,J ,mmal o{ Ftl111 opharmacolog y, 3 ( 198 1) Jl 3 · 335 : JI~ O El sevier .Scquo ia S .A ., Lau sannc - Printed in lhe Neihcrlands
ANALYSIS OF ALKALOIOS IN LEAVES OF CULTIVATED fERYTH llOXYLUM AND CHARACTERIZATION OF ALKALINE SURSTANCES USED DURING COCA CHEWfNG*
LAURENT RIVIER fnst ítu te o{ Plant Bio logy and Ph ysiolo¡.;y o f thl' University, 6 Place d e la Riponne, J 005 I...ausanne (S witze rland) ond D epart ment ol T oxico/o¡.!y, Karo fin :;;;k a lnslitule t. / M 0 1 Stochholm (Sweclen)
..;ummary
Severa! salven t.s were tested far the ex traction of the alkaloids in ··:rythroxylum coca. The resulting crude extracts were analyzcd by gas ·hromalography- mass spectrometry. Ethanol ex traction was found to be the ·nly quantitativc method presenting no artifacts. lt was established thal ncain P and cis- and trans-cinnamoylcocaine wcrc the endogenous alkaloids 1 E. enea leaves. From the severa] breakdown compounds arising during n1g-term extraction wi th H2S04 or CHC1 3 , ccgonine mc thyl ester was the nly alkaloid fully identificd ; ccgonidine methyl ester was tentativcly •ent.ified on the basis of its mass spectrum frah'Tllentation pattem. Quant.ification by mass fragmcntography of the three cndogenous nnpounds was performed using a stable-isotope dilution technique on dividua! !caves of single branches of E. coca, E. novugranatense and E. ·vogranatense var. truxillense. The relative amounts of thesc alkaloids anged with leaf age as WPll as betwcen species and varieties. The variation alkaloid levels between individual leaves was too grcat to allow the use uf ·' ratio between cocaine and the einnamoylcocainC' levels as a taxonomic ·rk er. The initial pH value uf 17 differcnt alkaline substances traditionally ·d during coca leaf chewing was mcasured after dissolution in H2 0; values gcd bctween 10.1and12.8. Buffer capacity was determined by titration h l!Cl. Three types of curve shapes were obtained which could correspond 'faOII, Na2 C03 and NaHC03 titration curves. One sample of alkaline ·erial had no buffer capacity at ali. Thc recovery and breakdown of thc ·tine contained in E. coca lcaf powder was monitorcd for onc hour at ''lis pHs at 37 "C. The levels of cocaine and bcnzoylecgonine did not ·1gc by more than 17% at any of the pHs test:c•d (6.0, 9.0 and 11.5). Jt concludC' d that the alkaline su bstances are mainly responsible for the
* Based on a paper presented at the Symposium on Erythrox y lnn - New Histo rical ~c i e ntifi c Aspecb;, sponsored by the Bolanical Museum o f Ha rvard University and de la Cultura del Ecuador, Quito, Equador, Deccmber 3 · 5, 1979. 314 transformation of the alkalo ids to free bases and not for a major hydrolysis of co c ain ~_. \ ------·------·- - - ·--- --
Introduction
Two closely related species o( thc genus F:ry throxy lum, namely E. coca Lam. and E. nouogranatense (M orris) Hieran., have long been cultivated in South America. For thousands of years Indians have used it as a mild stimu lant and medicine. The principal active constituent of the leaves of cultivated Erythroxy/um species is cocaine which is traditionally absorbed during chewing. Wh en the coca leaf is masticated, an alkaline suhstance is ge nerally added to the quid. Without su ch a su hstance the chewer does not experience fully the desired effects. lt has been suggested that when cocaine is ingested, as in the case of coca chewing, a complex metabolic p rocess takes place (Montesinos, 1965). Although it has been demonstrated that cocaine is present in the ph1sma of coca leaf chewers (Holmstedt et al., 1979), consider able controversy still exists concern ing the fate o f the alkaloids durin i; chewing. Variability in the alkaloid content of Erythroxylum species has long been recognized (Morris, 1889; de J ong, 1906). Many chemical analyses haw been carried out on cultivated coca lcaves of uncertain origin using methods that were at best only scmi-quantitative. Consequently thcre is a real lack of reliable data on the alkaloids of Erythroxylum species. The principal o bjectives sout(ht in this study werc : (1) to identify by gas chroma tography- mass spec trometry (GC--MS ) the principal coca alkaloid· present in well-preserved Erythroxylum leaf materials collect0d in Pcru and cultivated in North America; (2) to compare the efficacy of extraetion o f these alkaloids by different organic solvents and to determine tlw alkaloid levels in leaves by using a stable-isotope dílution technique ; and (3) to detcr mine the buffer capacity of alkaline substanccs commonly uscd during coca leaf chewing in Peru and Colombia, and to in vestigate thr stability o f the alkaloids in leaf material during incubation in aqucous buff<'rs under simu lated conditions of mastication.
Materials and methods
Chemícals Cocaine hydrochloride was obt.ained from the Karolinska Apo tekct, Stockholm, Sweden. Cocaine deuterated on the mcthyl-cster moicty (cocaine-d3 ) had been synthesized previously (Holmste
Alk alinc s11 bstances In gPneral, coca !paf ch(•Wf' rs add alkaline su bstances to thf' !caves - some times lime, some times vegctable aslws pn'pared in tllC' shape o f halls or rolls. Dcscriptions o f thc vo ucher specimens used in this study are given in TablE> 2. The cal is prPpared by buming calcium-containing rock fragm ents o r shells and grinding the remains in to fim' po wdPr. To form l/ipta, thc ashes of thc stalks of quinua (Ch eno podium quinua \Villd.) and cañih11a (Ch. pallidi· cau/c Aellen) cwo indigenous crops of th!' Peruvian Andes, ar!' mixed with water to form a paste which is formed in to small ¡,'Tey halls and dried in tlw sun. A small pi<'ce is brokf'n off and takf' n with each m outhful uf coca lf'avf's. O thf'r plant materials - su ch as plan ta in roo ts or cacao pods -- may also lw USf'd for thr samc purpose. In the Amazon, leaves of Cecropia spp. are burn('d and mixed with coca leaf powder (Schultes, 1957 ; llolmst.ed t e t al. , 1979 : l'lowman, 1981). l·:.draction mclhods Accura!A' measun •nwnt u f tlw total alkaloid con !A•n t in plan! maU:·rial should hL• obtained u s in ~ an extractio n proci>dun: that is as complPte as pussihl!'. Sine•• tlw cliff(•rent ulkaloids that havP so far ht'!'ll found in coca l1•av(•s vary in tlwir chemical natun' (bas0s, esters, acids and/o r PthPrs), th(•ir iw lation fro m biological rnaU:·rial is difficult. At the prcse 11 t time no singl(' nw thod o f ('Xlradion for ali lhe coca alkalo ids is availahl<'. Ouring this >ludy , six Px lractio n m('thods were t<'SlNI a11d com pan·d , using a singl<' col· ... ctio n o f ctriPd f;_ coca it'aVPS (vouchrr S[H•cimr n "lo. 7351) ¡;round to a ·1owd.,r: mr thods A . D wpn• qualit.ative and mcthocls E and F wPn· quanti ativt> .
.\ 1c tlwd A ( ..aci d ex tract "") Fo ur grams u f tlw l<>a f powd(•r wcrl' rxtrac t;:>d with 200 mi uf 1 N
12 S04 al room t!'mp<>rature for 20 h. ,\fter filtration un a G4 glass fil ter, lhP <1 lutiu 11 was brought to pll 9 hy th(• addit io n of solid Na2 C0 3 . Thc alkaloids •·n• ('Xtract.ed hy partitioning tlw aqueo us phase thn'e times wi th 100 mi u f .- hloroform- Mctho d H ("vo latile alkalo id ex tracl ") Four grams o f the leaf powder wPre m ix('d with 100 mi o f H 2 0 , 1 50 ml anhyd ro us glycerin and 20 g of NaOll (Fikenschrr, 1959); 100 mi of thP lution w1•rp (•vaporat.i'd un Plowman No.* Description 7351 Erythroxylum coca Lam. Air-dried leaves obtained from Empresa Nacional de la Coca, Lima, Peru, May 10, 1978, and mai~tained at - 20 º C in the dark until analyzed. 7611 E. coca Grown from see *Voucher specimens are deposited in the Herbarium of the Department of Botany, Field Museum of Natural History, Chicago, Illinois, and al the Botanicat Museum of Harvard Un iversity, Cambridge, Massachuselts , U.S.A. 3Ji eollected fraction was thcn extracted as in l\fothod A with thc chl oroform diethyl ether mixture at pH 9. Method C Four grams of thc leaf powder were extracted sequentially at room temperature with 200 mi of n-hexane (20 h), 200 mi of chloroform (20 h) and finally 200 mi of ethanol (20 h). The three fractions obtained were evaporated separately to dryness; the residues were then each redissolved in the corresponding solvent for GC- MS analysis. Method D ("crude ethano/ic ex tract ") Two grams of the leaf powder were cxtracted with 100 mi of boiling 96% ethanol for 15 min and left to cool for 4 h before filtration. The solvenl was evaporated as described previously and the residue was either redissolved in c thanol or silylated. Method E The isotope dilution technique using deuterated cocaine asan interna! standard (Holmstedt et al., 1977, 1979) was applied to small amounts (approx. 100 mg dry wcight) of plant material. Methud F In order to extract cocaine and its hydrolysis compounds, it was ncces sary to adapta method which has becn reported to givc 99% recovery for cocaine and 80% recovery for benzoylecgonine from uríne (van Mindcn and D'Amato, 1977). Bríefly, 100 mg dry weight of the leaf powd<>r werc left far a given time, ranging from O to 60 min, in 5 mi of the 0.4 M buffers used far the calibration of the pH-meter electrodes ( Radiometer, Copenhagen, Denmark). For extraction, the solution was brought to pH 1 with 2 N HCI. l"o ll owing centrifugation, the solids were removed and the supcm atant was .mmediately adjusted to pH 9.5 with a mix ture of salid NaHC03 and Na2 C03 '3:1, w/w). One milligram of cocaine-d3 was added and 10 mi of an ethanol·· ·hlorofarm (1:4, v/v) mixture was used to extract the bases and the ampho ·eric substances. The organic mixture was evaporated t o dryness at 60 ºC mder a fl ow o f nitrogen. The rcsidue was treated with 1 mi of an ice-cold olution of diazoethane in diethyl ether, rcduced to dryness and red issolv ed n 50 µ l of toluene for quantitative GC MS analyses. 'itration o f the basic substances Weighcd amounts ( 50 · 100 mg) of the alkaline materíals (see Table 2) ·rere d issolved as fully as possible by stirring them in 5 mi of freshly double istilled water for 5 min at room temperature. pH measurements and titra ·ons with 1 N HCI (Ti.tri.sol, Merck) were performed automatically by a tration assembly (Model TTA 60, Radiometer) ata flow-rate of 70 µl/min. -:ns Gas chromatography- mass spectrometry Qua/itative analyses The MS analyses were carried out o n a combincd c;c - MS instrument (Model 5985A, Hewlett-Packard, Palo Alto, U.S.A.). Typical GC conditions used throughout were: SP-2100 WCOT glass capillary column (10 m X 0.25 mm I.D.), te mperature isothermal at 100 ºC for 1 min, programmed at 10 "C/min to 240 "C. Split-mode injcctions of 1 µl were used ata helium pressure of 0 .5 kg/c m 2 giving a carriN gas-flow of 1 ml/min through th e column and a split ratio of 1:50. Thl' capillary column was connected directly to the MS source. The ion source parameters werc set routinely by using the Au totune program providing a 70 e V ionization cnergy and 300 µA emission curren t. The temperature of thc ion source was main tained at 200º ± 2 ºC. The resolution of the mass spectrome t.er was 2.5 times the rnass, when measured at half peak hcight. Repctitive scanning from m /z 40 to m/z 540 was perfonned at one sean per 1.3 scc. By using the Batch program, cach peak of the reconstrncted chrornatograrn that was biggcr than 1% ofthe base peak in the total rnn was further proc!'ssed for a library search. Identifications were based on similarity indices higher than 0.98 (absolutc identity had an index of 1.00) and by direct corn parison with authentic standards rnn under the sarne CC -MS conditions. Both thc rt>t.e ntion time and the mass spectrurn wcm critcria for thc identi fication of unknown compounds. Quantitative analyses GC conditions wcre similar to those usPd for the qualitativc analyses except that an oven temperature prograrn of 30 ºC/rnin was usPd. Samplc injection was carried out in the split-lcss mode ovcr 3 0 sec. By this m pans all the sample injectcd was transferred to the capillary column. F or th<: quan titative measurements, the GC- MS was operated in tlw sclcckd ion monitor ing mode, a technique also terrned rnass fra¡:,me1itoi.,rraphy ( M!') (Harnmar et al., 1969). For analyzing the extracts obtained using rn ethod E, the rnass spectrometer fil ter was set to record the ions at m/z 182.l and 303.2 for cocaine, 306.2 for cocainc-d3 , and 1 82.1 and 329.2 for both cis- and trans cinnamoylcocaine. Each mass was measured over a 150-rnsec period in each cycle. Calibration c urves for the accurate det<> rmination of the endogenous substances were obtained routincly by calculation of the response factors of each individual cornponent against cocaine-d3 as prcviously describt>d (Holm stedt e ta!_ , 1979). Whcn cocaine and benzoylecgoninc ethyl derivative were quantified in the extracts obtaincd using mcthod F, ions at m/z 18 2.1 and 303.2 werc c hosen for cocaine, 306.2 for cocaine-d3 , and 317.2 for he nzoylecgonine e thyl ester. Cocaine and benzoylecgonine ethyl ester werc quantificd from the ratio of ion intensities at m/z 303/306 and 317/306, respectivcly. ;319 Results and discussion Alhaloid content o{ E. coca The gc nus Erythroxylum is the only natural source of cocaine an d related compounds of the ecgonine type. Although minute quantities of cocaine have been detected to date in 17 wild species of the genus, including severa] Old World species (Aynilian et al., 1974; Holmstedt et al., 1977; Rivier and Plowman, unpublished data), only t wo main cultivated species - E. coca and E. novogranatense - have been found to contain large amounts of ecgonin e-related substances. As regards the distribution of the minar alkaloids in the genus, 12 such compounds have been reported for the culti vated coca (Raffauf, 1970). Besides cocaine (methyl-3¡3 -benzoyloxy-tropan-2(3-carboxylate; (1)) (Wi:ihler, 1862), other alkaloids have been isolated from leaves of cultivated coca. Pseudococaine (C-2cr,C-3¡J ) was found in coca leaves by Liebermann and Giesel (1890), but this compound is probably an artifact of the isolation procedure, arising from the action of alkali on cocaine during prolonged work-up (Archer and Hawks, 1976). Cinnamoylcocaine has been isolated from severa! coca leaf samples (Liebermann, 1889); Moore (1973) was the first to be able to id<>ntify thP cis (2) and trans (3) isomcrs by separating COO M, ... . M, C O O M, \¡__ \·" \¡. . \ H H e [ .' ) e i'<-~ ~..- o e ' e H ' O Q H . "; o Y) 111 121 COQ M, M, ~ OO M, N.H. . .·. " :e e '- -o e H " .-0 -»-j:" ( 3 ) 141 •em by gas chrornatography. Benzoylecgonine has been isolated from ·ruvian (de .Jong, 1940) and Colo mbian (Espinel Ovalle and Gusmán Parra, J71) coca leaves but not from the variety obtained from Java (de Jong, J40). It may also be an artifact , possibly arising from hydrolysis of co.caine \ rchcr and Hawks, 1976). De Jong (1940) has found that ccgoninc methyl ter (4) occurs naturally in Java coca leaves. Tropacocainc, thc benzoate ter of tropan-3-ol, occurs in Java (Willstae tter, 1896) and Peruvian (Hessc, 102) coca leaves. The hygrine group o f coca alkaloids is usually found in •• ethereal extract of a slightly alkaline p ercolate of coca leaves (Henry, 149). It is evident that the u se of a base for extracting the h ygrines raises •) question of the natural occurrence of these substances as such in living J mt material. 320 In the last 20 years, only two reports havc dealt with thc min or alkaloids in cultivated coca leaves. The first study was on commercial Java coca (probably E. nouogranatense var. nouogranatense) (Fikcnscher, 1959). Mostly cocaine and cinnamoylcocaine were found in the dried le avcs with small amounts of methylecgonine, hygrine, tropacocainc and traces of cuscohygrine and nicotine (Fikenscher, 1958, 1959; Hegnauer and Fikenscher, 1960). Urifortunately, ali identifications were based un Rr values in thin layer chromatography (TLC) and colour reactions with specific reagents; quantification was performcd on thin-layer spot areas. The sccond work was done on E. coca var nouogranatensis [sic]* (Espinel Ovalle and Guzmán Parra, 1971). Using cocainc as the single standard, these authors found 62% cocaine, 16% cinnamoylcocaine, 14% benzoylecgonine and 8% tropacocaine in the leaves. Identifications were based on comparison of melting points, odour of hydrolysis componcnts and infrared spectra which wcre recorded for each substance separated by paper and thin-laycr chromatography. Jt is impossible to evaluate a posteriori the importance of arlifacts in the data briefly reported above. It is also imposs ible to provc that sorne alkaloids found previously in coca migh t origina te from breakdown of alkaloids during poor drying conditions, long storage and/ or analytical work up procedures. In the prescnt work, the GC- MS results show clearly that th• crude ethanolic extract obtained using method D (sec Materials and mcthods) can be resolved on a capillary GC column without any furlhcr purification (Fig. 1). The reconstructed total ion chromatogram dísplays only ~ .·ew peaks. Ethanol extracts many compounds from the plant material: sugars, steroids, as well as alkaloids. Sugars do not appcar in tlw chromato¡;ram as tlwy are not sufficiently volatile. Whcn the extract was silylated, it was poss ible to find additional peaks in the chromato~>ram which were identified as tri methylsilylated (TMS) derivativcs uf m onosaccharidt>s such as glucose, man nose, galactose, xylose, etc. As far as ecgoninc alkaloids are concerncd ali est.ers are volatile enough for GC analysis. Acids such as benzoylecgonint• and ecgonine have to be derivatized; when the crudP ethanolic cxtract was silylated neither of these two su bstances could be detected as thcir TMS derivative. Cocaine (the main peak in the chromatogram) and both cis- and tra11s-cinnamoylcocaine wcre positively identified (Fig. 3A - C). None of the othervolatile alkaloids previously reported in the literalurc could be dctecU>d in the material used. In contrast, thc chromatograms of the intact or silylated acid cxtract obtained from the same material using method A (Fig. 2) were lcss complcx than the eructe ethanolic extract. This is due mainly to the additional parti tioning step included in this procedure, which eliminated most of the water· soluble compounds, in particular the sugars. In addition to the three sub stances found in the ethanolic extracts, ecgonine methyl ester (Fig. 30) was also identified. In the acid extract, the quantity of ccgonine methyl ester wa *Mostprobably l{ novogranatense var.nouograrwl <' nse (uide Plowman, 1979a an d b 321 :um: 1•J e l•)tl ' p) Fig. l. Total ion chromatogram of the GC MS analysis of the crude ethanolic extract of E. l'oca leaves (voucher specimen No. 7351). No peak appeared hefore 13.5 m in retention time under the GC conditions used (see text). Peak 1 was ident ified as cocaine. Peaks 2 a nd 3 were found to be cis- and trans-cinnamoylcocaine, respectively (see Fig. 3). No other su bstance of the ecgonine type has been d etected. The other peaks could not be fully identified when using the MS library search. The peak emerging at 25.3 min was fuund to be of the alkane type. ~ 0 0 ti ~) tl 300 , : ~·· . S~'') -? ?e 1eee 12ee \•)•) ~? 0 .. .1 . '.~tl % r:;;. 1 4 1 3 1 ¡1, ·I ~_JJ \ ¡.:• l.? 1 ~~ ¡:;e J " -- -' P •ttont 1•) n-· T111\ '!!' '111). r'l l 'ig. 2. Total ion chromatogram of the acid extract of E. coca (voucher specimen No. 351) run under the same GC--MS conditions as in Fig. l. The first peak appeared at 0.2 min retention time. Identities of peaks 1 - 3 were the same as in Fig. l. Peak 4 was ·lentified as ecgonine methyl ester (see Fig. 3). Peak 5 could not be fu ll y identified, but as postulated t o be related with ecgonidine m ethyl ester (see text). 322 estimated to be half that of cocaine. Ecgonidine mcthyl ester was t1rntatively identified from its mass fragmentation pattem (Jindal et al., 1978)*. How ever, lack of a reference substance meant that this identification could not be verified. Ecgonidine methyl ester, reported to occur in Java and Peruvian coca seeds (Henry, 1949) but not detected in the eructe ethanolic extract, was probably formed by the action of 1 N H 2 S04 during extraction using method A. Similarly, ecgonine methyl ester appeared during acid treatment and after storage of the leaves under certain conditions (Turner et al .. 1981). When organic solvcnts were used sequentially for extraction as in method C, complete ex traction of the alkaloids was difficult to achieve. With n-hexane, for example , cocaine and cinnamoylcocaine were only ! 50 ~ so . ¡ ... 4 ·------·-- - - , 1QO ···- T ivo :.i~I ! ! so r 'º j r. , .• , ' 1 ¡ . -\.-r-~..-· ,'...- _,+ -rrJ 200 'BO m¡, JO O ': 50 . J~<> ~ l' l .. ~ - · ~ ! ; -t..,._...... ,...1-~ --t -~ - t Llt::_ _:~ ::_ •·.. ' --- -. ·------y 'ºº 1 1 1 i ·¡l ·: ~ so ¡ 1 ¡ ¡ J. 1 ¡ ' " ..... ~ .. ' j'¡ ~ i { 1 1 " " 1 1 ; i ~ JJ.l, --r1------_:_ - ,--.- ---· --·+-- 50 100 150 2 50 300 5 0 100 ,~, o m¡ i. ;ioo 'ºº m¡' Fig. 3. Mass spectra (70 e V) obtained from (A) peak 1, (B) peal< 2, and (C) peak 3 or th1· GG- MS chromatogram represented in Fig. 1, and from (D) peak 4 or the GC MS chro· matogram reproduced in Fig. 2. The corresponding mass spectra of the reference sub stances were obtained under the same conditions (see text). + •Tue mass spectrum 3.t 70 eV of peak 5 or Fig. 2 was: m /z 181 (M~. 32%), 166 (M· -CH3, 10%), 152 (M · -N -CH3, 100%), 122 (M~ - COOCHa. 14%). 82 (22%) and 4 2 (20% ). 323 partially extractcd sin ce subsequent cxtraction with chloroform rcvealed additional amounts of these compounds. Prolongcd cxtraction with chloro farm resulted in sorne breakdo"'-n of cocaine since 15% of the alkaloids could be found as ecgonin e mcthyl ester. This amount, however, was smaller than the quantity faund in the acid extract (see Fig. 2). As shown previously (Holmstcdt et al., 1977, 1979; Turner et al. , 1979, 1981), and confirmed here, short cxtraction in boiling ethanol is the most efficient technique far minimizing the breakdown of cocaine and ensuring quantitative extraction of ali alkaloids of the ecgoninc type, as they are ali solu ble in this solvent. In an attempt to detect nicotine in E. coca leaves, the proccdure used for Java coca (E. nouogranatense var. nouogranatense) by Fikcnschcr (1958) was repeat.ed. No trace of nicotine was found in the plant material uscd in this study even using MF. Although a different species of Erythroxylum was used, unambiguous proof of thc occurrence of this alkaloid in the genus has still to be provided, since the single previous identification of nicotine was hasPd on TLC and colour reactions only. ({11an titaliue analysis of a/kaloids in Erythroxylum In this study, deuterated cocaine, labeled in thc methyl ester moie ty , was used asan interna! standard as previously reported (Holmstedt e t al., 1977, 1979). With ethanol as the extraetion solvmt (method E) no isotope ·xchangC' was observed to takc place anda simple linear relationship between hP appropria t.e ion intcnsity ratios and concentrations of cocaine, cis- and rans-cinnamoylcocaine and benzoylecgonine was obtained within the range ,f conc•mtrations used throughout this study. Analysis by !\1 F allowed thc determination of all components in a single 1j1·ction of the crudc ethanolic extract without further purification dueto '1t' specificity o f th<' ions chosen. Since prcvious GC runs of extracts showed iat. cocainc and the cinnamoykocaines do occur naturally (see ahove), MF ·ialysis wa5 carricd out during the retention times corresponding to these impounds ( Fig. 4) . ldentification of cocainc in the plant extracts was based 11 thc ratio of ion intensities m/z 3 03/182 occurring ata retention time i!'ntical to that of the interna! standard (R 1 = 6.2 min). Cis-cinnamoyl ":aine (R, = 7.3 min) was identifi ed from its mass spectrum which is almost •Jntical to that of the trans isomer which emerged subsequently from the C column (H , = 8.4 min). Consequently both isomers could be quantified sil y using thC' samP ions m/z 329 and 182. A mass fragmentowam from a 1gle coca \paf ••xtract is shown in Fig. 5. ~1 F is a very scnsitivc mrthod of detection: small pieces of plant 1!Rrial can be analyZ<'d successfully and nanogram quantitics of cocaine n be quantified ata 5% precision leve!. When working at higher concen :lions (in the prcsent assay, 100 · 500 ng were routinely injected) a 1.5% »cision was obtained far within sample variation. It is known that enzymatic conversion of cis-cinnamic acid to its trans >Jnt'r occurs in plants but no attempt t o correlate this reaction to cis-cin- 324 ¡1L,2 .. '32i .~ O• 819. 1,,\ Jl:l6.2 ,\1 O• 4 -?- ·? . J~ 3 . .? I '. I~ .. 2 lhi . .. S·Hl . ISZ. t :' Fig. 4. Mas.s fragme\itogram of authentic standards mixture consisting of 50 ng of co· caine, 100 ng of deuteratcd cocaine and 20 ng of trans-cinnamoylcocaine. The ions al miz 182.l and 303.2 were used to monitor cocaine, the ion at m/z 306.2 cocaine-d3, and the ions at miz 182.1 + 329.2 cinnamoylcocaine. Cocaine and cocaine-d3 emerged simul taneously at R 1 = 6.2 min and /rans-cinnamoylcocaine al H. 1 = 8.45 (see abscissa). Ion intensities are given as normalized for each channel individually (see ordinate) and area count..s in arbilrary units as calculated by the computer, were obtained for each peak. O• 33S. .. 138 . 3.?'1- .2 :"' 1\ O• 1415. ¡1, 306 .2 O• 3':11. Jli!3.2. ------ l'0.'5-1. 1\ O•• 25.?;" . O• 933 . l1 ¡, l:32 . 1 I '· 1 ' /\ Fig. 5. Mass fragmentogram of a 3-µI injection of a 5-ml ethanolic extract of the first unrolled leaf of E. coca (voucher specimen No. 6235) obtained by using method E to which 500 µg of cocaine-d3 had been added as interna! standard before extraction. Each channel was normalized to the highest peak of the run. Cis-cinnam oylcocaine emerged a! R, = 7.3 min. Correct retention times and miz 3031182 + 3291182 ratios are additional proof of the identity of cocaine and the cinnamoylcocaines, respectively. Quantification was based on calibration c urves of the miz 3031306 and 329/306 ratios (see text). 325 namoylcocaiiw has cvC'r b<'en kmpted. No isomerization of lrims-cinnamoyl cocainc was observed during cxtraction of the plant matRrial by thr present proccdurc. This has been confirmed by thc results obtained by Turner et al. (1981) using similar conditions. However, the stability of cis-c innamoyl cocaine has not been studie d in vitro and this is due probably to the diffi culties in obtainin g cís-cinnamoyl anhydride for synthcsizing thc correspond ing cinnamoylcocaine (Moore, 1973). Whcn considering ali thc rrsults for whole leaves obtained in the present study (Figs. 6 - 8), syskmatic conver sion of the cís-cinnamoylcocaine to the trans isomer seemed not to occur since variable amounts of both compounds were found in the different plant specimens. In fact, it can be assumed that if any isomerization did occur bdore or during work-up, both isomers would be present in ali samples. lf thr trans isomer was more stable than the cis isomcr, it would have been present in ali samplcs in which cinnamoylcocaine was A B E"' 1- :i:: 60 "UJ 3: > Q: o 'º . 4 4 .o. ~ "';;¡_ 1 k . i· 1- z UJ 1- z 300 o u 'ºº o o -' 100 "' "-' "' 4 5 , 2 3 4 LEAF NUMBER Fíg. 6. Dry weight and alkaloid contenl of individual leaves of lwo branches (A and ll) n E coca {voucher specimen No. 7611). 11te hatchcd areas represcntcd t h~ cocaine levels, the white and black a reas c is- and /rans -cinnamoylcocaine, respeetivcly . was unrolled and lf'al 10 was yellowed. Th!' alkaloids WC're al their highPst leve! in leaves 4 and 8. The relative amounts of cocaine ami c innamoyl cocaincs were very different to thosc found in the respective le aves of E. coa levels of cinnamoylcocaincs were higher in !caves 1 and 2 and decreascd graduaJly with leal age, in contrast t o cocaine. The highest ll'vel of cocainf' was found in leal 8. In E. nouogranatense var. truxillense, smaller Jeaf size is a conscquenc• of human selection under differing climatic conditions in thc dry arpas of northern Peru (Plowman, 1979b). The aJkaloid pattern in this variPty fol lowed the increase in leaf dry weight (Fig. 8). The proportion of cinnarno; cocaines was grpater in young leaves than in old ones when compared l< ' cocaine levels. 327 6 0 50 - m 4 0 t- 3 0 I "UJ 2 0 ;: >- 10 0: o o 10 5 00 m "- ....., 4 00 t- z íl UJ t- 300 z o 1 u 2 0 0 ~ íl 1 o 1~ i J ~ íl < 'ºº 1 ¡ . ! < ~ 11 íl 1: 1' 1 1 ~ ~1 o ¡ l 1 1 11 i. J1 10 LEAF NUMBER ig . 7 Dry w eight. and alkal oid contcnt of individual leaves o f onl' branch of E. novo· w rntr ns<' (voucher specimen No. 6 2 7 5 ). The system uscd for re prescnting each alkaloid the same as in Fig . 6 . In conclusion, a clcar diffrrence in thc alkaloid distribution betwcen ·aves of various agcs can be observed wh en comparing the thrce varieties ialyzed herc. This diffcrence could be attributed mainly to genelic factors ncc all thrce varicties wcre grown under identical conditions in the sarne •cality. Howevcr, variations within one species are of importance as shown, •r f'Xatnple, in Fig. 6 far E. coca. Consequently, it is not possiblc to use the lativc quantities of cocaine and cinnamoylcocaincs as chemotaxonomic .arkers at the prescnt stagc of our knowl edge, although tentative movcs in is direc tion have becn madc n'cently (Youssefi et al., 1979; Turner et al., 181). From thc point of view of biosynthesis, the prescnt data indicate that " coca leaf is certainly the sitc of a very active metabolism of thc alkaloids. lditional studies will be nccessary for a better understanding of their "tabolism befare any further conclusions can be drawn. 328 ( mg) ,.. "'Q o l----'--~~~~~~--'--~~~~~~~~~---j (,u9 ) .... z 1- z'" o u Q o.... < "'.... < LEAF NUMBER Fig. 8. Dry weight and alkalo id contenl of indivi dual !caves of o nc hranch of R. novo granatense var . truxillense (voucherspccimen No. 6235). Thesystem used for re presenting each alkaloid is the same as in Fig. 6. Characterization o{ the alkaline substances During chcwing of the leaves of Erythroxylum sp0cif's in the Andes or eating the powder preparation made from E. coca var. ipadu (Plowman, 1979a, 1981) by the natives of the Amazon, alkaline substances are frequen tly added. The purpose of this practice is not completely undPrstood. lt has been suggested that the base aids salivation in ex tracting the alkaloids in thc mouth from the leaf material and also that it could bP strong enough to hydrolyse cocaine into ecgonine derivativ''S of lcss potent psychoactivity (Zapata Ortíz, 1970). · As a first step in the understanding of the adion of these alkaline sub stances, their initial pH in solution and their buffer capacity were mcasurcd. Seventeen different alkaline materials of various origin were tested (Table 2). When starting with 50-mg amounts of material and dissolvin g them in 5 mi of distilled H2 0, the initial pH values ranged from 10 to 11, except for the cal samples which gave a pH ;;, 12.6. Reports on the chemical composition of these materials are scarce and imprecise. Gosse ( 1861) reported that llip ta contains CaC03 , MgC03 , KHC03 and sulfate, chloride and phosphate. Cruz Sánchez and Guillén (1949) found various tocra samples to cuntain " mainly potassium and calcium (30%) besides carbon, silica and carth (60%)". Such mixtures in a 1% solution gave pH values of 10.5 to 11.5. Baker and Mazess (1963) showed that llipta contains 12% of calcium anda lot of magnesium salts; one sample of cal de piedra was found to have 36% of calcium and another 42%, probably in the form of almost pure lime. From ali the curves obtained by titration with 1 N HCl of the various alkaline suhstances dis- Characteristics or Lh (' alka line substances used with coca lea í chcwing ··------·- ---- Voucher No. t-;am e Origin Starting Buffer Curve Plowman p Hb capacity e shaped 7602 Tocraª de Quinua Lima , Pe ru 10.5 0 .075 A (Chenopodium quinoa Willd .) 7603 Tocra de Platano (Musa sp. roots) Lima, Peru 10.6 0.35 B 7605 Cal de piedra Tocache, Hu ánuco, Peru 1 2. 7 1.52 c 7936 Llipta de haba (Vicia faba L.) Puno, Peru 11 .0 0.50 B 7937 Llipta de Quinua Puno, Pe ru 10.9 0.43 B 7938 Tocra de Quinua Andahuaylas, Peru 10. l 0 .30 Fl 7939 Tocra de Toknay Machente, Dept. Ayacucho, Peru 10.3 0.3.5 Fl 7942 Tocra Cerro de Paseo, Peru 10.4 0.015 A 7943 Mambe (from limestone ) San Agustin, Co lom bia 11.6 0.84 B 7944 Llipta de Quinua Cusca, Peru 10,6 0.59 B 7945 Llipta de Quinua Cusca, Peru 10.9 0.40 B 7946 Llipta de Cañihua Paucartambo, Cusco. Peru 10 .2 0.55 B (Chenopodium pallidicaule A ellen} í947 Llipta de Cacao Quillabam ba, Cu seo, Peru 10.4 0.83 R (Theobroma cacao L.) 7950 Llipla de Quinua a nd Cacao Espinar, Cusco, Pe ni 10.6 0 .34 B 7951 Leaf ashes of Cecropia \1itú, Vaupés, Co lo mbia 11 .0 0.15 A (8 ) (e(. sciadophylla Mart. ) 7953 Cal de Trujillo Truji !lo, Peru 12.6 1.30 C +B 7954 Cal de Mariscos Trujillo, Peru 12.8 1.12 C+B ª Tocra and Llipta are different local names for the same preparation. b0.1 M solutions of NaOH , Na2C03, NaHC03 and pure H20 gave pHs of 12.9, 11 .5, 8 .2 and 6.3, respectively. The starting p H corres ponds to the pH of a 5 mi solution from 50 mg of alkaline material. e Buffer capacity represents the quantities (expressed in mi) of 1 N HCl required to bring 100 mg of alkaline material dissolved in 5 mi of H20 to a pH of 5.0. dCurve shapes correspond to: (A) H20 titration ; (B) NaHC03 + Na2 C0 3 t itration; (C) NaOH titration by 1 N HCI (see Fig. 9). 330 ~--~-- -,------.------·· -~ - y--- . ---Cut di · -P1Pd1<1 /\ .;In·~ pH 11 G.·~c·c>p1il \ ,-- l l1p 1.1 1 l. ", \ \ \ ~\ \ \ \ \. ' n11 H C l. N Fig. 9. Tltration curve& obtained from selected alkaline suhstances uscd during coca leaf chewing . 'focra corresponds to vouchcr specim cn No. 7602, ( 'al de pir clra to No. 7605, Cccropia ashes to No. 7951 and Llipla de (}uúwa to No. 79-1 4 (sec Table 2). Buffer capacity (BC) as defined in Table 2 has been ~aphically represented [or Llipta 7944. solved in water, three distinct shapes wer(' observed (Fig. 9). Clcarly, cal de piedra had the largest buffer capacity. Its titration curve is similar to that obtained with NaOH. Thc llipta sample was characteristic of a mixture of Na2 C03 and NaHC03 . One other sample of ash hall (voucher No. 7942) had no buffer capacity at ali, as illustrated in Fig. 9. Two diffcring opinions are held for the action of the alkaline substanc<.» during coca chcwing. On thc onc hand, it is thought that thc usual addition of basic material facilitates the libcration of cocaine base from its conjugates with plant acids or other constitucnts. Thc sol u ble hydrochloride is o nl y formed in the stomach where it is absorhed (Salomon, 19,19). On thc other hand, a group of Peruvian workers hold the view that cocainP is broken clown to a considerable cxtent into benzoylecgonine and ecgonine before absorption (see Montesinos, 1965). The latter view srcmed to be strengthpned by the experiments of Nicshulz and Schmersahl (1969). Using pur<' cocaine base - which is almosl insoluble in water · and adding it to a solution of saturalPd Ca(OH)i for a period of 30 min at 37 ve, thesc authors found on thc basis of TLC analyse' that only 15% of the cocaine was hydrolyzed. The final product ohtained was ecgoninc only. In thc same study, when coca lPavrs were u sed instcad o· purc cocainc, the apparcnt breakdown of the cndogcnous alkaloid was mor" rapid. Such results lead sorne workers to believc that cocaine is degraded to ecgonine prior to uptake in the bloodstream ancl that this lattcr compound 1 the central alkaloid involved in coca chPwing pharmacology (Burchard, 197: 1976). Howcvcr, it is cvident that Nieschulz and Schmersal1l used C'xtrcmc values havl' b(• (• n chosen: pH 11.5, corresponcling to tlw pH nf a mixture of coca leaf powdcr, saliva ancl 5% cal de piedra ; pH 9 , corresponding to coca leaf powdt>r, saliva and 50% Cecropia asht>s; and pll 6, given by coca leaves ancl saliva alonc. lnstcad of saliva and alkalinc substanees, standard buffe r sol u ti o ns were used to ensure reproducibility of thc assays. In this experiment, the me thod for extracting cocaine ami benzoyl ccgonine had to be different fro m the one u sed for the total alkaloid analyses because of the presence of water in the incubation medium, preventing the use of ethanol as extracting solvent. Another solvent mixture was found to be satisfactory (Van Minde n and D 'Amato, 1977) and the ethyl ester of benzoylecgonine was synthesized far MF analysis. Cocaine and benzoyl pcgonine ethyl psf{) r did not separate well undC' r the GC conditions used but thcir respective molecular ions could be monitored for quantification without mutual in terfe rence . Ecgonine and ecgoninC' mcthyl este r, hreakdown pro Jucts o f cocainc h ydrolysis, wcre not quantified hcre bccause of the lac k of <•xt.raction metho ds satisfactory far the simultancous quantification of ali amphot.eric metabolit.es of cocaine in aqucous solution. Consequcntly, the rC'sults of the a nalysis of cocaine and only its first metabolitc, benzoyl l'cgonine. are re ported in Table 3. ~~vid e ntl y, cocaine is prcscnt throughout the duration of the assay whic h was limited to l h - - whPn chewed the !caves are usually not kept longcr in lhP c heeks (llolmstedt el al., 1979). A certain breakdown of cocaine occ urs durin¡.: incubation at ali the p!-1 valucs tested. llydrolysis at thc highcst pH (11.5) is limit.ed, howC'vcr, and does not cxceed 17%. lt is notcworth y that 0.7% more cocaine is found by incuhation at pH 9 far 30 min compared to l.IH' contro l. This indieatcs that an alkaline pl! dm•s not significantly in c rease tlw ex tractability of cocain!' from thP plant powdc r. This is in agre lo' mcnt -: labil ily of c.:oca inc in v<.i rious hufíers Incubalion lime Cocainc conlc ntb Benzoylecgonineb (min) (% ) content (%) ·11 6 .0 < o 100 100 c o c a powder 3 0 99 .3 97.9 60 92.9 100.8 ·ll 9 .0 30 100.5 99.2 1.·oca powdc r 60 9G.O 97.2 H 11.G ;¡o 85.9 103.3 coca powdcr 60 83.0 102.5 - ·------·------·------\faterial : l·.' ry th ruxylum coca Lam . from Pisac Market. Cusco, Peru ~ voucher No. Plow •:tll d al. 7133; air-dried leaves (see Holmstedt el al., 1979). ln iti al cocaine content was 0 .4 8 g per 100 g dry weight; initial henzoylecgonine contenl ;is 0.067 g per 100 g dry weighl; cinnamy lcocaine was not dctected. ;iH uf saliva= 5.97 _, 0 .07; pH uf coca lcaf powder = 5.8; pH of coca and Ce cropia ashes 8.55. 332 conditions which are not similar to the chewing practices used in the Andes or in the Amazon. More realistically, lower pH val u es should be u sed, as indi cated by the basic substances analyzcd hcre. In thc present study, thrce pi! with thc optimum pH value of 7 for the extraction of pure cocaine from water in to diethyl ether (Makisumi and Ota, 1959). Benzoylecgonine levels changed significantly bctween pi! 9 and 11.5, suggesting that its turnover is different at these pi! values. Consequently, the pf-1 value is decisive in these studies and completely different results can be obtained depending on the basicity of the medium used. Such a view corresponds to the unpublished data obtained recently with pure cocaine (Vitti, 1979): isflu and llipta wcre used for incubating synthetic cocaine, providing a pf-1of11.7 and 10.5, respectively. With isflu, Vitti found that 90% of the cocaine was destroyed in 20 min vía ecgonine methyl ester which was present in a five-fold higher amount than the other metabolites. In contrast, he also reported that with llipta 50% of the incubated cocaine was hydrolyzcd after 1 h, the main metabolite being benzoylecgonine. A GC method was used for quantification (Vitti, 1979). In the same kind of assays, Peruvian clwmists found carlier that incubation \hth tocra (5% solutions) at 37 ºC resulted in 2.5% and 5.5% breakdown over 30 ancl 60 min. (Cruz Sanchéz and Guillén, 1949). Finally, similar experiments with acidic medium (saliva, simulated gastric fluids, etc., in the pf-1 range 6 - 1.2) produced no brcakdown of cocaine (Montesinos, 1965; Vitti, 1979). The difficulty in these experiments is to simulatt• natural conditions as realistically as possible. Nieschulz and Schmersahl (1969) used an excessive amount of lime which could not be tolerated in thc mouth of coca chewers. Cruz Sanchéz and Guillén (1949) and Vitti (1979) used pure cocaine as start ing material which Gcncra.1 conclusion:; When using bo iling 95% ethanol for extracting the total alkaloids from E. coca, no artifacts were observed. By the use of capillary column GC- MS, such extracts were resolved wi thout further purification. Cocaine was found to be the main alkaloid together with cis- and trans-cinnamoylcocaine. No o ther minor alkaloid previously repor ted could be detected in the material use d . Alkaloid distribution among leaves o f various ages in E. coca, E. novo granatense and E. novogranatense var. truxillense could not be used as a chemotaxonomic marker since variations within leaves of o ne species ex cceded those between leaves of differe nt species. The buffer capacity of 17 alkaline substances u scd during coca leaf chewing was determined. When incubating coca leaf powde r at pH valucs corresponding to the alkaline substances, cocaine hydrolysis did not excecd 1 7% after 1 h at 37 ºC. The role of these alkaline substances during coca lcaf c hewing might be to provide an alkaline medium in thc mouth to libe rate cocaine from the plant material as thc free basf'. Ac kno wledgements Part of the research reported here was done during the Alpha Befo¡ l·:xpedition Phase VII, supported by the National Science F oundation Grant No. GB-37130). Support was also obtaincd from the Swedish Medica! Research Council, thc National Institute o f Health , th C' Wallenberg Founda ion, and by funds from the Karolinska lnstitutet. l wish to thank the Swiss '1ational Foundation for Scientific Research for a post-doctoral ¡,...-ant. Vouchcr and other specimens studiPd in this work have been p rovided 'Y Dr. T. Plowman. Sorne alkaloids used as refe re nce slandards have been t.'ceivcd from Professor R. H<'gnauer and Dr. L. H. Fikenscher. ·eferences rc\wr, S. ancl Hawks, H., The c hemistry uf cocaine ;rn d ils derivat.ivcs. In S. J. l\'lull (cd .). Cocaine : Chem ical. Bio lo¡:ical, Social ond Trcalment :1specls, CllC Pr Pss, Clev cland , 1976, pp. 15 - 34. \·nilian, G. H ., Duke, .J. A ., (il'ntncr, W. A. and Farnsworth, N. H., Cocai1w content o f Ery throx ylu m spccif>s . Journa/ o f Phurmaceu lical Scie nc e.~ .. 63 ( 197-t) 1938 - 1939 . . ke r , P. T. and Mazess, R . B., Calcium : unusual so urces in lhe highlan 1 rchard , H . E., Coca c h ew i n~: a new perspective. ln V . Ruhin (ed.), Ca nnabis and Culture. Mouton, The Hague, 1975, pp . 463 - 484. 334 Burchard, R. E., Myths of the saercd leal': 1':c olo¡.!ical p ers pectiv 1 ~ s on coca and pl'asant biocultural adaptation in Peru. Ph .lJ. Th csis, Indiana University , Bl oumingt.on, 1976, pp. 1 49 · 16 1. Ciu ffardi , E. , Dosis de alcaloides que ingieren los h abituados a Ja coca. ll<'v ista de Farma cologia y iUcdicina Experimen tal, 1 ( 1948) 216 - 231. Cruz Sánchez, G. an PJ ,,wman , T ., /\m azonian coca. Journal of Ethnopharmacology, 3 ( 1981 ) 195 · 225 Haff:rnf, H. F ., A !lcndboor. o f Al!~aloids and Alhu.luid·r.: or1lui11in¡; Pian is. \Viiey·f n t.er science, New York , 1970. lleens, E., La coca d e J ava, monograph ie historique, hotaniq uP, c him ique et pharm a· cologique, Ph.D. 1'hesis, Un ivcrsity of Paris , l 9 l 9a, pp. 1 · 1:)7. Recns, E ., La coca de Java. Bulletin des Sciences Pharmacologiques, 12 ( l 9 19b) 497 · 505. Salomon , P., PoLen tiali ty of the ne uro·st. im u lant effect o f cocaine by alkrtline substances. R evista de Farmacologia y M edicina Experimental, 2 (1 949) 114 (quoted by Zapa ta Ortíz, 1970 ). Schultes, R. E. , A n ew melhod of coca preparation in the colom bian Amazon. Botanical Museum Leaflets, Harvard Uniuersity, 17 (1 957) 24 1 · 246. T u rner, C. E., Ma, C. Y. a nd Elsohly , M. A ., Const il uents of Erythroxylu m coca l. Gas · chro m atographic analysis of coca ine from th ree locatio ns i n Pe ru. B ulle lin on Narcolics, 31 (1 979) 71 · 76. Turner, C. E., Ma, C. Y. an d Elsohly, M. J\ ., ConstiLuen ts of J::ry lhroxylum coca. IL l ;as chromatographic an alysis of cocaine a n d other alkalo ids in coc a leaves. ,fourrtal uf Ethnopharmaco/ogy , 3 (1 981) 293 · 298. Vitti, V ., Unpublished resuits p resen ted a t lhe 43rd lnte rnalional Congress of A m e rican· ists, August 15, 1979, Vancouver, Can ada. Von Minden , D. L . a nd D 'Am a to , N. A .; Simultaneous d elerm in ation of cocaine and bcn zoylecgonine in urin e by gas ·liquid chro m atograp hy. A nalytica/ Chem istry, 49 ( 1 977) 1974. 1977. Will stael ter, H. ., Ueber ein Isomer des Cocains. Berichte der Deutscherz Chemischen Gesel· ~c h an, 29(1896)221 6 . \V ó hl e r, F., Fortse tzung der U ntersuchung:e n über Coca a und d as Cocain. Ju stus Liebigs A nnaleri der Chemie, 1 21(1 862)372. Ynussefi, M., Cooks, R. G . and MacLaugh lin, J. L., ~fopping o f cocaine an d cinnamoyl cocaine in w h o le coca plan t t.issue by MIKES. Journal of the Anu::rican Chem ical Society, /01 ( 1979) 3400 · 3 ·102. Zapata Ortíz , V., T he che wing o f coca leaves in Peru . ln lernational Jou mal of thc Addic tions, 6 (1970) 287 · 291.