Planta (Berl. 103, 241-248 (1972) 9 by Springer-Verlag 1972

Isolation and Identification of the Gibberellins of Cucumis sativus and Cucumis melo* DELBERT I). HEltfPVrrLL, JR. **, L . R . BAKER, and H.M. SELL Departments of Biochemistry and Horticulture, Michigan State University, East Lansing, Michigan, U.S.A. Received October 22, 1971 Summary. Thin-layer chromatography, gas-liquid chromatography, and mass spectrometry were used to identify gibberellins isolated from mature seeds of both Cucumis sativus (cucumber) and Cucumis melo (muskmelon). The gibberellins were extracted and purified by organic solvent fractionation, paper and thin-layer chromatography, and crystallization. Seeds of C. sativus were found to contain gibberellins A1, A a, _4.4, and A 7 with A 1 the predominant species. Seeds of C. melo contained gibberellins A1 and A8 and a trace of A 5. Direct probe mass spectrometry of the gibberellins proved successful for identification purposes. Distinctive molecular ions and fragmentation patterns were obtained for each gibberellin.

Introduction

Individual gibberellins have been isolated and identified f r o m m a n y plant species and the occurrence of gibberellins in plants is p r o b a b l y ubiquitous. I n m a n y cases more t h a n one gibberellin occurs in a species suggesting either precursor-product relationships or unique functions for each individual gibberellin. Gibberellins A2, A,, A T, Ap, and A18, each of which lacks a C-13 h y d r o x y l (gibberellane numbering system), are particularly effective in inducing staminate flowers on gynoecious lines of C. sativus (Wittwer and Bukovac, 1962; Clark and Kenney, 1969). However, the p r e d o m i n a n t gibberellin of monoecious seedlings of this species is GA 1 (Hayashi et al., 1971). Gas-liquid and thin-layer chromatog r a p h y have also provided bruited evidence for traces of gibberellins A3, A 4, and A 7 in monoecious C. sativus seedlings (Hayashi et al., 1971; A t s m o n et al., 1968). Gynoecious lines of C. melo, a closely related species, do n o t produce staminate flowers in response to gibberellin application (Peterson, 1963), and the endogenous gibberellins of this species have not been reported. * Journal Article No. 5664 from the Michigan Agricultural Experiment Station. This work was supported in part by a grant from the Herman Frasch Foundation. ** Portions were taken from a thesis submitted in partial fulfillment of the requirements for the Ph.D. degree, Michigan State University, 1971 17 Planta(Berl.),Bd. 103

242

D.D. Hemphill, Jr., L. R. Baker, and H. M. Sell:

The purpose of this s t u d y was to determine the kinds of gibberellins occurring i n m a t u r e seeds of b o t h species a n d to i n v e s t i g a t e the use of mass s p e c t r o m e t r y for identification of n o n - d e r i v a t i z e d gibberellins. Mass s p e c t r o m e t r y is, ideally, a sensitive a n d definitive t e c h n i q u e for identification of m a n y compounds.

Materials and Methods Seed. Gibberellins were isolated from seeds of three C. sativus sex types; viz., the andromonoeeious variety "Lemon" (Dessert Seed Co., E1 Centro, Calif.), the monoecious line MSU 736, and the gynoecious line MSU 713-5. For C. melo, gibberellins were isolated from the gynoeeious line MSU 1 G and the andromonoecious variety "Rocky Ford" (Vaughn Seed Co., Chicago, Ill.). Isolation and Puri]ication o/Gibberellins. One hundred gram samples (2.0 kg total) of seed were homogenized for three minutes at 4~ C in 90 % methanol. The homogenate was stirred for twenty-four hours at 4~ C, then filtered under vacuum. The residue was then re-extracted as above. The combined extracts were evaporated to dryness and 5% sodium bicarbonate was added. The bicarbonate solution was partitioned three times against equal volumes of ethyl acetate. The bicarbonate layer was adjusted to pH 3.0 with dilute sulfuric acid and partitioned three times against equal volumes of ethyl acetate. Combined ethyl acetate layers were washed with distilled water, dried over anhydrous sodium sulfate, and evaporated to dryness. The gummy residue was dissolved in minimal ethanol or ethyl acetate and applied as a thin streak to ethanol-washed Whatman 3 MM paper (50 mg/chromatogram). The chromatograms (descending) were developed with isopropanol-ammonium hydroxide-water (10:1:1, v/v). Active compounds were eluted from strips of the chromatograms with 90% ethanol; volume was reduced; and the eluates were applied to Silica Gel G or Silica Gel F25a pre-coated preparative plates (Brinkmann Instruments, Inc.) for thin-layer chromatography (TLC). The TLC plate was developed with carbon tetrachloride-acetie acid-water (8:3:5, v/v), nonaqueous phase plus 10% ethyl acetate. Gibberellins A1, A2, A3, As and As did not migrate in this system, but much interfering material was removed. The residues at the origin and at l~f 0.5 to 0.7 (gibbereilins Aa, AT, A~) were eluted and rechromatographed with benzene-n-butanol-aceticacid (70:25:5, v/v) as the solvent system. Fractions eluted from this chromatogram were further purified, if necessary, by additional chromatography in these solvent systems followed by crystallization from ethyl acetate-petroleum ether. All chemicals used in the isolation were reagent grade. All solvents were redistilled prior to use and the solvent residues were bioassayed with negative results. Bioassays. Gibberellin-like activity was Iollowed at each step of the isolation and purification procedures by either a modified dwarf pea (Hayashi et al., 1971) or the barley half-seed bioassay (Jones and Varner, 1967). Instrumental Techniques. Samples were prepared for gas-liquid chromatography (GLC) by methylation with diazomethane. Mcthylated samples or mixtures of methyl gibberellin standards were injected into an F and M Model 400 Gas-Liquid Chromatograph equipped with hydrogen flame ionization detector and pyrex U column one meter long and 4.5 mm in diameter. The column packing was 1.5% QF-1 (fluorinated alkyl silicone polymer, D.C.) on 70-80 mesh Chromasorb (Analabs, Inc.). Column temperature was either 200~ or 180 to 225~ programmed at 2~ To obtain mass spectra, the highly purified gibberellin-like compounds

Isolation of Gibberellins

243

(ca. 1 ~g) were inserted into the direct probe inlet of an LKB 9000 single-focusing mass spectrometer with ionizing voltage of 70 eV and initial ion source temperature of 190~ Data output was obtained with an on-line digital computer system (Sweeley et al., 1970). Results

Purified fractions from seeds of different sex types of C. sativus all yielded spots at Rf 0.49 and 0.69 on TLC (benzene:n-butanol:acetic acid, 70:25:5) which corresponded to GA1 and/or GA 3 and GA4 and/or GA 7 standards, respectively. Purified fractions from seeds of different sex types of C. melo all yielded spots which corresponded with GA1 and/or GAa and a spot at Rf 0.65 which corresponded closely to a GA 5 standard, but could be A~ or A 7 as well. Subsequent characterization of the gibberellins utilized the andromonoecious seed extract for C. 8ativus and the gynoecious seed extract for C. melo due to the relatively large amounts of gibberellins in these extracts. Identi/ication o/Andromonoecious C. sativus GibbereUins The purified fractions of the anch'omonoecious C. sativus seed extract were combined, methylated, and yielded peaks with retention times of ca. 6, 7, 15, and 17.5 minutes on GLC (Fig. 1, lower curve). These peaks correspond closely to methylated gibberellin Aa, A 7, A1 , and A s standards, respectively (Fig. 1, upper curve). The compound corresponding with GA 1 is present in the highest concentration. These highly purified fractions were subjected to direct probe mass spectrometry (Fig. 2). The uppermost spectrum has a molecular ion at m/e 348, the molecular weight of GA1. Characteristic fragments useful in identifying the spectrum as that of a gibberellin include m/e 330 (loss of H~O), 312 (loss of 2 • 302 (loss of CH~02), and 284 (loss of H20, CH202). The spectrum corresponds closely to that for authentic GA1 (Pitel et al., 1971) and is consistent with that of the GA 1 methyl ester (Binks et al., 1969). The middle spectrum has a molecular ion at m/e 346, the molecular weight of GA3. Characteristic fragments include m/e 328, 310, 300, 284, 283 (M+-63, characteristic of gibberellins having 3,4 double bond; Binks et al., 1969), 282, 238, and 136. This spectrum corresponds closely to that of authentic GA 3 (Pitel et al., 1971). The bottom spectrum has a molecular ion at m/e 332, the molecular weight of GA4. Characteristic fragments include m/e 330, 314, 286, 270, 268, and 136. Again, this spectrum corresponds closely to that of authentic GA4 (Pitel et al., 1971). The suspected GA 7 fraction was not obtained in sufficient purity for direct probe mass spectrometry, but GLC (Fig. t) provided evidence for its presence in the extract. 17"

244

D.D. IIemphill, Jr., L. R. Baker, and It. JR. Sell:

Ai Detector Response

0

5

10 15 Retention time, min

20

Fig. 1. Gas-liquid chromatogram of a methylated gibberellin extract from andremonoecious C. sativus seed (lower graph) and a mixture of methylated gibberellin standards (upper graph). Column: QF-1,200~ C. Two ~l of a 1% alcoholic solution of the extract and standards were injected

Identification o/ Gynoecious C. melo Gibberellins TLC results indicated the presence of GA1 and/or GAa and possibly GA 5. The fractious suspected to contain A 1 and/or Aa had infrared spectra typical of gibberellins. One fraction had peaks at 1260, 970, and 770 cm -1 characteristic of GAa. The fractions were subjected to direct probe mass spectrometry (Fig. 3). The uppermost and middle spectra have molecular ions at m/e 348 and m/e 346, respectively. The spectra are essentially identical r the uppermost and middle spectra of Fig. 2, which were identified as GA1 and GAa. The bottom spectrum has a molecular ion at 330, the molecular weight of both GA5 and GA7. The peak at m/e 267 (M+-63), characteristic of methyl A T (Binks et al., 1969) and also expected for acidic GA~, is weak. The peak at 286 (M+-44) is very strong, a characteristic of the spectra of gibberellins which lack a C-3 hydroxyl (gibberel-

~"

3

i

0

o

I M

o I

I

() I

r

I

I

L F

M

I

I

m o I c~ I

(0

o

o

o

~-

:..--

) I

I M

I

I

I

i

I

i m

I

p.

D

0 r~

D. D. tIemphill, Jr., L. R. Baker, and H. M. Sell:

246 lO0-

i

S0-

8

60-

*'40-

20-

I '

60

100

140

190

220

260

300

34O

60

100

140

180

220

260

300

340

i

~ l

100-

60-

140-

60

Fig. 3.

The

mass

100 spectra

140

180

220

260

300

purified gibberellin-like compounds isolated from gynoeeions C. melo s e e d

of three

lane numbering system) such as GA 5. The spectrum of GA~ is characterized by strong peaks (Pitel et al., 1971) at 284, 223, and 44 which are weak in this spectrum. Thus, the extract appears to contain GA 5 rather than GAT.

Isolation of Gibberellins

247

Discussion Most mass spectrometry of the gibberellins has involved use of the methyl esters or methyl ester trimethylsflyl ethers. However, in the present study, direct probe mass spectrometry of non-derivatized gibberellins proved satisfactory for obtaining positive identifications of highly purified samples. Each gibbereUin produced distinct molecular ions and fragmentation patterns. However, combined GLC-mass spectrometry of derivatized gibberellins is preferable for impure samples. TLC, GLC, and mass spectrometry demonstrated that gibberellins A1, A3, A4, and probably A 7 occur in mature seed of C. sativus while mature seed of C. melo contains A1, A3, and A 5. Gibberellins A1, A3, A4, and A v also occur in seed of Echinocystis macrocarpa Greene (wild cucumber), another species of Cucurbitaceae (Elson et al., 1964). GA1 predominated in andromonoecious C. sativus, while gynoecious C. melo contained roughly equal amounts of GA 1 and GA3 and traces of GA 5. The finding that GA1 is the predominant gibberellin of C. sativus seed is in agreement with the report that GA1 is also the predominant species in six day old etiolated monoecious seedlings of C. sativus (Hayashi et al., 1971). This is in contrast to Phaseolus coccineus in which different developmental stages are marked by different kinds of gibberellins (Sembdner et al., 1968; Crozier and Audus, 1968). GA~ and GAv are most active in staminate flower induction in C. sativus (Wittwer and Bukovac, 1962), while exogenous GA3 application failed to induce staminate flowers on gynoeeious C. melo (Peterson, 1963). Thus, the occurrence of gibberellins A~ and A 7 in C. sativus, but not C. melo, may have some bearing on the apparent lack of a role for gibberellin in C. melo sex expression. However, C. sativus tissues may convert GA4 and GAv to compounds similar to GA1 (Atsmon et al., 1968). Thus, GAt and GAv may not be the actual species responsible for effects on flower sex expression in C. sativus.

The authors thank Mr. Jack Harten and Dr. Raymond ttammond, Department of Biochemistry, Michigan State University, for their assistance in obtaining and interpreting mass spectra. References Atsmon, D., Lang, A., Light, E.~. : Contents and recovery of gibberellins in monoecious and gynoeeious cucumber plants. Plant Physiol. 43, 806-810 (1968). Binks, R., MacMillan, J., Pryce, R. J. : Plant hormones VIII. Combined gas chromatography-mass spectrometry of the methyl esters of gibberellins A1 to Aea and their trimethylsilyl ethers. Phytochemistry 8, 271-284 (1969). Clark, R.K., Kenney, D.S.: Comparison of staminate flower production on gynoecious strains of cucumbers, Cucumis sativus L., by pure gibberellins (A3, A4, AT, A13) and mixtures. J. Amer. Soc. ttort. Sei. 94, 131-132 (1969). Crozier, A., Audus, L.J. : Distribution of gibberellin-like substances in light- and dark-grown seedlings of Phaseolus multi/lotus. Planta (Berl.) 83, 207-217 (1968).

248 D.D. ttemphill, Jr., L. R. Baker, and H. M. Sell: Isolation of Gibberellins Elson, G.W., Jones, D.F., MacMillan, J., Surer, P. J. : Plant hormones. IV. Identification of the gibberellins of Echinocystis macrocarpa by thin-layer chromatography. Phytoehemistry 3, 93-101 (1964). Hayashi, F., Boerner, D., Peterson, C.E., Sell, H.M.: The relative content of gibberellin in seedlings of gynoecious and monoecious cucumber (Cucumis sativus). Phytochemistry 10, 57-62 (1971). Jones, R.L., Varner, J.E.: The bioassay of gibberellins. Planta (Berl.) 72, 155-161 (1967). Peterson, C.E.: Gynoecious muskmelons for hybrid seed production. Abstract No. 332, 60th Ann. Meeting Am. Soc. Hort. Sci., Amherst, Mass. (1963). Pitel, D.W., Vining, L.C., Arsenault, G.P. : Biosynthesis of gibberellins in Gibberella /ujilcuroi. The sequence after gibberellin Aa. Canad. J. Biochem. 49, 194-200 (1971). Sembdner, G., Wetland, J., Aurich, 0., Sehreiber, K.: Isolation, structure, and metabolism of a gibberellin glucoside. S.C.I. Monograph No 81, 70-86 (1968). Sweeley, C.C., Ray, B., Wood, W.L, Holland, J.F.: On-line digital computer system for high-speed single-focusing mass spectrometry. ~ a l y r Chem. 42, 1505-1516 (1970). Wittwer, S.H., Bukovac, M. J. : Staminate flower formation on gynoeeious cucumbers as influenced by the various gibberellins. Naturwissenschaf~en 49, 305-306 (1962). Dr. D.D. Itemphill Department of Biochemistry Michigan State University East Lansing, Michigan 48823, U.S.A.