JOURNAL OF ~ I O L O G I C A LCHEMrSTRY Vol. 255, No. 3. Issue of April 10, pp. 3049-301056, 1980 Printed m I?.S A.

THE

The Biosynthesis of Cyanogenic Glucosides in Higher Plants CHANNELING OF INTERMEDIATES IN DHURRIN BIOSYNTHESIS BY A MICROSOMAL SYSTEM FROM S O ~ BICOLOR G ~ (LINN) ~ ~MOENCH* ~ (Receivedfor publication, October4, 1979)

Birger Lindberg M%ller+and Eric E. Conn From the Department of Biochemistry and Biophysics, University of California, Davis, California 95616

synthesis has been the low amounts of intermediates present (3, 7). Generally, only the in the enzymatic reaction mixtures tyrosine, N-hydroxytyrosine, p-hydroxyphenylacetal- compound used as substrate and the end product, p-hydroxdoxime, p-hydroxyphenylacetonitrile, and p-hydroxy- ybenzaldehyde, formednonenzymaticallyfromp-hydroxymandelonitrile. N-Hydroxytyrosine and p-hydroxy- mandelonitrile (3),are present in quantitieseasily detectable. phenylacetonitrile produced from L-tyrosine by micro- However, p-hydroxyphenylacetaldoxime doesaccumulate somes from seedlings of Sorghum bicolor are utilized when sorghum microsomes are prepared in the absence of more effectively as substrates than exogenously added mercaptoethanol and dialyzed under air instead of nitrogen N-hydroxytyrosine and p-hydroxyphenylacetonitrile. (3, 4). p-Hydroxyphenylacetonitrile accumulatesinsmall The minimum values f o r the channeling ratios are 25 quantities when p-hydroxyphenylacetaldoxime is incubated f o r N-hydroxytyrosine and 115 f o r p - h y d r o x y p h e n y l a - with themicrosomal system under anaerobic conditions in the cetonitrile. On the other hand, p-hydroxyphenylacetal- presence of NADPH (4). Accumulation of N-hydroxytyrosine doxime produced internally exchanges readily with exn g 14ogenously added ~-hydroxyphenylacet~do~me. These has only been observed in experiments u t ~ ~ carbon labeled tyrosine as substrate and large quantities of unlabeled results indicate that the biosynthetic pathway is catal y z e d b ytwo multienzyme complexes or by two multi- N-hydroxytyrosine as a trap (3, 8). In this report, data are functional proteins and explain why the rate of the presented which suggest that the sorghum microsomes conoverall sequential reaction starting from L-tyrosine is stitute a highly organized enzyme system exhibiting catalytic providing an efficient mechanism for greater than the rates of reaction initiated later in the facili~tion and thereby sequence with the known intermediates N-hydroxyty- channeling the flow of carbon from tyrosine into dhurrin. A rosine and p-hydroxyphenylacetonitrile. A t t e m p t s to preliminary report of these findings has already appeared(9). cross-link chemically the last enzyme in the p a t h w a y , EXPERIMENTALPROCEDURES a soluble UDP-glucose glucosyl-transferase, to the microsomal system were unsuccessful. ChemicaZs-Bis(meth~1)suberimidatewas synthesized from subemnitrile (IO). All other chemicals were synthesized or purchased as described earlier (3, 11, 12). ~-[U-“C]Tyrosine(specific activity, 440 mCi/mmol) and ~-[ring-2,3,5,6-’H]tyrosine (specific activity, 90 Ci/ T h e cyanogenic glucoside dhurrin (P-D-glucopyranosyloxy- mmol) were purchased fromNew England Nuclear, Boston,MA. ChemicalSynthesis of Radioactive Intermediates-Chemical S-p-hydroxymandelonitrile) is rapidly synthesized from L-tyrosine in seedlings of Sudan grass, Sorghum bicolor (Linn) synthesis of uniformly14C-labeled N-hydroxyt~sine,p-hydroxye, was carried out Moench. The dry, starch-rich sorghum seed contains no de- p h ~ n y l a c e t a l d o ~andp-hydroxyphenylacetonitr~e as earlier described with ~-[U-’~C]tyrosine as the starting material tectable dhurrin, butthe 3-day-old seedling contains15pmol (7). The N-hydroxytyrosinewas purified by ion exchangechromatogof dhurrin/g freshweight or approximately40 mg of dhurrin/ raphy (7). p-Hydroxyphenylacetaldoximeand p-hydroxyphenylaceg dry weight (1).Studies involving labeled precursors, trapping tonitrile were purified by preparative thin layer chromatography(7) by the GLC/GPC’ experiments, and cell-free enzyme systems have shown that and were radiochemically pure when analyzed ~ ~ as L-tyrosine is converted to dhurrin (Fig. 1)and that N-hydrox- procedure (11) and by scanning thin layer c ~ o m a t o with ytyrosine, p-hydroxyphenylacetaldoxime,p-hydroxyphenyla- radiochromatogram scanner(Packard model 7201). Analysis of Radioactive Intermediates-Reaction mixtures which cetonitrile and p-hydroxymandelonitrile are intermediates in did not contain N-hydroxytyrosinewere analyzed by thin layer chrothe pathway (2-5). A microsomal preparation from etiolated matography (3, 11).The TLC plates (Bakerflex 1B-F flexiblesheets) the conversion of L-tyrosine top- were prestreakedwithunlabeled standards to allowvisualization sorghum seedlings catalyzes hydroxymandelonit~e,i.e. all but thelast step in the pathwayunder ultraviolet light after c~omatography.Aliquots from biosyn( 2 ) . The latter compound is then converted to dhurrin bya thetic reaction mixtures were streaked directly on the plates. This resulted in immediate inactivationof the microsomal enzyme system soluble and specific UDP-glucose glucosyl transferase (6). judged by the linearity of the reaction with time (3). The TLC A major obstacle to unraveling the pathway of dhurrin as plates weredevelopedin benzene/ethyl acetate (51, v/v) and the * This project was supported by the Danish Natural Science Re- radioactivityin separated compoundswas determined by cutting search CouncilGrant 511-3988 and a Fuibright-Hays Act Fellowship appropriate areas into counting vials. The RFvalues obtainedwere: (to B. L.M.) and by NationalScienceFoundation Grant BMS- tyrosine, 0.W; p-hydroxybenzoic acid, 0.06;p-hydroxyphenylacetal7411997 and United States Public Health Service Grant GM-05301 doxime, 0.17; p-hydroxybenzaldehyde,0.34; and p-hydroxyphenyla(to E. E. C.). The costs of publication of this article were defrayed in cetonitrile, 0.42. part by the payment of page charges. This article must therefore be 18 U.S.C. Section hereby marked“advertisement”in accordance with ’ The abbreviations used are: GLC/GPC, gas-liquid chromatogra1734 solely to indicate this fact. phyigas propo~ional counting; Tricine, N-[T~s(hydroxymeth$ Present address, Department of Physiology, Carlsberg Labora- yl)methyllglycine;a l d o x ~ e , ~ h y d r o x ~ h e n y l a c e t a l d onitrile,px~e; tory, 10 Gamle CarlsbergVej, DK-2500 Copenhagen Valby,Denmark. hydroxyphenylacetonitrite;aldehyde, p-hydroxybenzaldehyde.

The biosynthetic pathway f o r the cyanogenic glucoside, dhurrin, involves the following intermediates: L-

3049

3050

Cyanogenic

Glucoside

Reaction mixtures containing N-hydroxytyrosine, which accumulates only under special conditions (3), were analyzed with a gas chromatograph (GLC) coupled to a gas proportional counter (GPC), as earlier described (11). This was necessary because N-hydroxytyrosine oxidatively decarboxylates, forming p-hydroxyphenylacetaldoxime, when analyzed by TLC (7). Ahquota from enzymatic reactions were lyophiliid to dryness and silylated by refluxing for 15 min at 90°C with an excess of N,O-his-(trimethylsilyl)-trifluoroacetamide in acetonitrile. An unlabeled standard mixture of t~ethyls~yl-de~vatized intermediates was also added. Aliquots were injected into the GLC/GPC equipped with a SP-2250 column. The temperature program consisted of an initial period of 6 min at 155”C, a temperature rise of 30”C/min, and a final period of 12 min at 1’75°C. The area of each labeled peak obtained was integrated electronically and the peak was identified by retention time and superimposition with the mass peak of the authentic standard. Mass and radioactivity tracings obtained using the GLC/GPC procedure have been published previously (3). The presence of dhurrin in microsomal reaction mixtures was also analyzed by use of the GLC/GPC procedure after trimethvlsilvlation of the sample. Since the molecul&weight of t~ethyls~y~ate~ dhurrin is large compared to those of the derivatized intermediates, the regular temperature program was expanded to include an additional temperature rise of 30*C/min and a final period of 12 min at 255%. With this program, dhurrin was eluted after 6.5 min at this final temperature. Detennirtation of Specific Activity-The commercial samples of L-[U-‘~C] and L-[ring-2,3,5,6-3H]tyrosine were diluted to suitable specific activities by addition of known amounts of unlabeled L-tyrosine. The diluted material was then recrystallized from ethyl alcohol-H’pO and its specific activity was determined by liquid scintillation counting of a weighed aliquot. The specific activity of intermediates synthesized from L-[II-‘*C]tyrosine was obtained from the specific activity of the b-[LJ-“C]tyrosine used by correcting for the loss of 1 or 2 carbon atoms during the synthesis of p-hy~ox~henylacetaldoxime or phydroxyphenylacetonitrile, andp-hydroxybenzaldehyde, respectively. Measurement of Radioactiuity-Radioactive samples (crystalline material, aqueous samples, or radioactive spots cut out from TLC plates) were measured after the addition of a total of I ml of Hz0 and 10 ml of a counting solution composed of 0.5 liter of Triton X-106, 1 liter of toluene, 7.50 g of 2,5-diphenyloxazole and 0.30 g of 1,4-bis[2(5-phenyloxazolyl)]benzene. All samples were kept in the dark for at least 2 days before counting in a Searle Mark II scintillation counter using the discriminator settings of Program 08 designed for samples containing both “H and I*C. Under these conditions, there was no spillover of “H into the ‘% channel and the cpm in the 14C channel was therefore converted directly to dpm of I%! by use of the external standard ratio. Correction for spillover of *% into the “H channel was made using a calibration curve obtained by counting a series of samples confining only “‘C. After subtraction of the ‘%.spillover, the remaining cpm in the “H channel were converted to dpm of “H by means of the external standard ratio. In enzymatic experiments, the .‘H and ‘% dpm values observed in the different metabolites were converted to nanomoles of metabolite by use of the known specific activity of the substrate employed. In this conversion, correction was made for the loss of 1 or 2 carbon atoms when products were formed from uniformly ‘%-labeled tyrosine. Seed Material and Microsomal Preparations-Seed of S. bicolor (Linn) Moench were obtained from Northrup, King & Co. Lubbock, TX. Tbe hybrid Sordan 70 was used in the investigation of the effect of dialysis time on the accumulation of intermediates (3). The hybrid Redland x Greenleaf was used in all other experiments. Microsomal preparations were obtained as earlier described (3), unless otherwise indicated. Experiments to Detect Enzyme-bound Intermediates-The production of bound intermediates was analyzed with the foilowing ‘%Zlabeled substrates: L-tyrosine (specific activity, 2.50 mCi/mmol), Nhydroxytyrosine (specific activity, 2.50 mCi/mmol), p-hydroxyphenylacetaldoxime (specific activity, 2.22 mCi/mmol), and p-hydroxyphenylacetonitrile (specific activity, 2.22 mCi/mmol). Each reaction mixture contained 0.2 pmol of ‘*C-labeled substrate, 0.3 pmol of NADP”, 3 pmol of glucose-6-P, 3 IU of glucose-6-P dehydrogenase, 3.0 ma of microsomal protein, and 50 Gmol of Tricine (pH 8.0) in a total Volume of 810 ~1: Aliquots (150 pl) were removed-at 5, IO, 30, and 60 min and pipetted into small plastic centrifuge tubes containing either 30 ~1 of 100% trichloroacetic or 206 ~1 of ethyl alcohol. Precipitated protein was isolated by centrifugation, and after five repetitive washings with 20% triehioroacetic acid or 60% ethyl alcohol, respec-

Biosynthesis

tively, the protein precipitate was resuspended in 1 ml of 1 N HCl and counted in a liquid scintillation counter. In a second series of experiments, with reaction mixtures identical to those described above and 30-min incubation time, the microsomal protein was recovered from the reaction mixtures by ultracentrifugation (100,000 X g for 30 min), washed, and counted as described above. The possible production of activated intermediates (i.e. coenzyme A derivatives or phosphates) was investigated with ‘%-labeled tyrosine as substrate. The reaction mixtures used were as described above, except that, in one experiment, 1.2 pmol of unlabeled N-hydroxytyrosine were added as a trap. After 30-min incubation, the reaction mixtures were transferred to an ice bath and an equal volume of 12 N HCl was added. After standing for 3 days at room temperature, the reaction mixtures were lyophilized and their content of labeled intermediates was assayed by the GLC/GPC procedure and compared with the content of unhydrolyzed reaction mixtures. N-Hydroxytyrosine is stable under these conditions (7) and therefore could be measured. p-Hy~xyphenylacet~doxime, however, undergoes some decomposition in acid and was therefore not quantitated. Alkaline hydrolysis could not be employed as N-hydroxytyrosine would decompose into other intermediates and produce ambiguous results (7). Experiments Involving Simultaneous Incubation of 3H- and 14Clabeled Substrates-The reaction mixtures contained 0.2 gmol of NADP’, 2 pmol of glucose-6-P, 4 IU of glucose-6-P dehydrogenase, 0.60 mg of microsomal protein, 15 pmol of phosphate buffer (pH 7.2), and 0.160 pmol of each radioactively labeled substrate (Table I) in a total volume of 202.5 $1. The specific activities of the labeled substrates were: r$H]tyrosine, 7.36 mCi/mmol; [“‘Clp-hydroxyphenylacetaldoxime, 6.39 mCi/mmol; [“‘Clp-hydroxyphenylacetonitrile, 2.49 mCi/mmol. In one experiment, the detergent, Triton X-106, was added to a final concentration of 0.05%. Aliquots (30 el) were taken at 2’ = 4, IO, 16,20,30, and 60 min and analyzed by the’ TLC procedure. Exoeriments Involvina Seuuential Incubation of 3H- and %‘lab&d Substrates-Thg readtion mixtures contained 0.3 pmol of NADP’, 3.0 pm01 of glucose-6-P, 4 IU of glucose-6-P dehydrogenase, 0.45 mg of microsomal protein, 25 pmol of phosphate buffer (pH 7.2), and 0.160 pmol of a radioactively labeled substrate in a total volume of 162.5 ~1. Aliquots (15 ~1) were taken at T = 2,4, 6, and 8 min. At 2’ = 8 min (Fig. 2), 0.160 pmol of a differently labeled substrate were added (40 ~1) and a second series of aliquots (25 ~1) were made at 2’ = 10, 12, 16, 18, and 20 min. The specific activities of the substrates used were as indicated above. All aliquots were analyzed by the TLC procedure. Experiments to Determine the Exchange between Internally Produced and Externally Added N-Hydrorytyrosine-Reactions were carried out in l-ml ampules. The reaction mixtures used contained 0.2 Fmol of NADP’, 2 ymol of glucose-6-P, 4 IU of glucose-6-P dehydrogenase, 2.0 mg of microsomal protein, 30 gmol of Tricine buffer (pH 8.0), and 0.160 pmol of each substrate (Table II) in a total volume of 380 ~1. The specific activities of the substrates were: [“‘Cltyrosine, 5.00 mCi/mmol; N-hydroxy[‘%]tyrosine, 3.37 mCi/ mmol. After 20-min incubation, the ampules were immersed in liquid nitrogen to stop any enzymatic reaction. Their contents were then lyophilized to dryness, silylated, and analyzed by the GLC/GPC procedure. Effect of Dialysis Time on the Accumufation of IntermediatesMicrosomes from etiolated sorghum seedlings (S. bicolor (Linn) Moench C.V. Sordan 70) were prepared without mercaptoethanol as earlier described (3) and aliquots (1 ml) were dialyzed against 20 mM Tricine buffer (pH 8.0) for different time periods (Fig. 3). At each time period, the microsomes were assayed in reaction mixtures containing 0.3 pmol of NADP’, 3.0 pmol of glucose-6-P, 4 IU of glucose6-P dehydrogenase, 2.0 mg of microsomal protein, 30 pmol of Tricine (pH 8.0), 0.060 gmol of r.-[‘4C]tyrosine (specific activity, 16.71 mCi/ mmolf in a total volume of 665 ~1. The incubation time used was 15 min and aliquots were analyzed by combined use of the TLC and GLC/GPC procedures (3, 11). Experiments to Cross-link the UDP-Glucose Glucosyltransferase to the MicrosomaE Complex by Use of an Imidate-Etioiated sorghum seedlings (10 g) were homogenized in a freshly prepared medium (20 ml) containing 3 mmol of triethanolamine, 7 mmol of NaCl, 0.3 pmol of mercaptoethanol, 0.2 mmol of bisfmethylfsuberimidate and 1 z of nolwinvlpolvpvrrolidone at pH 8.2. Microsomes were obtained by differential &ntrifugation,-as earlier described (3). In a control experiment, bis(methy1fsuberimidat.e was omitted from the homogenization buffer. The enzymatic activity of the microsomal preparations was assayed in reaction mixtures composed as described in the preceding section except that 2.0 pmol of

Cyanogenic Glucoside B i o s ~ ~ ~ ~ e s i s UDP-glucosewere added toeach reaction mixture.Thereaction mixtures were incubated for 30 min and then analyzed by use of the CGL/GPC procedure as modified for dhurrin detection. RESULTS

Analysis of Reaction Mixtures for Intermediates-When sorghum microsomes are incubated with radioactively labeled L-tyrosine orp-hydroxyphenylacetaldoxime(aldoxime)in the presence of NADPH, p-hydroxybenzaldehyde is the major product formed (Table I, Experiments 1 and 3). While the aldoxime and p-hydroxyphenylacetonitrile (nitrile) are established i n t e ~ e d i a t e sin the conversion of L-tyrosine to phydroxy-(S)-mandelonit~e, theyare present in concentrations 1 order of magnitude smaller. The low amounts of intermediates observed could be the result of such intermediates being enzyme-bound during biosynthesis. This possibility was ruled out by reisolation of the microsomal protein after incubation with radioactively labeled substrates in the presence of NADPH. The protein was isolated by precipitation with either trichloroacetic acid or ethanol or was recovered by ultracentrifugation of the incubation mixtures. In no case did the isolated protein contain radioactivity. Acid hydrolysis of the reaction mixtures alsodid not produce additional amounts of free intermediates. The results thus indicate that none of the intermediates are covalently bound to enzymes and also suggest the absence of quantitatively signifcant amounts of activated intermediates such as coenzyme A or phosphate esters. The formation of such activated intermediates would presumably also have been indicated by specific cofactor requirements of the microsomal system.

3051

The low amounts of intermediates present in the microsomal reaction mixtures (Table I) and the earlier reported difficulties in trapping both N-hydroxytyrosine (3) and phydroxyphenylacetonitrile (4) are understandable if the reactions of the biosynthetic sequence (Fig. 1) take place in a highly organized enzyme system so that intermediates produced internally are preferentially utilized by the next enzyme in the sequence. Such intermediates would not dissociate readily from the surface of the enzymes involvedand therefore would not exchange readily with intermediates added from outside. This possibility was investigated in experiments utilizing two substrates labeled with different radioisotopes. Experiments Utilizing S ~ m ~ Z t ~ n eAdditzon ous of Two Differently LabeEed Substrates-In one type of experiment, two substrates labeled with different radioisotopes (3H and I4C) were administered simultaneously to thesorghum microsomal system in the presence of NADPH. The amount of each substrate used was sufficient to saturate the enzyme system for incubation periods up to30 min. Aliquots taken at different time periods were analyzed for labeled intermediates by thin layer chromatography and the content of "H and I4C in each intermediate was determined by liquid scintillation counting. From the known specificactivities of the 'H- and I4C-labeled substrates employed, the nanomoles of intermediates formed both from the 'H- and the 14C-labeledprecursor were calculated separately. Data obtained after 20 min of incubation are presented in Table I (Experiments 5 and 6). Qualitatively similar data were obtained at 4, 10, 16, 30, and 60 min of incubation. When ["Hltyrosine and ['4C]p-hydroxyphenylacetaldoxime were usedsimultaneously as substrates (TableI,

TABLEI The metabolism of radioactively labeled substrates in microsomal rea.ction mixtures in the presence of NADPH Saturating amountsof each substrate (160 nmol) were used in all experiments. Data presented are for an incubation time of 20 min. For

further experimental details, see "Experimental Procedures." . . Composition of reaction mixture after incubation

Experiment

1 2

['HITyrosine r3H}Tyrosine+ 0.05%Triton X-100

3 4 5

['"C]Aldoxime

6

_"

--

Substrate(s)

[ 'H],Tyro-

VHIAl-

sine

doxime

128 130

3.6 10.6

,:,HINitrile

0.67 0.43

["HIAlde"C Aldoxhyde Le -. ." nmol

28 19.4

143

10.9

142

2.5 11.0

0.33

63 a2

5.7

1.68 143

st:c$iEi __

32 30 56

[I4C]Nitrile ['HITyrosine + ['dC]aldoxime ["HITyrosine + ['4C]nitrile __

Total sub['TIAldehyde

14.5

3.5 141 3.4

89 14.3

104 19 95 35

0 ' \ /

HO~CH2-CH-COOH-"o

CH2-CH-COOH +HO

YH

AH2

OH L-tyrosine

N-hydroxytyrosine p-

hydroxyphenylocetoldoxime

I HOOyH-5, I

glucose dhurrin

c "

N

-

p hydroxy mondelonitrile

p-hydroxybenzaldehyde

H O o C H 2 - C Ill

OH N

HO-@HO

FIG.1. Biosynthetic pathway for the cyanogenic glucoside dhurrin.

p-hydroxyphenylocetonitrile

+

HCN

3052

'

~ ~ a ~ o g Glucoside e n ~ c 3ios~n~~esis

Experiment 5), the molar 3H/"C ratio of the aldoxime isolated from the reaction mixture after 20 min of incubation was 0.133. The molar 3H/"C ratios for p-hydroxyphenylacetonitrile and for p-hydroxybenzaldehyde isolated after the same time periodwere slightly lower, namely 0.098 and 0.091, respectively. Thus, the aldoxime added externally competes effectively with aldoxime formedin situ from tyrosine during conversion of the amino acid intop-hydroxyphenylacetonitrile and p-hydroxybenzaldehyde and there is no channeling. Further analysis of Table I discloses that a totalof 69 nmol of p-hydroxybenzaldehyde were produced from ["H]tyrosine and ['4C]aldoxime incubated simultaneously with the sorghum microsomes (Table I, Experiment 5). When the two substrates were administered separately to the microsomal system (Table I, Experiments 1 and 3), a total of 117 nmol of aldehyde were formed. The main reason for the increased production of p-hydroxyben~ldehydewhen the substrates were administered separately is that saturating amounts of each substrate were used. Moreover,the ['4C]p-hydroxyphenylacetaldoximeused assubstrate in Experiment 5 acted as a trap for [~H]p-hydroxyphenylacetaldoxime formed from ['Hltyrosine in that experiment. A better measure of metabolic activity in these experiments is the amount of substrate which is utilized. Whenthe two substrates were administered separately (Table I, Experiments 1 and 31, a total of 136 nmol were metabolized. In Experiment 5, this value was 95 nmol. From the known composition of the reaction mixture in this experiment, it can be calculated that this lower value is obtained because a saturating concentration of aldoxime inhibits tyrosine metabolism 48%while a saturating concentration of tyrosine inhibits aldoxime metabolism only 25%. A similar experiment was performed in which ['Hltyrosine and ['4C]p-hydroxyphenylacetonitrile were administered simultaneously to the microsomal system (Table I, Experiment 6). The molar 3H/14Cratio ofp-hydroxyphenylacetonitrileand p-hydroxybenzaldehyde reisolated from the reaction mixture after 20 min of incubation was 0.012 and 1.32, respectively. These figures indicat,e that p-hydroxyphenylacetonitrile added externally is not able to compete effkiently with phydroxyphenylacetonitrileproduced in situfrom tyrosine during conversion of the latter into p-hydroxyben~dehydeby the microsomal enzyme system. A channeling ratio of 1.32: 0.012 or 112 may be calculated, and this means that not more than 1 molecule of p-hydroxyphenylacetonitrile (3H-labeled) exchanges with the pool of externally added nitrile ("C-labeled) when 112 molecules of ['Hltyrosine are converted to ['HIP-hydroxybenzaldehyde via["Hlp-hydroxyphenylacetonitrile. When saturating concentrations of tyrosine and p-hydroxyphenylacetonitrile are administered simultaneously to the microsomal system, a totalof 35 nmol are metabolized (Table I, Experiment 6). When the two substrates are administered separately, a total of 51 nmol are metabolized (Table I, EXperiments 1 and 4). From the composition of the reaction mixtures (Table I, Experiments 1, 4, and 6), it again can be calculated that the saturating concentration of p-hydroxyphenylacetonitrile inhibited tyrosine metabolism 46% while the saturating concentration of tyrosine inhibited p-hydroxyphenylacetonit~emetabolism only 10%. Experiments Utilizing Sequential Additionof Two Differently Labeled Substrates-In a second type of experiment, the microsomal system was initially incubated with only one substrate. Then, after a period of metabolism, the second and differently labeled substrate was added. Aliquots were taken during incubation to calculate the rateof conversion of each substrate intop-hydroxybenzaldehyde. Again, the concentrations of the substrates usedwere at saturation even after

metabolism and dilution of the reaction mixture upon addition of the second substrate. Preliminary experiments with each substrate incubated separately with the microsomal system also showed that the production of p-hydroxybenzaldehyde was linear with time. When [3H]tyrosine wasused as the f b t substrate, the rate of ["Hlp-hydroxybenzaldehyde production was 141 nmol/mg of protein/h (Fig. 2 4 ) . Upon addition of [ i 4 C ] p - h y ~ o x ~ h e n ylacetaldoxime at 8 min, the total rate of p-hydroxybenzaldehyde production from both compounds increased to 201 nmol/mg of protein/h. However, most of the p-hydroxybenzaldehyde now formed (191 nmol/mg of protein/h) was derived from the [14C]p-hydroxyphenylacetaldoximeadded at 8 min, and the total rate of p-hydroxybenzaldehyde production after addition of both substrates was still less than half the rate obtained whenp-hydroxyphenylacetaldoximewas administered initially as a single substrate (452 nmolimg of protein/ h) (Fig. 2 B ) . When, under the latter conditions, a saturating amount of t3H]tyrosine was added at 8 min (Fig. 2B), the total rate of p-hydroxybenzaldehyde production decreased only slightly to 414 nmolimg of protein/h. Radiochemical analysis of the product now formed disclosed that 402 nmol ofp-hydroxy~enzaldehydewere derived fromp-hydroxyphenylacetaldoxime while only 12 nmol originated from tyrosine. These two experiments (Fig. 2, A and B ) thus show that nearly all of the p-hydroxybenzaldehyde produced derives from p-hydroxyphenylacetaldoximewhen tyrosine and p-hydroxyphenylacetaldox~e are both present. This is the result expected if the [3H]p-hydroxyphenylacetaldoximeproduced in situ from [3H]tyrosine exchanges readily with the large quantity of externally added I4C-labeledp-hydroxyphenylacetaldoxime. Similar experiments were also carried out with [3H]tyrosine and ['4C]p-hydroxyphenylacetonitrile assubstrates. With ["Hltyrosine as thefist substrate and the later addition of a saturating amount of [i4C]p-hydroxyphenylacetonitrile,a decrease in total p-hydroxybenzaldehyde production from 142 to 106 nmol/mg of protein/h occurred (Fig. 2C). This addition of ['*C]p-hydroxyphenylacetonitrile only decreased the production of p-hydroxybenzaldehyde from ["Hltyrosine to 78 nmolimg of protein/h. When ['4C]p-hydroxyphenylacetonitrile wasused as the first substrate, the later addition of ['Hltyrosine at 8 min resulted in a totalproduction of 44 nmol of ["HIP-hydroxybenzaldehyde/mg of protein/h (Fig. 2 0 ) . However, the production of p-hydroxybenzaidehyde from ['4C]p-hydroxyphenylacetonitrile decreased dramatically at that time from 99 to 10 nmol/mg of protein/h. These results thus again indicate that p-hyd~oxyphenyiacetonitr~e produced in situ does not exchange to anysignificant extent with the pool of p-hydrox~henylacetonitrileadded externally. Microsomal Metabolism of N-Hydroxytyrosine-N-Hydroxytyrosine cannot be analyzedand quantitated by the TLC procedure used above (7). Therefore, a gas-liquid chromatographic procedure was employed which involvedthe separation of trimethylsilyl-derivatized intermediates and analysis of their I 4 C content by combustion to ['4C]C02 and continuous measurements in a gas proportional counter (11). Since the method could not be used for 'H counting, this study of N hydroxytyrosine metabolism involved a series of experiments in which the microsomal system was incubated separately or simul~neouslywith tyrosine and N - h y ~ o x ~ y r o s i naltere nately labeled with I4C (Table 11). In this way, the metabolism of the two compounds could be determined as well as the influence of the other substrate on the rate of metabolism determined. Table I1 (Experiments 1 and 2) shows that N-hydroxytyrosine metabohm is inhibited 65%by the addition of tyrosine.

3053

Cyanogenic Glucoside Biosynthesis 114C-aldoxime

4

0.

x

a: 20 n

8

12

16

20

8

12

16

20

C-nitrile

/3H-tyrorine

0

4

FIG. 2. Rate of p-hydroxybenzaldehyde production from differently labeled radioactive compounds. The experiments werestarted with the single compoundsindicated by the arrow at zero time. The second compound added in each of the four experiments (A,B, C, and D)is indicatedby an arrow at 8 min. For furtherexperimentaldetails, see “Experimental Procedures.”

r

>.

I I

15

CL

io 5

0

4

8

12

16

0 4 T I M E, min

20

TABLE11 Metabolism of N-hydroxytyrosineand tyrosine by the microsomal system Data presented are for an incubation time of 20 min. For experimental details, see “Experimental Procedures.” Experiment

Substratefs) ~

____-

_____ Unlabeled

14C-1~heled ””.

“C-Substrate utilized

1 2 3 4 ___”

[“ClAldehyde produced

.“___I-..-

N-Hydroxytyrosine N-Hydroxytyrosine Tyrosine Tyrosine

None Tyrosine None N-Hydroxytyrosine

~~~~~~~~~~~

lated

nmol

160 nmol 46

16 98

91

36 9 78 71

~

~

~

~

%,

0 0 0 3.6

29 10 61 57

Effect of Dialysis-It has previously been reported that However, the presence of N-hydroxytyrosine inhibited the metabolism of tyrosine only 7% (Table 11, Experiments 3 and when microsomes are prepared in the absence of mercapto4 ) . When 91 m o l of [“Cltyrosine were metabolized in the ethanol and dialyzed for 18 h, the major product obtained presence of a saturating, unlabeled N-hydroxytyrosine trap, upon metabolismof tyrosine isp-hydroxyphenylacetddoxime only 3.6 nmol were trapped as N-hydr~xy[‘~C]tyrosine (Tableinstead of p-hydroxybenzaldehyde (2, 3, 7). The effect of 11, Experiment 4 ) . This corresponds to a channeling ratio of dialysis time on this result was investigated more carefidly 25. However, 3.6 nmol is the minimum amount of N-hydrox- (Fig. 3) in the present study. With the experimental conditions ytyrosine produced in situ which exchanged with the exter- used, almost complete utilization of the tyrosine occurred. nally added pool because approximately 10%of the externally The enzymatic reaction is therefore not linear with time. added N-hydroxytyrosine pool was also metabolized during Further~ore,additional conversion of the accumulated pthe incubation (Table 11, Experiment 2). A total of 4.0 nmol hydroxybenzaldehyde into p-hydroxybenzoic acidwasalso of N-hydr~xy[’~C]tyrosine, therefore, had equilibrated with observed. It wasfound that p-hydroxyphenylacetddoxime than 14 h. the externally added N-hy~oxytyrosinepool during the ex- accumulated when the period of dialysis was longer periment. Thus, as with p-hydroxyphenylacetonitrile, inter- Microsomal preparations which accumulatedp-hydroxyphennally produced N-hydroxytyrosine exchanges only to a very ylacetaldoxime also did not metabolize p-hydroxyphenylacelimited extent with an externally added pooi of N-hydroxy- tonitrile.2 Mercaptoethanol is an inhibitor of some proteases tyrosine. The addition of a saturating amount of L-tyrosine (13). The accelerated inactivation of the enzyme system with inhibited N-hydroxytyrosine metabolism by 65% (Table 11, time in the absence of mercaptoethanol could thus be caused Experiments 1 and 2) and p-hydroxyphenylacetonitrile me- by proteolytic degradation. A second explanation could bethe tabolism by 10% (Table I, Experiments 4 and 6 ) . Even with facilitated oxidation of an essential sulfhydryl group in the this inhibition taken into consideration, high channeling of N- absence of mercaptoethanol. Preliminary experiments with hydroxytyrosine and p-hydroxyphenylacetonitrile are ob- different detergents which solubilizedthe microsomal system served. Contrariwise, p-hydroxyphenylacetdoxime readily exchanges with an externally added pool. ’B. L. Mgller, unpubiished results.

;

;

~

Cyanogenic Glucoside Biosynthesis

3054

/ I

iybenroic acid

?”Q

6-7

p-hydroxyphenylocetoldoxine

5 30 E 0

0

20

4

8

I2 (6 20 2824 Diolysis Time

32 hrs

FIG. 3. Effect of dialysis time on the accumulation of intermediates formed from L-tyrosine. For further experimental details, see “Experimental Procedures.”

resulted in complete loss of enzymatic activity. Treatment with a lower concentration of detergent (Table I,Experiment 2), which did not solubilize the microsomes, resulted in increased formation of aldoxime. From the datapresented above, it appearsthat theparticlebound enzyme system catalyzing the earlier enzymatic steps in the biosynthesis of dhurrin is highly organized.It is therefore surprising that the very last enzyme in the pathway, the UDP-glucose glucosyltransferase which converts p-hydroxymandelonitrile into dhurrin (Fig. l ) , is a soluble enzyme (6). We therefore questioned whether the g l u c o s y l t r a ~ f e r ais~ actually a part of the microsomal complex which dissociates during preparation of the microsomes. To study this question, the initial homogenization of the seedlings was carried out in triethanolamine buffer containing a bifunctional imidate, b~(methy1)suberimidate. The imidate will covalently link neighboring proteins if these have a free lysine €-aminogroup in suitable position, and might therefore possibly bind the glucosyltransferase to the microsomalcomplex. If this OCcurred, and if the enzymes werenot inactivated by the treatment, dhumn instead of p-hydroxybenzaldehyde should be the end product obtained. When the metabolic activity of such cross-linked microsomal preparations was examined, it was found that tyrosine was metabolized into p-hydroxybenzaldehyde at theSame rate as with untreated microsomes and that no dhurrin was formed. DISCUSSION

A multienzyme complex can be defined as an aggregate of different, ~ n c t i o n ~related y enzymes bound together by noncovalent forces into a highly organized structure (14). A multifunctional enzyme consists of a single polypeptide chain with multiple catalytic functions (15). In both systems, active centers catalyzing sequential reactions can form a composite active site which allows the intermediates to channel (16). A well characterized example of a channeled system is the soluble “aromatic complex” of Neurosporacrassa, which catalyzes five consecutivereactions in the shikimic acid pathway leading to thebiosynthesis of aromatic amino acids (17). This system wasoriginally thought to be a multienzyme complex containing five different polypeptides (18). Genetic analysis (19,20) and improved isolation techniques (21) have now shown that the aromatic complex is a single polypeptide

with five active sites (21). The soluble kaurene synthetase of higher plants (22) is another example of a bifunctional enzyme system capable of metabolic channeling (23). Due tothe limitation of experimental techniques, few equivalent studies on membrane-bound enzyme systems are available (14).However, phenylalanine ammonia lyase and two cinnamic hydroxylases, key enzymes in the biosynthesis of lignins, flavonoids, phenolic acids, coumarins, and stilbenes have recently been reported as assembled consecutive enzymes on microsomal and chloroplast membranes of plants (24, 25). Several lines of evidence suggest that theenzymes catalyzing the biosynthesis of dhurrin constitute an organized membrane-bound system capable of efficient metabolic channeling. First, the sorghum membrane preparation catalyzes a multistep conversion (2). Second, the formation of each individual intermediate in the reaction sequence can be shown but the experimental conditions required to demonstrate their formation varies (3, 4). Third, kinetic analysis shows that the particles preferentially utilize tyrosine and p-hydroxyphenylacetaldoxime instead of N-hydroxytyrosine and p-hydroxy( 3 , 4 ) . (The low rate of nitrile utilization is pheny~acetonit~le especially surprising since its hydroxylation to formp-hydrox~mandelonitrileis proposed as the next to last st,ep in the sequence (Fig. 1)). That the membrane system channels the flow of carbon from tyrosine into p-hydroxymandelonitrile is shown by the channeling ratios of 25 for N-hydroxytyrosine and 115 for p-hydroxyphenylacetonitrile obtained at saturating substrate concentrations. Similarly, when a saturating concentration of N-hydroxytyrosine or p-hydroxyphenylacetonitrile is added to amembrane preparation actively metabolizing tyrosine, the p-hydroxymandelonitrile subsequently produced still, to a significant extent, is derived fromtyrosine. These data thus demonstrate the channeling of the intermediates N-hydroxytyrosine and p-hydroxyphenylacetonitrile in dhurrin biosynthesis. By contrast, p-hydroxyphenylacetaldoximeproduced internally from L-t-yrosine exchanges freely with exogenous aldoxime at saturating substrate concentrations and there is no preferential utilization of either form. Moreover, when p-hydroxyphenylacetaldoxime is added to a membrane preparation actively metabolizing tyrosine, the p-hydroxymandelonitrile subsequently produced is derived mainly from the aldoxime (Fig. 2A). These results indicate that the sequence of reactions converting tyrosine into p-hydroxymandelonitrile is catalyzed by two bifunctional systems, the fist channeling the flow of tyrosine intop-hydroxyphenylacetaldoxime via Nh y d r o x ~ ~ s i nand e the second channeling the flowof phydroxyphenylacetaldoximeintop-hydroxymandelonitrile via p-hy~oxyphenylacetonitrile.This hypothesis would agree well with the observed accumulation of p-hydroxyphenylacetaldoxime after prolonged dialysis (Fig. 3) or after treatment with detergents (Table I, Experiment 2), indicating the specific inactivation of the second system. The conversion of p-hydroxyphenylace~doximeinto p - h y ~ o x ~ h e n y l a c e ~ n i t risi l e formulated as asimple dehydration reaction (Fig. 1). However, the reaction requires NADPH (2,4,26) and may be of complex nature. The catalysis of the reaction sequence from tyrosine t o p h y d r o x y m a n d e l o ~ t ~by e two separate enzyme SYStemS is also attractive from an evolutionary point ofview because aldoximes are believed to be the branch point from which either cyanogenic glucosidesor glucosinolates are formed (27). No plant has yet beenshown to contain both classes of compounds, but the enzyme system catalyzing the biosynthesis of the aldoxime may be similar in those plants which produce cyanogenic glucosidesor glucosinolates. Detailed analysis of the catalytic properties of the sorghum

Cyanogenic Glucoside Biosynthesis

3055

membrane enzyme system has raised several interesting ques- organization. Thus, p-hydroxyphenylacetonitrile,a lipophilic tions. For example, the apparentK , values for N-hydroxyty- compound containing the highly polarized nitrile group, may rosine and p-hydroxyphenylacetonit~evary greatly with dif- act as a detergent and strongly inhibit the metabolism of tyrosine (Table I, Experiment 6). Complex control mechaferent microsomal preparations while the values for V,,, remain constant (3). Results obtained with the same low nisms also regulate the enzymatic activities of the soluble substrate concentrations but different microsomal prepara- aromatic complex from N. c)%cssain that preincubation with 7-phosphate, the initial subtions were thus not reproducible. In the present study, this 3-deoxy-~-arabino-heptulosonate problem was overcome by the use of high concentrations of strate for the reaction sequence, dramatically decreases the each substrate to ensure saturation of the appropriate en- K,,, values of all five enzymatic reactions catalyzed by the zymes. However,data from several experiments indicate that complex (31). The preceding discussion raises questions which can probthe obtained channeling ratios are increased at low substrate concentrations. Values as high as 160 for N-hydroxytyrosine ably best be answeredby isolation and study of the individual and 280 for p-hydroxyphenylacetonitrilehave been obtained. enzymes involved.Unfortunately, attempts tocompletely solThus, the channeling ratios obtained with saturating substrate ubilize the membrane-bound enzyme system by a variety of concentrations may be minimum values. The reason for vari- detergents and by sonication results in loss of activity. p-Hydroxy-(S)-mandelonitrileis converted into dhurrin by ation in the ability of the membrane preparation to metabolize N-hydroxytyrosine and p-hydroxyphenylacetonitrileis not a soluble UDP-glucose glucosyltransferase (6). This is conknown. Oneexplanation could bethat themetabolism of these sistent with the observed formation of p-hydroxymandelonitwo intermediates is restricted either to partly broken enzyme trile as thefinal product of the microsomal system. Attempts complexes or to a proteolytically degraded multifunctional to obtain dhurrin as the final product by cross-linking the protein. Such modified systems may be present in differing glucosyltransferase to the microsomal system by use of a bifunctional imidate wereunsuccessful. This may indicate amounts from one preparation to the next. The more efficient utilization of N-hydroxytyrosine and p- thatthe glucosyltransferase has a subcellular localization hydroxyphenylacetonitrile produced in situ from tyrosine may which is different from that of the membrane-bound system. and also be related to differences in the ability of the various Another explanation could bethat the glucosyltransfer~e substrates to penetrate the sorghum membrane system (24). the membrane-bound system are oriented in such a way that Since the microsomal preparation consists mainly of closed no free amino groups are available for cross-linking(32).Yet, vesicles (28), tyrosine andp-hydroxyphenylacetaldoximemay a third explanation could be that theglucosyltransferase was exchange freely through the membranes of these vesicles, inactivated by the imidate treatment. If, in viuo,the glucosylwhile N-hydroxytyrosine and p-hydroxyphenylacetonitrile transferase is not in close proximity to the membrane-bound may exchange only to a very limited extent. If metabolism of enzyme system, the p-hydroxy-(S)-mandelonitrileobtained the latter two compounds takes place in the interior of these may partly dissociate into p-hydroxybenzaldehyde and cyavesicles, these active sites would be inaccessible for exoge- nide before enzymatic glucosylation and a mixture of the nously added N-hydroxytyrosine and p-hydroxyphenylace- enantiomeric cyanohydrins wouldbe obtained. Since the tonitrile. Each vesicle would thus function as a microcom- UDP-glucose glucosyltransferase is stereospecific for the (5')partment for conversion of tyrosine or p-hydroxyphenylace- enantiomer (and only dhurrin is formed) (6), stereochemical taldoxime, or both, to p-hydroxymandelonitrile. The validity considerations do not exclude free diffusion and dissociation of such a model would depend on a high degree of polarity in of the cyanohydrin before glucosylation. However, considering the assembling of the vesicles frombroken bits of membranes the strict stereochemical control of the preceding steps (33), formed during cell rupture. Electron microscopy of thin sec- dissociation and loss of some of the p-hydrox- ande el on it rile tions of such membrane vesicles generally indicates a high through racemization seems extravagant unless these procdegree of polarity (29, 30). On the other hand, some of the esses have a specific biological role such as facilitating transsorghum membrane data argue against a simple membrane port across a membrane required for final storage of the model based on permeability barriers. Thus, the channeling cyanogenic glucoside (34). ratios decrease when the tyrosine concentration is increased Acknowledgment-We thank Professor Peder Olesen Larsen, even though a saturating concentration of tyrosine does not result in an in situ concentration of p-hydroxyphenylacetoni- Chemistry Department., Royal Veterinary and Agricultural Univertrile sufficient to saturate the catalytic site for p-hydroxy- sity, Copenhagen, for a critical review of the manuscript. phenylacetonitrile utilization. This is indicated by the increase REFERENCES in p-hydroxybenzaldehyde production upon addition of exog1. Akazawa, T.,Miljanich, P., and Conn, E. E. (1960) Plant Physiol. enous nitrile (Fig. 2C). Therefore, by increasing the tyrosine (Bethesda) 35,535-538 concentration, a concomitant increase in p-hydroxybenzalde2. McFarlane, I. J., Lees, E. M., and Conn, E. E. (1975) J . Biol. hyde formed from tyrosine, i.e. a higher channeling ratio, Chem. 250,4708-4713 wouldbe expected according to a model based on simple 3. Maller, B. L., and Conn, E. E. (1979) J. BioZ. Chem. 254,8575membrane p e ~ e a b i l i t y . 8583 4. Shimada, M.,and Conn, E. E. (1977) Arch. Biochem. Biophys. Unknown control mechanisms may also influence the cat180, 199-207 alytic properties of the sorghum microsomal system. Thus, 5. Meller, B. L., and Conn, E. E. (1978) in Biological Oxidation of the ultimate rates of p-hydroxybenzaldehyde production obNitrogen (Gorrod, J., ed) pp, 437-442, ~:lsevier/North Hotland served upon sequential addition of either tyrosine and pPubiishing Co., New York hydroxyphenylacetaldoxime (Fig. 2, A and B ) or tyrosine and 6. Reay, P. F.,and Conn, E. E. (1974)J. Biol. Chem. 249,5826-5830 p-hydroxyphenylacetonitrile (Fig. 2, C and D) were different 7. Maller, B. L. (1978) J. Labelled Compd. Radiopharm. 14, 663671 depending on which of the two substrates was added fist. 8. Msller, 8. L., and Conn, E. E. (1977) Proc. West. Pharmacol. The inhibitory effects observed upon simultaneous addition SOC.20, 103-107 of two substrates to themicrosomal system were also complex 9. Mdler, B. L., and Conn, E. E. (1978) Plant Physiol. (Bethesda) (Tables I and 11). While competition for the same active site Gl(suppl.), 85 obviously influences the reactions observed, addition of large 10. McElvain, S. M., and Schroeder, J. P. (1949) J. Am. Chem. SOC. amounts of an intermediate could also modifythe membrane 71,40-46

3056

BiosynthesisGlucosideCyanogenic

11. Meller, B. L. (1977) Anal. Biochem. 81,292-304 12. Moller, B. L., McFarlane, I. J., and Conn, E. E. (1977)Acta Chem. Scand. Ser. B Org. Chem. Biochem. 31,343-344 13. Smith, E. L. (1951) Adu.Enzymol.12, 191-257 26. 14. Ginsburg, A,, and Stadtman, E. R. (1970) Annu. Reu. Biochem. 39,429-472 15. Kirschner, K., and Bisswanger, H. (1976) Annu. Reu. Biochem. 45, 27. 143-166 16. Davis, R. H. (1967) in Organizational Biosynthesis (Vogel, H. J., Lampen, J. O., and Bryson, V., eds) pp. 303-322, Academic Press, New York 17. Gaertner, F. H., Ericson, M. C., and DeMoss, J . A. (1970) J . Biol. Chem. 245, 595-600 18. Jacobson, J. W., Hart, B. A,, Doy, C . H., and Giles, N. H. (1972) Biochim. Biophys. Acta 289, 1-12 19. Giles, N. H., Case, M. E., Partridge, C. W. H., and Ahmed, S. I. (1967) Proc. Natl. Acad. Sci. U. S. A . 58, 1453-1460 20. Case, M. E., and Giles, N. H. (1976) Mol. Gen. Genet. 147,83-89 21. Gaertner, F. H., and Cole, K. W. (1977) Biochem. Biophys. Res. Commun. 75,259-264 22. Frost, R. G., and West, C. A. (1977) Plant Physiol. (Bethesda) 59,22-29

23. Fall, R. R.,and West, C. A. (1971) J. Biol. Chem. 246, 6913-6928 24. Czichi, U., and Kind, H. (1977) Planta (Berl.) 134,133-143 Eur. Biochem. SOC.55,49-61 25. Kind, H. (1979) Fed. Moller, B. L., and Conn, E. E. (1976) 26th Symposium andAnnual Meeting of the Phytochemical Society of North America, Van8 to 11, 1976, Phytochemical couuer, British August Columbia, Society of North America, p. 14 Ettlinger, M. G., and Kjaer, A. (1968) Recent Adu. Phytochem. 1, 59-144 28. Saunders, J. A., Conn, E. E., Lin, C . H., and Shimada, M. (1977) Plant Physiol. (Bethesda) 60,629-634 29. Kreibich, G., and Sabatini, D. D. (1973) Fed. Proc. 32,2133-2138 30. Losa, G. A,, Weibel, E. R., and Bolender, R. P. (1978) J. Cell Biol. 78,289-308 31. Welch, G. R., and Gaertner, F. H. (1976)Arch. Biochem. Biophys. 172,476-489 32. Sun, T.-T., Bollen, A,, Kahan, L., and Traut, R. R. (1974) Biochemistry 13,8302-8308 33. Rosen, M. A., Farnden, K. J. F., Conn, E. E., and Hanson, K. R. (1975) J. Biol. Chem. 250,8302-8308 34. Kojima,M., Poulton, J. A., Thayer, S., and Conn,E. E. (1979) Plant Physiol. (Bethesda) 63, 1022-1028