Isolation of 2-fluorocitrate produced by in vivo dealkylation of 29-fluorostigmasterol in an insect.

A novel pro-insecticide, 29-fluorostigmasterol, is proposed to cause mortality due to release of fluoroacetate during side chain dealkylation. The 29-3H-labeled substrate was fed to third instar tobacco hornworms (Manduca sexta) and erythro-2-fluoro-[2-3H] citrate was isolated in 0.012% yield by ion-exchange, silica gel, and reverse-phase chromatography of the tricarboxylic acid, trimethyl ester, and trimethyl ester benzoate, respectively. The less toxic 29-fluoro-[29-3H]sitosterol did not provide sufficient labeled fluorocitrate to allow isolation, while a more toxic 16-3H-labeled 16-fluorofatty acid gave nearly 1% conversion to labeled fluorocitrate. This is the first direct chemical evidence for the fate of the two carbons removed during phytosterol dealkylation in an insect. It is also the first use of labeled fluoroacetate precursors to identify labeled 2-fluorocitrate as an in vivo metabolite of these precursors.

To obtain further support for the proposed metabolic activation for the 29-fluorosterols, the 29-3H-labeled analogs were required. We also required an efficient method for the isolation of trace quantities of labeled 2-fluorocitrate from a crude insect homogenate. We report herein the synthesis of 29fluoro-[29-3H]sitosterol (2), 29-fluor0-[29-~H]stigmasterol (4), and (E)-16-fluoro[16-3H]-9-hexadecenoic acid as substrates for in uiuo conversion to 2-fluor0- [2-~H]citrate. The * This work was supported by Grant AI-17031 from the National Institutes of Health and by Grant PCM-8011159 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
isolation of fluor0- [2-~H]citrate from insects fed these substrates is described and a comparison of the phytosterol dealkylation and fatty acid 8-oxidation pathways for fluorocitrate production in Manduca larvae is presented.

MATERIALS AND METHODS
General-Unless otherwise noted, materials were obtained from commercial suppliers and used without further purification. CH,Cl, was distilled from CaCl, and stored over 4 A molecular sieves. Hexane, ethyl acetate, and methanol were Fisher HPLC grade and used without further purification. Sodium [3H]borohydride was purchased from New England Nuclear. All reactions were performed under N,. IR spectra were determined with a Perkin-Elmer Model 727 instrument and are reported in wave numbers (cm"). 80 MHz 'H-NMR and 20 MHz 13C-NMR spectra were obtained on an CFT-20 spectrometer. 300 MHz 'H-NMR were obtained in CsD6 or CDCll using a Nicolet NT-300 instrument. Proton-noise decoupled 3H-NMR spectra were obtained by J. Balschi (Stony Brook) at 106.7 MHz on a modified Varian XL-100 spectrometer. Chemical shifts are expressed in parts per million downfield from internal tetramethylsilane.
TLC was performed using MN Polygram Si1 G/UV 254 (4 X 8 cm) TLC plates. Flash chromatography (11) on nonradiolabeled material was performed under N, pressure on Merck Silica Gel G (400-230 mesh) using hexane/EtOAc mixtures by applying pressure with a pipette bulb. The chromatograms of nonradiolabeled materials were visualized with an ethanol/vanillin/H2S04 reagent. Visualization of radiolabeled material was accomplished at 254 nm followed by staining with 12. Radioactive samples were counted in a Packard Tri-Carb liquid scintillation counter using an Omnifluor/toluene mixture for organic solvents and Biofluor (New England Nuclear) for aqueous samples. Counting was 52-57% efficient as determined by quench curves, and all counts were corrected using automatic external standardization.
HPLC' was performed using a Waters M6000 pump and a Whatman CS PXS 10/25 column with detection at 254 nm. Gas chromatography was performed using a Varian 3700 capillary instrument equipped with a flame ionization detector and one of the following columns: Durabound DB-1701 (0.25 mm X 30 m) or Durabond DB-5 (0.25 mm X 30 m) (J&W Scientific) at the temperatures indicated for citrate analyses and at 280 "C for sterol analyses.
Synthesis of Labeled Substrates (Fig. 2)-29-Fluor0-[29-~H]sitosterol (2) was synthesized in five steps from 29-hydroxysitosterol isomethyl ether, an intermediate in the synthesis of the unlabeled 29fluorositosterol (2). First, a CH2C12 solution of the 29-alcohol 1 (162 mg, 0.354 mmol) was added to a suspension of 110 mg (0.531 mmol) of pyridinium chlorochromate in 5 ml of CHpC12. The mixture was stirred 1 h at 20 "C, diluted with ether, and filtered through a short silica gel column to remove chromate salts. The crude aldehyde was purified by flash chromatography on 230-400 mesh silica gel under Nz pressure by elution with 5% ethyl acetate/hexane to yield 140 mg (88%) of aldehyde. 'H-NMR showed a characteristic RCHO resonance at 69.75 (t, 2 Hz) and a carbonyl 13C resonance a t 6202.2 ppm.
(E)-16-Fluoro-[16-3H]hexadec-9-enoic acid (6) was synthesized in five steps from methyl-(E)-16-hydroxy-9-hexadecenoate following procedures described previously (12). First, a solution of 225 mg (0.833 mmol) of hydroxyester 5 was oxidized with 350 mg (1.6 mmol) of PCC in 25 ml of CHZCl,. After workup as described above and evaporation distillation (120 "C/O.l torr), TLC and GLC homogeneous aldehyde ester was obtained in 90% yield: 'H-NMR (CDCl3) 69.73 The aldehyde (48 mg, 0.180 mmol) was dissolved in absolute ethanol and added to a freeze-degassed solution of 100 mCi of sodium borotritide (0.7 mg, 5.4 Ci/mmol) in 2 ml of ethanol in a vacuumsealed break-seal apparatus. After 2 h at 20 "C, a solution of 0.6 mmol of unlabeled NaBH, in 1.5 ml of ethanol was added and the reaction was stirred an additional 30 min at 20 "C. Acetone was added to react with excess hydride and the volatiles were removed in uacuo. The residue was taken up in 2 ml of 0.5 N HzSO, and 2 ml of hexane, extracted three additional times, and the combined hexane layers were washed (saturated NaCl), dried (MgSO,), and concentrated on 3 g of 230-400 mesh silica gel. Elution with 10% ethyl acetate/hexane gave 29.8 mg of TLC homogeneous alcohol ester with a specific activity of 826 mCi/mmol (total of 91 mCi).
To a solution of 29.8 mg of the tritiated w-hydroxy ester in 10 ml of dry CH2C12 at 0 "C was added 0.5 ml of freshly distilled N-(1,1,2-Fluorocitrate from 29-F1uorostero.h in Insects trifluoro-2-chloroethyl)-N,N-diethylamine. The reaction was stirred at 0-20 "C for 2 h, the solvents were removed in uacuo, and then 1 ml of 2-propanol in 2 ml of 5% ethyl acetate/hexane was added and stirred 10 min. This mixture was washed (saturated NaCl), dried (MgS04), and filtered through silica gel to give 6.5 mg (45 mCi) of the desired tritium-labeled, fluorinated ethyl ester (due to ethanol transesterification catalyzed by basic impurities). Spectral data were analogous with those of the previously reported unlabeled methyl ester (12). In addition, the proton-decoupled 3H-NMR showed a doublet, 64.05, J = 51.2 Hz for the RCHFT tritiofluoromethyl group. Two major 3H-containing fluorinated byproducts at higher R F were also isolated but were not further characterized.
To an aliquot of 4.5 mCi of this ester in methanol were added 21 mg of the unlabeled fluoro ester, 30 mg of KzC03, and 100 ~1 of HzO.
The hydrolysis mixture was stirred 16 h a t 20 "C, concentrated in UQCUO, and partitioned between 4% aqueous HCl and hexane. The hexane layer was applied to flash silica minicolumn and 14.5 mg (3.6 mCi) of the tritiated fluoroacid 6 (specific activity, 72 mCi/mmol) were eluted with 10% ethyl acetate in hexane.
Incubation and Isolation Procedures-M. sexta larvae were reared on a long day photoperiod (17:7, light to dark), from eggs provided by the USDA-ARS Insect Physiology Laboratory at Beltsville, MD, as described previously (14). Each labeled substrate (2, 4, or 6) was diluted with unlabeled carrier to give a final specific activity of 62 mCi/mmol after recrystallization. Dimethyl formamide solutions (1 pl) containing approximately lo6 dpm of each substrate were injected perorally into each of 20 third instar hornworm larvae (0.15 g/larva). After 8-h incubation at 26 "C, larvae were frozen in duplicate sets of 10 larvae/set.
Fluorocitrate isolation was optimized in preliminary runs using the labeled w-fluoroacid 6 as the substrate (Fig. 3). Thus, a set of 10 larvae were homogenized in 20 ml of 1:l ethyl acetate/water to which 1 mg of (-)-erythro-fluorocitrate-tris(cyclohexylammonium) salt had been added as a carrier. The residue was resuspended in three additional 10-ml portions of 1:1 ethyl acetate/water, and the combined aqueous fractions (about 30 ml) were passed through a 2-ml column of 200-400 mesh Dowex AG1-X8 (formate form) to bind to organic acids. After washing with water (5 ml) and 3 N formic acid (5 ml), the tricarboxylic acids were eluted with 2 N ammonium formate (10 ml). The eluate was acidified (pH < l ) , lyophilized, methylated contained trimethyl fluorocitrate along with the trimethyl esters of citrate and isocitrate as determined by capillary gas chromatography (Durabond DB-5,0.25 mm X 30 m, 130 "C).
The in vivo data for Munducu larvae fed unlabeled 2, 4, 6, and (-)-erythro-2-fluorocitrate (1) showed a 300-to 1000-fold higher toxicity for fluorocitrate relative to 29-fluorostigmasterol. Thus, we estimated that less than 0.1% of the administered labeled 29-fluorostigmasterol 4 (e.g., less than 10 ng) would be converted to the lethal fluorocitrate stereoisomer. Unlabeled fluorocitrate was thus required as a carrier to facilitate chromatographic detection and improve recovery of labeled material. In preliminary experiments following this protocol, mass recovery of trimethylfluorocitrate benzoate exceeded 70%.
Initial experiments with labeled 2,4, and 6 employed older 5th instar M. sextu larvae, and the labeled compounds were administered by impregnation into small blocks of artificial diet (15). These experiments gave suboptimal incorporations into labeled fluorocitrate, due to the enormous bulk (10 g) of the insect and relatively low dealkylation rates for this stage. We recently determined that sterol dealkylation rates for [29-3H]sitosterol vary during the insect life cycle, from a maximum of 7.5 nmol of sterol/g fresh weight insect/h in 3rd instar larvae to 0.57 nmol/g/h in 5th instars and 0.18 nmol/g/h in pupae (15). These assays employed a peroral injection technique followed by a simple partition assay to measure aqueous tritium-labeled metabolites of a [29-3H]phytosterol. Thus, the experiments reported here employed were optimized in terms of 1) peroral injection to standardize equantity of labeled precursor administered, 2) use of 3rd instars to maximize phytosterol turnover, and 3) use of 10 insects/replicate to minimize the effects of individual variation on the incorporation results.
A key requirement for our isolation of labeled fluorocitrate was that the ultimate HPLC step would resolve the erythro-

TABLE I Purification of erythr~-Z-fluoro-[Z-~H]citrate from precursors injected perorally into third instar hornworm larvae
For two replicates of 10 third instar M. senta each, 1 day after apolysis. Each insect received 1 pl of dimethyl formamide containing one-tenth of the indicated amount. After 8-h incubation at 26 "C, insects were frozen and stored at -17 "C until homogenization. enantiomer, which has an identical retention time to the toxic (2R,3R)-enantiomer, was used for the HPLC standard for the erythro-fluorocitrate.
and threo-2-fluorocitrate diastereoisomers from each other as well as from nonfluorinated citrates. For this reason, the initial stages of purification were designed not to discriminate among these compounds and increasing discrimination was achieved by increasing the degree of replacement of polar groups with less polar derivatives. At the final stage, HPLC conditions for this separation were critical; use of a CI8 (octadecy1)-capped silica rather than C, (octyl), or deviation by more than 6% from the 34% CH3CN/H20 mixture resulted in loss of base-line resolution of the trimethyl (fluoro)citrate benzoates. The final separations showing UV absorbance of the benzoate chromophore and radioactivity due to the tritium label are illustrated in Fig. 4.
The (-)-erythro stereoisomer has been shown to account for all observed toxicity, being over 300-fold more toxic than the other stereoisomers (8). Indeed, in both vertebrate systems in vivo and with purified enzymes in vitro (bacterial citrate lyase and mammalian ATP citrate lyase), only the (-)-erythro-(2R,3R) isomer is produced (8,10). The absolute stereochemistry has been unambigr,lously assigned by total synthesis from masked oxaloacetate (13) and by enzymic means (16). In this insect, therefore, the erythro form is the exclusive product of the condensation of fluoroacetyl-CoA with oxaloacetate. That this is the correct (2R,3R) enantiomer of erythro-2-fluorocitrate can only be inferred from the high toxicity of the small quantity of fluorocitrate produced in vivo.
The data in Table  I demonstrate (1,15). In fact, of the two 29-fluorosterols only the 100-fold more toxic 29-fluorostigmasterol 4 led to an isolable quantity of labeled fluorocitrate. A hypothetical pathway can be proposed which integrates the known aspects of sterol dealkylation (or fatty acid P-oxidation) and fluorocitrate synthesis with the 29-fluorophytosterol toxicity and labeled citrate isolation results (Fig. 5). In a formal sense, P-oxidation produces fluoroacetyl-CoA directly. This circumvents one inefficient step in the lethal synthesis of fluoroacetate itself into fluorocitrate (V,,, for fluoroacetate < 1% that of acetate) (8,10). The next step, condensation with oxaloacetate to give (2R,3R)-fluorocitrate, also has a V,,, for fluoroacetyl-CoA of 0.3% that of acetyl-coA (8). Moreover, fluoroacetyl-CoA can enter many of the pathways, e.g. thiolase, malate synthase, and acetyl-coA carboxylase, for which acetyl-coA is the starting material (10). Thus, the 1% conversion from a 16-fluoroacid 6 to erythro-fluorocitrate is in accord with our best expectations.
For the 29-fluorophytosterols, the two-carbon fragment released during dealkylation is formally fluoroacetaldehyde. This requires conversion into fluoroacetyl-CoA prior to the condensation to give fluorocitrate. This conversion should involve an aldehyde dehydrogenase and could proceed directly to the CoA derivative or via oxidation, hydrolysis, and reactivation at a later time. Although the toxicities of fluoro- ethanol and higher even-carbon homologs (and their corresponding aldehydes) are known (17), the enzymatic conversions have not been examined in detail. In any event, the dealkylative pathway leading to fluorocitrate is considerably more tortuous than that for p-oxidation, and has at least five separate enzymic steps in which the substitution of fluorine for hydrogen could exert an adverse electronic effect on K,,, or VmaX of the reaction. In particular, the desaturation-oxidation-fragmentation has been shown to be sensitive to 29fluorine substitution. Lower fluoroacetate-produced toxicity is observed for the 22,23-saturated sterols than for the 22,23unsaturated analogs (1, 2, 15). In addition, one expects fluoroacetaldehyde to be less efficiently oxidized than acetate. The observed conversion of labeled sterol 4 to fluorocitrate in 0.012% yield, nearly 100-fold lower than for the fluorofatty acid 6, is reasonable given the numerous opportunities for enzymic bottlenecks. That any fluorocitrate at all can be isolated testifies to the efficiency of sterol utilization in this phytophagous insect, which increases in weight 1000-fold during its 2-week development from egg to 5th instar.