, fumonisins (or related compounds) might also be therapeutically useful in diseases where defects in sphingolipid turnover cause cells to accumulate high levels of sphingolipids.