Proceedings of the National Academy of Sciences Vol. 68, No. 1, pp. 20-24, January 1971

Hypoglycin A: A Specific Inhibitor of Isovaleryl CoA Dehydrogenase

KAY TANAKA, EDITH M. MILLER, AND KURT J. ISSELBACHER Department of Medicine, Harvard Medical School and Massachusetts General Hospital, Boston, Mass. 02114 Communicated by Herman M. Kalckar, August 27, 1970

ABSTRACT Evidence is presented for the specific A or its in vivo and in vitro inhibition of isovaleryl CoA dehydro- hypoglycin metabolite, methylenecyclopropylacetic genation by hypoglycin A and its derivative, a-keto- acid, and also 4-pentenoic acid (4-PE) were shown to inhibit methylenecyclopropylpropionic acid. a-Methylbutyryl long-chain oxidation; whereas oxidation of straight CoA dehydrogenation was also impaired, but the degree short chain fatty acids including butyrate, hexanoate, and of inhibition was much lower. Isobutyryl CoA dehydro- octanoate was not inhibited genation was not inhibited. 4-Pentenoic acid inhibited (15-18). none of these reactions. It is concluded that isovaleryl Based on the data cited above plus the similarity in chemical CoA is dehydrogenated by a specific enzyme, isovaleryl structure between isovaleric acid and methylenecyclopropyl- CoA dehydrogenase, contrary to previous assumptions acetic acid we postulated that hypoglycin A or its metabolite that it is dehydrogenated by green acyl CoA dehydrogenase. is a inhibitor of CoA The The present specific isovaleryl dehydrogenase. concept agrees with our previous findings in studies have shown this to be the case. isovaleric acidemia, a genetic disorder in which a specific present defect of isovaleryl CoA dehydrogenase was observed. It AND was also demonstrated that isovaleric acidemia can be MATERIALS METHODS induced in experimental animals by the administration Hypoglycin A and a-ketomethylenecyclopropylpropionic of hypoglycin A. Furthermore, some symptoms of "the acid (KMCPP) were generous gifts from Dr. P. H. Bell, vomiting sickness of Jamaica" appear to be due to iso- Lederle valeric acid accumulation secondary to the ingestion of Laboratories. hypoglycin A. Fasted male Sprague-Dawley rats weighing about 150 g were used throughout. Isovaleric acidemia is a genetic defect of metabolism (1-4) in which isovaleric acid accumulates in blood in large In vivo experiments amounts. Based on the biochemical findings in this hereditary Amino acids were given by stomach tube 30 min after intra- disorder, we proposed that isovaleryl CoA is dehydrogenated muscular injection of the inhibitor. The blood samples were by a specific enzyme, isovaleryl CoA dehydrogenase (1, 3) and drawn by heart puncture 150 min after injection of the inhibi- that this enzyme is deficient in patients with isovaleric aci- tor. Short chain fatty acid analyses were performed by gas- demia (Fig. 1). It was previously believed that isovaleryl CoA, liquid chromatography (GLC) (1). butyryl CoA, and hexanoyl CoA were dehydrogenated by a single enzyme, acyl CoA dehydrogenase (5, 6). In vitro experiments It has subsequently been shown that oxidation of leucine Incubations were for 3 hr at 37°C. The tissues and the incuba- is inhibited by hypoglycin A (7, 8). In contrast, oxidation of tion medium were homogenized after acidification with 0.5 ml and are not significantly inhibited by this of 2 N H2SO4 and extracted with 20 vol of chloroform-meth- compound (8). However, the specific step in the pathway of anol 2:1. The filtered extract was separated into two phases leucine metabolism that is inhibited by hypoglycin A has not by mixing with 0.2 N NaOH (0.2 vol of the extract). Un- been precisely identified. labeled carrier was added to the homogenate before extraction. Hypoglycin A is an unusual having the structure Only the upper alkaline layer, which contains most of the of a-aminomethylenecyclopropylpropionic acid (9). It has water soluble salts of carboxylic acids and acidic conjugates, been extracted from unripe fruits of ackee grown in Jamaica was analyzed. The steam-distillable fractions after acid hy- and has been identified as the cause of the "vomiting sickness drolysis were designated as total fractions; the fractions steam- of Jamaica" (10). Extreme (11) and depletion distilled without acid hydrolysis were referred to as free acid of liver glycogen (12) have been observed in the cases of this fractions. Conjugated fractions were calculated from the differ- disease. These same in vivo effects have been observed in ex- ence between the total and the free acid fractions. For the perimental animals after injection of hypoglycin A (13, 14). total fraction, the residue after evaporation of the upper layer Several groups of investigators subsequently concluded that was redissolved in 2 ml of 6 N HCl and hydrolyzed for 17 hr the hypoglycemia and depletion of glycogen were due to the at 1100C. The hydrolysate was alkalinized with 2.5 ml of 5 N decreased resulting from impairment of NaOH, reacidified with 5 M o-phosphoric acid to pH 2.8-3.0, long chain fatty acid metabolism (15-17). In these studies, and then steam-distilled (1).N-isovalerylglycine in the incuba- tion medium was analyzed by GLC (3). Teflon tubing (2 mm Abbreviations: 4-PE, 4-pentenoic acid; KMCPP, a-keto- X 1 ft) cooled with dry ice-acetone, was used to trap the methylenecyclopropylpropionic acid; iVGly, N-isovalerylglycine. effluent.

20 Downloaded by guest on October 2, 2021 Vol. 68, 1971 Isovaleryl Dehydrogenase Inhibition 21

NH2

(CH3)2CHCH2-C H - COOH LEUCINE

11 (CH3)2 CHCH2C COOH a -KETOISOCAPROIC ACID

+ GLYC/NE (CH3)2 CHCH2-COOH (CH3)2CHCH2-CO-S-CoA - (CH3)2 CHCH2CO-NHCH2COOH ISOVALERIC ACID ISOVALERYL CoA N-ISOVALERYLGLYCINE

W g(KMCPP Action Site)

(CH3)2C= CH-CO-S- CoA ,G-METHYLCROTONYL CoA FIG. 1. Scheme of the site of action of a-ketomethylenecyclopropylpropionic acid in the pathway of leucine oxidation.

RESULTS mM glycine. Therefore, further analyses were made only on In vitro studies the group where glycine was added. 1. Inhibition of Leucine Oxidation and Fractionation of the 2. Identification of Inhibition Product as N-isovalerylgly- Products by Steam Distillation. We assumed that if leucine cine. To identify the product of the inhibited reaction as iVGly oxidation is inhibited at the step of isovaleryl CoA dehydro- an aliquot of one incubation medium with inhibitor was genation, isovaleryl CoA might alternatively be converted to analyzed by silicic-acid chromatography (3) (Fig. 2). The N-isovalerylglycine (iVGly), as has been observed in pa- radioactivity peak of the product corresponded with that of tients with isovaleric acidemia (Fig. 1) (3). Therefore, liver unlabeled iVGly (assessed by titration). Much less radioactiv- slices were incubated in 2 ml of Krebs-Ringer bicarbonate ity was observed around the region of iVGly in the same chro- buffer containing 2.8 mM and 0.76 mM D,L-[2-14C]- matographic system from an incubation medium without in- leucine, with and without glycine. As shown in Table 1, ["4C]- hibitor. Therefore, the fractions of the main radioactive peak CO2 production was inhibited almost 90% by 0.70 mM for each sample were combined and the product was recrystal- KMCPP, a transamination product of hypoglycin A. At lized several times (Table 2). The iVGly crystals from the in- the same concentration of 4-PE ["4C]C02 production was cubation medium with the inhibitor showed a constant, high inhibited only 16% (Table 1). In the presence of KMCPP, specific activity through four recrystallizations. In contrast, the increased radioactivity in the conjugated fraction was 60 almost all of the radioactivity of the control was lost in the times the control when the incubation was performed in the mother liquor from the first recrystallization, and the specific presence of 1.5 mM glycine. Radioactivity in the total steam- activity in the iVGly crystals decreased with each subse- distillable fraction was also increased in the presence of 1.5 quent recrystallization. In a subsequent experiment, it was

TABLE 1. Incorporation of 14C into carbon dioxide and steam-distiUable fractions from D,1,[2-14C]leucine [14C]C02 14C Incorporation into steam-distillable fractions Glycine Inhibitor Produced Inhibition Total Free Conjugated [amol] [Mmol] [nmol] % [nmoll O 0 67.1 4- 1.3 14.8 -- 0.4 11.3 - 0.5 3.5 -- 0.9 0 KMCPP 1.40 8.7* it 0.8 87 43.6t±- 5.0 38.0*-- 0.6 5.1 a 4.4 3.0 0 88.8 ± 0.7 9.8 ±- 0.7 9.0 ±- 0.5 0.8 ±- 0.4 3.0 KMCPP 1.40 10.4* It 0.5 88 66.2*-±7.0 22.2* It 2.0 44.0* 4- 7.8 0 0 63.0 ±- 3.1 0 4-PE 1.40 52.9 ±- 1.1 16

Each incubation mixture contained liver slices (about 100 mg) in 2 ml of Krebs-Ringer bicarbonate buffer containing 5.6 jsmol glucose. The amount of the substrate was 1.5 lsmol (1 1sCi). Results are expressed as the mean ± standard error of four experiments. Numbers marked are significantly different from corresponding controls: * P < 0.001; t P < 0.005, as appraised by Student's t test. Downloaded by guest on October 2, 2021 22 Medical Sciences: Tanaka et al. Proc. Nat. Acad. Sci. USA

100- CONrROL KMCPP 80- l Iz~ N - Isovolerylglycine ci 60-

-a 40- T 0C1 e z [ 201 hd 15 5 10 15 20 5 10 15 20 0 5 10 Io 5 l0

TuBE NUMBER MINI] TES (Left) FIG. 2. Silicic acid chromatography of incubation product after addition of carrier N-isovalerylglycine. Each incubation me- dium contained liver slices (about 100 mg) in 2 ml of Krebs-Ringer bicarbonate buffer containing 5.6 ,mol of glucose, 3.0 /mol of glycine, and 1.5 /Amol (1 ,.Ci) of D,L-[2-'4Clleucine. Four-tenths of each homogenate was used for the analyses. 20 mg of unlabeled N-isovaleryl- 40 glycine was added as a carrier to the aliquots of the extracts. Elution solvents were chloroform-methanol mixtures as follows: 100:0, ml; 99: 1, 60 ml; 98: 2, 80 ml; 96.5: 3.5, 60 ml; 95: 5, 40 ml. 10-ml fractions were collected. N-isovalerylglycine was eluted in the 9X: 2 fraction. 1.5 ml of each fraction was used for titration and 4 ml was counted for radioactivity. The small peak of radioactivity in the con- trol was not N-isovalerylglycine (see recrystallization studies, Table 2). (Right) FIG. 3. Radio gas chromatography of the incubation products. Aliquots of the same samples as in Table 1 were analyzed. Only the samples that contained 3.0 /Amol of glycine were analyzed. 500 jug of unlabeled N-isovalerylglycine was added as carrier; the peak seen in the figure is N-isovalery'glycine. Results are expressed as mean ±t standard error of four experiments. The difference in the N-isovaleryl- glycine fraction was significant (P < 0.005) by Student's t test. confirmed by GLC, that the iVGly fraction was highly labeled [14CIC02 productionwas inhibited 60% with 2.1 mrNI KTMCPP, in the four incubations containing KMCPP, whereas radio- but radioactivity did not increase significantly in the steam- activity of this fraction was very low in the control group (Fig. distillable fractions. By GLC it was found that '4C-incorpora- 3). tiol) into isobutyric acid was not significantly different from control incubations (Table 3). KMCPP showed inhibitory 3. Demonstration of Radioactivity in the Isovaleryl Moiety effects (20%) on isoleucine oxidation at a concentration of 2.1 of iVGly. Incorporation of 14C into the isovaleric acid moiety mMA. Analysis of the incubation by GLC showed accumulation was confirmed by GLC of short-chain fatty acids after acid of labeled a-methylbutyrate (Table 3). hydrolysis. As shown in Table 3, most of the radioactivity was found in the isovaleric-acid fraction in the samples from in- TABLE 3. GLC analyses of the incubation products after cubations containing KMCPP. Very little label was found in acid hydrolysis this fraction in the control group. Essentially no activity was found in the j3-methylcrotonic acid fraction in either group. KMCPP Acetate Isovalerate i3-Methyl- and Valine. Substrates added crotonate 4. Effect of KMCPP on Oxidation of Isoleucine (.amol) (JAmol) (nmol/100 g tissue) To test the specificity of the KMICPP inhibition, experiments using valine and isoleucine as substrates. D,I, [2-14C]- were performed ± ± At the concentration of 0.7 mM KMCPP, [14CICO2 produc- Leu,1.5 0 1.0 0.1 0.1 1.6 0 tion from was not significantly inhibited. DALE [2-14C] - D,[-14-4C]valine Leu,1.5 1.4 1.0 ± 0.1 18.1* i 1.6 0 a-Methyl- TABLE 2. Specific activity of N-isovalerylglycine upon repeated butyrate Tiglate recrystallization L [U_14C] _ Ile,1.5 0 1.6±0.1 0.2±0 0 Control KMCPP (1.4 umol) L [U-14C] _ Mother Mother Ile,1.5 4.2 2.3 ± 0.2 3.8* ± 0.5 0.3 ± 0 liquor, liquor, Meth- Crystals, total Crystals, total Isobutyrate acrylate specific radio- specific radio- activity activity DL [4-14C] - Recrystal- activity activity ± 0.2 3.3 ± 0.8 3.4 ± 0.5 lization (cpm/mg) (cpm) (cpm/mg) (cpm) Val,1.5 0 2.5 DIL [4-14C] - 1st 650 2216 18 1477 Val,1.5 4.2 1.5 ± 0.2 5.1 ± 0.5 S.6 ± 1.1 2nd 636 844 14 64 3rd 631 584 9 31 Each incubation mixture contained liver slices (about 100 mg) 4th 623 694 in 2 ml of Krebs-Ringer bicarbonate buffer containing 5.6 Mumol glucose and 3.0 Mmol glycine. Results are mean ± SE of four One-half of each sample from tubes 7-11 of Fig. 2 were com- experiments. Numbers marked are significantly different from bined. 10 mg of iVGly was added before the first recrystallization. corresponding controls: *P < 0.001 (see Table 1). Downloaded by guest on October 2, 2021 Vol. 68, 1971 Isovaleryl Dehydrogenase Inhibition 23

TABLE 4. Effect of amino acids, administered orally with and without inhibitors, on plasma short-chain fatty acids (mg per 100 ml)* Amino acid Inhibitor Number of (60 mg/100 g) (mg/100 g) experiments Isobutyrate n-Butyrate Isovalerate n-Hexanoate None None 4 0.06 ±t 0 0.05 i 0 0.05 ±4 0.01 0.15 i 0.03 None Hypoglycin (10 mg) 4 -0.06 ± 0 0.71 ±- 0.18 2.32t i 0.36 0.33 ± 0.11 Leu None 5 0.01 ±0 0.03 ±0 0.10 ± 0.02 0.12 ±00 Leu Hypoglycin (10 mg) 5 0.06 ±t 0.01 1.16 ±t 0.08 19.2t i 1.1 0.44± 0.01 Leu 4-PE(30mg) 4 0.02±0 0.05±0 0.06±0.01 0.14±0 Ile None 4 0.06 ± 0.02 0.06 ±E 0.02 0.23t 4 0.04 0.12 ±00 Ile Hypoglycin (10 mg) 4 0.06 ± 0.01 0.45 ± 0.04 6.59tt ±L 0.58 0.28 ± 0.01 Val None 4 0.26 ± 0.04 0.14 ± 0.03 0.18 ± 0.03 0.26 ± 0.05 Val Hypoglycin 4 0.33 ±00.04 0.47 ±00.05 0.94 ±00.12 0.49 ±00.03

Data on the direct metabolite of each amino acid and n-butyrate are described. No significant change was observed in the amounts of propionate and n-valerate. No f3-methylerotonate, tiglate, or methacrylate was found in significant amounts after amino acid administra- tion. * Values expressed as mean ±J SE for four experiments (except n = 5 for Leu, none and Leu, hypoglycin experiments). t P < 0.001; t a-methylbutyrate.

Comparison of inhibitory effect of hypoglycin A and 4-PE acid were found in urine samples of rats after these treatments in vitro was made using uniformly labeled ileucine, L-iso- (unpublished results); this finding is similar to that in patients leucine, and L-valine. Hypoglycin A (1.4 mM) inhibited with isovaleric acidemia (3, 19). The amount of leucine added [14C1C02 production from L-leucine most strongly. i-Isoleu- as substrate in the in vitro experiments was small. Therefore, cine oxidation was inhibited less, while L-valine oxidation was most of the isovaleryl CoA that could not be dehydrogenated unaffected. In contrast, 4-PE, up to 4.2 mMA, did not inhibit because of the inhibition was readily conjugated with the ex- leucine oxidation significantly. cess of glycine (3). In the in vivo experiments, this conjugation In vivo effects of hypoglycin A on metabolism of L-leucine, system appears to be overloaded with the amount of leucine L-isoleucine, and L-valine given. As a result, isovaleryl CoA is hydrolyzed resulting in an The administration of each amino acid to rats at a dose of 60 accumulation of free isovaleric acid in the blood, again similar mg per 100 g of body weight resulted in no significant increase to the findings in patients with isovaleric acidemia (1, 3, 19). of the corresponding short-chain fatty acid in the plasma. Although isoleucine oxidation was inhibited at the a-methyl- However, when hypoglycin A was given (10 mg per 100 butyryl CoA dehydrogenation step, the degree of the inhibi- g of body weight) 30 min before ileucine, the plasma content tion was significantly lower. No inhibition of isobutyryl CoA of a branched-chain pentanoic acid increased up to 200 times dehydrogenation occurred either in vivo or in vitro. It may be that of the control group (Table 4). In a second control group reasonable, therefore, to assume that CoA esters of these that was given only hypoglycin A, the same branched-chain branched-chain fatty acids are dehydrogenated by three differ- pentanoic acid increased 45-fold. The product was identified ent dehydrogenases, and that isovaleryl CoA dehydrogenase is as isovaleric acid on a DOP-PA column (1). specifically inhibited by hypoglycin A or its metabolites. This When isoleucine was given after hypoglycin A administra- conclusion agrees with the findings in isovaleric acidemia tion, a branched-chain pentanoic acid increased 30-fold. By where a more specific decrease in isovaleryl CoA dehydro- the use of a DOP-PA column, this peak was identified as a- genase has been observed (1, 3). methylbutyric acid (1). In both the leucine and isoleucine Bressler and co-workers have suggested that the inhibition experiments, there was also some increase in plasma n-butyric of long chain fatty acid oxidation by methylenecyclopro- acid. Note that in the L-valine-hypoglycin group, in which pylacetic acid and 4-PE is due to the ability of these inhibitors isobutyric acid should have increased if the same kind of to form nonmetabolizable esters with CoA and (-)-carnitine inhibition had occurred, no significant increase in isobutyric (17, 20). As a result, free CoA and (-)-carnitine become less acid was found. available. Since (-)-carnitine is necessary for the oxidation When these experiments were carried out using 4-PE as an of long-chain fatty acids in mitochondria, but not for short- inhibitor (30 mg per 100 g of body weight), followed by ileu- chain fatty acid oxidation, this has been proposed as the basis cine administration, no significant increase in plasma iso- for the specific inhibition of long-chain fatty acid oxidation valeric acid was observed (Table 4). while short chain fatty acid oxidation remains unaltered (21). Methvlenecyclopropylacetic acid and 4-PE share two DISCUSSION chemical structures in common, namely a five-carbon chain The acctumulation of N-isovalerylglycine in the medium when and a vinyl group separated by two carbons from the car- liver slices were incubated with [2-'4C]leucine and KMCPP boxyl group. These structures are believed to be necessary for indicated that leucine oxidation was inhibited by KM\CPP at their hypoglycemic action (9) and, also, for their inhibition of the isovaleryl CoA dehydrogenation step. This inhibition was long-chain fatty acid oxidation (17, 20). However, there is a further confirmed by in vivo experiments showing the accumu- striking difference in the structure of these two compounds, lation of isovaleric acid after the administration of hypoglycin namely the presence of the ring. The inhibitory A and leuciine. N-Isovalerylglycine and fl-hydroxyisovaleric effect of hvpoglycin A on isovaleryl CoA dehydrogenase ap- Downloaded by guest on October 2, 2021 24 Medical Sciences: Tanaka et al. Proc. Nat. Acad. Sci. USA

pears to be, in part, due to the presence of a cyclopropane ring 2. Budd, M. A., K. Tanaka, L. B. Holmes, M. L. Efron, N. Engl. J. Med., 277, action. In our preliminary ex- J. D. Crawford, and K. J. Isselbacher, since 4-PE failed to show this 321 (1967). periments, cyclopropane carboxylic acid, which is analogous 3. Tanaka, K., and K. J. Isselbacher, J. Biol. Chem., 242, to isobutyric acid, only slightly inhibited dehydrogenation of 2966 (1967). isobutyric acid in vivo. It appears, therefore, that the methy- 4. Newman, C. G. H., B. D. R. Wilson, P. Callaghan, and lenecyclopropane group is responsible for the inhibition of L. Young, Lancet, ii, 439 (1967). 5. Green, D. E., S. Mii, H. R. Mahler, and R. M. Bock, J. isovaleryl CoA dehydrogenation. Biol. Chem., 206, 1 (1954). The inhibition of isovaleryl CoA dehydrogenase by hypo- 6. Bachhawat, B. K., W. G. Robinson, and M. J. Coon, glycin A has important clinical implications. One of these is J. Biol. Chem., 219, 539 (1956). the experimental production of isovaleric acidemia. In addi- 7. Patrick, S. J., and L. C. Stewart, Can. J. Biochem., 42, 139 (1964). tion, the present findings provide a pathophysiological ex- 8. Posner, B. I., and M. S. Raben, Biochim. Biophys. Acta, planation for some of the symptoms of the "vomiting sickness 136, 179 (1967). of Jamaica." In this disease, profound hypoglycemia occurs 9. Sherratt, H. S. A., Brit. Med. Bull., 25, 250 (1969). following the ingestion of unripe ackee fruit; the hypoglycemia 10. Hassal, C. H., and K. Reyle, Biochem. J., 60, 334 (1955). has been attributed to ingested hypoglycin A. Some of the 11. Jelliffe, D. B., and K. L. Stuart, Brit. Med. J., i, 75 (1954). patients with this disorder die in spite of glucose infusions. 12. Hill, K. R., G. Bras, and K. P. Clearkin, West Indian This lack of response has been attributed to irreversible brain Med. J., 4, 91 (1955). damage occurring during the duration of hypoglycemia prior 13. Chen, K. K., R. C. Anderson, M. C. McCowen, and P. N. to the glucose infusions (11, 12). However, it would appear Harris, J. Pharmacol. Exp. Ther., 121, 272 (1957). 14. Feng, P. C., and S. J. Patrick, Brit. J. Pharmacol., 13, 125 that increased concentrations of isovaleric acid induced by the (1958). ingested hypoglycin A probably account for some of these ob- 15. von Holt, C., M. von Holt, and H. Bohm, Biochim. Bio- servations. Congenital isovaleric acidemia is sometimes fatal phys. Acta, 125, 11 (1966). (4). It is also noteworthy that when cats and dogs are given 16. Enteman, M., and R. Bressler, Mol. Pharmacol., 3, 333 hypoglycin A, they manifest symptoms similar to those found (1967). 17. Corredor, C., K. Brendel, and R. Bressler, Proc. Nat. in isovaleric acidemia (depression, vomiting, and ataxia) Acad. Sci. USA, 58, 2299 (1967). without any significant hypoglycemia (13). 18. von Holt, C., J. Chang, M. von Holt, and H. Bohm, Biochim. Biophys. Acta, 90, 611 (1964). This work was supported in part by grants from The John A. 19. Tanaka, K., J. C. Orr, and K. J. Isselbacher, Biochim. Hartford Foundation, Inc. and the National Institutes of Health Biophys. Acta, 152, 638 (1968). (AM-01392 and AM-03014). 20. Brendel, K., C. Corredor, and R. Bressler, Biochem. Biophys. Res. Commun., 34, 340 (1969). 1. Tanaka, K., M. A. Budd, M. L. Efron, and K. J. Issel- 21. Corredor, C., K. Brendel, and R. Bressler, J. Biol. Chem., bacher, Proc. Nat. Acad. Sci. USA, 56, 236 (1966). 244, 1212 (1969). Downloaded by guest on October 2, 2021