Plant Physiol. (1975) 56, 626-629

Comparative Study of the Composition of the Seeds of Ripening Momordica charantia and Tomatoes1

Received for publication March 11, 1975 and in revised form July 17, 1975

DELIA B. RODRIGUEZ, TUNG-CHING LEE, AND CLINTON 0. CHICHESTER Department of Food and Resource Chemistry, University of Rhode Island, Kingston, Rhode Island 02881

ABSTRACT MATERIALS AND METHODS The total carotenoid concentration of the seeds of Mo- Materials. The Momordica charantia seeds were taken from the mordica charantia rose about 100-fold from the immature fruits used in our previous research (17). The fruits represented to the ripe stage. The massive increase was almost exclu- four stages of maturity: immature, mature-green, partly ripe, and sively attributable to , which accounted for 96% ripe. For each maturity stage, the seeds from at least five fruits of the of the ripe seeds. The carotenoid pattern were analyzed separately. of the seed was found to be drastically different from that Ripe cherry tomatoes, Lycopersicon esculentum var. Cerasi- of the pericarp. The seed, which contained fewer carote- forme (Dun.) A. Gray, were purchased from a local fruit market. noids, had a total concentration 12 times greater than that The cherry variety was used because the seeds were easier to sep- in the pericarp at the ripe stage. The acyclic lycopene se- arate. The seeds from 900 g offruits were separated from the peri- lectively accumulated in the seed, whereas the cyclic caro- carp and washed with water on a sieve to remove juice and tenoids, cryptoxanthin, and 13-, were residual pulp. The water was drained and the seeds were air dried the principal pigments of the ripe pericarp. The seed of ripe on the sieve for a couple of hours. Four 30-g samples were taken tomatoes reflected the qualitative pattern of the whole and analyzed separately. fruit. The total carotenoid concentration was, however, In addition to the authentic carotenoids used for comparison much lower and the lycopene content was particularly low. in the preceding study, lycoxanthin and rubixanthin were ex- ,3-Carotene, having a comparatively high concentration, tracted and purified from Solanum dulcamara and rose hips, emerged as the major pigment of the seed. respectively. Extraction and Separation of Carotenoids. The carotenoids of the seeds of both fruits were extracted and saponified in the same manner described for the pericarp (17). The saponified pigments from the ripe and partly ripe Momordica samples were separated initially on a column of alumina (activity grade III), using 1 to 5% (v/v) ethyl ether in petroleum ether as solvent. Fraction 4 Our previous study (17) showed the variations in the carotenoid was rechromatographed on a MgO-HyfloSupercel (1:2) column composition of Momordica charantia fruits at the immature, ma- developed with 25% (v/v) acetone in petroleum ether, resulting ture, partly ripe, and ripe stages. A number of interesting features in the formation of five bands (fractions 4-a to 4-e). If necessary, in the carotenoid pattern were observed. One of the most notable these bands could be purified on a silica gel plate (Quanta Gram changes was the synthesis of cryptoxanthin at the onset of ripen- TLC plates QIF, Quantum Industries, New Jersey) with 3%C (v/v) ing, making this pigment responsible for the orange color of the methanol in benzene as the developing solvent. The pigments ripe pericarp. extracted from the immature seeds separated into four bands on In sharp contrast to the pericarp, the ripe seeds of Momordica the alumina column. These fractions consisted of single caroten- charantia are brilliant red in color. Presently we have determined oids and thus needed no further separation. The topmost fraction the carotenoid distribution in the seeds at different stages of de- of pigments from mature samples consisted of three components velopment not only to account for the striking color contrast be- when rechromatographed on a silica gel plate developed with 3% tween the pericarp and seeds, but also to provide the basis for (v/v) methanol in benzene. As in the previous study, the colorless detailed studies on carotenogenesis and the control mechanisms eluate preceding the fluorescent band was collected involved. and examined for the presence of . Ripe commercial tomatoes have a characteristic red pericarp The tomato seed carotenoids were fractionated as described for and yellowish seeds and thus represent the reverse of the Momor- the whole fruit (15) except that lycopene was subsequently sep- dica charantia color pattern. Although the carotenoids of the arated from the on an alumina column (activity whole tomato fruit have been the subject of numerous investiga- grade III) developed with 1 to 5% (v/v) ether in petroleum ether. tions for many years now, the seed carotenoids per se have not Identffication and Quantitative Determination. The carotenoids been given much attention. We have also analyzed the carotenoid were identified by the absorption spectra, polarity (TLC RF composition of the tomato seeds as a basis of comparison. values), co-chromatography with authentic samples, and by re- sponse to chemical tests such as acetylation, methylation, dehy- dration, epoxide tests, and iodine catalyzed isomerization. The ' This work was supported by United States Public Health Services details of the identification procedure and the quantitative meas- Grant 5R01-FD-00433 to C. 0. C. Contribution No. 1579 of the urement ofthe various carotenoids have been described previously Agricultural Experiment Station of the University of Rhode Island, (17). The extinction coefficient of,-carotene was used in the calcu- Kingston, R. I. lation of the concentration of tomato seeds. An 626 Plant Physiol. Vol. 56, 1975 SEED CAROTENOIDS 627 additional chemical test was used for the identification of fraction as readily as isocryptoxanthin. Oxidation with nickel peroxide 4-e (lycoxanthin). Allylic oxidation with nickel peroxide was produced a monoaldehyde without altering the absorption accomplished according to the method of Liaaen-Jensen and maxima. Hertzberg (10). None of the carotenoids isolated from the seeds of Momordica charantia gave a positive reaction to the epoxide tests. RESULTS AND DISCUSSION The immature seeds contained only phytofluene, (3-carotene, , and traces of lycopene. At the mature stage, traces of Carotenoid Composition of Momordica charantia Seeds. The cryptoxanthin and rubixanthin were detected in addition to seeds of the ripe and partly ripe fruits contained nine carotenoids phytofluene, (3-carotene, lutein, and lycopene. found mainly in the seed coat (Table I). Fractions 1, 2-a, 2-b, 3, The results of the quantitative analysis at four stages of ma- 4-a, 4-b, 4-c and 4-d were identified, as described in our previous turity are presented in Table II. The seeds showed greater varia- paper (17), as phytofluene, ,B-carotene, D-carotene, lycopene, bility in the levels of the individual pigments than the pericarp. zeinoxanthin, cryptoxanthin, lutein, and rubixanthin, respec- The stage of maturity was determined by the visual inspection of tively. In addition to the characteristics previously discussed, the fruit, not the enclosed seeds. Lycopene, in particular, in- fraction 4-d was also shown to co-chromatograph with rubixan- creased so rapidly and substantially in the seeds during ripening thin (3-hydroxy-y-carotene) isolated from rose hips on the silica that a slight delay in harvesting could introduce large differences gel sheets, developed with either 3% methanol in benzene or 20% in its concentration without being manifested in the fruit color or ethyl acetate in methylene chloride. Iodine catalyzed isomeriza- fruit cryptoxanthin which did not accumulate to such high levels. tion shifted the maxima to shorter wavelengths, confirming that It could still be possible that the seed carotenoids were innately the pigment was not gazaniaxanthin, the cis isomer of rubixan- more variable in their concentration. thin. Rubixanthin was identified by Goodwin (3) as a pigment Massive synthesis of carotenoids accompanied the ripening characteristic of rose hips and its structure was ascertained by process in the seeds of Momordica charantia. This is a definite Brown and Weedon (2) and Arpin and Liaaen-Jensen (1). departure from other pericarp-enclosed seeds which are generally Fraction 4-e was identified as lycoxanthin (16-hydroxylyco- devoid of carotenoids or contain relatively small amounts of pene). The spectral absorbance resembled that of lycopene but carotenoids. The total carotenoid concentration in the Momordica the RF values were much lower and were indicative of a hydroxy seeds rose about 100-fold from the immature to the ripe stage. carotenoid. The pigment ran as a single spot with lycoxanthin Carotenogenesis was most active at the transition from the ma- purified from Solanum dulcamara on the silica gel sheets developed ture to the partly ripe stage. Although the number of carotenoids with either of the two solvent systems. The lycoxanthin from increased from four in the immature seeds to six at the mature Solanum dulcamara has been characterized and its structure con- and nine at the partly ripe and ripe stages, the enormous increase clusively determined by Markham and Liaaen-Jensen (11). Frac- in carotenoid concentration was almost entirely attributable to tion 4-e also exhibited the chemical reactions reported for lyco- lycopene which comprised 96% of the ripe seed carotenoids. xanthin (9, 11). Acetylation produced a monoacetate derivative. Lycopene increased from near zero at the immature stage to an The pigment could be dehydrated with acidic chloroform but not average of 261 Aglg in the ripe seeds. Table I. Carotenoids of Ripe and Partly Ripe Momordica charantia Seeds

RF Values on Silica Gel Absorption Spectra in Fraction Identification Petroleum Ether Chemical Reactions Comparison with Authentic Carotenoids 3% Methyl E 20% AlooAlcoholinnC6H6 Ethylin CH2CI2Acetate nm I Phytofluene 330, 345, 365 Carrot and tomato phytofluene 2-a 1B-Carotene (425), 448, 476 0.97 Hoffman-La Roche 13-carotene 2-b c-Carotene 377, 399, 424 0.97 Carrot and tomato i-carotene 3 Lycopene 443, 469, 501 0.84 Tomato lycopene 4-a Zeinoxanthin 421, 443, 472 0.56 0.84 + acetylation -methylation -dehydration 4-b Cryptoxanthin (425), 448, 476 0.44 0.82 + acetylation Lemon cryptoxanthin -methylation -dehydration 4-c Lutein 419, 442, 471 0.21 0.64 + acetylation Hoffman-La Roche lutein + methylation -dehydration 4-d Rubixanthin 435, 457, 486 0.4 0.79 + acetylation Rose hips rubixanthin -methylation -dehydration

4-e Lycoxanthin 442, 468, 499 0.3 0.45 + acetylation Solanum dulcamara lycoxanthin + hydration + nickel peroxide oxidation 628 RODRIGUEZ, LEE, AND CHICHESTER Plant Physiol. Vol. 56, 1975

Small increases were seen in lycoxanthin, rubixanthin, and level in the seed declined during ripening in contrast to the upward cryptoxanthin. Neither zeinoxanthin nor c-carotene were de- fl-carotene trend in the ripening pericarp. The disappearance of tected at the immature and mature stages but both were found in lutein in the pericarp as ripening progressed was not seen in the the partly ripe and ripe seeds. Phytofluene was present at all seed where lutein maintained an almost constant low level stages in trace amounts, while lutein was detected at a fairly throughout development. Rubixanthin, which was detected only constant low level throughout the development period. 3-Caro- in trace amounts in the ripe and partly ripe pericarp, was found tene content showed a definite downward trend. in measurable concentrations in the ripe and partly ripe seeds. A comparison of the results obtained in this and our previous Lycopene started to accumulate in the seed at the mature-green study (17) reveals dramatic differences both in the qualitative stage, before the synthesis of cryptoxanthin in the pericarp, which and quantitative carotenoid composition of the seeds and pericarp took place at the onset of ripening. Consequently, the appearance of Momordica charantia (Table II). The ripe pericarp has a more ofthe red color in the seed preceded the development ofthe orange complex carotenoid composition than the seeds. Fifteen carote- color in the pericarp. The seed not only reached a much higher noids were found in the pericarp as compared to only nine in the maximum concentration, but also started to accelerate carotenoid seeds. Phytofluene, fl-carotene, c-carotene, zeinoxanthin, crypto- production earlier in the development process. This came as a xanthin, lutein, lycopene, and rubixanthin were detected in both surprise since the seeds are enclosed by the pericarp and are thus the pericarp and the seeds. a-Carotene, flavochrome, 5,6-mono- shielded from sunlight which is known to stimulate carotenoid epoxy-fl-carotene, mutatochrome, 6-carotene, y-carotene, and synthesis. It is commonly observed, for example, that the peel of zeaxanthin were found exclusively in the pericarp whereas lyco- fruits is more highly pigmented than the pulp. xanthin was isolated only from the seeds. Carotenoid Composition of Tomato Seeds. In the tomato, the In spite of the wider variety of the pigments in the pericarp, difference in pigment composition between seeds and pericarp the total carotenoid concentration of the seed surpassed that of was not as striking as in the case of Momordica charantia. The the pericarp by a factor of about 12 at the ripe stage. The ca- contrast was consistent with the observation that the pericarp rotenoid accumulation at the later stages was therefore much more and the seeds are morphologically more distinct in the Momordica rapid in the seed than in the pericarp. At the immature stage, the than in the tomato. Qualitatively, the tomato seeds contained es- pericarp had twice as much total carotenoid content than the sentially the same carotenoids reported for the tomato fruit. seed. However, the carotenoid concentration was much lower; the Whereas the pericarp derived its orange color from cryptoxan- lycopene content was extremely low (Table IL). Commercial to- thin, the intense red color of the seed was attributable to the mato varieties contain 70 to 120 MAg of lycopene/g of fresh fruit massive amounts of lycopene deposited in the seed coat. The con- (7, 14). The 8.2 Ag/g concentration obtained for the seeds fell far centration of the other carotenoids was far too small to contribute below this range. The fl-carotene content (12.9 ug/g seed), on to the seed color. Lycopene was also found in the pericarp but at the other hand, was comparatively high. In Rutgers tomato, barely detectable amounts. Surprisingly, cryptoxanthin appeared where the lycopene concentration was found to be 69 ,ug/g fresh first in the seed (mature-green stage) where it never reached fruit, fl-carotene was only 4 ,Ag/g (7). Summer sunrise tomato, substantial levels. which contains 526 MAg of lycopene/g of dry wt of fruit, has only Other less notable differences were observed. The fl-carotene 50 ug of f-carotene/g of dry wt of fruit (15). The light color

Table II. Comparisoni of Carotenzoid Compositionz of Momordica charanitia Seeds and Pericarp and Tomato Seeds

Momordica charantia Seeds Caro*enoid I______FullyPericarpRipe RipeSeedsTomato Immature green Mature green Partly ripe Fully ripe

| lAg/gl % glgg' % Ag/lg % ig/gl %Agg/g2g/g3lg Phytofluene trace trace trace trace trace 0.4 1 .7 ce-Carotene 0.4 ± 0.1 0.6 2.5 3-Carotene 1.1 +t 0.4 39.3 1.2 I 0.4 8.3 trace trace 5.3 4- 2.3 12.9 54.4 13-Zeacarotene trace 0.3 1.3 c-Carotene trace trace trace y-Carotene trace 0.8 3.4 Lycopene trace 11.8 It 13.8 81.4 244.9 I 54.3 97.1 261.0 + 60.2 96.3 0.5 +t 0.1 8.2 34.6 Zeinoxanthin 0.5 - 0.4 0.2 0.4 ± 0.2 0.1 0.6 -- 0.2 Lutein 1.7 + 1 60.7 1.5 + 0.6 10.3 1.1 ± 0.7 0.4 1.1 4 1 0.4 trace Cryptoxanthin trace 0.8 ± 0.4 0.3 1.9 + 1.5 0.7 13.7 ± 5.4 Zeaxanthin [ 1.5 ± 0.1 Rubixanthin trace 2.3 + 0.3 0.9 2.9 -- 0.7 1.1 trace Lycoxanthin 2.7 ± 1.3 1.1 3.6 -- 1.5 1.3 Total xanthophylls 1.7 1.5 7.4 9.9 15.8 0.54 2.1 Total 2.8 14.5 252.3 I 270.9 22 23.7

lEach value is the mean of five samples. The average fresh weight of the seeds was 6.1 g at the immature stage, 12 g at the mature green stage, 9.8 g at the partly ripe stage, and 11.3 g at the fully ripe stage. 2 Taken from Rodriguez et al. (17) for comparison. The pericarp also contained traces of flavochrome, mutatochrome, 5,6-mono- epoxy-,B-carotene and 6-carotene. 3 Each value is the mean of four samples. 4 Identity of the xanthophylls was not determined. Plant Physiol. Vol. 56, 1975 SEED CAROTENOIDS 629

(yellow) of the tomato seeds as compared to the red pericarp can 2. BROWN, B. 0. AND B. C. L. WEEDON. 1968. Rubixanthin and gazaniaxanthin then be attributed to two factors: (a) low total carotenoid con- Chem. Commun. No. 7: 382-384. 3. GoODwiN, T. W. 1956. Studies in carotenogenesis. 19. A survey of the polyenes in centration, and (b) very low lycopene content. a number of ripe berries. Biochem. J. 62: 346-352. It is obvious from our results that the carotenoid pattern of the 4. GoODWxI, T. W. AND M. JAMIXORN. 1952. Biosynthesis of in ripening seeds and pericarp of Momordica charantia and tomato are en- tomatoes. Nature 170: 104-105. different. Whereas the Momordica seeds surpass the peri- 5. GoODwIN, T. W. AND S. PHAGPOLNGARM. 1960. Studies in carotenogenesis. 28. tirely The effect of illumination on carotenoid synthesis in French bean (Phaseolus carp in total carotenoid concentration and contain abundant vulgarni) seedlings. Biochem. J. 76: 197-199. amount of the acyclic lycopene, the tomato seeds have a much 6. HORVATH, G., J. KIssimoN, AND A. FALUDI-DANIEL. 1972. Effect of light inten- lower total carotenoid and lycopene concentration than the sity on the formation of carotenoids in normal and mutant maize leaves. Phyto- chemistry 11: 183-187. pericarp. 7. KARGL, T. E., F. W. QUACKENBUSH, AND M. L. TOMES. 1960. The carotene-poly- The presence of the early precursor phytofluene in both peri- ene system in a strain of tomatoes high in delta-carotene and its comparison carp and seeds of Momordica, in addition to the diversity of the with eight other tomato strains. Proc. Am. Soc. Hort. Sci. 75: 574-578. corresponding carotenoids, suggests that independent biosyn- 8. KAY, R. E. AND B. PHINNEY. 1956. Plastid pigment changes in the early seedling thetic may operate in different structures within the leaves of Lea moys L. Plant Physiol. 31: 226-231. pathways 9. LIAAEN-JENSEN, S. 1971. Isolation, reaction. In: 0. Isler, ed., Carotenoids. Birk- same organ. No conclusion can be drawn, however, until the hauser Verlag, Basel. pp. 61-188. problem is studied at the enzymic level. 10. LLkAEN-JENSEN, S. AND S. HERTZBERG. 1966. Selective preparation of the lutein Although there were some disparities in the responses reported, monomethyl ethers. Acta Chem. Scand. 20: 1703-1709. and showed some indica- 11. MARKHAM, M. C. AND S. LIAAEN-JENSEN. 1968. Carotenoids of higher plants. I. studies on leaves, seedlings, cotyledons The structure of lycoxanthin and lycophyll. Phytochemistry 7: 839-844. tions of sunlight-stimulated synthesis of cyclic carotenoids (5, 6, 12. MEREDITH, F. I. AND R. H. YOUNG. 1969. Effect of temperature on pigment de- 8, 16, 19). Conversely, red blush grapefruits placed under reduced velopment in red blush grapefruit and ruby blood oranges. Proc. Int. Citrus light ("shade") consistently had slightly higher lycopene concen- Symp. 1: 271-276. to has 13. MEREDITH, F. I. AND R. H. YOUNG. 1971. Changes in lycopene and carotene con- trations than those exposed sunlight (12, 13). Temperature tent of 'redblush' grapefruit exposed to high temperature. Hortscience 6: 233- also been found to affect fruit carotenogenesis. In tomatoes 234. ripened above 30 C, lycopene synthesis was drastically inhibited 14. PORTER, J. W. AND R. E. LINCOLN. 1950. I. Lycopersicon selections containing a whereas ,8-carotene synthesis was unaffected (4, 18). In red blush high content of carotenes and colorless polyenes. II. The mechanism of carotene for was biosynthesis. Arch. Biochem. 27: 390-403. grapefruit, the optimum temperature lycopene production 15. RAYMUNDO, L. C., A. E. GRIFFITHs, AND K. L. SnapsoN. 1967. Effect of dimethyl about 32 to 35 C (12, 13). It is apparent, however, that the effect sulfoxide (DMSO) on the biosynthesis of carotenoids in detached tomatoes. of sunlight and temperature cannot account for the large dispar- Phytochemistry 6: 1527-1532. ity of lycopene and cryptoxanthin distribution between the seeds 16. REBEIZ, C. A. 1968. Dark and light carotenoid accumulation in etiolated and and is not with greening cucumber cotyledons. Magon Inst. Rech. Agron. Publ. (Serie Sci.) and pericarp of Momordica, certainly compatible 23: 1-10. the data on the tomato seeds in which the reverse situation is 17. RODRIGUEZ, D., L. C. RAYMUNDO, T.-C. LEE, K. L. SrMPsoN, AND C. 0. CHI- found, with the acylic lycopene being concentrated in the pericarp. CHESTER. 1975. Carotenoid pigment changes in ripening Momordica charantia fruits. Ann. Bot. In press. LITERATURE CITED 18. TOMES, M. L. 1963. Temperature inhibition of carotene synthesis in tomato. Bot. Gaz. 124: 180-185. 1. ARPIN, N. AND S. LIAAEN-JENSEN. 1969. Carotenoids of higher plants. II. Rubi- 19. WOLF, F. T. 1963. Effects of light and darkness on biosynthesis of carotenoid pig- xanthin and gazaniaxanthin. Phytochemistry 8: 185-193. ments in wheat seedlings. Plant Physiol. 38: 649-652.