Alkaloid Variability in aestivum from Wild Populations Liliya Georgievaa, Strahil Berkovb, Violeta Kondakovaa, Jaume Bastidab, Francesc Viladomatb, Atanas Atanassova, and Carles Codinab,*

a AgroBioInstitute, bul. Dragan Tsankov 8, 1164 Sofia, b Departament de Productes Naturals, Biologia Vegetal i Edafologia, Facultat de Farma`cia, Universitat de Barcelona, Av. Joan XXIII s/n, 08028 Barcelona, Catalonia, . Fax: +34934029043. E-mail: [email protected] * Author for correspondence and reprint requests Z. Naturforsch. 62c, 627Ð635 (2007); received March 2/April 3, 2007 Leucojum aestivum (summer snowflake) is a species used for the extraction of galan- thamine, an acetylcholinesterase inhibitor for the treatment of Alzheimer’s disease. Extracts from collected from 18 Bulgarian populations and from shoot-clumps obtained in vitro from 8 different populations showed variations in their alkaloid composition. Nineteen alka- loids were detected in the studied samples by GC-MS. Typically, the alkaloid fractions of L. aestivum bulbs were dominated by galanthamine type compounds, but lycorine, haeman- thamine and homolycorine type alkaloids were also found as dominant compounds in some of the samples. Extracts from the shoot-clumps obtained in vitro were found to contain galanthamine or lycorine as main alkaloids. The galanthamine content ranged from 28 to 2104 μg/g dry weight in the bulbs, and from traces to 454 μg/g dry weight in the shoot-clumps. Key words: Leucojum aestivum, in vitro Cultures, GC-MS

Introduction tive compounds, but it could also influence the Leucojum aestivum L. (), known technological process for the extraction of galan- as summer snowflake, is a threatened plant species thamine. that is currently used as a commercial source of At the moment, the studies on the alkaloid vari- galanthamine in Bulgaria. It is gathered from the ability of leaves and bulbs of L. aestivum from dif- natural habitats for industrial purposes; this causes ferent populations have been carried out only by increasing problems with depletion of the wild TLC (Stefanov, 1990). More recently, the galan- populations. The in vitro cultures of galanthamine- thamine content of shoots obtained in vitro of this producing have also attracted attention as plant has been determined by HPLC (Diop et al., an alternative source for the production of this 2006). Capillary GC-MS has been proved to be a compound (Codina, 2002; Diop et al., 2006). Gal- useful and reliable method for the rapid separa- anthamine is a reversible acetylcholinesterase in- tion and identification of compounds in complex hibitor which is marketed as hydrobromide for the mixtures of Amaryllidaceae alkaloids (Kreh et al., treatment of Alzheimer’s disease (Maelicke et al., 1995; Berkov et al., 2005). In contrast to HPLC 2001). and TLC techniques, capillary GC-MS allows the The biochemical variability in plants of the fam- simultaneous separation and identification of ily Amaryllidaceae has not been widely studied. In around 20Ð25 alkaloids (Kreh et al., 1995). A sen- a recent investigation on the alkaloids of Galan- sitive MS detector can be very useful for the un- thus elwesii plants by GC-MS we found a high ambiguous identification of compounds in trace level of intraspecific variability in their alkaloid amounts or for searching galanthamine and re- patterns (Berkov et al., 2004). With regard to L. lated alkaloid metabolites in samples obtained aestivum, galanthamine, lycorine and homolyco- from a few milligrams of plant material (Berkov rine chemotypes were found in different popula- et al., 2007). tions in Bulgaria (Stefanov, 1990). Reasons for this As a part of our ongoing studies on the alkaloid phenomenon in Amaryllidaceae species are un- diversity in amaryllidaceous plants, we report here known. The biochemical variability in summer for the first time variations of the major alkaloids snowflake could serve as a source of novel bioac- of in vitro shoot-clumps obtained from plants of 8

0939Ð5075/2007/0900Ð0627 $ 06.00 ” 2007 Verlag der Zeitschrift für Naturforschung, Tübingen · http://www.znaturforsch.com · D 628 L. Georgieva et al. · Alkaloid Variation in L. aestivum wild populations of Leucojum aestivum. Addition- maintained at 23Ð25 ∞C with a 16 h light (at ca. ally, the alkaloid pattern of 35 individual bulbs col- 2.0 klx fluorescent cool light intensity) and 8 h lected from 18 Bulgarian populations was ana- dark photoperiod. lyzed by GC-MS providing valuable information for the selection of galanthamine-rich genotypes. Alkaloid extraction Material and Methods Both the bulbs and the in vitro shoot-clumps Plant material were cut and dried at 60 ∞C until constant weight. Plants of Leucojum aestivum L. (Amaryllida- Dried samples (200Ð400 mg) were powdered in a ceae) were collected at the flowering stage (MayÐ mortar and extracted with 5 ml of methanol for June) during two years, 2004 and 2005, from 2 lo- 12 h. After filtration, the plant residues were cations of the north-east part of Bulgaria (district rinsed with methanol (2 ¥ 5 ml) and the combined of Shumen) and 16 locations of the south of the methanol extract was evaporated under vacuum. country (Table I). They were placed in soil (pots) The dry extract was redissolved in 3% H2SO4 until drying of the aerial parts (in summer), and (4 ml) and defatted with diethyl ether (3 ¥ 5 ml). then the bulbs were cleaned and kept in a dark After basification to pH 9Ð10 with 25% ammonia, and cool place. The plants collected in 2004 were the alkaloids were extracted with chloroform planted in the winter of 2005 in a greenhouse, and (3 ¥ 5 ml). The organic solvent was dried with an- their bulbs collected in summer and stored until hydrous Na2SO4 and then evaporated. The dried alkaloid extraction (October 2005). Vouchers were alkaloid fractions were redissolved in 250 μlof deposited in the Herbarium of Institute of Botany methanol containing 200 μg/ml of codeine as inter- (Bulgarian Academy of Sciences, Sofia). nal standard.

In vitro cultures GC-MS analysis The in vitro cultures were initiated from bulbs collected in 2004 (Table I). The bulbs were rinsed The GC-MS analyses were recorded on a Hew- under a stream of water and cut in “twin-scales”. lett Packard 6890+ MSD 5975 instrument (Hew- The explants were treated with 70% ethanol for lett Packard, Palo Alto, CA, USA) operating in 30 s and sterilized with 0.1% HgCl2 for 3 min. the EI mode at 70 eV. A HP-5 MS column (30 m ¥ Then the explants were rinsed with distilled water 0.25 mm ¥ 0.25 μm) was used. The temperature (three times), placed in Petri dishes on solid B5 program was: 100Ð180 ∞Cat15∞C minÐ1, 180Ð medium (Gamborg et al., 1968) supplemented with 300 ∞Cat5∞C minÐ1, and 10 min hold at 300 ∞C. vitamins, and kept in the darkness. The culture Injector temperature was 250 ∞C. The flow rate of medium used for the initiation and maintaining of carrier gas (helium) was 0.8 ml minÐ1. Split ratio the cultures contained 30 g/l of sucrose, and it was 1:20. 1 μl of the methanol solutions were in- was solidified with 6 g/l of agar and adjusted to jected. pH 5.7Ð7.8. Casein hydrolysate (0.5 g/l), 2,4-di- Galanthamine, sanguinine, narwedine, N-for- chlorophenoxyacetic acid (2,4-D, 1 mg/l), adenine mylnorgalanthamine, epinorgalanthamine, epigal- (2 mg/l), and glutathione (10 mg/l) were added to anthamine, homolycorine, 8-O-demethylhomoly- the starting culture medium. corine, 11-hydroxyvitattine, lycorine, haemanth- The in vitro cultures were subcultured during amine and hordenine were identified by compar- four months on full Murashige and Skoog (1961) ing their GC-MS data and RI (Kovach retention medium supplemented with 1 g/l Ca(NO3)2, indexes) values with those of authentic com- 0.5 mg/l benzylaminopurine (BAP), 0.01 mg/l in- pounds isolated in our laboratory. Methyltyram- dolebutyric acid (IBA), and 2.93 mg/l paclobutra- ine, ungiminorine and lycoramine isomers were zol. The subcultures were done at 2.5 month inter- identified by comparing their mass spectral frag- vals on the MS medium described above, but mentation pattern with standard reference spectra without paclobutrazol. During the subcultures, the from NIST 05 database and literature data (Kreh formed shoot-clumps were cut to increase the et al., 1995; Suau et al., 1988). Codeine was used as number of explants, and the newly formed shoot- an internal standard (IS). The mass spectra were clumps were separated. The in vitro cultures were deconvoluted by AMDIS 2.64 software (NIST). L. Georgieva et al. · Alkaloid Variation in L. aestivum 629

Table I. Plant material of L. aes- Voucher Population Sample Year of Location District tivum used for alkaloid analysis. (SOM-Co) code code collection

Wild plant (bulbs) 1137 Svilengrad Haskovo I I 2004 J 1138 Biser Haskovo J1 2004 J2 K 1139 Ljubimets Haskovo K1 2005 K2 L 1143 Gradina Burgas L1 2005 L2 M 1146 Goritsa Burgas M1 2005 M2 N 1141 Vinitsa Plovdiv M1 2005 N2 O 1133 Osmar Shumen O1 2005 O2 P 1140 Kochovo Shumen P1 2005 P2 Q 1146 Kosharitsa Burgas Q1 2004 Q2 R 1134 Arkutino Burgas R1 2004 R2 S 1135 Veljiov vir Burgas S1 2004 S2 T 1145 Sozopol Burgas T1 2004 T2 U 1136 Lozenets Burgas U1 2004 U2 V 1142 Chernozem Plovdiv V1 2005 V2 W 1144 Prisad Burgas W1 2004 W2 X 1150 Jambol Jambol X1 2005 X2 Y 1149 Blatets Sliven Y1 2005 Y2 Z 1148 Palauzovo Burgas Z1 2005 Z2 In vitro culture (explant) Svilengrad Haskovo I IIV 2004 Biser Haskovo J JIV 2004 Lubimets Haskovo K KIV 2004 Arkutino Burgas P PIV 2004 Arkutinoa Burgas P PIVS 2004 Sozopol Burgas T TIV 2004 Lozenets Burgas U UIV 2004 WIV Prisad Burgas W1 2004 a WIV In vitro culture initiated from 2 seeds.

RI of the compounds were recorded with a For comparison of the samples, the relative al- standard n-hydrocarbon calibration mixture (C9Ð kaloid content in the alkaloid fractions was calcu- C36) using AMDIS 2.64 software (NIST). lated using the peak area of the single compound 630 L. Georgieva et al. · Alkaloid Variation in L. aestivum related to the sum of the peak areas of all com- tations but different retention times, were identi- pounds (100% method). fied by comparison of their mass spectra with those reported in the literature (Kreh et al., 1995) Galanthamine quantification and with mass spectra of an authentic compound The calibration graph was obtained injecting so- from NIST 05 database. In addition, lycoramine lutions of 5, 10, 20, 50, 75, 100, 200, 400, 600 and has 19 RI units more than galanthamine (3) 1000 μg/ml of galanthamine containing 200 μg/ml whereas the lycoramine isomer has 30 units more of codeine by triplicate. The ratios of the peak than lycoramine. The same difference in RI units areas of selected ions (TIC mode) of galanthamine was reported for these compounds by Kreh et al. (m/z at 286) versus those of the internal standard (1995). The difference in the RI values of these codeine (m/z at 299) were plotted against the cor- compounds reported in the literature and that responding concentration of galanthamine to ob- found by us is due to the different column (HP-5 tain the calibration graph. MS) used in our study. The assignment of the hy- droxy group at C-3 in the position of 11- Results and Discussion hydroxyvitattine (14) was tentative by analogy to that of haemanthamine (13) due to the equal RI Alkaloid identification and mass spectral fragmentation of this compound Nineteen compounds in the alkaloid fractions of and its 3α-epimer named hamayne (unpublished both L. aestivum bulbs and shoot-clumps obtained data). To our knowledge, compounds 1, 2 and 4 in vitro showed mass spectral fragmentations typi- had not been previously reported in Leucojum cal for Amaryllidaceae alkaloids and tyramine aestivum plants. type protoalkaloids (Table II). The compounds identified in these plant materials belong to the galanthamine (3Ð10), lycorine (15, 19), haeman- Alkaloid pattern in bulbs of wild plants thamine (13, 14) and homolycorine (11, 12, 16Ð 18) type alkaloids, as well as to the tyramine type Owing to both the ontogenic and organ specific protoalkaloids (1, 2). Sixteen of them were identi- variations of alkaloids in plants (Stefanov, 1990; fied by comparison of their mass spectra and RI Elgorashi et al., 2002), in this work we used only with those of authentic compounds as described in dormant bulbs, thus the plant material being in the Material and Methods. Tyramine type compounds similar ontogenic stage. The number of alkaloids are no typical Amaryllidaceae alkaloids, but they detected in the samples of the different bulbs (and have also been found in other plant families. Com- populations, respectively) varied greatly, from two pound 1 has M+ at m/z 151 which is 14 mass units to twelve compounds (Table II). The alkaloid pat- less than hordenine (2). It was identified as N- terns of the L. aestivum bulbs here studied were methyltyramine by comparison of its mass spec- dominated by compounds coming from a para-or- trum with that of a standard compound from NIST tho’ (galanthamine type) and ortho-para’ (lycorine 05 database. Compound 7 showed a mass frag- type) oxidative coupling of O-methylnorbelladine mentation characteristic for galanthamine type al- (Fig. 1). Typically, the bulbs were found to contain kaloids. Ungiminorine (19), an alkaloid previously mainly galanthamine (3), epinorgalanthamine (6), isolated from L. aestivum (Kobayashi et al., 1985), narwedine (10) and lycorine (15). Ungiminorine was identified by comparison of its mass spectrum (19) was found to occur in many populations as with that reported in the literature (Suau et al., well. Significant differences in the ratios of the in- 1988). Compounds 11 and 12 have no molecular dividual alkaloids in the alkaloid fraction of each ions but they have base ions at m/z 109 which is population were observed. Galanthamine was the characteristic for homolycorine type alkaloids with dominant compound in most of the bulbs, and with a double bound at Δ3,4 and no substitution at C-2 the exception of the samples from the Chernozem (Kreh et al., 1995). So, compounds 7, 11, and 12 population, galanthamine generally represented were not possible to be identified solely by GC- more than 50% of the total alkaloids, reaching up MS without isolation and structure elucidation by to 98% in the samples of the Prisad population. other spectroscopic methods. The percentage of lycorine also varied in a wide Lycoramine (5, RI 2423) and its isomer (8,RI range, from traces up to 50% of the total alkaloid 2453), showing very similar mass spectral fragmen- content in sample Q1 (Kosharitsa location). L. Georgieva et al. · Alkaloid Variation in L. aestivum 631 632 L. Georgieva et al. · Alkaloid Variation in L. aestivum L. Georgieva et al. · Alkaloid Variation in L. aestivum 633

In comparison to the rest of the samples, those ers accumulated lycorine as main compound. Also, of the Chernozem population differed significantly the two in vitro strains initiated from bulbs of the regarding both the alkaloid skeleton types and al- Prisad location showed a remarkable difference in kaloid profile (Table II). Thus, compounds coming their alkaloid pattern, one of them producing from a para-para’ oxidative coupling of O-methyl- mainly galanthamine (approx. 100% of the total norbelladine (haemanthamine type alkaloids) alkaloids, sample WIV2) and the other producing were found to occur together with alkaloids com- mainly lycorine (59% of the total alkaloid fraction, ing from ortho-para’ and para-ortho’ oxidative sample WIV1). Some shoot-clumps showed an al- couplings. In addition to lycorine type alkaloids, kaloid pattern similar to that of the wild plants, the products of the ortho-para’ oxidative coupling like those obtained from the Svilengrad and Vel- were also accumulated as homolycorine deriva- jiov vir locations, whereas the plantlets obtained tives in those plants. The main alkaloid was hae- from the Biser population did not show such a cor- manthamine (13), whereas lycorine type alkaloids relation. The content of galanthamine was found (lycorine) were detected only as traces. Instead of to range from traces to 454 μg/g DW in the shoot- galanthamine, plants of this population accumu- clumps obtained in vitro. The accumulation of this lated a higher amount of its 4,4a-dihydroderiva- compound in both the intact plants and the shoot- tive, lycoramine (5), in the bulbs. The ratio be- clumps obtained in vitro is in accordance with pre- tween galanthamine, haemanthamine and homo- viously reported data for summer snowflake (Ste- lycorine type alkaloids was almost equivalent. In fanov, 1990; Diop et al., 2006). contrast to the rest of the samples, the percentage The results obtained in the present work are in and amount of galanthamine found in the alkaloid agreement with those reported recently by Diop fractions of the bulbs from the Chernozem popu- et al. (2006), who studied the variation in galan- lation were significantly lower. The galanthamine thamine accumulation in explants of L. aestivum content of the bulbs varied in a wide range, from grown on culture media with different combina- 28 to 2104 μg/g dry weight (DW). tions of growth regulators. These authors found a In a previous study, we found a population of L. remarkable variation in the content of this alka- aestivum in the west-north region of Bulgaria loid among the explants they obtained, but they which contained over 50% of epinorgalanthamine did not determine the levels of other alkaloids in the alkaloid fraction of the bulbs at the flower- from this plant species. Our results also showed a ing stage (Berkov et al., 2005). Thus, up to the mo- variation in the galanthamine content in the in ment, galanthamine, lycorine, haemanthamine and vitro shoot-clumps initiated from different popula- epinorgalanthamine have been found to occur as tions of L. aestivum but grown on the same culture the main alkaloids in L. aestivum bulbs. The occur- medium (only one combination of growth regula- rence of different alkaloid biosynthetic pathways tors). Furthermore, the results of the GC-MS analy- in summer snowflake explains the chemotypes sis showed, as mentioned above, not only a varia- found in this plant species. Such a biochemical de- tion in the galanthamine content but also in the viation in the alkaloid profiles has also been re- alkaloid profile among the in vitro cultures initi- ported for another amaryllidaceous plant, Galan- ated from different genotypes, and even among thus elwesii (Berkov et al., 2004), as well as for different strains. Despite galanthamine, the alka- plants of other families, like Fabaceae (Genista lo- loid pattern is also an important feature which can belia; Kirch et al., 1995) and Asteraceae (Senecio be used to optimize the composition of the culture jacobaea; Macel et al., 2004). medium aiming to the manipulation of the biosyn- thetic pathway of the plants and also for the selec- tion of in vitro-obtained explants with specific al- Alkaloid pattern in the in vitro obtained plantlets kaloid profiles. The six alkaloids identified in shoot-clumps of In former studies, it was found that the produc- L. aestivum obtained in vitro are shown in Ta- tion of galanthamine in L. aestivum plants can be ble II. Galanthamine and lycorine were found to influenced by genetic factors, growth conditions be the main alkaloids, whereas the rest of the com- and/or chemical composition of the soil (Stefanov, pounds occurred in trace amounts. It is notewor- 1990; Poulev et al., 1993; Gorinova et al., 1993). thy that some of the shoot-clumps obtained in However, the reasons for this variation in the alka- vitro produced mainly galanthamine, whereas oth- loid profile of wild summer snowflake and of 634 L. Georgieva et al. · Alkaloid Variation in L. aestivum

Fig. 1. Biosynthetic relationships of the alkaloids identified in L. aestivum bulbs and in vitro cultures. L. Georgieva et al. · Alkaloid Variation in L. aestivum 635 shoot-clumps obtained in vitro are unknown. Due plants and in vitro cultures, is necessary to obtain to the importance of summer snowflake as a com- galanthamine-rich for the extraction of mercial source of galanthamine, there is a need this pharmaceutically interesting substance. for further investigations, on the one hand on the factors determining the alkaloid variations in both Acknowledgements the in vitro cultures and intact plants, and on the other hand on the variations of the alkaloid pat- This work was partially financed by the Gener- terns within and among wild populations. The vari- alitat de Catalunya (2005SGR-00020). S. Berkov ations in both the alkaloid content and alkaloid thanks the Spanish Ministerio de Educacion y profile of the in vitro shoot-clumps (cultivated un- Ciencia for a research fellowship (SB2004-0062). der the same conditions) and wild plants suggest The authors also thank Dr. Asuncio´ n Marı´n, Ser- that the genetic factor plays an important role in veis Cientificote`cnics, University of Barcelona the alkaloid biosynthesis in this plant species. In (Faculty of Pharmacy) for technical assistance dur- this respect, an intensive selection, both in wild ing the GC-MS analyses.

Berkov S., Sidjimova B., Popov S., and Evstatieva L. Kirch J., Viet M., Wätzig H., Greinwald P., and Czygan (2004), Intraspecies variability in alkaloid metabolism F. (1995), Alkaloid variation in Genista lobelia s. l. in Galanthus elwesii. Phytochemistry 65, 579Ð586. Biochem. Syst. Ecol. 23, 635Ð643. Berkov S., Pavlov A., Ilieva M., Burrus M., Popov S., Kobayashi S., Kihara M., Yuasa K., Imakura Y., Shingu and Stanilova M. (2005), GC/MS of alkaloids in Leu- T., Kato A., and Hashimoto T. (1985), Alkaloidal con- cojum aestivum plants and their in vitro cultures. Phy- stituents of Leucojum aestivum L. (Amaryllidaceae). tochem. Anal. 16,98Ð103. Chem. Pharm. Bull. 33, 5258Ð5263. Berkov S., Bastida J., Viladomat F., and Codina C. Kreh M., Matusch R., and Witte L. (1995), Capillary gas (2007), Analysis of galanthamine type alkaloids by chromatography-mass spectrometry of Amaryllida- capillary gas chromatography-mass spectrometry in ceae alkaloids. Phytochemistry 38, 773Ð776. plants. Phytochem. Anal. (in press). Macel M., Vrieling K., and Klinkhamer P. (2004), Varia- Codina C. (2002), Production of galanthamine by Nar- tion in pyrrolizideine alkaloid patterns of Senecio ja- cissus tissues in vitro. In: Medicinal and Aromatic cobaea. Phytochemistry 65, 865Ð873. Plants Ð Industrial Profiles: Narcissus and Daffodil Maelicke A., Samochocki M., Jostok R., Feherbacher (The Genus Narcissus) (Hanks G., ed.). Taylor and A., Ludwig J., Albuquerque E. X., and Zerlin M. Francis, London and New York, pp. 215Ð241. (2001), Allosteric sensitization of nicotinic receptors Diop M., Ptak A., Chre´tien F., Henry M., Chapleur Y., by galanthamine, a new treatment strategy for the and Laurain-Mattar D. (2006), Galanthamine content Alzheimer’s disease. Biol. Psychiatry 26, 279Ð288. of bulbs and in vitro cultures of Leucojum aestivum. Murashige T. and Skoog F. (1961), A revised medium for rapid growth and bioassays with tobacco tissue Nat. Prod. Commun. 1, 475Ð479. cultures. Physiol. Plant. 15, 473Ð497. Elgorashi E., Drewes S., and Van Staden J. (2002), Or- Poulev A., Neumann B., and Zenk M. (1993), Enzyme gan to organ and seasonal variation in alkaloids from immunoassay for the quantitative determination of Crinum macowanii. Fitoterapia 73, 490Ð495. galanthamine. Planta Med. 59, 442Ð446. Gamborg O. L., Miller R., and Ojima K. (1968), Nutrient Stefanov J. (1990), Ecological, biological and phyto- requirements of suspension cultures of soybean root chemical studies on natural populations and intro- cells. Exp. Cell Res. 50, 151Ð158. duced origins of snow flake (Leucojum aestivum L.) in Gorinova N., Atanassov A., Stojanov D., and Tencheva Bulgaria. D.Sc. Thesis, Sofia, Bulgaria (in Bulgarian). J. (1993), Influence of the chemical composition of soil Suau R., Go´ mez A., Rico R., Va´zquez-Tato M., Castedo on the galanthamine content in Leucojum aestivum.J. L., and Riguera R. (1988), Alkaloid N-oxides of Ama- Plant Nutr. 16, 1631Ð1636. ryllidaceae. Phytochemistry 27, 3285Ð3287.