Interference with Valine and Isoleucine Biosynthesis by Cyclic Hydroxamic Acids

Transcription

1 JOURNALIOF BACTERIOLOGY, Nov., American Society for Microbiology Vol. 9, No. Printed in U.S.A. Interference with Valine and Isoleucine Biosynthesis by Cyclic Hydroxamic Acids ROBERT W. HOGG, CHITRA S. BISWAS, AND HARRY P. BROQUIST Laboratory of Biochemistry, Department of Dairy Science, University of Illinois, Urbana, Illinois Received for publication 8 June 196 ABSTRACT HOGG, ROBERT W. (University of Illinois, Urbana), CHITRA S. BISWAS, AND HARRY P. BROQUIST. Interference with valine and isoleucine biosynthesis by cyclic hydroxamic s. J. Bacteriol. 9: The introduction of a hydroxyl group in a number of common barbiturates, a substitution which converts these compouinds to cyclic hydroxamic s, gives rise to compounds which inhibit growth of Escherichia coli. The toxicity of these hydroxybarbiturates appears to be associated, at least in part, with interference with valine and isoleucine biosynthesis, as a combination of these amino s counteracts their toxicity. A subinhibitory level of 1-hydroxy--ethyl-- isopropylbarbituric (hydroxyipral) was counteracted either by valine or by its early precursor, a-acetolactate, and led to a study of the effect of these cyclic hydroxamic s on acetolactate synthesis in a cell-free enzyme system of E. coli. In this system, the parent barbiturates and their respective hydroxy derivatives were only moderately active in blocking acetolactate synthesis. Detailed kinetic studies of the most active compound, hydroxyipral showed no obvious relationship to the substrate or cofactors of the system. The inhibitory effects of hydroxyipral, either on growth of E. coli or in the acetolactate-forming system, could not be counteracted by Fe++, but the toxic effect of aspergillic and o-phenanthroline in these instanices was reversed by Fe++, which implies an iron involvement in the acetolactate-forming system of E. coli. Aspergillic, an antibiotic produced by (Safir et al., 193). It was subsequently discovered that these cyclic hydroxamic s, Aspergillus flavus (White and Hill, 193), has been shown to be a pyrazine cyclic hydroxamic like aspergillic, were antibacterial agents (Dutcher and Wintersteiner, 19) (Fig. 1). and that they appeared to interfere with the Its alkyl substituents suggest a relationship to synthesis of valine and isoleucine in Escherichia the branched-chain amino s and, indeed, coli (Biswas and Broquist, Bacteriol. Proc., p. it has been demonstrated (MacDonald, 1961) 93, 196). The details of these findings are the that isoleucine and leucine are precursors of aspergillic. Although this antibiotic possesses subject of this communication. a potentially useful antibacterial spectrum, it is MATERIALS AND METHODS quite toxic (White and Hill, 193) and hence of Source of compounds. Aspergillic was kindly little interest clinically. provided by J. R. Dutcher, Squibb Institute for Some years ago Safir, Hlavka, and Williams Medical Research, New Brunswick, N.J. The (193) sought to prepare compounds in the N'-hydroxybarbituric derivatives listed in barbituric series, analogous in structure to Table 1 were kindly given to us by S. R. Safir, Lederle Laboratories Division, American Cyanamid aspergillic. By condensing appropriately substituted malonyl chlorides with benzyloxy Co., Pearl River, N.Y. Subsequently, additional urea and subsequently removing the benzyl 1-hydroxy--ethyl--isopropylbarbituric was group by hydrogenolysis, barbituric s were required, which was prepared as described by Safir obtained with the added feature of a et al. hydroxyl (193). DL-a-Acetolactate was synthesized group attached to a nitrogen atom, e.g., as in by the procedure of Krampitz (198). All other compounds used in the investigation were from structure B, Fig. 1. This substitution confers on commercial sources. the molecule the )roperties of a cyclic hydroxamic Microbiological procedures. E. coli strain. Such compounds were found, however, to Crookes and strain B were maintained by monthly have only a low order of pharmacological activity transfers on nutrient agar. Unless otherwise indi- 16 Downloaded from on November 1, 18 by guest

2 166 HOGG, BISWAS, AND BROQUIST J. BACTERIOL. O CH3 OH3 '11H,Cs 'C-sH NO,c ~ cz. NH C-C CH3 HO ~C CH3 1 NH N H (A) VALINE (B) HYDROXY-IPRAL a H HO"N,'CC',,C,Cs CH,CH 3 I H3Csc "IN CH3 H CH (C) 1I ASPERGILLIC ACID FIG. 1. Structural relationships of valine and cyclic hydroxamic s. cated, E. coli Crookes was used. When desired, stock cultures were transferred to ml of sterile medium (Davis and Mingioli, 196) in -ml Erlenmeyer flasks and placed on a New Brunswick incubator-shaker for 18 hr at 3 C. One drop of a 1:1 dilution of the broth cultures served as inoculum for microbiological assays. Assays were made in -ml Erlenmeyer flasks with test compounds and media (Davis and Mingioli, 196) in a final volume of ml. After incubation (3 C) with shaking for 18 hr, the yield of bacteria was estimated turbidimetrically in a Spectronic- colorimeter at 6 m,i Ċell-free acetolactate-forming enzyme system. CH3COCOOH thiamine pyrophosphate, flavin adenine dinucleotide, MIg+ CH3COHCOOH + CO COCH3 The acetolactate-forming system, comprising the initial step in the synthesis of valine in E. coli and Salmonella typhimurium (Umbarger and Brown, 198), has been studied. The requirement of the system for flavin adenine dinucleotide (FAD), briefly reported by Strmer and Umbarger (196), has been confirmed and has been considered as an integral part of the enzyme system. E. coli cells (strain Crookes) were grown in 1 liter of Davis and Mingioli (196) medium in Fernbach flasks at 3 C in a New Brunswick rotary shaker for 1 hr, at which time the cells had reached the early log-phase growth. The cells were collected by centrifugation, washed in 3 ml of. M phosphate buffer (ph 8), and centrifuged again. The packed, washed cells from liters of medium were suspended in to 8 ml of. M phosphate buffer (ph 8), and were subjected for min of treatment with a Branson Sonifier, model S-7. The ruptured cells were centrifuged at, X g for min to remove cell walls and debris. Protein was determined by the method of Lowry et al. (191). The complete reaction mixture contained the following: 1 umoles of Mg+, 1,ug of thiamine pyrophosphate (TPP), ug of FAD, 1 jimoles of phosphate buffer (ph 8), test compounds,. ml of enzyme containing. mg protein, j,moles of pyruvate, and water to 1.9 ml. In actual practice, the complete system less pyruvate was allowed to come to 37 C before addition of substrate to initiate the reaction. The reaction was allowed to proceed for 1 min, and then was terminated by the addition of.1 ml of 36 N HSO and heated at 6 C for 1 min. The precipitated protein was removed by centrifugation, the supernatant liquid was neutralized with sodium bicarbonate, and a.-ml sample was removed for acetoin determination by the method of Westerfeld (19). RESULTS Table 1 indicates that barbituric and several, -dialkyl derivatives commonly used pharmacologically are very weak inhibitors of growth of E. coli. However, the introduction of the hydroxyl group to the number 1 nitrogen atom of certain ot these barbiturates results in a marked enhancement of antibacterial activity. Nonetheless, the naturally occurring cyclic hydroxamic, aspergillic, was still one order of magnitude more potent in antibacterial activity. The feeble antibacterial activity of hydroxyphenobarbital may reflect the inability of this derivative, with its phenyl substituent, to enter the cell. Attempts were made to reverse hydroxybarbiturate toxicity by natural materials and by use of selected metabolites. Such experiments led to the discovery that casein hydrolysate (1 mg/ml) counteracted the toxicity of these compounds, and that the effect of casein hydrolysate could be attributed to its combined content of valine and isoleucine. The effects of these branched-chain amino s on hydroxybarbiturate inhibition in E. coli are illustrated in Table. Valine plus isoleucine did not alter the toxicity of aspergillic for growth of E. coli. In other experiments not shown in Table, valine alone was more effective than isoleucine in reversing low levels of hydroxybarbiturate inhibition, but at higher levels a combination of both amino s was always required for optimal growth. In the experiment in Table 3, for example, the toxicity of 1,umole of hydroxyipral, an amount resulting in only partial inhibition of growth, was counteracted not only by valine, Downloaded from on November 1, 18 by guest

3 VOL. 9,196 ANTIBACTERIAL ACTION OF CYCLIC HYDROXAMIC ACIDS TABLE 1. Barbituric s, cyclic hydroxamic s, and growth of Escherichia coli Additiont to basal medium Barbituric, -Diethylbarbituric (barbital) -Ethyl--isopropylbarbituric (ipral) -Ethyl--phenylbarbituric (phenobarbital) 1-Hydroxy-, -diethylbarbituric (hydroxybarbital) 1-Hydroxy--ethyl- -isopropylbarbituric (hydroxyipral) 1-Hydroxy--ethyl- -phenylbarbituric (hydroxyphenobarbital) Aspergillic Amt Jmoles Growth (absorbance at 6 mu) Amt for 1% inhibition pumoles > > ca >. Microbiological assay conditions as described in t For convenience, these compounds are referred to in the text by the abbreviated terms given parenthetically after the strict chemical nomenclature. TABLE. Effect of valine and isoleucine on the toxicity of cyclic hydroxamic s in Escherichia coli Expt no.t I II III IV V Inhibitor Hydroxyipral Hydroxyipral Aspergillic Amt Mmoles.. Amt of test compound DL but by certain of its precursors available for testing, e.g., a-keto-isovalerate and a-acetolactate. The competition between the inhibitor and valine or its precursors appears noncompetitive, which implies that the hydroxybarbituric function as growth inhibitors by interfering with the synthesis of acetolactate. The synthesis of acetolactate and acetohydroxybutyrate, which are precursors of valine and isoleucine, respectively, is thought to be catalyzed by the same enzyme (Umbarger and Brown, 198; Leavitt and Umbarger, 1961). If the hydroxybarbiturates interrupt E. coli metabolism at this point, it becomes apparent why their toxicity is over- Hydroxybarbital Hydroxyphenobarbital L-ISO- D- leu- Valine cine,umoles,umoles Growth (absorbance at 6 mia) Microbiological assay conditions as described in t Experiment I with E. coli strain B; Experiments II to V, strain Crookes. TABLE 3. Effect of valine and valine precursors on the toxicity of hydroxyipral in Escherichia coli Amt (Amoles) of Metabolite Amt hydroxyipral pmoles None.... 7t.1. DL-Valine ce-keto-isovaleratet DL-a-Acetolacetate Microbiological assay conditions as described in t Growth expressed as absorbance at 66 m,. $ Added aseptically after autoclaving. Downloaded from on November 1, 18 by guest

4 168 HOGG, BISWAS, AND BROQUIST J. BACTERIOL. TABLE. Inhibition of the acetolacetate-forming TPP, FAD, Mg++ system ( Pyruvate acetolacetate + CO) in a cell-free extract of Escherichia coli Crookes by cyclic hydroxamates and related compounds Addition Amt Inhibition,moles % Barbital Hydroxybarbital Ipral Hydroxyipral Phenobarbital Hydroxyphenobarbital 3 1 Aspergillic... o-phenanthroline.... Enzymatic assay conditions as described in V SxlIO3M VALINE APYRUVATE xlo3m) X-- x 1-3M HYDROXYIPRAL I,/(FAD x 1M) C.C C. IxlO M HYDROXYIPRAL / I ZfPYRUVATEX 13M) G7- ox-.3 S x 1 M HYDROXYIPRAL 7x / XTPP x 1- M) FIG.. Double reciprocal plots showing the nature of the inhibition of the acetolactate-forming system by valine or hydroxyipral. 1/V is reciprocal of acetoin formed under conditions as described in come by the combination of valine and isoleucine (Table ). The synthesis of acetolactate from pyruvate in a suitable cell-free enzyme system from E. coli was studied. Both the parent barbiturates and the hydroxybarbiturates under investigation showed significant inhibition of the system, although relatively high levels of these compounds were required (Table ). For example, 1,umoles of the barbiturates or their respective hydroxybarbiturates gave from to 3% inhibition of acetolactate synthesis under the conditions of the experiment. Hydroxyipral, which was far more potent as a growth inhibitor than its parent barbiturate, ipral (Table 1), was only about twice as active as ipral in blocking acetolactate formation (Table ). Aspergillic was the most potent cyclic hydroxymate in inhibiting the system, which was of particular interest since the antibacterial effect of this antibiotic is not reversed by isoleucine plus valine (Table ). The possibility was considered that the effects of these cyclic hydroxamic s on growth (Table 1) and on the acetolactate-forming system (Table ) might be due at least in part to the binding of an essential cation such as Fe++(+). Indeed, o-phenanthroline effectively inhibited acetolactate formation (Table ). This observation led to the experiments in Tables and 6. A detailed kinetic study of the relationship between hydroxyipral and the substrate and cofactors of the acetolactate-forming system was performed to gain some insight into the mechanism of inhibition of this system by hydroxybarbiturate. The results expressed as double reciprocal plots are summarized in Fig.. In confirmation of the findings of Umbarger and Brown (198), valine effectively inhibited the acetolactate-forming system in E. coli and was reversed competitively by the substrate, pyruvate. The relationship between hydroxyipral and pyruvate was not as clear-cut, however, and the kinetics appeared to be mixed in nature. When the substrate was kept constant and the cofactors FAD and TPP were varied, the kinetics of hydroxyipral inhibition were essentially noncompetitive (Fig. ). Table indicates that the toxicity of hydroxyphenobarbital, aspergillic, and o-phenanthroline for growth of E. coli is reversed by Fe+ but not by Mg++; however, Fe+ did not alter the toxicity of hydroxyipral. In the acetolactateforming system (Table 6), Fe+ actually enhanced hydroxyipral inhibition, but the inhibitory action of aspergillic, o-phenanthroline, and, to a lesser extent, of a, a-dipyridyl was overcome by Fe+. Table 6 strongly suggests that there is an iron requirement in the acetolactate-forming system of E. coli, but that hydroxyipral blocks acetolactate formation by a Downloaded from on November 1, 18 by guest

5 VOL. 9, 196 ANTIBACTERIAL ACTION OF CYCLIC HYDROXAMIC ACIDS 169 TABLE. Effect of Fe++ and Mg+ on the toxicity of cyclic hydroxamates and o-phenanthroline in Escherichia coli Growth Inhibitor Amt (FeSO) Fe++M : (MgSO (anbsorb ance pmoles,imoles,umoles Hydroxyipral Hydroxypheno barbital Aspergillic o-phenanthroline Microbiological assay conditions as described in Inhibitor Hydroxyipral Aspergillic o-phenanthrolinie a, a-dipyridyl Amt 1 1 Fe++ Mg (FeSO) (MgCI) Inhibition Enzymatic assay conditions as described in mechanism other than interference with any cationic requirement in the enzyme system. The possibility was considered, in view of certain structural features shared by hydroxyipral and valine (Fig. 1), that hydroxyipral might TABLE 7. Inhibition of the acetolactate-forming system by valine and related compounds Addition to complete system Amt Inhibition j,moles % None... L-Valine L-Valine amide L-Valine hydroxamate L-Acetyl valine Isovaleric L-Isoleucine L-Leucine Enzymatic assay conditions as described in mimic valine and bind to acetohydroxy synthetase at a site vulnerable to feedback inhibition by valine. From this standpoint, it appeared instructive to consider the structural alterations of valine that could be tolerated without elimninating its effective inhibition of the acetolactate-forming system. Table 7 shows that the carboxyl group of valine may be blocked, as in valine hydroxamate or valine amide, without eliminating effective inhibition. If the amino TABLE 6. Reversal of cyclic hydroxamates and iron group of valine is acylated, or absent, as in isovaleric, such compounds have no effect on chelator inhibition of acetolactate-forming system by Fe+ and Mg++ the enzyme system. These latter findings could II~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ be Additions interpreted to mean that hydroxyipral in- (ulmoles/ml) hibits the acetolactate-forming enzyme system at loci other than the feedback site, for, if carbon- of hydroxyipral is viewed as analogous to the a carbon of valine, it can be seen (Fig. 1) that carbon lacks an amino substituent. Indeed the closely related amino s, isoleucine and leucine, wvere markedly less active than valine in inhibiting acetolactate synthesis (Table 7). DISCUSSION Certain aspects of the foregoing experiments were disappointing in that the mechanism of action of hydroxyipral in disrupting valine and isoleucine synthesis could not be fully explained at the enzymatic level. This may have been due in part to the crude state of the cell-free extract; however, repeated attempts to purify the system proved unsuccessful. The finding (Table 3) that the toxicity of low levels of hydroxyipral could be overcome by either valine or its initial precursor, acetolactate, gave a strong indication that hydroxyipral might be blocking acetolactate (and acetohydroxybutyrate) formation. However, in the acetolactate-forming enzyme system hydroxyipral was only twice as active as ipral, whereas hydroxyipral is at least 1 times more potent than ipral in inhibiting growth of E. Downloaded from on November 1, 18 by guest

6 17 HOGG, BISWAS, AND BROQUIST J. BACTERIOL. coli (Table 1). The possibility that hydroxyipral might be a repressor of synthesis of acetohydroxy synthetase is presently being considered. Despite certain gross structural similarities between valine and hydroxyipral (Fig. 1), the latter should not be viewed as an antimetabolite of valine, for it is not a competitive antagonist of valine in growth experiments (Table 3), and kinetic studies of its inhibitory action on acetolactate synthesis revealed significant differences when compared with valine (Fig. ). The possibility that hydroxyipral, a cyclic hydroxamic, might be exerting its effect by complexing with an essential cation, particularly Fe+, was not borne out by experiment, but led to the finding that certain iron chelators, e.g., o-phenanthroline and a,a-dipyridyl, together with aspergillic, blocked acetolactate formation and that such inhibition was reversed by Fe-. If aspergillic interferes in general with iron metabolism in cells, it becomes understandable why valine and isoleucine alone failed to significantly alter the toxicity of this antibiotic for E. coli (Table ). The finding that there is an iron requirement in the acetolactateforming enzyme system of E. coli takes on added significance in light of the recent recognition of the involvement of FAD in this enzyme system (Strmer and Umbarger, 196). ACKNOWLEDGMENT This investigation was supported by grant GB from the National Science Foundation. LITERATURE CITED DAVIS, B. D., AND E. S. MINGIOLI. 19. Mutants of Escherichia coli requiring methionine or vitamin B1 J. Bacteriol. 6:17-8. DUTCHER, J. D., AND. WINTERSTEINER. 19. The structure of aspergillic. J. Biol. Chem. 1: KRAMPITZ, L Synthesis of a-acetolactic. Arch. Biochem. Biophys. 17:81-8. LEAVITT, R. I., AND H. E. UMBARGER Isoleucine and valine metabolism in Escherichia coli. X. The enzymatic formation of acetohydroxybutyrate. J. Biol. Chem. 36: LOWRY,. H., N. ROSEBROUGH, A. L. FARR, AND R. J. RANDALL Protein measurement with the Folin phenol reagent. J. Bol. Chem. 193: 6-7. MACDONALD, J. C Biosynthesis of aspergillic. J. Biol. Chem. 36:1-1. SAFIR, S. R., J. J. HLAVKA, AND J. H. WILLIAMS Synthesis of compounds related to the barbituric s. J. Org. Chem. 18: STRMER, F. C., AND H. E. UMBARGER The requirement for flavin adenine dinucleotide in the formation of acetolactate by Salmonella typhimurium extracts. Biochem. Biophys. Res. Commun. 17:87-9. UMBARGER, H. E., AND B. BROWN Isoleucine and valine metabolism in Escherichia coli. VIII. The formation of acetolactate. J. Biol. Chem. 33: WESTERFELD, W. W. 19. A colorimetric determination of blood acetoin. J. Biol. Chem. 161: 9-. WHITE, E. C., AND H. H. HILL Studies on antibacterial products formed by molds. I. Aspergillic, a product of a strain of Aspergillus flavus. J. Bacteriol. :33-. Downloaded from on November 1, 18 by guest